The embankment damProceedings of the sixth conference of theBritish Dam Society held in Nottingham on12-15 September 1990

The British Dam Society

Conference organized by The British Dam Society (formerly known as^&NCOLD)

Organizing Sub-Committee: J. R. Claydon (Yorkshire Water), I. M. Stevens (YorkshireWater), Dr A. I. B. Moffat (University of Newcastle upon Tyne), N. Tyler (Secretariat,British Dam Society)

A CIP catalogue record for this book Is available from the British Library.ISBN: 978-0-7277-1647-7First published 1991

© Authors, 1991, unless otherwise stated.All rights, Including translation, reserved. Except for fair copying, no part of this publication may be reproduced, stored in a retrie-val system or transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise, without theprior written permission of the Publications Manager, Publications Division, Thomas Telford Ltd, Thomas Telford House, 1 HeronQuay, London E14 4JD.

Papers or other contributions and the statements made or the opinions expressed therein are published on the understandingthat the author of the contribution is solely responsible for the opinions expressed in it and that Its publication does not necessarilyimply that such statements and or opinions are or reflect the views or opinions of the organizers or publishers.

Published on behalf of the organizers by Thomas Telford Ltd, Thomas Telford House, 1 Heron Quay. London E144JD.

Contents

The Geoffrey Binnie Lecture. P. A. BACK 1

The state of the art1. Design and performance of the forty mile Coulee East Dam on a soft

clay foundation, B. G. CHIN, D. M. DAVISON. E. K. KLOHN. R. P. BENSONand J. W. CAMPBELL 7

2. The application of new techniques in the design of the two high dams

in South West China. F. DAFENG 15

3. The use of low grade rockfill a t Roadford Dam. A. a WILSON and J. D. EVANS 21

4. A perspective of the art of the embankment d a m in South West Asia.

W. J. CARLYLE 29

6. Instrumentation of the Mrica Dam. i. w. BURTON and P. A. S. FERGUSON 41

Discussion 47

Tailings dams7. The safety of tailings dams and lagoons in Britain, A. D. PENMAN

and J. A. CHARLES 53

8. Tailings dams of the copper mining plant Elatzite after eight years ofoperation, c. B. ABADJIEV and A. A. KARADIMOV 59

9. Waste retention embankments on soft clay. A. R. GRIFFIN, C. K. PENG

and D. L GUDGEON 63

10. Tailings deposition predictive computer modelling, p. WATERMEYERand N. A. THOMPSON 71

11. Geotechnical aspects of the construction of tailings dams- two European studies, M. CAMBRIDGE and R. H. COULTON 77

12. Spillway systems for tailings dams. P. J. FORBES, S. A. CALE

and L F. CLELLAND 85

13. Clay mining waste disposal problems - central and peripheral.C. D. ROUTH. A. STREET and C. F. J. GRIGG 93

15. Gale common ash disposal scheme - concept , design and construction.E. T. HAWS. P. L. MARTIN and R. A. N. ORANGE-BROMEHEAD 99

16. Gale common ash disposal scheme - planning, environment, operationand restoration, J. A. DENNIS, D. J. HILLIER and H. T. MOGGRIDGE 105

Discussion 113

Risk, hazard and safety17. Evaluation of dam safety at a series of hydropower dams including risk

assessment, D. s. BOWLES, L. R. ANDERSON, T. F. GLOVER,G. S. TARBOX, R. B. WAITEand P. E. YIN AU-YEUNG 119

18. Safety considerations with existing embankment dams and in theirraising, D. j . KNIGHT, E. A. JACKSON, J. HALCRO-JOHNSTONand C. MAKINSON 127

20. Woodhead Reservoir - remedial works. R. w. CHALMERS 135

21. The Bureau of Reclamations new downstream hazard classificationsguidelines, D. J. TRIESTE 141

22. Mitchell's House reservoirs - investigations, monitoring and remedialworks. B. H. ROFE, D. B. WICKHAM and J. K. HOPKINS 145

Discussion 151

Environment and research23. The design and operation of flood storage dams for recreational uses.

J. B. ELLIS, M. HALL and D. L HOCKIN 155

24. The use of close-range photogrammetry for reservoir embankmentmonitoring, J. K. HOPKINS, D. B. WICKHAM and D. M. STIRLING 159

25. Accommodating rare floods over embankments and steepreinforced channels. G. R. POWLEDGE, T. J. RHONE and P. E. CLOPPER 165

26. Deformation of Ramsden dam during reservoir drawdown andrefilling, P. TEDD, J. A. CHARLES and J. R. CLAYDON 171

27. The routine monitoring of embankment dam behaviour.A. C. ROBERTSHAW and T. N. DYKE 177

28. Embankment dam behaviour: the contribution of geo-chemistry.A. MACDONALD and J. M. REID 185

29. Reservoirs - a legacy of opportunity, K. E. COXON. J. E. EMBERTONand R. M. PAGETT 193

Discussion 199

Papers 5,14 and 19 were withdrawn

The Geoffrey Binnie Lecture

P. A. BACK, Sir Alexander Gibb and Partners

Before I turn to the topic of my lecture itwould be appropriate to say a few words aboutthe man whose name we especially remember today- Geoffrey Binnie.

I consider it a great privilege to be invitedto give this - the first Geoffrey BinnieLecture and to recall briefly just a fewaspects of his distinguished career and theconsiderable contribution he made both as anindividual and also through his illustriousfirm to the craft of dam building - and indeedto the profession of Civil Engineering.

Geoffrey Binnie was born in 1908 and died in1989. He came from a long line ofdistinguished engineers - both his father andgrandfather being eminent in the profession andwho together founded the firm of Binnie &Partners in 1902. Geoffrey was at Charterhouseand then Trinity College Cambridge. He wasborn partially deaf and the need to overcomethis disability which might have adverselyaffected his career perhaps gave him that extradetermination to succeed. He proved himself amost able young engineer and after variedexperience in Switzerland and Hong Kong and theU.K. he was appointed a Partner in his firm in1939.

On the outbreak of war he volunteered for theRoyal Engineers and served in North Africa andthe Middle East from 1940 - 1945. At the endof the war he returned to the U.K. to find hisfirm reduced to no more than a handful ofstaff. From that low point Geoffrey was at theforefront of the reconstruction of his firmwhich imder his leadership went from strengthto strength and was to achieve an internationalreputation for excellence in the field of damengineering.

Two projects in particular stand out:

- The Dokan Dam in Iraq and Mangla Dam inPakistan

The Dokan Dam marked a watershed in thedevelopment of new computational methods forthe analysis of arch dams.

Mangla Dam was at the time (1957) the singlelargest project undertaken by a consultingengineer. An organisation was set up withBinnie & Partners as Project Consultants andwith Geoffrey Binnie as Project Partner. Thisis not the place to go into any detail on theMangla project, but it is worthy of note that

the project was carried through to a mostsuccessful conclusion and much of the creditfor this was undoubtedly due to Geoffrey Binnieand the wise and firm, but always courteousleadership that he provided.

Today we honour his memory and the traditionsof engineering excellence which marked hiswork.

I have chosen as a title for my lecture:

THE ULTIMATE DAM

No doubt this title will be considered theultimate in cheek.

How can there be such a thing as the ultimatedam? Surely every dam is a once-off structuredesigned to meet the very specific needs of aparticular site.

Very true - but perhaps only half true.

We all know, I am sure, that we each bring tothe design of a dam our own particularpreconceptions - our own idiosyncrasies - ourown foibles and prejudices. We say that wealways consider all options - but find strangeto say that every project we are involved intends to have a certain type of dam. And ifanother designer were involved it would almostcertainly be a different type of dam.

I say that we delude ourselves if we believethat any one of us is truly impartial.

Terzagi, Cassagrande, Penman - EmbankmentdamsAndre coyne, Serafim - Arch damsBarry Cooke - Rockfill dams with upstreamconcrete faceSchreder, Dunston, Hollingsworth - Rollercompacted concrete

I certainly have to confess my own partiality!

I have been known to wax lyrical about thesublime beauty of a thin arch dam compared withthe splodge of an embankment dam. Now there'sprejudice for you.

To parody the words of Dryden in Absalom &Achitophel:

THE GEOFFREY BINNIE LECTURE

"For dams they've built of every sizethat engineers and damsmiths could devise.But each to his favoured typewill ever return - no matter what thesite."

If we could just for a few moments put asideour prejudices and look at the whole spectrumof dam engineering - where amongst the hugevariety of choices may future dams tend to go?What should be the hallmarks of a dam designedfor the 21st Century and hopefully be stillthere - like Buck Rodgers in the 25th Century.

I have selected 10 criteria which in my viewwould be desirable for the ultimate dam. Notevery dam could meet all 10 - but the more thatcan be met - the safer and longer the expectedlife of the dam. The ultimate dam would, bydefinition, meet all the criteria.

I am well aware that in putting forward thesecriteria I am also setting out my ownprejudices. I am sure that many of you wouldproduce a different list. There is scope herefor much argument - and fun.

But when I have produced my list I would thenfinally examine what type of dam these criteriadrives us toward - and whether that type shouldthen be the initial starting point for each newsite we encounter - and we back away step bystep only when other considerations force us todo so.

So here are my criteria:

1) Materials of construction are verydurable.

2) Flood routing is independent of power oroperator.

3) The dam will not be destroyed byovertopping.

4) There is redundancy in the structuralbehaviour of the dam.

5) The structure can accommodate reasonablesettlements and deformations.

6) The structure is resistant to internalerosion.

7) The structure is capable of withstandingsubstantial seismic shock.

8) The structure should be highly resistantto destruction by acts of sabotage orbombs.

9) The reservoir behind the dam can be drawndown to at least half height andpreferably emptied within a few weeks.

10) The dam satisfies the generation test.

There is no special significance in the order -all 10 criteria are important - but clearlythere are many dams which can only satisfysome.

You may be puzzled by my 10th criteria - thegeneration test. More about that later - butthat is one criteria which every dam built inthe future should satisfy.

So a few words about the list.

D Materials of construction are very durable

This, as indeed many of the criteria Ihave listed, almost goes without saying.But it is of course of quite fundamentalimportance. If we get this wrong, thereis almost no other alternative than toabandon the dam and start again. Witnesswhat we are having to do at Maentwrog inNorth Wales. That dam is suffering fromAlkali Aggregate Reaction and is having tobe replaced after only 60 years - 60 yearsshould be as nothing in the life of a dam.

When I am asked how long should a dam lastI always say its life should be indefinite

- it should be seen in terms of geologicaltime. And in geology 100,000 years isrecent.

A few years ago I was asked to report on adam that had failed in Sri Lanka withgrievous loss of life. The dam was infact 1400 years old and the part that ledto failure was the sluiceway, built by theRoyal Engineers some 115 years ago. DerekKnight and I were amazed to see what thoseearly Sri Lankan engineers had achieved.They built better than they knew and theybuilt for keeps. Short termism is amodern scourge.

2) Flood routing is independent of power oran operator

It is a well known statistic that failureto correctly route a flood past a dam hasbeen perhaps the single most common causeof dam failure.

The reasons for maloperation of thespillway gates are almost invariably forone of two reasons:

(1) The power supply has failed, makingit impossible to open the gates - or

(2) The operator has not responded intime for an incoming flood.

The dam is then overwhelmed by the flood.

Clearly the safest way to deal with thisproblem is to have no gates at all - but asimple ungated overspill. But this can bevery costly in terms of lost storage dueto the allowance that has then to be madefor the flood surcharge.

One way round this problem which we usedon the Victoria Dam in Sri Lanka is anautomatic gate which opens without theneed for power or an operator and respondsexactly to the needs for optimum floodrouting. And the method is entirelymechanical. There is no dependence on

BACK

sophisticated computer hardware orelectronics.

3) The dam will not be destroyed byovertopping

This follows straight on from the previouscriteria - but goes still further.

Spillway design is linked to statisticalor other projections of what might happenin the future. And it is very easy to getthe projections wrong. Take for instancethe Manchu II dam in India. In the spaceof 20 years the projections and thenmodified projections were proved wrong andin the end resulted in a five foldincrease in the design flood with ofcourse the attendant dramatic change inthe design of spillways.

In the middle of building Kariba weexperienced what till then was the 1000year flood which, of course, was then nolonger the 1000 year flood. We had to add50% to the capacity of the spillways.

All this emphasises the need for a certainhumility in our attitude to design floodsand surely demonstrates the benefit ofconstructing a dam which, even if we getour flood projections wrong, will not failif the worst happens and the dam isovertopped.

Does this mean that embankment dams wouldnever be able to satisfy the criteria.Not at all- but it puts pressure onembankment engineers to come up with asolution to this problem. The answer maylie in suitable geomembrane fabrics - ortorpedo netting type reinforcement or someother device which protects the downstreamface from being washed out by overtopping.That is the challenge.

4) There is redundancy in the structuralbehaviour of the dam

Redundancy is a very good thing. With allour cleverness we are still quite capableof getting it wrong - of not visualisingcorrectly the possible mode of failure -or failing to take account of a particularloading configuration which might arise inthe life of the structure - such as asevere earthquake or deterioration offoundation conditions. Redundancy enablesthe loads to be carried in more than oneway and can significantly enhance thesafety of the structure.

A typical example is a concrete gravitydam. Almost invariably these are builtwith a straight axis and therefore slidingresistance on the foundations is the onlymechanism preventing failure. Introduce agentle curve and immediately other forceswould be brought into play to help carrythe load if sliding resistance provedinadequate. Why do we ever build agravity dam with a straight axis?

a curved gravity dam unless there areother overriding considerations.

In considering redundancy, however, wemust be very careful not to deludeourselves as to the overall safety. Thesecondary resisting mechanism frequentlyonly comes into play when the firstmechanism has failed. We cannot add thetwo mechanisms together as they will infact be overcome one by one. It is liketearing along the (Jotted line.

5) The structure can accommodate reasonablesettlements and deformations

Perhaps the single factor in which ouruncertainties are greatest is in thefoundation of the dam - which is as much apart of the structure as the edifice abovethe foundations.

We inevitably have to make assumptions asto how the foundation will behave - andour assumptions are more likely to beinaccurate there than anywhere else. Itis therefore very desirable that the damcan accommodate deformations withoutsuffering unduly. Or that if thefoundations yield, other mechanisms comeinto play which transfer the load safelyelsewhere - and the dam's safety is notjeopardised.

One type of dam wh^ch is particularly goodat redistributing stress is the thin archdam but there is of course a limit to howmuch stress re-distribution can betolerated.

The Kariba dam is a case in point. Atthis dam the North bank and up to half ofthe South bank the foundations of the damwere on excellent rock. However, the upperhalf of the South bank was composed ofquartzite with thick bands of clay. Thequartzite itself was good and at first wetried to jet out the clay but without muchsuccess. Our concern was that thequartzite and clay would deform under loadrelatively easily and the loads wouldtherefore be reflected downwards to thehard relatively unyielding gneiss whichmight then become overstressed.

Paradoxically, the dam might then fail atits strongest point. So it was that wedecided to construct four massiveunderground buttresses to carry the thrustthrough the quartzite and into theunderlying gneiss. Of course, we then hadto completely ignore any contribution fromthe quartzite in carrying the loads fromthe dam.

6) The structure is resistant to internalerosion

If a dam is to have an indefinite life allthe material that goes to make it shouldbe in a state of stable equilibrium.

If I am a designer I will always insist on

THE GEOFFREY BINNIE LECTURE

Internal erosion, however slow and whetherit be as a result of chemical action or ofphysical movement of particles representsan ultimate threat to the security of thedam. We must satisfy ourselves thereforethat such phenomenon as piping, ordispersivity, or leaching or AAR or anyother phenomenon which is changing theinternal condition of the structure willnot happen. For concrete damsf as ourknowledge of the chemistry of aggregatesand cements grows, so too are we coming torealise the considerable problems we canface in completely satisfying thiscriterion.

7) The structure is capable of withstandingsubstantial seismic shock

This may seem obvious, but past practicein dam design has frequently treatedseismicity with scant respect.

Now that we have much more sophisticatedcomputational methods to hand we know thatthe old approach of treating the seismicload as a pseudo static phenomenon is verymisleading and can underestimate the realeffects of an earthquake verysignificantly.

For example, in the case of an arch dam weanalysed - using first a dynamic approachand then a pseudo static method for thesame peak acceleration, the dynamic methodgave a threefold higher maximum stress ina completely different part of thestructure.

In considering what seismic effects todesign against, one naturally tends tolook at past history for guidance - inmuch the same way as we do for floods.There is a growing view in some quartersthat such an approach may be misleading.And that indeed some of the greatestshocks that are waiting to happen is whereprevious seismic activity has been low.

I understand that in the USA wheretraditionally all the emphasis has been onthe Pacific Seaboard, some seismologistsbelieve that the Eastern Seaboard could bein for even greater trouble.

I have just returned from a visit withDerek Knight to the Philippines where theyhad a Magnitude 7.7 earthquake in Julyclose to two of their major dams. We wereasked to inspect the dams and advise onremedial works.

It is of interest to note that on one ofthe dams with an upstream concrete face,not as the water barrier, but in place ofrip rap protection, the concrete face wascrushed in one area and pulled apart inanother. Undoubtedly, had the concreteface been the water barrier, the dam wouldhave been breached.

The construction of the concrete face wasnot exactly as it would have been for useas a water barrier, and reinforcing steel

did not pass through the joints in theslab - but the degree of damage to thisconcrete certainly raises questons in mymind as to the seismic performance of damswhich depend on such concrete faces as themain water barrier.

8) The structure should be highly resistantto destruction by acts of sabotage orbombs

It is a sad commentary on the age in whichwe live that such a criteria has to beincluded. But included it should bebecause undoubtedly a dam that isvulnerable to easy destruction couldbecome the target of terrorists - witnessthe agonies that the security people arehaving over the Channel Tunnel.

The dam type which obviously is mostvulnerable to sabotage is the buttress ormultiple arch. In some countries suchdams are no longer allowed to beconstructed. And if they already existthe space between the arches have beeninfilled. However, it may well be that itis still an economic form of constructioneven allowing for such infilling.

9) The reservoir behind the dam can be drawndown to at least half height anctpreferably emptied within a few weeks

I am surprised that, as far as I know, nocountry has actually legislated such arequirement and that is has been left tothe designer - or the owner as to whetheror not there is any significant drawdownfacility.

When one considers the regulations whichcontrol the nuclear industry it issurprising that dam safety is stillgenerally treated so perfunctorily. Itsurely must be a matter of normal safetyand common sense, that once impoundingbegins behind a dam, it is possible toreverse the process if a problem arises.Otherwise all control is lost and we canbecome helpless spectators to a possiblecatastrophe.

It is perhaps because we have not had thisrequirement imposed upon us that we havehad to be so conservative in our designsto ensure that when uncontrolledimpounding takes place nothing will gowrong. We actually might save money byspending more at the outset on ampledrawdown facilities.

10) The dam satisfies the generation test

This criterion has nothing to do with thebasic design or safety of the dam - butshould be applied to determine whether thedam should be built at all.

We dam designers and builders do not enjoya very good press. We are frequentlythought of as despoilers of naturebringing destruction to flora and fauna

BACK

- threatening the habitat of man andbeast alike

- upsetting the delicate balance ofnature and generally causing a netdisbenefit to the community

Sometimes these strictures are true - andsometimes patently false. But we doourselves and our profession enormous harmif we do not bring a truly critical eye toour activities and learn to differentiatebetween the good and the bad scheme and toresolutely oppose the bad. I am deeplytroubled by those who in their enthusiasmsweep aside all criticism as if it wasalways uninformed, and blindly support dambuilding as if it was always a good thing.

Now in this day and age whether a scheme goesahead or not is often decided by Accountantsand Bankers applying their economic criteria ofcost benefit analyses and internal rates ofreturn to assess the merits of the project.But I believe such decisions are far tooimportant to be left to Accountants andBankers. Their horizon is limited to somethinglike 30 years. Nothing that happens beyondthat time has any real significance for them.But we are building structures to last forhundreds if not thousands of years.

And so I believe we must introduce another testto be satisfied that what we are doing is good- and I have called this the generation test.

This is an approach I first put forward twoyears ago at a Seminar in the United States onthe subject of River Basin Development inAfrica.

Consider two development options - one with a12% internal rate of return - the other with6%.

In 30 years time the 12% option may havecreated a desert - by say reducing the watertable catastrophically.

The 6% option, on the contrary, may havecreated a thriving community. Yet by stricteconomic criteria, the choice is made to go forthe 12% option - because it satisfies therepayment criteria.

Such a limited and clearly inadequate basis forguiding decision making will no longer beacceptable. Evaluation of a project must lookbeyond the immediate economic goals - it mustlook to succeeding generations.

Planners must metaphorically be made to standbefore their children 30 years on and tell them"this is what we have done". Here is thisdesert or here is this fertile land - for you"Only if this generation test is passed, shoulda scheme be considered worthy of support.

Finally, with these 10 criteria before us whatwill the ultimate dam be like. We must ofcourse assume that the 10th criteria hasalready been satisfied.

My ultimate dam would be built with relativelylow grade RCC in the form of an arch gravitystructure without joints, with sloping upstreamand downstream faces and with an impermeablemembrane on the upstream face, possibly createdby a clay filled geosynthetic mattress which isprotected by sand and random rockfill. I wouldtake no special measures to make the RCCimpermeable, indeed I would wish it to berelatively porous to avoid uplift and ensurethat the phreatic surface is close to tailwaterlevel.

Such a dam would be designed for overtoppingand could accommodate reasonable settlementwithout affecting its ability to store water.It would be highly resistant to internalerosion or earthquake damage or sabotage andwould of course have to include low leveloutlets to permit drawdown. The dam could beovertopped safely during construction so thecost of diversion would be minimised. Whetheror not gates are provided at the crest or asimple overspill is adopted is a matter ofchoice depending on the required storage volume- but if gates are used they should be capableof operation without power or a controller.

Mr. Chairman, I am aware that much of what Ihave said is highly controversial and I daresay if 100 of us were asked to define ourultimate dam, there would be 100 differentanswers. For what it is worth I have given youmine but I don't expect that I will ever havethe opportunity to build it. Life is not likethat and with dams the evolution of methodologyis, probably with good reason, very very slow.

1. Design and performance of the forty mile Coulee East dam on asoft clay foundation

B. G. CHIN, D. M. DAVISON, E. K. KLOHN, R. P. BENSON, and J. W. CAMPBELL KlohnLeonoff, Calgary, Canada

The Forty Mile Coulee East Dam is a 28 m high earthfill embankment founded on up to 60 m of highlyplastic, soft clays. Large foundation movements were measured during construction. Moreover, highpore pressures exceeding predictions by up to 35% were recorded, despite the benefits of a testfill and the (then) conservative assumptions used in design. This paper describes the constructionperformance of the East Dam and highlights some of the key trends of instrument readings. Theresults of finite element analyses to match the field behaviour are also shown, which demonstratedthat undrained yielding of the clay can result in horizontal stress increases much larger thanthose assumed in a linear elastic model. Neglect of this behaviour will lead to unconservativeestimates of the pore pressure response in the field.

INTRODUCTION1. As part of continuing efforts to alleviate

problems of cyclic drought patterns, the Alberta(Canada) government commissioned construction ofthe Forty Mile Coulee Project to supplyadditional water and to improve water deliverytimes to the farmlands in southern Alberta (Fig.1). The project includes two 28 m high earthdams(East and West Dams); a 10 km 86 000 000 m3reservoir; a spillway; an inlet chute; a 750 mlong inverted syphon; and a 20 m3/sec pumpstation(Fig. 2). Construction was staged over a twoyear period from 1986 to 1987, and the reservoirwas filled to full supply level in 1989. Thetotal cost of the project was $55 million.

2. Foundations at the damsites comprise softlacustrine clays up to 60 m and 35 m thick at theEast and West Dams, respectively. Duringconstruction, the West Dam performed generally asexpected. However, the East Dam performance waspoorer than anticipated, requiring designmodifications to enlarge the toe berms forconstruction prior to raising the embankment toits ultimate height. Vertical and horizontalmovements in the foundation were up to 1.6 m and0.3 m respectively at the end of construction,and have continued slowly after construction(2.5 m and 0.5 m to May 1990). Moreover,unusually high pore pressures exceedingpredictions by up to 35%, coupled with minimal orno dissipation, raised concerns with respect tostability during reservoir operations. Finiteelement studies were undertaken to investigate

MEDICINE HAT

FIGURE 1 - LOCATION PLAN

The embankment dam. Thomas Telford, London, 1991

l~— INLET SYSTEM^-EMERGENCY SPILLWAY

TWIN BARREL SYPHON

FIGURE 2 - FORTY MILE COULEE PROJECT SITE PLAN

the key factors controlling field behaviour, andto evaluate future performance. Based on theresults of this study, and because the downstreamimplications of a dam breach are minimal, thereservoir was allowed to be filled in acontrolled manner. Completed to full supplylevel in mid-1989, the performance of the EastDam has been satisfactory to date.

3. This paper describes the performance of theEast Dam foundation during construction,highlights some of the key trends of instrumentreadings, and presents typical results of thefinite element studies. In particular, porepressures predicted by finite element analysesare shown to provide a good match to fieldresponse, and are compared to original estimatesto highlight the inaccurate (and unconservative)predictions obtained from linear elastic modelsfor deformable foundations. Space limitationspreclude a thorough discussion of all data andanalytical studies carried out. Nevertheless,sufficient detail is presented to illustrate theimportance of the observational approach to thedesign and construction of large embankments onsoft clays.

FOUNDATION CONDITIONS4. Formation of prairie coulees typically

resulted from ice marginal channels during thelast glaciation, which eroded through the glacialdrift into bedrock. Deposition began as themajor flows subsided, with soils derivedprimarily from adjacent bedrock and eroded uplandglacial drift. Coalescing slopewash colluviumformed along the coulee walls below bedrockoutcrops, while lacustrine clays and silts

STATE OF THE ART

820

800-

1 760

740-

720-

CREST EL. 818m

//r GROUND ELEVATION^r

E300 E400 E500 E600 £700 E800 E900EASTINGS (m)

FIGURE 3 - STRATIGRAPHiC PROFILE ACROSS VALLEY

deposited behind either fans spreading across thecoulee or ice dams from local glacial advances.The Forty Mile Coulee was downcut by up to 90 m,then infilled with up to 60 m of sediments.These sediments comprise highly plasticlacustrine clays near mid-valley, interfingeringwith sandy colluvial soils toward either side ofthe coulee. On the south side of the coulee, atthe East Dan, a thick, beached sand depositunderlies the clays above a near horizontal shelfin the bedrock surface (Fig. 3).5. Liquid and plastic limits of the clays

range from 50% to over 100% and 18% to 25%respectively. Natural moisture contents varyfrom 20% to 35% and liquidity indices from 0.1 to0.3. Clays of higher plasticity (liquid limit90% to over 100%) are more predominant and muchthicker at mid-valley. Towards the coulee walls,the clays are less plastic (liquid limit 50%±)and contain more frequent sand layers as a resultof interfingering with the colluvial soils.

6. Consolidation tests and pocket penetro-meter readings, indicated that the clay had beenoverconsolidated to a depth of about 18 m. Theprimary process of overconsolidation at this siteis believed to have been by desiccation. Atypical soil log at mid-valley, and the inferredsite stress history are shown on Fig. 4.

EMBANKMENT SECTION7. The effective stress method was used for

design of the dam. Initial predictions of porepressures were based on Henkel's(1) pore pressureequation, using pore pressure parametersestimated from Law and Bozozuk( } and total stress

750-

TH81-4008 (MID-VALLEY)50% 100%

TV (kPo)0 100 200 300 400 500

1 3 0 -

50-

EFFECTIVEOVERBURDEN

PRESSUREb) PRECONSOUDATION PRESSURE vs DEPTH

"UNDRAINED" STRENGTH (kPa)0 100 200 300 400 500

~r - C . 3 REPRESENTINGP NORMALLY

CONSOLIDATEDCLAYS

10-

£20-

4 0 -

5 0 -

c) "UNDRAINED" STRENGTH vs DEPTHQ) SOIL LOG (BASED ON POCKET PENETROMETER READINGS)

FIGURE 4 - TYPICAL SOIL LOG AND STRESS HISTORY

changes from linear elastic solutions* KMeasurements beneath a 12 m test fill constructedin 1982 (Fig. 5) showed good agreement with theabove predictions, and therefore the same methodwas applied for design of the main dams.Dissipation parameters were derived from the testfill data, but modified to account for possiblyslower dissipation beneath the much wider base ofthe East Dam.

8. Despite the benefits of a test fill, and asa safeguard against unfavourable pore pressurebehaviour, provisions were included in thecontract documents to construct toe berms ifrequired. The initial design of berms was basedon the conservative assumptions of no dissipationusing undrained pore pressures. Furthermore,the observational method(4) was selected as anappropriate approach to design and construction.Therefore, the necessary instrumentation wasincluded to permit ongoing monitoring andevaluation during construction.

9. The final design tendered for constructionhad 8H:1V slopes, and included an upstreamimpervious clay zone, a downstream random zone,a vertical chimney drain/filter, a drainageblanket/finger drain system along the downstreambase, and toe berms (Fig. 5). The wide plasticupstream zone and the vertical chimneydrain/filter were included to provide protectionagainst cracking due to the large deformationsanticipated.

10. As a result of the observational method,higher than expected pore pressures were able tobe identified during construction, and alloweddesign changes to be made part way throughconstruction to enlarge the toe berms. Theseberms were constructed prior to raising theembankment to its full height. The as-constructed geometry at the end of constructionis shown and compared with the original design onFigs. 5 and 6.

FOUNDATION PERFORMANCE DURING CONSTRUCTIONGeneral

11. The instrumentation program included 136pneumatic piezometers in the foundation andembankment fill; 17 pneumatic settlement sensors;6 overflow manometer settlement gauges; 3 Sondexsettlement gauges; and 6 telescoping typeinclinometers socketed into bedrock, which hadsettlement rings attached to measure incrementalsettlements in the foundation. Some of these hadbeen installed for the test fill. In addition,66 surface monuments (of which 25 were abandonedas a result of berm construction) and 8standpipes downstream of the dam were installed.Typical layouts of the instrument types are shownon Fig. 5. Locations of the instrumentedsections are shown in plan on Fig. 6.

12. Because of space restrictions, onlyselected data relevant to the foundationbehaviour are presented. For reference to thefollowing discussions, typical time plots of porepressure responses, settlements and horizontalmovements are shown on Fig. 7.

Pore Pressure Response13. All piezometers exhibited a steady rate of

rise during construction, reaching a well definedbreak in the response curve after each

PAPER 1: CHIN , DAVISON, KLOHN, BENSON A N D CAMPBELL

820

810

800

780

770

U/S S0N0EX SETTLEMENT -

CHIMNEY DRAIN-

UPSTREAM IMPERVIOUS CLAY ZONE -

ORIGINAL OESIGN -r

DOWNSTREAM RANDOM ZONE. D/S

INCUNOMETER - EBI-5 / " B E R M IN ORIGINAL DESIGN

xG8

x G 9

x G 1 0

xG7

AS-CONSTRUCTED BERM

DRAINAGE BLANKET AND FINGER DRAIN

TM111

N1200 N1150 N850NT 050 N1000 N950 N900NORTHINGS (m)

FIGURE 5 - TYPICAL DAM SECTION AND INSTRUMENTS

N800 N750

LEGEND

• SETTLEMENT GAUGE

x PIEZOMETER

T SURFACE MONUMENT

TOE OF BERM

FIGURE 6 - EAST DAM KEY PLAN

construction stage (Fig. 7). Piezometers nearthe toes also exhibited a decrease in the rate ofrise, as construction moved further away from thepiezometers. Some piezometers continued to riseslowly after construction, with most piezometersnear mid-valley showing minimal or no dissipationto date (May 1990).

14. The maximum piezometric surface measuredat the end of construction, together withoriginal de_sign projections, are shown on Fig. 8.The field B values (defined as the ratio of theexcess pore pressures to the total verticalstress) varied from about 0.5 near the toes up toabout 1.2 near the centreline.

15. Further insight into the pore pressurebehaviour can be gained from Fig. 9 which showsthe piezometer data as a function of embankmentload. For reference, a 45° line is included to

represent the case where the excess porepressures are equal to loading. The followingpoints are highlighted:• Piezometers G6 and G8 installed at 10 m depth

showed an initial low response to loading,followed an increasing, almost linear responseparalleling the 45° line.

• Piezometers G7, G9 and G10 installed at orbelow 20 m depths exhibited a high responseimmediately at the start of construction,nearly matching the 45° line. They deviateaway from this line at higher dam heights,however, this may be due to the "stress bulbeffect".

• Piezometers G8, G9 and G10 located directlybeneath centreline exhibited an increase inresponse steeper than the 45° line above the300 kPa stress level.16. The first two observations generally

confirm the site stress history established bylaboratory tests. That is, the upper parts ofthe clay formation behaved initially as anoverconsolidated soil, followed by an undrained,normally consolidated response. The stresses atwhich this change in behaviour occurred were65 kPa and 105 kPa, which are within the range ofpreconsolidation pressures measured in thelaboratory. The deeper clays, however, behavedas a normally consolidated soil from thebeginning of construction.

17. The third observation implies that duringthe final stages of construction, the incrementalpore pressure increases were greater than theincremental increase in vertical stress. Thisbehaviour is believed to be attributed to localyielding of the clay, as a result of beingcritically stressed.

800

; 530'

FOUNDATION POREPRESSURES (TYP.)

EMBANKMENT ELEVATION AT CENTERLINE

NOTES:

1. SEE FIGURE 5 FOR INSTRUMENT LOCATIONS.2. TM111 INSTAUEO AFTER BERM CONSTRUCTION.

/- EBI-5- VECTOR SUM OF A AND B AXES COMPONENTS'

-SG2SETTLEMENTS

DOWNSTREAM-^.HORIZONTAL v ^ .MOVEMENT , S . - -^*^C~

- S G 1

H ' J ' J ' A ' S ' O ' N ' O ' J ' F ' M ' A ' H ' J ' J ' A ' S ' O ' N ' C ' J ' F ' M ' A ' M ' J ' J ' A ' S ' O ' N ' O ' J ' F ' U 1 /1986 1987 1988

FIGURE 7 - TIME PLOT OF TYPICAL INSTRUMENTATION READINGS

2000

5 0 0 '

STATE OF THE ART

-£840-1— 830-O 8 2 ° "g 810-§ 800-3 790 J

PREDICTED PIEZOMETRICSURFACE (NO BERM)

^ - MAXIMUM MEASURED^"^ PIEZOMETRIC SURFACE

(END STG 2)

- 8 3 0- 8 2 0- 8 1 0- 8 0 0

SETTLEMENTS DUE TOTEST FILL (SINCE 1982)

END STG -END STG 2- PREDICTED

ULTIMATE LSETTLEMENTS

hsoodh1600dh2400E

N1200 N1150 N1100 N1050 N1000 N950 N900 N850NORTHINGS (m)

FIGURE 8 - PIEZOMETRIC SURFACE AND SETTLEMENT PROFILE AT MID-VALLEY

N800 N750

300-

250-

200-

S.150-

100-

50-

Qc

FIGURE

Vertical

Q D P7 20m• • P9 20m0 O P10 25m

) 50 100

\r-\ v

x

/A

/*'/y

^-A A P8 10m

•x '^^-— X X P6 10m

150 200

AXH (kPa)

9 - PORE PRESSURE VS

Deformations

250 300 350

EMBANKMENT STRESS

18. Foundation settlements exhibited animmediate response during construction, followedby time-delayed consolidation after construction(Fig. 7). Profiles of the measured settlementsfor various time periods, and the predictedultimate values are shown on Fig. 8.

19. Fig. 10 shows the immediate settlements(i.e. during construction periods only) as afunction of embankment load. By considering onlythe immediate settlements, the slope of theresponse curve can be viewed as indicative of anaverage undrained deformation modulus of theentire foundation. As shown, the trend issimilar to that observed for the shallow piezo-meters within the overconsolidated portions ofthe foundation. Thus, beyond a critical stresslevel, an initial low deformation response isfollowed by a large deformation response, and theaverage foundation stiffness was successivelydecreased. This can be attributed to a pro-gressively greater proportion of the upper zonesbehaving as a normally consolidated soil as thecritical stress levels were reached. The entirefoundation is speculated to have reached anormally consolidated state at 200 kPa, which isconsistent with the maximum preconsolidationpressure measured in the laboratory.

Horizontal Movements20. Typical horizontal movements during

CHAHOE IKJ VERTICAL STRESS (kPa)O O O o~ O O o

tO DO T-

FIGURE 10 - IMMEDIATE SETTLEMENT VS SURFACE LOAD

construction are shown on Fig. 7. Analysis ofhorizontal deformations is extremely complex, andindeed is less understood in geotechnique com-pared to settlement behaviour. From a construc-tion perspective, it must be ascertained if themovements reflect normal behaviour, or there isan impending failure. Horizontal deformationnear the toe is a sensitive indicator ofinstability. In this respect, toe movements atthe East Dam were nominal during construction andhave since ceased, confirming the embankment'sstability. However, lateral displacements ininclinometers at midslope were found to be nearthe upper bound of values previously experiencedfor stable embankments on similar soils in theCanadian prairies, and are continuing (althoughat a decreasing rate). The depth of movementsalso appeared to coincide with the critical slipsurface determined by limit equilibrium analysis.Thus, a detailed review was undertaken in anattempt to better understand the mechanics of themovements, and to confirm the foundationstability.

21. The review was carried out in two parts.Firstly, other case histories were reviewed toprovide a comparison with the movements at theEast Dam. The results suggested that theconstruction movements were not unusual, giventhe deformation characteristics and thicknessesof the foundation materials involved. Secondly,the characteristics of the shear strains wereanalyzed with respect to the vertical strains inthe foundation. It was thought that an analysisof this type would help determine if themovements were occurring under drained orundrained conditions. Movements in inclinometerEBI-5 are illustrated on Fig. 11. Part a) showsthe cumulative movements versus depth forseparate time periods. Part b) compares theshear strains to the vertical strains determinedfrom settlement rings attached to the casing.Part c) shows the time history of the two straintypes at the depth of maximum movement. Thefollowing key points are highlighted:• The major movements occur over a 6 m to 7 m

thick zone, located at about 25 m depth.Above this zone, the shear strains in theclays are minimal.

• The pattern of vertical strains is anapproximate mirror image of pore pressureisochrones, consistent with double drainage inone-dimensional consolidation theory. Local"peaks" in the vertical strain distributionare likely indicative of intermediate drainageboundaries. It is noted that vertical strainswithin the middle of the clay formation arealso minimal.

• Maximum shear strain is located near the base

10

PAPER 1: CHIN , DAVISON, KLOHN, BENSON AND CAMPBELL

790

745

VERTICAL STRAIN

SHEAR STRAIN

100 200(mm)

EBI-5 A-AXIS —•DOWNSTREAMa) CUMULATIVE HORIZONTAL MOVEMENTS b) VERTICAL & SHEAR STRAIN PROFILES (END OF CONSTRUCTION)

1 A ^ S ' 0 ' N ' D | J ' F1987 ' 1988

c) VERTICAL ic SHEAR STRAIN VS TIME

VECTOR SCALE100mm

1. OCT.27/86 (END STG 1)2. MAY 6 /87 (START STG 2)3. JUN.2/87 (END STG 2)4. SEPT. 10/87

DOWNSTREAM

d) DIRECTION OF HORIZONTAL MOVEMENT

FIGURE 11 - INCLINOMETER DEFORMATIONS - EBI-5

of the clay, immediately above a thick sandlayer. Significant vertical strains are alsorecorded at this depth, which suggests thatthe shear strains can be attributed to thedevelopment of consolidation.Fig. 11 c) shows that both the verticalstrains and shear strains increased in asimilar manner during stage 1 construction,followed by a time dependent increase over theshutdown period. During stage 2, however, theshear strains are further increased,accompanied by very little additional verticalstrain. This change in behaviour suggeststhat the early shear strains were related toconsolidation, while the shear strains in thefinal stages of construction likely occurredunder undrained conditions. A review of theconstruction history indicated that thetransition from "partially-drained" to"undrained" behaviour occurred when theembankment was about 3 m to 4 m below ultimateheight.

22. A different perspective of the deformationbehaviour was obtained through a review of thedirections of movements (Fig. 11 d). The resultsfurther illustrate the influence of the deep sandlayer on the horizontal movements. As the sandis located on the south side of the coulee (Fig.3), the major movements tended toward the sandwhere, as shown by the analysis of verticalstrains in the foundation, greater consolidationis occurring. During stage 2 construction, themovements apparently swung more normal to the damaxis. Thus, the direction of movements appearsto be consistent with the interpreted "undrained"behaviour of the shear strains near the end ofconstruction. Following construction, however,the movements have swung back toward the sandlayer, suggesting a progressively greaterinfluence from consolidation. It is noted thatthe inclinometers located further north and awayfrom the influence of the sand consistentlyshowed movements perpendicular to the dam axis,

as would be normally expected.23. The above observations indicated that any

unfavourable (undrained) shear strains in thefoundation occurred near completion of theembankment, when the toe berms were already inplace. The fact that minimal movements wererecorded at the toe confirmed that the berms weresuccessful in mitigating any potential forinstability that may have existed. Of interest,the shear strains have continued at a steadilydecreasing rate after construction, and there ispresently no evidence of unstable behaviour inthe foundation.

FINITE ELEMENT STUDYGeneral

24. The instrument data have confirmed thecomplexity of soft clay behaviour beneathembankments. To develop a basis for evaluationof future performance and to clarify the variousaspects of clay behaviour, finite element studieswere undertaken to match the observed performanceduring construction.

25. Stress-deformation analysis for the EastDam was carried out using FEADAM 84(5), whichsimulates the non-linear and stress dependentstress-strain properties of a soil by ahyperbolic function. Both the embankment filland foundation clays are quite ductile and arewell-suited for analysis with the hyperbolicmodel. Derivation of the hyperbolic parameterswas based on consolidated-undrained triaxialtests and oedometer consolidation tests,supplemented with data from the technicalliterature. Parameters were assigned todifferent soil layers to account for over-consolidation and/or partially-drained, drainedor undrained behaviour, based on actual behaviourinferred from the instrumentation data. Fig. 12shows the finite element grid used for analysisof the dam at mid-valley. Construction wassimulated by applying the dam in successive liftincrements.

11

STATE OF THE ART

o) FINITE ELEMENT GRID & SHEAR STRAIN CONTOURS (%)

740 J

-820

-800'g '

-780 |

-760 3

- 7 4 0

80 120 160 200DISTANCE (m)

320 360

800

FINITE ELEMENTPREDICTION (SG1)

MEASURED (SG1) -

MEASURED (EBI-5)

^—FINITE ELEMENT^ ^ PREDICTION (EBI-5)

=-790-

790 795 800 805 810 815CREST ELEVATION (m)

b) DEFORMATION VS FILL HEIGHT

820750

- D/S

FINITE ELEMENTPREDICTION (EBI-5)

MEASURED (EBI-5)

600100 200 300 400 500HORIZONTAL MOVEMENT (mm)

c) HORIZONTAL MOVEMENT VS DEPTH AT END OF CONSTRUCTION

FIGURE 12 - STRESS DEFORMATION ANALYSIS

26. FEADAM 84 does not account for porepressures and it was necessary to apply theconstruction lift increments as a single stage.This approach was considered reasonable in viewof the minimal pore pressure dissipation observedduring winter shutdown. For comparison to thecomputed deformations, however, only the portionsof the movements which occurred duringconstruction were used. Some fine-tuning of themodel was necessary in order to achieve asatisfactory match to field performance.

Comparison of Deformations27. Fig. 12 shows the key results for

comparison. Part a) superimposes on the finiteelement grid, the contours of the computed shearstrains in the foundation. Direct comparison ofthe strain magnitudes may not be applicablebecause of the coarseness of the finite elementgrid compared to the depth interval measured bythe inclinometers. However, the overall patternof the contours indicates the largest strainsoccurred beneath the central portion of the damdecreasing towards the toe. This is in agreementwith the observed behaviour, where theinclinometer movements at mid slope were greaterthan those at the toe. Moreover, the contoursreflected a preferential zone of movement atabout 20 to 25 m depth, consistent with themovement zone recorded in inclinometer EBI-5.Parts b) and c) compare the computed horizontalmovements at various embankment heights, and thecumulative movements with depth at the end of

construction, to those measured in the field.Again, it is evident that a satisfactory matchwas achieved.

28. Comparison of vertical deformations (Fig.12 b) showed that the measured settlements wereslightly higher than those computed. This can beexpected, however, due to the locally greaterconsolidation at intermediate drainageboundaries, which was not accounted for in theanalyses. Nevertheless, the general trend ofcomputed settlements during dam raising isconsistent with that observed.

Foundation Stresses and Pore Pressures29. Fig. 13 compares the stress increases

computed by finite element analyses to thosecomputed from linear elastic solutions. Thecomparison is made for three piezometer locations(10, 20 and 25 m depths) beneath centreline. Asshown, the vertical stresses from both methodscompare very well in each case. The horizontalstresses at the shallow 10 m depth are also inagreement. However, there are significantdifferences between the horizontal stresses atthe 20 m and 25 m depths, where the finiteelement results are up to 100% greater. Ofinterest, the major zones of horizontal movementsalso occurred at these depths.

30. Fig. 14 shows the recalculated undrainedpore pressures using the stresses from FEADAM 84in HenkeTs pore pressure equation, for the twodeep piezometers beneath centreline (G9 and GIO).Also shown are the measured pore pressures and

—

//

r

790 800 810 820 790 800 810 820ELEVATION DAM CENTERLJNE

STRESS AT 10m DEPTH (G8)

500-

£,300-

200*

100-

0 -

/

/

<

1/AT, 7

/// „

AT3

790 800 810 820 790 800 810 820ELEVATION DAM CENTERLJNE

STRESS AT 20m DEPTH (G9) LEGEND

UNEAR ELASTIC^-

500-

400-

300-

100-

/

/

_3

/

/

!

i1 AT, /.

//

/

I 1

r

790 800 810 820 790 800 810ELEVATION 0AM CENTERUNE

STRESS AT 25m DEPTH (GIO)

FINITE ELEMENT

FIGURE 13 - COMPARISON OF FINITE ELEMENT AND LINEAR ELASTIC STRESSES

12

PAPER 1: CHIN , DAVISON, KLOHN, BENSON AND CAMPBELL

500

400

"5*300

3 200

100

FINITE ELEMENT - ^

ex

MEASUREDO G9x G10

UNEARELASTIC(G9)

UNEARELASTIC(G10)

795 800 805 810CREST ELEVATION (m)

815 820

FIGURE 14 - COMPARISON OF PORE PRESSURE PREDICTIONS

those predicted by linear elastic methods forcomparison. As shown, the pore pressurescomputed from linear elastic solutions signif-icantly underestimated the actual pore pressuresat the end of construction. However, the finiteelement solutions not only correctly predictedthe final pore pressures, but also the rate ofpore pressure generation during construction.Although not shown, the finite element model alsoprovided a satisfactory history match of porepressure generation for the remaining piezometersalong the dam section analyzed.

CONCLUSIONS AND COMMENTS31. The East Dam case history has provided an

excellent illustration of the value of theobservational approach to the design andconstruction of large embankments on soft clayfoundations. Successful completion of theproject within budget and on schedule may nothave been achieved without adopting thisapproach.

32. The performance of the East Dam hasconfirmed the complexity of soft clay behaviourbeneath embankments, and re-affirmed the need toaccount for the stress history and yieldbehaviour of the clays as well as the anticipatedtypes of loading, in prediction methods. Despitethis complexity, the present study has shown thatit is possible, with the advent of moderncomputers and advanced analytical tools, toaccurately forecast settlements, horizontalmovements and pore pressures concurrently.

33. One of the most important aspects of thefinite element studies was to demonstrate thatundrained yielding of the clay can result inhorizontal stress increases which are much largerthan the corresponding stresses for claysdeforming in a linearly elastic manner. Neglectof this behaviour can lead to an unconservativeestimate of pore pressures, as experienced at theEast Dam.

34. The finite element studies presented inthis paper were implemented because of concernsfor embankment stability during operations, dueto minimal dissipation of pore pressures and the

slow rate of increase in the factor of safetyafter construction. The calibrated finiteelement model was subsequently used to evaluateand predict the performance during reservoirfilling. Because of space limitations, it wasnot possible to include the data collected duringthe filling periods in this paper. However, theobserved behaviour essentially conformed to thefinite element predictions and since reachingfull supply level in mid-1989, the embankment hasperformed satisfactorily. Monitoring of theembankment is continuing on a reduced basis.

ACKNOWLEDGEMENTS35. The Forty Mile Coulee Project was funded

by the Alberta Heritage and Savings Trust Fund,and administered by Alberta Environment. Thepermission of Alberta Environment to publish thedata is greatly appreciated. Design of theproject was carried out by Klohn Leonoff Ltd. inassociation with W-E-R Engineering Ltd. andAssociated Engineering Alberta Ltd. Theconstruction contract was awarded to KiewitManagement Limited. Project review board membersfor Alberta Environment, Mr. P. Rivard, Dr. E.Brooker and Mr. L. Swan, and AlbertaEnvironment's Mr. J. Thiessen, provided thenecessary stimulation and encouragementthroughout construction, and their contributionsto successful completion of the project are fullyacknowledged. Finally, special thanks is givento Dr. Morgenstern who provided suggestions andan independent review of the analytical studies,and to Mrs. S. Housken and Mr. C. Baron for theirassistance in preparing this manuscript.

REFERENCES1. HENKEL, D.J. The Shear Strength of SaturatedRemoulded Clays. Proceedings, ASCE SpecialtyConference on Shear Strength of Cohesive Soils,Boulder, Colorado, 1960, pp.533-554.2. LAW, K.T. and BOZOZUK, M. A Method ofEstimating Excess Pore Pressures BeneathEmbankments on Sensitive Clays. Canadian Geo-technical Journal, 1979, vol. 16, pp. 691-702.3. GRAY, H. Stress Distribution in ElasticSolids. Proceedings, First InternationalConference on Soil Mechanics and FoundationEngineering, 1936, Vol. 2, p. 157.4. PECK, R.B. Advantages and Limitations of theObservational Method in Applied Soil Mechanics.Geotechnique, 1969, Vol. 19, No. 2, pp. 171-1875. DUNCAN, J.M., SEED, R.B., WONG, K.S. andOZAWA, Y. FEADAM 84: A Computer Program forFinite Element Analysis of Dams. GeotechnicalEngineering Research Report No. SU/GT/84-03,Dept. of Civil Engineering, Stanford University,

13

2. The application of new techniques in the design of two highdams in South West China

F. DAFENG, Ministry of Water Conservancy and Power, People's Republic of China

The paper describes two high dams located in the southwest region of China. The author puts empha-sis on presentation of the design of the Lubuge rockfill dam with a saprolite core and the designof the Tianshengqiao I concrete face rockfill dam, which reflect the modern design level of damsin China.

I. LUBUGE ROCKFILL DAM WITH A SAPROLITE COREThe Lubuge hydropower project is located on theHuangni river, a tributary of the Nan Pan riverbetween Yunnan Province and Guizhou Province inthe southwest of China. It was designed by theKunming Hydroelectric Investigation and DesignInstitute, Ministry of Energy and Ministry ofWater Resources. The headworks and the power-house complexes of the project are being con-structed by a Chinese construction force, theFourteenth Construction Bureau (FCB). The Tai-sei Corporation was awarded the contract tobuild the hydraulic system.

1. IntroductionThe reservoir formed by the dam has a totalstorage capacity of ].11x108 m^, which is capa-ble of seasonal regulation only. The total in-stalled capacity of the project is 600 MW andits mean annual output is 28.5x10** KWH.The construction of the project started at

the beginning of 1983. The first two generatingunits were put into commissioning in December,1988 and September, 1989 respectively. Theother two generating units are scheduled to beput into commissioning by the end of 1990.The project comprises three major structure

groups, namely, the headworks complex, the hy-draulic system and the powerhouse complex.A dam and three spillway structures are the

main components of the headworks complex. Thedam is of rockfill type with an impervious sa-prolite core, 103.8 m high and 2 17 m long atits crest. A parapet wall, 1.2 m high, is pro-vided along the dam crest. The surface spillwayis constructed on the left abutment. The leftbank spillway tunnel has its upper portion un-der pressure, while its lower portion, whichcoincide with the diversion tunnel, is of freeflow. The right bank spillway tunnel is usedfor river diversion during dam construction,flood releasing, flushing and emptying of thereservoir. These spillway structures form aflood releasing system at high, intermediateand low levels to provide a total discharge ca-pacity of 10,092 mVs (P.M.F.); The hydraulicsystem, located on the left bank, consists ofa power intake, a power tunnel, a differentialsurge shaft, two penstocks, two bifurcations,

The embankment dam. Thomas Telford, London, 1991

a power house, an auxiliary power plant, amain transformer and switchgear room and atailrace gate chamber, and four tailrace tun-nels. The power tunnel is 8.0 m in diameterand 9,387 m in length. The surge shaft has anupper pool. The two penstocks have a diameterof 4.6 m each and the two bifurcations get intothe power house with an inclination. The mainpower house and the auxiliary power plant, themain transformer and switchgear room and thetailrace gate chamber are parallel to eachother. The main power house accumulates fourturbo-generator units with an installed capa-city of 150 MW each.For the layout of the headworks complex, see

Fig. 1.

2. Embankment material of the damA. For the cross section of the dam, see

Fig. 2.B. Impervious core material of the dam:

At the initial stage of the construction, theimpervious core material of the dam was pre-viously planned to obtain from a borrow area,13.7 Km away from the dam site, where a largeamount of slope wash and residual soil of wea-thered dolomite is deposited. Its natural moi-sture content is higher than the optimum moi-sture content by 8.6-10% and the content ofmedium-sized clay particles (less than 0.005mm) in the soil accounts for 60%. It oftenrains and is wet in the project area and thereare only 65 sunny days in a year, which wouldbring about difficulties to construction of

1. Earth and rockfill dam2. Right bank spillway tunnel3. Surface spillway

4. Left bank spillway tunnel5. Centre line of flushing tunnel6. Centre line of power intake7. Diversion tunnel

Fig. 1. A layout of the headworks complex

15

STATE OF THE ART

of the dam. So it is necessary to mix 40% sand1. Dam axis2. Core axis3. Impervious core made of

weathered material4. Rockfill shell5. Filter

6. Fine rockfill7. Excavated spoil8. Impervious inclined wall of

cofferdam9. Transition zone

10. Consolidation grouting11. Curtain grouting12. Concrete cushion & clay over it

13. Upstream slope protection14. Parapet wall15. Normal storage water level

Fig. 2. A cross section of the Lubuge earth &rockfill dam

and gravels into the soil in order to improvethe properties of the material and the con-struction conditions. Thus it can be seen thatthis kind of soil material needs certain treat-ment, which would not only be costly, but alsotake too much time. Furthermore, the haulingdistance is long. It is not economical to usesuch soil as core material. After careful in-vestigation and systematic tests done both inlaboratory and at field a nearby borrow area ofresidual soil of weathered sandstone and shalewas established as acceptable core material. Itis only about 3 Km away from the damsite andhas following properties:

a. The soil material used for the core in-cludes Quaternary saprolite, 2-3 m in thick-ness, soil of completely weathered sandstoreand shale, 3-5 m in thickness, fragments ofcompletely weathered sandstone and shale, 2-4 min thickness, and blocks of completely wea-thered sandstone and shale, 1-3 m in thickness.b. The soil material is mineralogically com-

posed of illite and kaolinite. The chemicalanalysis has shown that it contains more freesilicon, aluminium and iron, and SiO2/Al2C>3 is1.33-2.c. The natural moisture content and the void

ratio of the weathered blocks of the materialare 20-4 1% and over 40% respectively, whichare much higher than those of the unweatheredblocks. Meanwhile their compressive strength islower than 3.0 MPa under their moisture contentof more than 10%. It indicates that the rocksare highly weathered and subject to be brokenunder pressure.d. The test results have shown that the dis-

persion of the soil is zero. It is consideredas a non-dispersive clay.e. There is no much difference between the

natural moisture content and the optimum moi-sture content of the soil, so no special treat-ment is needed. The average content of clayparticles of the material is 24%. It can bewell compacted by roller and convenient forthe embankment construction.f. The soil has a permeability coefficient

of ixl0~7 cm/s, a effective strength of the in-ternal friction angle, 32° and cohesion, 0.034MPa, a compressibility coefficient of 0.014-0.027 cm^/lON, which can satisfy the technicalrequirements for the embankment construction.It can be seen from the above that the above-

mentioned soil material is characterized by lowpermeability, high shear strength, low compres-sibility and good compactibility, and it issuitable for construction of the embankment. It

is considered an ideal material for the damembankment.However, attention should be paid to the fol-

lowing problems during construction of the damembankment:a. Control of the gravel content in the

course of compaction: The content of gravelsshall be within the limit of 30-50% after com-paction. The compaction test has proven thatthe average content of gravels is 65.7% beforecompaction, while it is 38.3% after compaction.The breaking ratio is 40%.

b. The soil material must be after compa-ction homogeneous, no seepage passage to creatdue to concentration of gravels is allowed.The inspection pits excavated after the fieldcompaction tests after mixing the materialsduring borrowing have shown that the fillinglayers are well bonded and the gravels are sur-rounded by fine-grained materials, and no holeswere found in the compacted layers.c. In order to increase the breaking ratio

of the weathered materials, it is advisable toemploy heavy roller wet compaction.C. Filter materials:Although the above-mentioned soil material

has significant advantages, it has poor plasti-city, low tensile strength (0.003-0.0078 MPa)and is liable to disintegration, which will re-sult in cracking in the dam core. Of course, itis not easy to prevent the core of the dam fromcracking, but we have to protect the core ma-terial against erosion and to make the core ma-terial be capable of self-sealing. So properfilters are required to protect the core mate-rial in case the core is cracked.The core of the Lubuge dam is provided with

two filters each on both sides to protect thecore material against erosion and to transitboth strength and deformation of the core. Thefilter material was made from crushed dolomiteand limestone. A part of screened river-bedmaterials was also used for the filters.A core protection test against cracking by

using filters was carried out. For the test,the simulated crack in the core is 2 mm wide,7 cm long and 15 cm deep, the max. hydraulicgradient is 100, and five gradations of filtermaterials were adopted. The test results haveshown that the seepage discharge for each testis stable and the finer the filter material is,the less the seepage discharge is. The seepagedischarge decreased with time. It indicatesthat the cracks are gradually self-sealed. ThePervious material of D20 less than 2.5 mm willmeet the requirements for the filters.D. Rockfill material:Material excavated from the surface spillway

and tunnels were used for rockfill. Some dolo-mite and limestone rocks were also quarriednear the damsite. The grain size gradation ofthe rockfill material was determined accordingto the results of the calculations and theblasting tests. The material excavated from thetunnels is of fine grain size, with a contentof grain particles of less than 5 mm in sizeaccounting for about 30%, from which testsamples were taken for testing. The tests re-veal that its permeability coefficient isixl0~2 cm/s and its shear strength, the inter-nal friction angle is 40-42°. Both of them can

16

PAPER2:DAFENG

meet the technical specifications. Therefore,the content of the fine rockfill material witha grain size of less than 5 mm should not ex-ceed 30%, while that of less than 0.1 mm shouldnot be more than 5%.After being compacted the dry density of the

rockfill material should be 20.6 KN/m^ and voidratio, 25%.

3. Analysis and calculation of the damThree-dimensional and two-dimensional stressstrain analysis were conducted for the dam. Theresults of the calculations have revealed:A. In the middle of the core, is 60% of

the soil pressure applied at that point andthere is an arching effect, which decreased toa certain extent after impounding.B. is of tension in the places where both

abutments are close to the river bank slopes.This indicates that cracks might occur in thoseplaces.C. During impounding of the reservoir, the

stress level (shear stress/shear strength) ofthe core is 0.5, which is considered safe, How-ever, the stress level of the dam shell in-creases and the safety is lowered.Seen as a whole, the designed dam cross se-

ction and the dam materials selected are sui-table for such a dam. Local cracks might occur,as long as the filter material could be proper-ly prepared and its thickness and grain sizegradation could meet the required technical re-quirements, the dam would be safe.

4. Foundation treatment of the damThe dam core is founded on moderately weatheredrocks and a reinforced concrete cushion, 0.5-1.0 m thick, has been installed between thebedrock surface and the core of the dam inorder to protect the core bottom against con-centrated seepage and contact erosion. Blanketgrouting was performed under the cushion tokeep the foundation complete and integral. Thegrouting holes, 5 m deep each, were arranged inrows, 3 m apart. Under the dam foundation agrouting curtain has been provided in threerows. The maximum depth of the grouting holein the middle one is 89 m and they are spacedat 2 m, while the depth of the holes in the twoside ones, which are spaced at 3 m and 2 m fromthe middle one on both sides, ranges from 12 mto 15 m. In order to prevent seepage around thedam, grouting curtain extends 121 m in the leftbank and 93 m in the right bank respectively.All the grouting was coducted from ground sur-face or grouting galleries.

5. Construction of the damA. 5.7 m3 WA600 loaders, 20 t HD205 and 15 t

Mitsubish dump trucks were used for borrowingand transporting the weathered materials forthe core construction. Bulldozers or levelerswere employed for leveling. The layer is 25 cmthick. Compaction was carried out by SPF-84 vi-brating tampers, with 12 passes. Before sprea-ding soil materials, bulldozers were used toroughen the surfaces so as to provide a goodcontact between two layers.B. Filter materials: After dumping materials

over the dam by dump trucks, it would be le-veled with a bulldozer. The layer is 50 cm

thick. A BW217D vibrating smooth roller wasused for compaction, with 2 passes under staticpressure.C. Rockfill materials: CAT769C 32 t trucks

were used to transport rockfill material fromthe stock pile areas or the quarry site to thedam, and bulldozers or hydraulic backhoes wereemployed for leveling material. The layer is80-100 cm thick. Compaction was done by BW217Dvibrating smooth roller, with 8 passes afterwatering.

6. InstrumentationIn order to monitor a safe operation of the damand to verify the design as well as the, con-struction quality of the dam, a number of de-formation gauges, seepage pressure meters,earth pressure cells and pore water pressurecells have been installed in the dam embank-ment. Because the dam has not yet been underoperation for a long time, the water level ofthe reservoir has never reached the elevationas high as the normal storage water level, theobservation results obtained are quite limited.However, the analysis of the observation datacollected so far has shown that the dam fun-ctions normally and all the parameters proveto be up to the design specifications.

II. THE TIANSHENGQIAO I CONCRETE FACEDROCKFILL DAMThe Tianshengqiao I hydropower project, theuppermost one of the cascade hydropower deve-lopments on the Hongshui river, is located onthe Nan Pan river along the border betweenGuizhou Province and the Guangxi Zhuang Auto-nomous Region. This hydropower project is de-signed by Kunming Hydroelectric Investigationand Design Institute under Ministry of Energyand Ministry of Water Resources. The prelimi-nary design of the project has already beenexamined and approved by the state. The tenderdocuments are being prepared at present.

1. IntroductionThe Tianshengqiao I hydropower project is theuppermost one of the hydropower developmentson the Hongshui river. After completion of thedam, the water reservoir formed will have astorage capacity of 102.57x108 m3, which is ca-pable of a long-term regulation. The total in-stalled capacity of the project will be

Fig. 3. A layout of the Tianshengqiao I pro-ject complex

17

STATE OF THE ART

1,200 MW and its mean annual output will be52xl08 KWH. In addition, the total capacity ofthe three constructed hydropower stations, whichare located downstream of the Tianshengqiao Ihydropower project will be increased by 640 MW,and their firm capacity and annual output willbe increased by 880 MW and 41xlO8 KWH resptly.Therefore the construction of the TianshengqiaoI hydropower project will bring about great be-nefit.The Tianshengqiao I hydropower project com-

prises the following major structures:A concrete faced rockfill dam, 178 m high

and 1,137 m long at its crest;A surface spillway on the right bank, de-

signed for 1,000 years frequency flood with acorresponding discharge of 15,282 mVs, checkedwith P.M.F. with a corresponding discharge of21,750 m3/s;

An intermediate outlet on the right bank,which will be used for emptying the reservoirand for river diversion during construction ofthe project;

A hydraulic system on the left bank, con-sisting of four power tunnels and four pen-stocks to convey water to four turbo-generators;

A surface power house to accommodate fourturbo-generators with an installed capacity of300 MW each; and

Two diversion tunnels in the left bank.During the first year after river closure,flood will overtop the dam, during the secondyear, the two diversion tunnels will meet thecriterion for 300 years frequency flood. Afterthat the flood protection criteria will beraised year by year.It will take 7.5 years to put the first gene-

rating unit into commissioning from commence-ment of construction of the project. The totalconstruction period of the project will be 9years. For the layout of the dam complex, seeFig.3.

2. Design of the dam cross sectionA. For the cross section of the dam, see

Fig.4.B. Zoning of the dam: Zoning of the dam is

specified as follows:Zone I - impervious material, upstream of

the concrete face;Zone II - cushion material, downstream of

the concrete face; and. Zone III - rockfill

791

615V

1. Dam axis2. Concrete face slab3. Plinth

5. Curtain grouting6. Parapet wall7. Slope protection8. Bedrocks

4. Consolidation grouting 9. Normal storage water level

Fig. 4. A cross section of the Tianshengqiao Iconcrete face rockfill dam

a. Zone I: The Tianshengqiao I concreteface rockfill dam is the highest one of thiskind in the world, there will be an imperviousearth membrane adjacent to the bottom, upstreamof the concrete face. Its height will be aboutone third of that of the dam embankment. Theearth membrane is divided into two subzones,namely, IA and IB. The former will be providedon purpose to protect the concrete face againstseepage or to seal cracks and open joints incase it cracks, and it will be made of sandyloam or silty clay or clay and directly placedagainst the concrete face and plinth, while thelatter will be used to stabilize the former asa counter weight, and excavated spoil can beused for this subzone, which does not have spe-cial requirement for selection of material.b. Zone II: It is a cushion under the con-

crete face. Provision of Zone II is to providea reliable and smooth foundation surface tosupport the concrete face, to uniformly transitwater pressure to the rockfill as well as toretain water before placing of the concreteface slab. The cushion will be protected witha bituminous emulsion coating on its surface.When the concrete face cracks, soil particlesin Subzone IA and Zone II will not be washedaway by water, thus reducing leakage of thedam. For these reasons, the material in Zone IIshall be well graded and its permeability coef-ficient shall be 10""3-10~^ cm/s so that it cannot only function as a filter to Subzone IA,but also be capable of protecting against see-page. The width of the cushion will be deter-mined according to the requirements for con-struction. Its thickness will be 3 m and can beincreased at its bottom.c. Zone III: It is a rockfill zone, which

is divided into four subzones, i.e. IIIA,IIIB, IIIC and HID. Zone III is so arrangedthat the compressibility and the permeabilityof the rockfill in all the subzones shall gra-dually be increased towards downstream.Subzone IIIA: As a transition zone between

Subzone IIIB and Zone II, it will prevent par-ticles in Zone II from being taken into voidsin the rockfill. So its grain size gradationshall meet the requirement for filtering. Sub-zone IIIA, 5 m thick, will be made of screenedmaterials obtained from Subzone IIIB.

Subzone IIIB: It is a main rockfill zone,which will resist the water pressure appliedover the dam. The magnitude of its deformationwill have direct influence on deformation ofof the concrete face. So it shall be well gra-ded and drained, and also require a high com-pression modulus. Hard rocks excavated fromthe spillway or from the quarry site will beused for this zone.

Subzone IIIC: Materials excavated from thestructures on the left bank and from the inter-mediate outlet on the right bank will be placedon the downstream side of the dam axis abovethe downstream water level to form this zone.Subzone HID: It is the most downstream

part of the rockfill embankment. Its defor-mation has no much influence on the concreteface. It has no strict requirement for itsthickness and grain size gradation. Duringconstruction of the dam, excessive size stoneswill be pushed to the downstream slope of the

18

PAPER2:DAFENG

dam by bulldozers and trimmed manually to forma slope revetment, nice in appearance.

Materials in Different Zones

o 2 10015 r

Zone

IA

IBII

IIIA

Material

riverbed alluviumsilty & sandysoil or clay fromthe borrowrandom materialswell-graded cru-shed sand & gra-vels

screened well-graded coarse-grained material

Thick.of

a layer

(m)

0.30.3

0.3

0.3

Compaction

ordinarycompactionroller

dittovibratingroller, 4passes hori-zontally, 6passes frombottom to topover slopeunder staticpressure

vibratingroller, 4passes

ITIB well-graded lime-stone excavationfrom the spillway

IIIC sandstone andmudstone or lime-stone excavationfrom the tunnels

H I D limestone exca-vation from thespillway

0.9

0.9

ditto

ditto

ditto

For the gradation curves of the materials fordifferent zones, see Fig.5.

3. Concrete face slab and plinthThe concrete face is the major impervious stru-cture of the dam. It must be very reliable andmeet the requirements for both strength andimpe rme ab i 1 i t y.

The thickness of the concrete face is 0.3 mon its top and then gradually increases fromtop to bottom as per the following formular:

T = 0.3 + 0.0035H

where T is the thickness of the concrete faceand H is the vertical height below the damcrest.

The concrete face will be divided into 7 1blocks along the length of the dam axis, 16 mwide each. In order to reduce the joints causedby change of temperature and let them developin a uniform way, a layer of steel reinforce-ment arranged in two directions will be in-stalled in the concrete face, with a rate ofreinforcement of 0.4% in each direction. Theconcrete face will be concreted in two stages.The first stage concrete wil not be placeduntil the dam embankment has been filled ashigh as 112 m. The second stage concrete willbe placed after the rockfill has been raisedto the crest of the dam.

The width of the plinth will be one fifteenth

1000 100 10d: mm

0-1

Fig. 5 Grain size gradation curves of the Tian-shengqiao I dam materials

of the water head. The plinth will have threedifferent widths, namely, 10 m, 8 m and 6 m,with their corresponding thickness of 1.0 m,0.8 m and 0.6 m respectively. The plinth,which will be provided with two-directionalreinforcement in its upper part, with a rate ofreinforcement of 0.3%, will be founded over themoderately weathered rock formations, connectedwith its foundation with dia. 30 mm dowels,which will be driven into the foundation 3 mdeep, spaced at 1.5 m. Beneath the plinth con-solidation grouting to be arranged in rows,2 m apart, will be performed, 15 m deep, withholes spaced at 2 m. In addition, a row of cur-tain grouting, 80 m deep, is to be installed.200# concrete, with a permeability require-

ment of S8 will be used for both concrete faceand plinth.

4. JointsIn order to simplify the arrangement of theconcrete face, only peripheral joints and ex-pansion joints will be provided for the Tian-shengqiao I dam.A. Peripheral joints: The peripheral joints

will be installed between the plinth and theconcrete face, which will be founded over thebedrocks and the rockfill respectively. Thewater load and the self-weight of the rockfillwill result in different displacements of theperipheral joints. It is expected that thedisplacement of the peripheral joints will havesignificant influence on permeability of thedam. Having consulted the details of the peri-pheral joints in the Foz Do Areia dam, theTianshengqiao I dam will also be installed withthree kinds of water stops, i.e. copper waterstops at the bottom of the concrete face, pla-stic water stops in the middle and bituminousmastic on the top, to be covered with rubberstrips, which will be fixed by bolts into theconcrete, so that the expansion joints could befilled by mastic in case they are opened. Inthe mastic, chloroprene rubber tubes will beembedded. In the joints there will be bitumen-coated wooden boards to prevent the concretefrom breaking, when the concrete face is undercompression.

B. Expansion joints: The expansion jointsrefers to the joints between two blocks of theconcrete face. After impounding of the reser-voir, the rockfill will be somewhat displacedand most of the concrete face will also be un-der pressure, only on both abutments the con-crete face will result in tension due to com-pression of the rocks on both banks. Based on

19

STATE OF THE ART

the analyses and calculations as well as theexperience from other projects, 27 open joints,i.e. 17 in the right abutment and 10 in theleft abutment, will be provided. Compressiblejoints will be installed in the middle. Thecompressible joints will only be provided withcopper strip water stops, with a bituminouscoat over the joint surface, while the openjoints shall be filled with bituminous masticon their tops in addition to the water stops.

5# Parapet wallBased on the max. flood water level and theheight of waves, a vertical parapet wall, 4.7mhigh, i.e. 1.0 m above the dam crest, will befounded on the rockfill, with its upstreamface connected to the concrete and with waterstops inside. Provision of the parapet wallwill make it possible to reduce volume of theupstream rockfill and to increase the width re-quired for sliding formworks.

REFERENCES!. Kunming Hydroelectric Investigation andDesign Institute Instruction of the TechnicalDesign of the Lubuge Hydropower Project.2. Kunming Hydroelectric Investigation andDesign Institute Engineering Properties of theWeathered Materials Used as Impervious Mate -rials for the Lubuge Earth and Rockfill Damand Selection of the Compaction Standards forthe Dam Construction.3. Kunming Hydroelectric Investigation andDesign Institute Instruction of the Prelimi-nary Design of the Tianshengqiao I Hydropowerproject.4. COOKE J.B. and SHERARD J.L. Design of Con-crete Faced Rockfill Dam II.5. PINTO N.L.de S., FILHO P.L.M. and MAURER E.Foz do Areia Dam - Design, Construction and Be-haviour .

20

3. The use of low grade rockfill at Roadford Dam

A. C. WILSON, Babtie Shaw and Morton, Glasgow, UK, and J. D. EVANS, South WestWater Services Ltd, Exeter, UK

Roadford Dam impounds a reservoir for South West Water Services Limited which will improvewater supplies to various areas of Devon. The dam is constructed of low-grade rockfill and anasphaltic concrete membrane on its upstream face provides the waterproofing element. Siteinvestigations with an exploratory quarry and trial embankments enabled the characteristics ofthe rockfill material to be examined. Minerals in the rocks may cause degradation of the fillmaterial within the dam embankment. The results of testing and the possibility of degradationwere taken into account when designing the embankment. The paper presents the results ofthe investigations and describes the design and construction of the embankment.

INTRODUCTION1. Roadford Dam forms a 37000 Ml reservoir

to alleviate water supply deficiencies inPlymouth, South West & North Devon.Impounding commenced in October 1989 and bySpring 1990 the water level had risen towithin 6.5m of TWL where the volume stored isalmost 60% of the total capacity.

2. The dam is an embankment, 4lm high and430m long and is formed of 1,000,OOOnr oflow-grade rock fill obtained from a borrowquarry within the reservoir basin, about 600mupstream of the dam.3. Waterproofing is by means of an

asphaltic concrete membrane laid on theupstream face and connected to a cut off andinspection gallery structure at the upstreamtoe which surmounts a grout curtain up to 40mdeep.4. The overflow consists of a bellmouth

spillway tower discharging into a reinforcedconcrete culvert constructed in thefoundation of the embankment. An integralculvert provides access to the draw-off towerwhere pipework and valves allow water to bedrawn off from three levels.5. A comprehensive range of geotechnical

instruments are built into the embankment tomonitor its performance. These includepiezometers in the foundation and embankment,vertical settlement and horizontalextensometers, pressure gauges abovestructures and electrolevels to measuresettlement of the asphalt membrane.

INVESTIGATIONS6. Site investigations were undertaken in

1975 and 1977 which confirmed that the sitewas suitable for the construction of a damand enabled preliminary designs to beprepared for presentation at a PublicInquiry. In fact, the Public Inquiry was re-opened on a number of occasions between 1978

and 1983 to consider various topics beforethe water Order was finally authorised.7. A more detailed investigation commenced

late in 1983 which included trial pits,boreholes and seismic surveys together withthe excavation of a trial quarry and theconstruction of trial embankments.Assessments of the physical properties of thefill were made by both laboratory and in situtests and the chemical properties were alsoinvestigated.8. When the exploratory quarry was opened

the material was classified into three typesrelating their depth below ground level andweathering characteristics. Laboratory testscomprised index property determinations,particle size analyses, unconfinedcompression tests on cored rock samples, rocksoundness tests, compaction tests todetermine optimum moisture content,permeability testing, direct shear in a 300mmx 300mm shear box, assessment of shearstrength by triaxial compression apparatus,consolidation testing by Rowe cell and X-raydiffraction. In situ tests included densityby sand and gravel replacement methods.Permeability was measured by soakaway andwell permeameter tests, both analysed bymethods described in the USBR Earth Manual.9. The geology at the dam site and borrow

quarry consists of sedimentary rocks of theCrackington Formation, part of the CulmMeasures of the upper Carboniferous age. Therocks comprise rapidly alternating mudstones,siltstones and sandstones which have beenconsiderably folded and even overfolded.Faulting has also occurred and sheardeformation is common, giving a closelyfractured structure. Near the surfaceweathering is extensive in the mudstones togive a considerable depth of clay typematerial, but above the sandstone there is acomparatively thin layer of gravel andcobbles.

The embankment dam. Thomas Tetford, London, 1991 21

STATE OF THE ART

10. A review of the results indicated:-1) that the most weathered material wouldnot be suitable to form a core within theembankment.ii) an embankment could be constructed of theless weathered materials obtained from depthsgreater than about 4m below original groundlevel.

11. In addition to standard tests carriedout on small size samples, large scale testswere carried out at the Building ResearchEstablishment to determine the properties oflarger samples of rockfill. These includedinvestigations of compressibility in 152mmand lm diameter oedometers (see Fig. 1),strength by drained triaxial compressiontests in 230mm diameter apparatus (see Fig.2) and the influence of fines on permeabilityusing a 150mm diameter permeameter.

CHEMICAL DEGRADATION12. Early in the investigations it was

noted that the rockfill, in particular themudstone fraction, contained minerals whichare prone to degradation by chemicalweathering. The principal mineral componentsof quartz and silicates are cemented into asolid matrix by iron sulphide (pyrite), ironbearing carbonate (siderite) organic carbonand occasionally calcite. The mostvulnerable of these is the iron sulphide.Mudstone samples from the borrow quarry werefound to have a mean pyrite content of almost1% by weight.

13. Weathering of the sulphides in anoxidising environment will produce acidityand increased sulphate content. The acidicconditions may attack and alter clay mineralsand leach out elements. Ultimately thestrength of the rock may be affected to suchan extent that calcareous mudstone may bereduced to a clay or calcareous sandstonereduced to sand.

APPLIED VERTICAL STRESS ( K N / m 2 )

0 500 1000 1500

1 -

6-

7-

""fcfcW..,of construction ^ ^ V K

l e v e l 1 2 0 0 m A O D f Settlement Gauge

14. The rate at which these processesproceed is the subject of much debate*Estimates ranging from a geological timescale, apart from the few metres near thesurface, to only a few years, if conditionswithin the embankment permit vigorousactivity of various sulphur reducingbacteria. The evidence from local conditionsand from embankments formed from similarmaterials elsewhere indicates thatdegradation is likely to proceed at arate between these two extremes. Degradationmay affect the properties of the dam fillmaterial and make the water drainingfrom the dam unsuitable for discharge to theunpolluted River Wolf.

15. Analyses and tests have shown that itis mainly the fresher mudstone rocks withinthe dam fill material which are mostvulnerable to long continued degradation, themore weathered rocks having a considerablylower potential for such degradation.As the completely weathered residual soilsoverlying the site represent the climax ofthe weathering process they will have similarproperties to the rockfill in the damembankment if it degrades completely. Theresults of tests on the most weatheredmaterial gives a measure of these properties.These indicate, perhaps surprisingly, thatthe material is at least as strong as therockfill obtained from greater depths in theborrow quarry.

EMBANKMENT DESIGN16. The stability of the dam has been

assessed for a number of design conditionsincluding end of construction, in-service andrapid draw-down. The effect of earthquakeforces of 0.05 g in addition to the first twocases was also considered. The analyses werecarried out using the GEOCOMP-SLIPSYSTprogram for both circular and non-circularfailures. In undertaking the analyses theinfluence of a variety of possibleconfigurations of planes of weakness in thefoundation were considered. A number ofdifferent face slopes were examined within therange 1:2.0 to 1:2.5. Satisfactory factors

- BRE TRIAXIALS » d = 2 22 Mm3

- . - Yd s 1 9 6 t / m l

DESIGN STRENGTH.IN-SERVICE ^*m/^

>>-

CONSTRUCTION TEST ENVELOPE

INVESTIGATION STAGE100mm OIA TRIAXIALS

Inundation

700 tOO 600 800

MEAN STRESS I 6'. 6JI7 kN/m*

Fig 2 Drained triaxial test results

Fig 1 Large oedometer test results

22

PAPER 3: WILSON AND EVANS

of safety were found for both upstream anddownstream faces having slopes of 1:2.25 andthese are the slopes to which the embankmenthas been constructed, except at the extremedownstream toe where it slackens to 1:2.5.

17. Roadford is considered to be ahomogeneous dam. The strength parametersused in the design reflect the propertiesappropriate to the material type,the confining pressure within the embankmentand its state of degradation as shown inTable 1. At the end of construction the fillmaterial was considered to be undegraded witha shear strength depending on the confiningpressure within the body of the dam. Underthe in-service condition it is assumed thatthe fill material is saturated and degraded.

Table 1 Design shear strengths

Design Case

End of Construction

In service

0f

deg

4032

3529.5

kN/m2

031

031

ConfiningPressurekN/m

0 - 150>150

0 - 230> 230

18. The permeability of the rockfill inthe embankment at the end of construction wasexpected tp lie within the range of 1 x 10to 1 x 10"" m/s. No constructional porepressures were anticipated.

EMBANKMENT CONSTRUCTIONProgramme

19. The £16.lm contract for the dam was letto Alfred McAlpine Construction Ltd. inFebruary 1987. The construction period was31 years with the proviso that the dam besufficiently far advanced for impounding tocommence before the winter of 1989. Thepriority during the first season was theconstruction of the overflow culvert as itwas to be used for the temporary diversion ofthe River Wolf. Work also proceeded on thecut-off and grout curtain and other -structures. Eventually over 50,000 m ofconcrete was placed from the site plant.

20. Exactly a year after work began onsite the river diversion was achieved inMarch 1988 which allowed the embankmentfoundations to be prepared and the drainagelayer placed in the valley bottom.

Embankment construction commenced in June and1,000,000m of rockfill material was placedin a 5 month period. Also during this seasonthe 30m high valve tower and the 15m highparallel sided section of the overflow towerwere constructed by slipforming techniquesduring a 10 day continuous working period.

21. After the completion of the embankmentat the end of October the filters anddrainage layers were placed on the upstreamface during the winter period. Dry and warmweather allowed the 25,000m of site mixedasphaltic concrete membrane to be placed inan eight week period during the Spring of1989.

22. By the end of July 1989, three monthsahead of target, the dam was ready forimpounding. However at that time the flow inthe river was below the 9 Ml/d compensationwater requirement and it was not until thedrought was broken on 20th October thatimpounding was able to commence. During theintervening period most of the outstandingwork was finished and the completioncertificate was issued at the same time asimpounding, nine months ahead of contractcompletion date.

Prevention of Pollution23. As a result of commitments made to

the downstream riparian owners, the contractstipulated that no pollution of the RiverWolf should occur as the result of damconstruction activities. To achieve this,the contractor constructed a series ofsettling lagoons into which all water fromexcavations and construction operation wasdirected.

Constructional Aspects24. The specification required that material

to be used for the embankment would be derivedfrom rock as defined in Table 10 of BS 5930as "Fresh", "Slightly Weathered" or"Moderately Weathered". The weathering gradewas to be applied to the rock mass, based onobservation of about 100m of face, and notto the individual layers within the mass.

25. To ensure that the embankment would nothave areas of uncontrolled ranges ofpermeability, two alternative approaches wereadopted (i) material with differentpermeability properties should be placed indifferent locations within the dam or (ii)blending should be carried out so that auniform material could be placed at anylocation. If option (i) were adopted it wasdesirable to expose the full range of stratain the quarry at the earliest opportunity sothat both the more and less weatheredmaterial could be worked simultaneously anddirected to the most appropriate part of the

23

STATE OF THE ART

embankment. However it was always acceptedthat whatever method was adopted the moreweathered material from the top of the quarrywould form the lower part of the damembankment.

26. When construction commenced thecontractor proposed that he would selectmaterials from the quarry and direct them tothe appropriate parts of the embankment.Excavation commenced from benches followingthe hillside contours which resulted in somedifficulty in identifying the top of suitablematerial and in a preponderance of weatheredstrata. The use of the BS 5930 definitionsalso gave rise to some discussion regardingthe meaning of 'soil1 within the moderatelyweathered description. In addition theweather susceptibility of the weatheredmaterial gave rise to concern that fill withproperties poorer than had been assumed inthe design may be incorporated in thestructure.

Operation of the Borrow Quarry27. Excavation was initially carried

out by face shovel loaders, assisted by aripper, but these became less effective asthe less weathered and harder rocks wereexposed. To keep up output an increasingnumber of more powerful rippers wereintroduced and the material pushed towardsthe loaders. This method of workingproduced a well blended and consistentmaterial and very little direction todifferent locations in the embankment wasnecessary. At the peak of operations therewere Komatsu 455, 355 and 155 and CaterpillarD9 rippers with two attendant Komatsu 155blades. The loading was done by twoCaterpillar 245 face shovels and Caterpillar245 and 235 and Ackerman backacters.

28. Although the material appearedconsistent, regular estimates were made ofthe rock types within the blend because ofthe effect a large imbalance in theproportion of either mudstone or sandstonemay have had on permeability. Theseexaminations revealed some significantvariations but overall the ratio of sandstoneto mudstone was 50:50. The sandstonestrength was measured by point load testswhich showed that over 75% was 'very1 or'extremely'strong with strengths in excessof 100 MN/m .

29. As the quarry developed, a large andextremely hard outcrop of sandstone wasrevealed in the centre of the area.Excavation of this material would have beenvery difficult and would have yieldedmaterial significantly different in characterto the fill that had already been produced.It was therefore decided on both economic and

technical grounds to work around the outcropand extend the quarry to the east.

30. From the quarry workings, samples weretested regularly for specific gravity,moisture content and grading. The fill wasgenerally drier than expected with theoverall average moisture content being only4.4? with the range between 0.6% and 13.4%.After ripping and loading the maximum size ofrocks was rarely greater than 0.5m but nogradings of the full range of particle sizeswas undertaken. From the grading of thefraction below 106mm it was concluded thatthe fill had a slight excess of fines whichwould give maximum stability and strengthwith minimum compressibility.

31. On completion, the embankment fill willbe protected by the upstream membrane, thecrest road and the grassing and drainage ofthe downstream face. This configuration willreduce ingress of air and water thusreducing the rate of chemical degradation bythe oxidation of pyrites.

32. Regular samples from the fillindicated an average sulphur content of0.53%, although one test on fresh mudstonegave a value of 2.46%. Tests carried outusing BS 1377 Test 11(A) showed thefill to be consistently acidic with the pH inthe range between 4.33 and 6.60. X-raydiffraction and X-ray fluorescence testsconfirmed that minerals susceptible tosignificant decay were only present in smallquantities.

Embankment Compaction33. After commencement of the contract, but

prior to embankment construction, furthertrials were carried out to investigate theeffect of differing layer thicknesses,numbers of passes and types of roller. Thetrials were carried out on a 'more weathered1

material (predominantly sandstone) and a'less weathered' material (predominantlymudstone). The trial showed thatthe permeability was at a slightly highervalue than had been assumed, also that alayer of fine material was formed aftercompaction of each layer and the addition ofwater caused this surface to become slurried.

34. Compaction of the rockfill was to amethod specification which required a 450 ramfinished thickness layer to be compacted by 8passes of a roller. Should it be necessary tochange the number of passes or the layerthickness a formula was provided to enableany increase or decrease in cost to becalculated. To achieve the required standardthe contractor provided three smooth drumvibratory rollers, an ABG MAW 173 of 4800kg/m and two Bomag BW 10s of 5350 kg/m whichwere towed by Caterpillar D6 bulldozers.

24

PAPER 3: WILSON AND EVANS

35. Haulage from the quarry was undertakenby a fleet of 12-15 Caterpillar D35v>articulated dump trucks. The fill was tippedabout 3m back from the advancing face of thelayer and bulldozed forward over the edge toassist the distribution of fines within themix. It was noted that the tracks of themachines caused an additional breakdown ofthe rock which further increased the finescontent. Prior to placing a new layer, thesurface was scarified using a multi-tineripper to break up any fcrust1 formed byrolling.

Testing of Embankment Rockfill36. During embankment construction holes,

800mm cube, were excavated in the fill toobtain the density by gravel replacementmethods. Initially the results indicatedlower densities and higher air voids thenmight have been expected and the number ofroller passes was increased from 8 to 12.This produced only about 1% improvementin density and trials were carried out toestablish the point of 'refusal1. Levellingof layers of rock after each pass of theroller showed that very little compactionoccurred after the 8th pass; beyond this thematerial started to breakdown and a Tbowwave* sometimes appeared ahead of the roller.It was concluded that 8 passes was sufficientto achieve a rockfill compacted virtually torefusal.

37. There were 149 test holes excavated,giving a rate of one hole per 6650m.- of fill.The mean bulk density was 2159 kg/m with arange between 2675 and 1820 kg/m giving adry density of 2070 kg/m at an averagemoisture content of 4.4$. Contrary to whatmight have been expected it was found thatthe higher densities occurred with lowermoisture content as is illustrated in Figure3.

38. It was expected that the materialwould be highly weather susceptible but onlyon a few occasions was it necessary to stopplacing during or after periods of heavyrain. However the weather caused materialchurned up by the wheels and tracks ofconstruction plant to become a 'stickyslurry1. It was thought that this couldrepresent an extreme condition such as mayresult if severe degradation of the rockfilloccurred. A shear box test gave particularconfidence with results 0* of 29° (peak) andof 25.5° (residual) for material with amoisture content of 18.5%.

2200-

2100-

2000-

e (means of 10 gravel

i

e

ii

replacement tests)

" ~ 1Iiiii^-0verall mean

-—^1(2070 kg/m3@4 4 %i

^ JNStandardx deviations

2 3 4 5 6 7 8MOISTURE CONTENT (%)

EMBANKMENT ROCKFILL

Fig 3 Dry density/moisture content relationship

39. Over the period of construction routinesamples of the fill were taken for triaxialtests and large shear box tests. These testswere carried out on the fraction of materialbelow 20 mm and below 37.5 mm and thereforeunderestimated the strength of the actualmaterial in the embankment. In general theresults from the triaxial tests approximatedclosely to those obtained in the initialinvestigations but those from the shear boxwere lower by about 5%» It was concludedthat the reason for this was that theparticle size of 37.5mm was too great for the100mm thick specimens resulting inunrepresentative density and compaction.

40. On completion of density measurements,the 800mm cube holes were filled with waterand a falling head permeability test carriedout. The permeability was calculated fromthe soakaway formula given in C I R I Areport No.113, modified to allow for thegravel in the hole and for differinghydraulic gradients in the sides and base ofthe hole. The average permeability obtainedwas l.*5 x 10*" m/s with .the range between 1.8x 10"3 m/s and 2.8 x 10~' m/s. Thus the fillis within the range assumed in the design.In the event of severe damage to themembrane, most of the water will drainthrough the fill to the underdrainage layer.There will be only a modest downstream flowthrough the fill material and the embankmentwill remain stable under these conditions.Collapse Settlement

41. There was some concern that the placingof comparatively dry mudstone within the fillcould result in collapse settlement. Thisoccurs when the point contact betweenindividual particles crush and compress due

25

STATE OF THE ART

to weakening of the rock by wetting. Inaddition dry rockfill has a strengthcontribution from pore water suction whichdisappears when it becomes wet. Toinvestigate the susceptibility of the rock tocollapse compression, tests were carried outwith the Building Research Establishment's lmdiameter oedometer. Samples of 20 mm downfill were heavily compacted and inundatedwhile subject to a vertical stress. Theresults indicated that collapse compressionshould not be a major problem because a highstandard of compaction was being achieved onsite. Also in spite of the low moisturecontent of the fill it was considered thatthe addition of water during placing wouldnot greatly reduce the collapse settlementunless sufficient was added to almost fullysaturate the rockfill.

Sandwaste42. China clay sand waste was used

downstream of the concrete cut off tominimise differential settlement under theasphalt concrete membrane. This was the samematerial that had been used for theconstruction of Colliford Dam (ref. 1) whereit was found that settlements were minimal.The properties of the sand were confirmed bya test in the BRE lm diameter oedometer whereit was found to be one third thecompressibility of the rockfill. (See Fig. 1)As this test was at a density below thatactually achieved in the field, thecompressibility should be an underestimate ofits true value. ~

43. A total of about 40,000m of sand wasdelivered by road from Parsons Park Pit andcompacted in 250mm layers using a towedvibrating roller of 3400 kg/m width.Initially 12 passes were used, based on planttrials, but this was increased to 16 passesafter the densities showed a highvariability. The routine sand replacement ^tests gave a final mean density of 1842 kg/mat an average moisture content of 6.7%. Thiswas about 2% less than the mean of 1880kg/m achieved at Colliford, probably becauseof different compaction plant operating incomparatively small areas.

Finite Element Analysis44. A number of finite element analyses

were carried out by Professor P R Vaughan toconfirm the likely behaviour of the embankmentat the interfaces between the gallery,sandwaste and rockfill. The propertiesassumed for the materials were those obtainedfrom the site investigations and used in thedesign calculations. In practice thoseactually achieved during construction weregenerally found to be superior. The analyses

confirmed that there were no zones of highstress which could lead to local and perhapsprogressive failure and also that stresslevels would be reduced following impounding.The incorporation of the sand waste zone hasa major influence in reducing differentialmovements adjacent to the cut off. In factthe displacements between the embankment filland the sandwaste were shown to be greaterthan at the interface with the cut-offstructure.

45. The analyses also provided predictionsof the likely settlement at the locations ofthe various instruments installed in theembankment. During construction thesettlement measured in the embankmentcorrelated closely with the finite elementanalysis, (see Fig 4) During a period ofheavy rain in February/March 1989 someevidence of collapse compression was notedconsistent with water percolating downwardsthrough the fill. The observations haveshown the compressibility of the fill to bebetter than the oedometer tests on a denselycompacted fill (see Fig. 1)

46. Horizontal movements indicating aspread of about 80 mm were recorded withinthe embankment by the end of construction.This was about half that predicted by thefinite element analysis. Surveys of thedownstream face of the dam have confirmed themovements indicated by deformationinstrumentation installed in the dam.

Finite element analysis predictionAs measured from vertical settlement gauge

Reservoir level 120mAOD(14-3-90)

constructionelevation

131mA00(2810-86

120 160 200 240 280 320

SEITLEMENT (mm)

Fig 4 Settlements on embankment centre-line

26

PAPER 3: WILSON AND EVANS

REFERENCE Consultants and Technical Advisors who have1. JOHNSTON T A an<* EVANS J D Colliford all played their part. The authors wouldDam sand waste embankment and asphaltic like to acknowledge the permission of Southconcrete membrane, Proceedings of the West Water Services Limited to publish thisInst.Civil Engineers 1985, Vol 78, Aug.689- paper and are appreciative of the help given709. by their colleagues.

AcknowledgementsThe credit for the successful completion ofthe project must go to Contractors,

27

4. A perspective of the art of the embankment dam in South WestAsia

W. J. CARLYLE, Binnie and Partners, Redhill, UK

SynopsisBinnie and Partners with their associated firms in Malaysia and Hong Kong have been responsible fora large number of embankment dams in range of height up to 70 m formed of residual soil on weatheredrock foundation. Some 20 such dams have been completed in the last decade and have been put intoservice with no significant problem arising either during the construction or the first fillingphase. As a result of this rather satisfactory situation, little or nothing has been published onthe techniques that have evolved into a standard form of practice. At present there are 5 majordams in the final stages of detail design or under construction in Malaysia alone, the highest ofthese being the Upper Muar dam in Negeri Sembilan with a height of 55 m. The particular aspect ofinterest to dam engineers is the behaviour of the residual soils when placed in humid tropicalconditions, usually at moisture contents well in excess of optimum. The difficulty of prediction ofconstruction pore pressures because of the nature of the tropically weathered residual soil and thepossibility that post-placement chemical bonding of the clay particles may effect the prototype porepressure response. The evolution of the design has resulted from the observation of wellinstrumented dams over the decade. Working practices relating to the inclusion of drainageblankets, the allowable rate of increase in height of the fill have evolved on the same basis. Thefoundation conditions for most of these dams consist of rock weathered to various states and theappropriate treatment of the foundation to avoid excavation of large amounts of in-place residualsoil is fundamental to the economic development of the dam sites. This matter will be dealt withfully in the paper, being a combination of cutoff construction and grouting the different zones ofweathered rock in place.

INTRODUCTION1. The residual soul derived from tropical

weathering of the rock mass in situ provides asource of excellent material for theconstruction of embankment dams. By far themost common soil is the product of weatheredgranite but the extrusive volcanics such as therhyolites found in Hong Kong, and sedimentarywill also yield soil albeit more variable inmechanical properties.

2. The process of weathering and theerosion of the product by rapidly downcuttingimmature streams results in sites having freshrock exposed in the stream bed and depths ofweathering increasing in the abutments. Such aprofile is ideal for the earth embankment damwhich can accommodate the variabledeformability of the foundation.3. The strength characteristics of the soil

permit the use of conventional central coredams having relatively economical externalslopes, typically on 3.5 upstream and 1 on 3downstream. Thus the dam will have only 45%more fill volume than a typical rockfill damwith central core and the unit cost of the fillwill be significantly less.

4. Other favourable features are theproximity of the borrow pits to the dam, theusual absence of deep alluvium at such sitesand the proximity of fresh rock in or near thethalweg. This permits economical arrangementsof draw-off and diversion culverts or tunnels.

5. There is usually a range of weatheringwhich provides by selection suitable fill forthe zones of the earth dam; more clayey soilsfor the core and more sandy soils for the outershells. The sand and gravel materials requiredfor filters and drainage are not normallyavailable from river bed alluvium and arecommonly obtained by crushing quarried rock onsite.

CLASSIFICATION OF WEATHERING6. There are various schemes of weathering

classification which describe the weathering ofthe rock mass in relation to the deteriorationof the rock crystals and the effect ofweathering on the discontinuities andboundaries of the rock mass. The state ofweathering is usually subdivided in grades fromfresh rock (Grade I) to fully decomposed rocktermed residual soil (Grade VI) and followloosely upon the work of Moy (1955).

7. Table 1 is drawn from BS 5930 1981 withan additional description column drawn fromBinnie and Partners' standard specificationwhich gives some added practical aids toidentification.8. In large cuts in weathered granite it

may be possible to identify the whole range ofweathering from the surface downwards to freshrock. It may be equally common to go straightfrom Grade V to Grade I in certain

The embankment dam. Thomas Telford, London, 1991 29

STATE OF THE ART

Table 1. Seals of weathering grades of rock i

Description 1 Specification

Fresh No visible sign of rock materialweathering; perhaps slight dis-coloration on major discontinuitysurfaces

Slightly Discoloration indicates weatheringweathered of rock material and discontinuity

surfaces. All the rock materialmay be discoloured by weathering

Moderately Less than half of the rock materialweathered is decomposed or disintegrated to

a soil. Fresh or discoloured rockis present either as a continuousframework or as corestones

Highly More than half of the rock materialweathered is decomposed or disintegrated to

a soil. Fresh or discoloured rockis present either as a discontinuousframework or as corestones

Completely All rock material is decomposedweathered and/or disintegrated to soil. The

original mass structure is stilllargely intact

Residual All rock material is converted tosoil soil. The mass structure and

material fabric are destroyed.There is a large change in volume,but the soil has not beensignificantly transported

Strong, hard rock; cannot be scoredwith a hand knife; and felsparsfresh

II Strength approaches that of freshrock; sharp edges cannot be paredbut may be scored, with difficulty,with a hand knife; ground mass oforiginally dark coloured volcanicrocks bleached to reddish brown orpale grey and plagioclase felsparsmay be decomposed

III Pieces the size of NX drill corecannot be broken by hand; sharpedges of fine grained rocks can bepared and the rock scored with ahand knife; bleached to pale brownor white and felspars considerablydecomposed

IV Pieces the size of NX drill corecan be broken and crumbled in thehands, fine grained rocks caneasily be pared with a hand knife;does not disintegrate when soakedin water and can sometimes berecovered as cores by carefuldiamond drilling, using watercirculation, but is often lost;bleached or stained yellowishreddish-brown

V Still has a recognisable rocktexture; the original felspars aredecomposed to clay minerals; itwill disintegrate when immersed inwater and often cannot be recoveredas cores by normal diamond drillingmethods. Pale or yellowish reddish-brown clayey-silt

VI Completely decomposed by weatheringin situ and with rock texturedestroyed; red or yellow clayey-silt containing over 20Z clay-sizefraction; the sand content willdepend on the original proportionof quartz-phenocrysts

+ Residual soil is used throughout the paper as a term to describe the excavated f i l l from other lower grades.

Table 2. Soil Properties

SMng Nun

Plover CoveMain dam

Upper Peirce

tangat

Semeniyih

Core

Shoulders

Foundation

CoreShoulders

Core

Shoulders

Shoulders

Core(colluvium)ShouldersFoundation

Core(placed)Shoulders(dumped)

CoreShoulders

Core

Shoulders

Foundation

Average values

Proctor compactionDry density I Optim

g/cm3 m.c. X

1.64

1.77

1.751.81

1.72

1.73

1.75

1.75

21

15.5

1615

18.5

18

17

16

fn-situDry density «.c.

g/cni3 X

1.58

1.72

1.59

1.681.73

1.64

1.67

1.71

1.70

1.84

1.841.76-1.84

1.70

1.78

18.5 )16.5 )

21

17.5

19

18.5

18.5

16.3

Proportion offine materiel

c 0 . 0 0 2 M *34X

< 0 .002M*6-48X

< 0 .002M*13-28%

< 0.002m*2-22X

< 0 . 0 0 2 M *

22%< 0.002iiM

11X

< 0.002m*13X

< 0.002*1*7X

c 0 . 0 6 M *

28-73Xc 0 . 0 6 M *

7-61X

Design values

21.5(21.5)9.6

(0-40.7)9.6

(4.8-28.7)

24.0

0(26.3)0

(30.6)

12(24)14(19)19

1414-19

30*(30*-35')

35*(32'-43«)

35*(30'-39*)

31*

30'(33')35'(36')

31'(32* )31"(33«)29*

31"28*-30*

29

32.5

32.5

k

1 x 10 y

e.1 x 10° )

c. 1 K 10'7 )

«2/ve«r

7.8(2.2-8.9)22.3

(11-670)

22.3(22.3)

(C.130)

84

22.3

4.2

1146

11

11

11X

71

51

3833

57

43

65

43

31

31

62

HX

37

28

2*22

32

12

34

33

22

22

33

* Values in parentheses are measurement averages on fill samples or field tests

30

PAPER4:CARLYLE

circumstances and to find all grades ofweathering juxtaposed for no obvious reason.

SOIL IN PLACEFoundations for embankment dams9. Where granitic rocks form the foundation

of the dam the weathering profile will usuallyshow considerable variations in depth with thegreatest depth of weathering on the abutmentswhere the embankment loading is least.

10. In very general terms undisturbedresidual soil has relatively high shearstrength and low compressibility, thus thefoundations are not generally a problem for thedesign but careful exploration is needed todetermine variations which may lead to locallyhigher construction pore pressures in thefoundation and differential settlement of theembankment•

11. Foundation exploration by drillholes canbe difficult. It is vital to distinguishbetween weathered rock in place and materialwhich has been displaced by slope instability.Displace granite core boulders up to 7.8 mdiameter have been recorded at one dam site.Bedrock in place must be proved by good corerecovery in Grade I to II rock to a depth of atleast 10 m from the surface of the rock.

12. The foundation exploration in theweathered zone is made using shell and augersoft ground boring equipment and obtaining U100samples. It is difficult to make satisfactorycore recover even with triple tube core barrelsin Grave V and Grade VI weathered granite. Theboring is stopped when penetration cannot beachieved without chiselling and the hole iscontinued with rotary diamond drillingequipment in H or N size. Even then it isdifficult to achieve a satisfactory coringresult in Grade III rock, although this is thezone which is critical for foundationpermeability.

13. The use of geophysical exploration toinfill detail between trial boreholes anddrillholes has been disappointing in itsapplication to weathered rock profiles,probably because the seismic velocity are toosevere.

Soil properties14. The range of mechanical properties for

residual soils varies between the Grade V to VIclay residual soil and the Grade III sandysoil. In the former case there will have beensignificant loss of the constituents of theparent rock and the residue will be arelatively dense clay soil with a plasticityindex in excess of 20. In the case of theGrade III soil there will be a relativelyhigher strength and stiffness, consolidationbeing rapid due to the residual fissuringprobably resulting from the joint system in theoriginal rock mass. In either case theresidual soil behaves as a normallyconsolidated clay but chemical bonding mayexplain strengths generally higher than that ofnormally consolidated sandy silty clays.

15. It is difficult to give typical valuesof foundation strengths or settlement

characteristics because of the variedweathering profile and the difficulty ofobtaining representative samples of adequatesize to represent the mass behaviour of thesoil.

16. Typical soul properties are given inTable 2. Shear strength parameters adopted fordesign are typically 0' = 32° C = 12 kPa. Forconsolidation calculations cv values varybetween 20 and 150 m2/year and a usual valueadopted as a conservative design value is 50m2/year. Measured settlements of thefoundation under dams up to 60 m in height havebeen under 200 mm. The foundationconsolidation is relatively rapid. A checkwould normally be run with C = 0 for deepsurfaces.

17. Construction pore pressures are notnormally a problem. For design purposes ruvalues of 0.2 are normally assumed but theseare rarely reached in practice. Most dams showlittle or no post construction foundationsettlement.

Reduction of permeability - Cutoff or grouting18. As recently as 1965 the 57 m high Lower

Shing Mun dam, Hong Kong, was provided with a 3row grout curtain in fresh rock in the riverbed and a 2 m wide concrete filled cutofftrench in the residual soil in the abutments.This cutoff trench which extended to a depth of36 m at the abutments provided an idealpositive cutoff, particularly in the transitionzone between the more clay weathered rock(Grade IV) and the fresh rock (Grade I to II).It would now be regarded as prohibitivelyexpensive and would not normally be employed.

19. Although the permeability of mostresidual soil is low, of the order of k 10"5

m/s, there are zones of higher permeabilityparticularly in the transition between theGrade III and the fresh rock Grade I or II. Asthe foundation is often a weak point in thedam, it must be treated to improve itsresistance to erosion and reduce the seepagelosses by a combination of grouting and properdrainage provisions in the downstream area ofthe dam.

20. Thus there are two problems in thedesign of an appropriate grouting scheme. Totreat the fresh rock by reduction of thefracture permeability requiring a difficulttechnique. No scheme for foundation groutingcan be considered complete at the design stage,the detail of the geological anomalies will berevealed as the drilling and groutingprogresses and the site work must therefore becontrolled by those responsible for the design.

21. The most critical area of any foundationgrouting is the shallow contact areaimmediately below the core. At shallow depththe rock fissures are open, the possibility todevelop high grouting pressures is least and oncompletion the hydraulic gradient is at itsgreatest. Where the foundation of the corearea is predominantly rock and is highlyfractured, it is desirable to place a concretegrout cap or sprayed concrete layer on the corearea. This performs two essential functions in

31

STATE OF THE ART

addition to providing a good working surfacefor the blanket grouting operations:

(i) By preventing surface leakage itallows the development of reasonablegrouting pressures in the near-surface fissures. Even a sprayedconcrete layer 6 cm thick can resistthe uplift developed by the grout inrelatively small fissures,

(ii) It isolates the erodible base of thecore from the high velocity flowwhich may develop in isolatedungrouted near-surface fissures.

22. Where the rock surface is massive butwith occasional fractures or fault zones thesurface can be treated by slush grouting andoccasional dental concrete. Where the corearea is on weathered rock (Grade III) it isimpossible to treat the surface in this manner.Nonetheless the core contact area will be acritical zone for permeability. In such casesa minimum of 6 m of core fill is placed overthe surface before grout injection of the corecontact area and the fresh rock below.

23. The treatment of this shallow zonerequires multiple rows, usually 3 or 5 rows ofgrout holes with a nominal spacing of 3 m. Inthe abutments where the height of dam isreduced and the core width narrowed groutingcan be reduced to 3 rows and the extensionabove into the abutments beyond the end of thedam is usually continued as a single rowcurtain.

24. Most rock foundations are treated bycement grouting at depth. This usually takesthe form of a 3 row grout curtain. The depthof the grout curtain should be determined bythe rock conditions at depth. It is oftenrequired that the Lugeon values should be lessthan 5 and that the profile of rockpermeability should be determined by coredinvestigation holes taken to a depth at leastequal to the height of the dam. If Lugeonvalues are as low as 5 or below, there may belittle or no grout take and if the dam can bedesigned to eliminate risk of damage due toseepage there may be a case for eliminating thegrout curtain altogether. However a groutcurtain is an effective tool for detailedexamination of rock foundation and it shouldnot be omitted without good reason. Normalcement grout can be employed to grout rockfissures down to a width of less the 0.5 mm,the lower limit is usually taken to be about0.2 mm

25. In formulating an appropriate grout forsealing fine fissures a considerable differenceof opinion has arisen amongst various experts.On the one hand there is the need to ensurepenetration of fine fissures. This is governedby formulating a grout with sufficiently lowviscosity but which will not have too high apercentage of bleed, ie separation of the waterfrom the grout in suspension. On the otherhand, the grout must be stable in the fissureswith good durability.

26. Cement grout is a suspension of finelyground particles of cement in water. In finefractures the cement particles will tend not to

enter the fracture and the larger particleswill act as filters for the smaller particles*It has been suggested that the fracture sizeshould be three times the diameter of themaximum particle size in the cement grout. Formost cements this would lead to a minimumfracture width of around 0.3 mm.

27. It is important to formulate a cementgrout which is adequately durable. Clearly,thick grouts with a water/cement ratio of 2:1or 3:1 will have a low proportion of bleedwater and will form strong stable grouts.

28. In the past, very much thinner groutswere used down to water/cement ratios of 12:1or even thinner for good penetration of finefissures but in certain cases that have beenquoted recently, grout curtains have seriouslydeteriorated and it is now commonly acceptedthat thin grouts may not be stable.

29. To minimise the percentage of bleedwater from the grout it has been the practicefor some years, particularly in Europe, to addbentonite to the mix. This in its turn hassome disadvantages in increasing the viscosityof the mix and thereby reducing its penetrationof the fissures.

30. It would now seem that good practice isto go for a relatively thick grout mix lessthat 5:1 water/cement ratio and in certaincases to add bentonite equivalent to 2.5% ofthe cement by weight to improve the ability ofthe mix to keep the cement in suspension.

31. The grouting of the Grade III weatheredrock beneath the layer of core fill in place isusually performed as stage grouting with apacker set in an 85 mm PVC pipe taken throughthe fill and grouted into the surface of theweathered rock. Stages of this process areillustrated in the accompanying diagram (Figure1.).

32. The grout mix is progressively thickenedfrom 5:1 to 1:1 injecting 2-3 batches at eachstep and then proceeding to the next thickermix until refusal is achieved. Batch volumesare given in Table 3 - Grout mixes.

33. In conclusion, it can be stated that aproperly constructed grout curtain must bebased on the determination of insitupermeability and the detailed probing of thegeological situation as the grout injectionprogramme proceeds. The depth of the groutcurtain should be determined by thepermeability conditions rather than a fixeddepth. The grout curtain should be composed ofat least three rows of grout holes to create asignificant thickness of grout curtain.

Grouting pressure34. Grout pressures are controlled to

prevent hydraulic fracture or uplift. In thegrouting of the overburden or weathered rock,pressure at the injection gauge is controlledat 6 kN/m2 per metre depth. Taking account ofthe grout column to the stage being injected,this ensures that the total grout pressure doesnot exceed 85% of the nominal overburdenpressure at this point. In certaincircumstances the minor principal stress can beless than this pressure and it is possible that

32

Ta

ble

3

. G

rou

t M

ixes

(a)

Cor

e fi

ll

TC

on

tact

gro

uti

ng

(b)

Dec

omp

osed

an

d

Sta

ge

„w

eath

ered

ro

ck

gro

uti

ng

(o)

Roc

k

85m

m

PVC

• p

ip«

N,

Sta

ge

an

do

r p

ack

er

50n

n P

VC

0,

pip

e ^

pa

cker

gr

ou

tin

gx

Fre

eh

rock

1.

Dr

ill

and

in

sta

ll

85 a

m p

.r.c

. p

ipes

in

(*

)

dr

ill

into

(b

) an

d g

rou

t co

nta

ct

of

(a)

an

d (

b).

2»

Dr

ill

to

rock

an

d s

tag

e g

rou

t (b

).

3.

Dr

ill

and

in

sta

ll

50 a

n p

.v.c

. p

ipe

in (

b).

4.

Dr

ill

into

(c

) in

}m

len

g-t

h,

sta

g*

and

pa

cker

gro

uti

ng

in

3<a

len

gth

to

fr

esh

ro

ck,

or

pa

cker

gro

uti

ng

in

JB

len

gth

fr

om

fres

h

rock

w

hen

ro

ckis

g

oo

d.

'/h

ere

core

fi

ll

in d

ire

ct

con

tact

w

ith

ro

ckp

roce

du

re

(2)

and

(3

) w

ere

om

itte

d.

Wat

er/C

emen

tra

tio

by w

eigh

t

8:1 6:1

4:1

3:1

2:1

1:1

2:1

2:1

2:1

2:1

2:1

• 5%

Ben

ton

ite

•so

luti

on in

hydr

ated

fo

rm(L

itre

)

12-5

12.5

25 25 25 25 25 25 25 25 25

Vat

er(L

itre

)

187.

5

137.

5

175

125 75 25 75 75 75 75 75

Cem

ent

(Kg)

25 25 50 50 50 50 50 50 50 50 50

San

d

X

(Kg)

- - - - 2*5

5.0 10 25 50

Rem

ark

One

to

three

ba

tch

es

of

ea

ch

m

ix

to

be

use

d

for

inje

cti

on

dep

en

din

g

on

gro

ut

tak

e,

sta

rti

ng

wit

h

the

thin

ne

st

mix

u

nti

l

re

fusa

l.

*

FIG

UR

E

1

- SO

WM

AU

T

OF

DR

ILL

ING

A

ND

CS

OO

TIH

G

• D

ry

weig

ht

of

be

nto

nit

e e

qu

iva

len

t to

2.5

f<

by w

eig

ht

of

cem

ent,

x

san

d

5

to

100&

by

weig

ht

of

cem

en

t.

+

Defi

ned

a

s g

ro

ut

tak

e o

f Ie

98

tha

n

1 ll

tre

/min

/^m

le

ng

th.

5 m TO

©I ! !

®

LfiL

Drill 114a« 0

grout hole

through the compacted core

fill material by dry

auge.rlng or wash boring or

tricone.

Roc

k

fczL

.fi"

J-JL

I

Inst

all

85m

m ^

PVC

pip

ean

d sl

eeve

gr

out

the

grou

t h

ole.

Was

h th

roug

h th

e gr

out

in th

e PV

C p

ipe

and

dri

ll

300m

m l

engt

h be

lovr

cor©

bas

e u

sin

g NX

dri

ll b

it.

DIA

CR

AM

MA

TIC

PR

ESEN

TATI

ON

O

F D

RIL

LIN

G

AND

GR

OU

TIN

G.

Dri

ll

an

d g

TO

ut

bla

nk

et

zone

st

ag

e b

yst

ag* (

repea

t ©

an

d©

Red

rill

th

roug

h th

e bl

anke

tgr

outi

ng z

one

to t

he

rock

-con

tact

.

Gro

utin

g o

f th

e co

re/

fou

nd

atio

n co

nta

ct

surf

ace.

Dri

ll

thro

ugh

blan

ket

zon

e.C

rout

bla

nket

zo

ne b

yse

ttin

g pa

cker

in

sid

e7V

C p

ipt

m

Install 50mm 0 PVC and sleeve grouted

from the surface to the blanket zone/

rock contact.

Drill into rock using AJC (4801m)

bit.

Perform rock grouting.

Stage by stage rock grouting.

Repeat @

and @

.

2 m TO O

STATE OF THE ART

vertically orientated hydraulic fracture canoccur.

35. For fissure grouting in rock a gaugepressure of 14 kN/m2 per metre depth isregarded as an acceptable pressure (See Figure1).

Stability in excavation36. Weathered rock including Grade V and VI

residual soil, is remarkably stable in theshort term and this permits the use of verysteep slopes in temporary cuts required for theconstruction of the permanent works. It is notunusual to have permanent cut slopes with abatter of 1 on 1 to a height of 30 m whenforming tunnel portals and spillways but thesemust be properly drained and provided with aprotective membrane to prevent surface erosionand softening by infiltration of rain. Otherstabilising measures include horizontaldrainage pipes and rock anchors.

37. Such considerations are particularlyimportant in designing the diversion works forresidual soil dams. The depth of weatheringoften inhibits the use of tunnelling for streamdiversion because there is a high cost in thepreparation of the tunnel portals fortunnelling lengths usually less than 200 m.Because the fresh rock is often at or nearstream bed level in the centre of the dam, itis usually more economical to adopt a cut andcover culvert construction for the diversionworks. This can usually be placed on rockGrade I or II and the risk of differentialsettlement or seepage past the culvert can belimited.

38. Such an arrangement was made for theSeletar dam in Singapore when the invert of thetunnel was designed to accommodate 150 mm ofdifferential settlement. In the event, thesettlement was a fraction of this.

FILL MATERIALRemoulded soil characteristics

39. The residual souls derived from tropicalweathering are close to ideal for damconstruction.

40. The fill is usually easy to dig fromwithin the reservoir area near to the dam site,by shovel or scraper, although occasional ribsof rock and isolated core boulders may beencountered. After transportation to theembankment there is usually no difficulty indischarging the fill from the earth mover andspreading it in layers of the requiredthickness. Equipment to break up lumps is notusually necessary although harrowing to aidmoisture content control is effective.

41. The natural moisture content cangenerally be adjusted without difficulty andafter delay due to rainfall, filling canusually be recommenced without delay sincemoisture loss is rapid in dry weather.

42. Customary characteristics of the fillare outlined below.

Grading43. The particle size distribution of a

residual soil is typically very broad, usuallyhaving from a trace to 30% clay and oftenincluding all intermediate fractions up toabout medium gravel sizes. The percentage ofmaterial finer than sand (i.e. clay and silt)may range from less than 10% to about 80%depending on degree of decomposion. (Figure2).

44. Since the weathering process progressesfrom the ground surfaces downward, the finergrained (more decomposed) material tends to befound as an upper layer lying over the coarsergrained soil, though ribs of more resistantrock or zones where weathering has progressedlaterally from faulted or fissured rock may

2 -Ul ^ i

£

4

t

/

s7

/

ooo

1

.. p

y>

Q

_ > -

10

«

>T» BRITISH STANOARO SIEVE SIZES MM

? *t

S j

.:tr* c

<

•i

S

*

\

y

A

0 - *

7 -

A

/

/

-

/I/

a x }

i

i

? ? a

>

f

i0 *O0MM

J CLAY f i r * j M»d>om [

SI ITUtdlum f Co*fM Fln«SANO ±GRAVEL

OeS4.ES BOULDERS

Semenyih Dam: Core envelope P-Q; Shoulder envelope R-S.

Figure 2. Particle size distribution

36

PAPER4:CARLYLE

result in marked variation in depth ofdecomposition•

Atterberg limits45. The wide range of soil gradings to be

found in decomposed rock useful for embankmentfill are reflected in the results of liquid andplastic limit tests.

46. Liquid limits of between 40% and 60% arecommon, with plastic limits of between 30% and40%. However, liquid limits may range on anysite from below 30% to over 100% and plasticlimits may be between "non-plastic" and 40%.

Natural moisture contents47. At most sites we have found that the

natural moisture content of the fill in theborrow areas has been close to, or a littlebelow the optimum moisture and some wateringhas been necessary to raise the moisturecontent to the specified values, includingreplacing the moisture lost in evaporationduring excavation, transporting, placing andspreading it on the embankment. At some sitesit has had to be acknowledged that much of theavailable fill will be too wet to allow 95% ofthe Proctor maximum density to be reachedunless the moisture content is reduced. Ithas, from study of the Proctor compaction testcurves, been possible to specify a range ofmoisture contents and to require the fill to becompacted to say 98% of the Proctor testdensity at the placement moisture content. Bythis means satisfactory fills have beencompacted into place under practicablecontrols.

48. It is observed that residual soils arecapable of being handled and compacted often athigher moisture contents with respect to theProctor optimum than specified. This can betolerated if the stability due to the porepressure rise is not critical and thesubsequent consolidation settlement istolerable.

Strength49. The shear strengths of residual soils

are generally appreciably higher than theirparticle size distribution and in particulartheir clay content would suggest to the soilmechanics engineer familiar with soil depositsof transported materials in temperatelatitudes.

50. As an example, recent tests on aresidual soil derived from basaltic lavasshowed it to have 33% to 40% clay sizedparticles and 40% to 44% silt sized particles.Liquid and plastic limits were about 62% and25% respectively. Given these figures aneffective strength angle of shearing resistance0' in say the range of 20° to 25° might beguessed for the material based on experienceonly of sedimentary soil deposits originatingfrom unweathered rock.

51. However high quality consolidateddrained triaxial compression tests with porepressure measurement on remoulded specimensshowed its shear strength to be 0' = 34° andC = 0 in effective stress terms.

52. The values for 0' for tropicallyweathered granitic and volcanic residual soilsgenerally lie in the range 27" to 37°. Forpreliminary design of embankments, before testresults can be obtained it is usuallyconservative to adopt shear strengthparameters: 0' = 30°, C = 0. There willusually be sufficient of the coarser higherstrength fill int he dam cross-section tocounter the effects on stability of any finermaterial of lower strength (below 0' = 30)material that is used and which is likely to bedirected into a core zone.

Compressibility and permeability53. The remoulded residual souls in the

compacted state are generally of lowcompressibility. Consolidation is rapid andpost construction settlement is very low.

54. Compacted core fills of clayey residualsoil can usually be expected to have apermeability of the order of 1 x 10~9 m/s,which is perhaps 2 orders lower than isnecessary to keep seepage losses through coresof conventional design to insignificant rates.

55. Even in Plover Cove dam, Hong Kong,where decomposed rock was placed in the coreunderwater without compaction, Guildford andChan (1969) report than large-scale boreholetests indicated average permeability (k value)of 1 x 10"6 m/s which corresponded to thedesign value.

DAM DESIGNGeneral

56. In general, the embankment dam designhas developed to allow the most economical useof the locally available residual soildistributed so that the more clayey material isavailable for the core and the more coarsegrained residual soil is as a sandy clay isused in the shoulders of the dam.

57. The proportions of core width to heightare generally selected to have a ratio of 0.5.This allows a core having low hydraulicgradients and a good base width on the corecontact.

58. The dissipation of pore pressure duringconstruction is sufficient to maintain asatisfactory margin of safety against slidefailures or excessive deformation duringconstruct ion.

59. With the generally high value of cv ofthe order of 100 mVyear or more, dissipationof construction pore pressure is relativelyrapid. End of construction ru (U/ h) willnormally be less than 0.6 (Ayer Itarn dam) butmore often is significantly lower than this,0.27 (Lower Shing Mun dam). Some residual porepressure in the core has an advantage inensuring that the core can respond in adeformable manner to differential stresses inthe embankment during construction.

60. The outer shell of the dam is commonlyprovided with drainage blankets for rapiddissipation of construction pore pressure.Spacing of blankets relates to theconsolidation rate of the fill and thepermitted rate of raising the embankment. A

37

STATE OF THE ART

spacing of 10 m is a usual value.61. The pore pressure response in the core

of a typical residual soil dam, is affected bythe relative stiffness of the outer shell andthe resultant redistribution of the verticalstresses to less than the theoreticaloverburden stress.

Drainage provisions62. The drainage provisions are

complementary to the good treatment of the damfoundations. Foundation relief wells may incertain circumstances be required and a basedrainage mattress is essential and must beproperly detailed to take care of foundationirregularities. Inadequate construction of thebase drain, chimney drains or filters has beena significant cause of dam malfunction in thepast.

63. It is now generally accepted that thebest all-round filter is medium sand gradedwith a D50 size 2 mm and a D10 size 0.2 mm.Such a filter will be appropriate for any typeof fine grained soil core. It should be ofadequate width to allow for any conceivabledisplacement. A least width of 3 m will giveand adequate margin for intermixing with theadjacent zones during compaction withoutimpairing the integrity of the filter. Thecontrol of the filter during construction is asimportant as that of the core for the securityof the dam. The base drainage is controlled bya filtered drain to take care of foundationseepage.

Surface protection - Upstream slope64. The wind strength is low in Peninsula

Malaysia although very high in Hong Kong.65. Riprap is the usual and most economical

form of protection for embankment dams andthere have been few reports of any damage tothe revetments. Maximum wind and wavepredictions are based on Saville's methodtaking into account the fetch increaseappropriate to long narrow valleys.

66. The significant wave height for UpperMuar dam is 0.5 m with small fetch of < 1 km.

67. In Hong Kong the higher wind speeds andlarger reservoirs have resulted in maximum waveheights of up to 2.3 m.

68. Riprap design is rarely a criticalfactor because of the abundance of granite rockof suitable dimension. Minimum requirementsfollow CIRIA report No 61 which providescriteria for the grading of riprap related tothe slope angle and the significant waveheight. For "no damage" criteria in the 50year event, the D50 riprap must be equal orgreater than Hs the significant wave heightderived from the 50 year short period windevent. For Upper Muar riprap was 0.8 m thick

on a graded rock base,kg.

'50 of riprap was 150

Surface protection - Downstream slope69. In the high tropical rainfall areas

protection of the exposed downstream slope ofresidual soil is important. Frequent bermdrains and substantial mitre drains are

38

essential as is careful establishment andmaintenance of grass slope cover.

CONSTRUCTION CONTROL70. To achieve the objective of producing

the most economical embankment from the localmaterials, the specification for constructioncontrol is written to minimise restriction ofthe flow of embankment material from the borrowpits, and to accommodate both the variation inweathering profile within the borrow pit andthe normal tropical humid climate, particularlyin Peninsula Malaysia.

Selection71. The first consideration is to obtain

sufficient core material from the Grade V toGrade VI weathering profile. Such materialgenerally has 25% at least passing 0.06 mm(silt size) and should generally have aplasticity index in excess of 20%. Suchmaterial exists in the borrow pit at or aboutProctor optimum moisture content. It isgenerally necessary to increase the watercontent of the fill to between 2% and 5% aboveProctor optimum.

72. The core fill is generally placed andcompacted to produce a layer thickness of 200mm and compacted to achieve 95% Proctor maximumdry density or 100% Proctor dry density at theplacement moisture content.

73. The material for the shoulders of thedam can be from the remaining weatheredmaterial until progressive increase in thehardness of the rock prevents its use as anormal fill material.

74. The moisture content control in theshoulder fill is generally -2% to +3% with samedensity control as the core.

75. The rate of placement is controlled at ageneral rate of 3 m in a four week period butthe placement of fill must be modified oradditional drainage blankets added if the porepressure response exceeds design values. Awide range of compaction equipment is suitablefor residual soil fill.

76. It is normal to specify the adequacy ofthe plant by the following criteria:

(i) The total output of compactionequipment at the Site shall be atleast 4000 cubic metres per day ofmaterial measured when compacted tothe specified in situ dry density.

(ii) One heavy-duty disc harrow shall beprovided.

(iii) Rubber tyred rollers shall each weigha minimum of 400 kN and shall eachhave a minimum of 4 wheels locatedabreast. They shall be capable ofexerting average ground surfacepressures on a plane firm surfacegreater than 550 kN per square metre.They shall be designed so each wheelwill carry equal loads whentraversing uneven ground.

(iv) Vibrating rollers shall have a staticweight of not less than 100 kN and ofnot less than 33 kN per metre widthof the roller. The frequency of

PAPER4:CARLYLE

vibration of the rollers shall beadjustable from 20 to 30 cycles persecond so that compaction can becarried out at or above the criticalfrequency of each type of fillmaterial.

77. Adequate watering equipment is essentialto maintain fill moisture content and it isnormal to specify at least three 4 m3 capacityself propelled rubber tyred bowsers having acapacity of 330 1/minute.

MONITORING THE BEHAVIOUR OF THE DAM78. When planning the instrumentation of

embankment dam it is important to limit thearray of instruments to those necessary toobserve the construction behaviour of theembankment and its long term performance underreservoir full conditions. The instrumentsthat are installed should be designed to giveanswers to specific questions that are posed bythe designers and owners of the dam.

79. During the construction of residual soilembankment dams it is essential to measure porepressure developed in the core and the shoulderand, if it is compressible, the foundation. Itis our practice to use hydraulic piezometerswith low air entry tips for this function.

80. No attempt will be made in this shortpaper to define standards or types ofinstallation.

81. To monitor the consolidation of the fillduring placement and the deformation of thefoundation, a limited number of verticalsettlement gauge tubes should be installed.Although many of our dams have employed theUSBR type of mechanical cross-arm settlementgauge, we now routinely use the magneticvariety. It has to be accepted that anyvertical settlement gauge tube will cause aconsiderable problem when compacting fill roundthe tube. This cannot be avoided but greatcare must be taken to supervise theinstallation.

82. Finally, as construction proceedssurface monuments must be installed to monitorfinal settlement and deformation. It isimportant to install the control stationsbefore the commencement of construction andinstall monuments as soon as berms aredeveloped on the downstream slope particularly.

83. Tha piezometric readout equipment mustbe installed in a robust, preferably buried,chamber under controlled temperature and

humidity conditions beyond the limits of theembankment. If the whole installation isproperly installed and serviced during theconstruction period when there are highlyqualified staff available on site, then it isreasonable to expect a very long life for suchinstruments so that the long term behaviour ofthe embankment can be monitored.

84. Dam safety requirements demand thatthere are periodic, preferably annual,surveillance reports prepared for the dam fromthe date of completion and first impounding,particularly for the first five years on anannual basis.

85. An instrumentation system should bedesigned to ensure that the readings can easilybe made by technically qualified staff and aremeaningful to the owner's operation andmaintenance staff.

REFERENCE1. BEAVAN et al. Construction Pore Pressuresin Clay Cores of Dams.2. BRITISH STANDARD 5930: 1981.3. CARLYLE W. J. 1965. Skek Pik Dam. ProcInst Civ Engrs, London, Vol 30, p557.4. CARLYLE W. J. 1988. Wave damage toupstream slope protection of reservoirs in theUK. Reservoir Renovation 88. BNCOLD Sept.5. CHUNG Ho. 1965. Lower Shing Mun main dam.Eng Soc of Hong Kong.6. GUILDFORD C. M. & CHAN H. C. 1969. Somesoils aspects of the Plover Cove Marine Dam.Proc 7th Int Conf on Soil Mechanics &Foundations Engineering, Mexico, p291.7. LITTLE A. L. 1967. The use of tropically-weathered soils in the construction of earthdams. Proc Third Asian Regional Conf on SoilMechanics and Foundation Engineering, p35.8. TEO CHOK BOO. 1987. Klang Valley WaterSupply. Sungai Semenyih Dam. Journal AssocCons Engrs, Malaysia. Second quarter, p8.9. TEO CHOK BOO. 1985. Drilling & Groutingof Semenyih Dam. Prof Exam Document, Inst CivEngrs London (unpublished).10. VAIL A. J. The Ayer Itam Dam 1963.Journal of the Singapore-Malaya Joint Group,Vol III, March, p41.11. VAIL A. J. 1970. The Seletar Scheme,Singapore. Water & Water Engineering, Nov.12. WESLEY L. D. 1990. Influence of Structureand Composition of Residual Soils. Journal ofGeotechnical Engineering, Vol 116, No 4. ASCE.

39

6. Instrumentation of the Mrica dam

I. W. BURTON and P. A. S. FERGUSON, Halcrow, Swindon, UK

SynopsisMrica Main Dam is situated on the Serayu River in Central Java where it provides storage for a180MW power station. The crest of the Main Dam is 800m long and its maximum height above thefoundation is 110m. The types of instrument, their installation and operation, for monitoringthe behaviour of the main dam, appurtenant structures and excavated slopes are described,together with procedures for collecting the readings and processing the data. Some aspects ofthe behaviour of the main dam during impounding are discussed and compared with predictedbehaviour.

INTRODUCTION1. Mrica Hydro-Electric Power Project was

commissioned by the Indonesian electricityauthority, Perusahaan Umum Listrik Negara(PLN). The dam is situated on the SerayuRiver in Central Java at an altitude of about200m where it provides storage for a 180MWpower station.

2. The geology of the area is characterisedby Recent and Pleistocene deposits overlyingTertiary basement volcanics, the lattercomprising andesitic and basaltic lavas andtuff agglomerates. Overlying the weatheredcrust of the basement volcanics, thePleistocene deposits consist of tuffsandstones and agglomerates up to 80m thickand are of lacustrine and alluvial origin.These Pleistocene deposits have weathered toclays up to 35m thick. Recent deposition hasyielded a mantle up to llm thick of gravelsand boulders under a clay surface.3. The river at the damsite runs in a deep,

steep-sided gorge and the crest of the main(rockfill) dam, at el. 235m, is 800m long ata maximum height above the foundation of 110m.Subsidiary high level embankment dams, up to15m high and some 6km long, raise thereservoir rim along each flank.

4. The power house sits in a deepexcavation immediately downstream of the maindam on the left bank, while the spillwayoccupies a similar excavation on the rightbank (see Fig 1).

Power house

Diversion tunneloutlets

Access road

Diversionintakes

Fig. 1. General Layout

5. Halloysitic, residual clay for the highlevel embankment dams and the core of the maindam was obtained from borrow areas within theimpounded area and immediately downstream ofthe dams. The core of the main dam issupported upstream and downstream by rockfillshells formed mainly from quarried andesitebut including a zone of "rockfill soil" in thedownstream shell to utilise selected materialarising from the excavations for the spillwayand power house. Transitional filtersupstream and downstream of the core wereformed using crushed rock from the sameandesite quarry.

The embankment dam. Thomas Telforcl, London, 1991 41

STATE OF THE ART

6. The main dam foundation in the rivergorge was excavated to sound rock over thewidth of the core. On the higher, flatterground above the river gorge the dam isfounded on clay and constructed to flatterslopes. A grout curtain some 40m deepprovides a cut-off beneath the dam foundationacross the river gorge section (see Fig 2).

Residual

Rockfii]

Rockfil!cofferdam

Upstream filter

TWL + 23>x ^ x y + 2

' • j f c - . T ; - - - ^ ^ * ^

a • a

a < i ^ • •

Standpioes ingroutec tunnel.

Downstream filter

35 ^ /KOCKTIII

v / Rcckfii. soil

LEGEND.• Piezometer.• "otal pressure guage.• Settlement cell.A Surface monument.\ :nc!>ometer casing.

Fig. 2. Cross sections of main dam showingdam construction and arrangements ofinstruments

7. The Project was designed by SWECO AB ofStockholm in association with Engineering andPower Development Consultants of Sidcup andWiratman & Associates of Jakarta.

8. Construction took place during 1983 and1989 with reservoir impounding during May toSeptember 1988. The contractors were aconsortium of Shanska, Balfour Beatty andAsea.

9 . Construction supervision was carried outby PLN with advice and assistance from theSupervising Engineers, Sir William Halcrow andPartners in association with Beca WorleyInternational of Auckland and P.T. Citaconasof Jakarta.

PURPOSE OF INSTRUMENTATION10. Instrumentation for the main dam wasinstalled for the following three reasons:

- during construction, to provideinformation that could be used in checkingthe validity of some of the designparameters adopted

-during impounding of the reservoir, tocheck actual behaviour against thatpredicted and to give warning of anyserious deviations

- during operation, to monitor the long termperformance of the dam and confirm itssatisfactory functioning

11. The frequency and form of measurementand reporting were tailored to suit therequirements of each of the phases describedabove and procedures and computer softwaredeveloped accordingly. Local staff weretrained during the construction phase inmeasuring the instruments and reducing thedata for regular prompt reporting, using thecomputer programs developed. These staff havebeen transferred to the operations branch ofPLN and will continue to be responsible forthe long term monitoring of the behaviour ofthe Main Dam and appurtenant structures.

LAYOUT OF INSTRUMENTATION12. Instruments were installed at foursections along the main dam and includedpiezometers, total pressure gauges, settlementcells and inclinometers. Surface monumentsand accelerpgraphs were provided along thecrest of the dam. A typical arrangement ofinstruments is shown in Fig 2.13. Tubes from the remote reading hydraulicand pneumatic instruments were led to terminalpanels in three instrument houses on thedownstream slope of the dam where readingswere measured manually using portable digitalreadout units.14. Pneumatic piezometers, standpipes andsurvey monuments were also installed along thehigh level dams, beneath the foundation of thespillway and in the cutslopes around thespillway and power station.15. A trial embankment, which wasincorporated into the left bank high leveldam, included pneumatic and hydraulicpiezometers and readings on these continuedthroughout the construction and impoundingperiod.

INSTALLATION AND PERFORMANCE OF INSTRUMENTS16. The instruments were installed under thesupervision of the Supplier's technician whovisited site for each episode of installationin the main dam and just before impoundingbegan. Damage to equipment by constructionactivity and instrument failures did occur andexperience in this is described in thefollowing paragraphs. It confirms the need toinstall redundant instruments wheremeasurements are crucial and provide anadequate stock of spare instruments.17. Instruments in the main dam wereinstalled in clusters, as shown in Fig 2.Each cluster consisted of piezometers,pressure gauges and settlement cells so as topermit calculation of effective stresses andru values.

PAPER 6: BURTON AND FERGUSON

Table 1. Summary of instruments installed

Location

Main Dam

High Level Dams

TrialEmbankment

Spillway

Power StationArea

Pneumatic

Piezometer

72

18

20

59

Hydraulic

Piezometer

11

-

5

-

Standpipe

Piezometer

-

1

-

25

28

Total

Pressure

34

-

-

-

Settlement

Cell

11

-

-

-

Inclino-

meter

5

-

-

-

Surface

Monument

48

8

-

35

23

TOTAL 169 16 34 11

Piezometers18. Piezometers in the main dam foundation

were installed in boreholes with their tipssurrounded by sand sealed on top with abentonite cement plug. Those in the core wereplaced in intimate contact with the clay byforming a hole for the tip with a dummy probe.Three types of piezometers were used:19 . Pneumatic piezometers were chosen by the

Designer to be the principal instrument formeasuring pore pressures, mainly for reasonsof ease and convenience of use. Theseinstruments were not particularly sturdy,especially in the dam foundation where thefailure rate exceeded 50%. Of the 72pneumatic piezometers installed in the maindam, only 46 were still in operation atproject completion. Possible reasons for thisrelatively high failure rate includeconstriction in the tubes due to largesettlements; silt in the tubes followingrepair after construction damage; poor readingtechnique on instruments with long tubinglengths; and equipment malfunction.20. Hydraulic piezometers were installed inthe main dam embankment as back-up instrumentsfor the pneumatic piezometers. Theseinstruments have been robust and reliable. Ofthe 11 hydraulic piezometers installed in themain dam, 8 were still in operation at thecompletion of the project.21. Standpipe iezometers consisting of

Casagrande-type porous pots and standpipeswere installed in boreholes to monitor longerterm changes in the groundwater levels,particularly around the tunnels, along thedownstream toe of the main and high level damsand on cutslopes in the downstream area.

Total Pressure Gauges22. Pressure gauges of the oil filled,

hydraulic-pneumatic type were installed intrenches in the main dam embankment. Theywere subsequently backfilled and compacted byhand.23. At the lower levels of the dam, the

gauges were placed in rosettes of three, onevertical, one inclined at 45°, and onehorizontal, the faces of each being parallelto the dam axis. At the highest level ofinstrumentation (el 210m) on the instrumentedsections at each abutment, a fourth gauge,perpendicular to the dam exis, was added inorder to measure longitudinal stresses.24. Of the 34 pressure gauges installed inthe main dam, 30 were still in operation atthe end of the period of maintenance andlittle difficulty has been experienced withthem. Unfortunately, two of the gauges nolonger functioning are those that wereinstalled perpendicular to the axis of the damat each abutment.Settlement Cells25. Pneumatic settlement cells wereinstalled in a manner similar to that forpressure gauges. Good results, consistentwith those from the plate magnets on theinclinometers, were obtained until the readoutunit malfunctioned, unfortunately during theimpounding period, and had to be returned toUK for repair. Substituting the read-out unitfor the pneumatic piezometers gave wildlyerratic readings and a consequent gap in dataduring the long repair time.

43

STATE OF THE ART

Inclinometers26. Inclinometers were installed at two

sections of the main dam, in the clay core(tw) , in the upstream filter zone (one) and inthe downstream shell (two). That in theupstream filter was inclined parallel to theupstream face of the core.27. Three-metre lengths of plasticinclinometer casing were built into theembankment and extended as filling proceededusing telescopic couplings to allowcompression along the casing. A biaxialtorpedo was supplied and readings were storedin a cartridge readout unit and transferreddirectly into the site computer.28. Ring magnets placed over the casings and

embedded in the fill material enabledsettlements to be measured using a reed switchprobe.29. Difficulties were experienced in

maintaining the alignment and orientation ofthe inclinometers during construction. Thetubes in the core and upstream filter zonewere worst affected and when the embankmentlevel had reached el 187m the torpedo wasunable to pass below approximately el 170m and142m respectively in these tubes. A toleranceof 50m radius on installed alignment wasimposed and measurement of tube twist bysurvey was introduced.30. In April 1988 a twistmeter was broughtto site and a series of twist readings weretaken on each inclinometer, but the resultsshowed poor repeatability and it was concludedthat the twistmeter measurements could not berelied upon for any of the inclinometers.Consequently the inclinometer deflectionsthemselves are of very limited use. Worstaffected is the inclined inclinometer where anincorrect figure for the orientation of thetube will clearly produce very large errors inapparent displacements.31. With the exception of one tube (blocked

at about el 200m) all the casings coul betraversed by the probe for the plate magnetsand useful information on settlements withinthe embankment was obtained.32. The reason for the excessive twist in

the inclinometer casing was not establishedbut it is possible that storage arrangementson site were a contributory factor.

Seismic Accelerographs33. Seven strong motion accelerographs were

supplied but completion of the dam crest afterthe start of impounding meant that those onthe main dam could not be installed beforeimpounding started. Several problems with theinstruments themselves, such as tracealginment, time code generator and batteryfailure, were also encountered. As a resultvery little information was obtained duringthe impounding period although two tremors didoccur.34. By project completion all of the seven

units were installed and working correctly.Mains power supply was introduced wherepossible and a spare battery system where not.Surface Monuments35. Surface monuments were surveyed using

EDM tachymetric equipment with readings storedin a data logger which was downloaded directlyinto the site computer. In the absence ofreliable data from the inclinometers thesemeasurements provided the only information onlateral movements of the dam. The onlyproblem experienced was occasional damage dueto construction activity and, as with theaccelerographs, completion of the dam crestafter the start of impounding meant that someof the monuments were not available during theearly part of the impounding period.

Seepage Measurement36. Seepage measurement at the toe of the

main dam in the river valley was by means ofa vee-notch weir fitted with an automaticlevel recorder.37. The readings proved to be rather

variable, being directly influenced byrainfall and, apparently, by the flow fromrelief wells under the downstream shell.However, the maximum flow with the reservoirfull has not exceeded 30 1/min which isconsidered very satisfactory.38. Seepage into the grouting tunnel and

construction adits was collected in sumps andpumped out. This seepage was measured byflowmeters on the discharge pipelines.

DATA COLLECTION AND PROCESSING39. Data collection was carried out by PLN

laboratory and survey departments under theguidance of the Supervising Engineer. Withthe large number of instruments involved, thisrequired a full time team. The technicians inthis team participated in the supervision ofinstallation and were given instruction onreading techniques by the Supplier'stechnician during his visits to site.

44

PAPER 6: BURTON AND FERGUSON

40. Each day's results were delivered to thecomputer section for processing and plottingusing the site computers. Initially, themicro-computers were two Commodore machinesand the necessary programs were written inBasic by the Supervising Engineer's staff.Later, an IBM-compatible micro was acquiredand a spreadsheet program was used on it forsome of the instruments. The computer sectionwas required to work a double shift to keeppace with the supply of data.

Monitoring during construction41. All instrument readings and survey

measurements were made weekly while seepagemeasurements were made daily. Results wereplotted and reported weekly with copiesdistributed to the PLN Site Project Manager,Site Supervising Engineer and the Designer'ssite representative. Copies were forwarded tothe Supervising Engineer's head officemonthly.42. Readings from piezometers in the mainand high level dams were plotted against timeto show piezometric elevation. Embankmentlevel and ruvalues (pore pressure ratios) werealso shown on the same plot.43 . Total pressures (in kPa) measured by the

pressure cells in each cluster were plottedagainst time. Also shown on these graphs werepore pressure, ru value and theoreticaloverburden pressure for the cluster.44. Settlements measured by the settlementcells were plotted against time. Thosemeasured by the ring magnets on inclinometerswere plotted against elevation, settlements onfour consecutive dates being plotted on eachgraph.45. Data from survey of surface monuments

were presented as changes in easting, northingand elevation and plotted against time.

Monitoring during impounding46. During impounding, weekly instrumentreadings and survey measurements continued asbefore, but selected piezometers (in the maindam foundation, power station area, tunnelsand beneath the spillway) were read dailyalong with the seepage measurements. Dailyinspections were also made by PLN engineersand inspectors, with advice from thesupervising Engineer. The inspectionsencompassed the downstream toe and slopes ofthe main and high level dams , cuts lopes aroundthe power station and spillway, and the valleysides immediately downstream of the main dam.

47. The daily inspections were recorded onforms, purpose made by the SupervisingEngineer, and the daily measurements weresummarised on a single sheet that wasdistributed daily. The measurements were alsoplotted daily by hand and these graphs wereavailable for inspection in the PLNinstrumentation office.48. The weekly computerised plots were

modified as necessary to provide informationappropriate to what was required duringimpounding. Plotting of ru values wasdiscontinued and reservoir level was includedin place of embankment elevation. A new datumwas used for measurement of settlement.Typical plots of piezometrikc level are shownon Fig 3.

••PPH1H5C —RESERVOIR LEVEL |

250Ptezometrcc Elevation (m)

200-

100-

Fig. 3. Main Dam - Typical PiezometricLevels at elevation 145m

Longterm monitoring49. Maximum intervals between readings ofinstruments have been recommended by thedesigner for checking the longterm performanceof the dam and appurtenant structures.Measurements of flow and seepage arestipulated to be at weekly intervals and ofsurface monuments at 3-monthly intervals. Allother instruments are to be read monthly.50. The measurements are to be plotted in

the same manner as was adopted duringimpounding. As stated earlier, theinstrumentation team has been transferred tothe operational branch of PLN at Mrica andwill continue to carry out the measurementsand plot the results. "Guard" values for eachof the instruments have been supplied by theDesigner and a review by Engineering Staffwill ensue automatically if they are exceeded.

45

STATE OF THE ART

BEHAVIOUR OF MAIN DAM51. The instrumentation has confirmed that

the dam has generally behaved withinacceptable limits and as predicted by theDesigner. Only particular aspects will bedescribed herein, notably the deformation ofthe dam and the stresses in the core followingimpounding.52. The Designer carried out finite element

analyses of the main dam as a part of thedesign process, with the purpose of assessingstresses in the core and the risk of hydraulicfracture. Since it had not been possible totest the shoulder materials ultimately used toderive deformation moduli, ranges of valueswere assumed and three cases were analysed.The analyses showed that the design wasreasonable and gave predicted stresses in thecore and deformations in the shoulders that itwas hoped would be verified by the instrumentsin the course of impounding.53. This aim was partly confounded by thefailure of the inclinometers to providereliable results of actual deformations andthe malfunctioning of the settlement cellreadout unit during impounding. It wasparticularly disappointing that deformationsin the upstream shoulder were not caputredduring impounding. However, settlements fromthe pneumatic settlement cells and the platemagnets on the inclinometers were reliablymeasured during construction and were used toreview predicted settlements thereafter, whichwere partly verified by measurement of thesurface monuments; and the date from thesurviving piezometer, pressure gauge andsettlement cell clusters were used to comparemeasured stresses in the core with thosepredicted. The results of these lattercomparisons may be summarised as follows,together with the aid of Figs 4 and 5:

- - Vertu:oL Stress — Reservoir Pressure - - Horizontal StrejIncUned Stress - - Pore Pressure

.St ress (KPa)

Jan Feb ' Mor Apr May ' Jun ' Jul Auq Sep Oct Nov D

I - - P o r e Pressure --• IndLned Stress

--Horizontd Stress —Reservoir Pressure

Jan ' Feb ' Mar ' Apr ' May ' Jun ' Jul ' Aug ' Sep ' Oct Nov De

Fig. 5. Stress at Elevation 210m onSection IV

total principal stresses measured in thecore show good agreement with thosecalculated for one of the finite elementcases analysed

effective shear stresses mobilised arehigher and predicted, possibly partlybecause the finite element analyses tendedto underestimate pore pressure

there is no rick of hydraulic fracture inthe lower levels of the core as,additionally, the high pore pressures therewill tend to rule this out, as noted bySherrard (Ref 1)

at high levels in the core and particularlyat the abutments (where there areinstrumented sections 2 and 4), there areindications of low effective stresses witha consequent risk of hydraulic fracture.It is unfortunate that most of thecasualties amongst the pressure gaugesoccurred at high elevation and at theabutments.

REFERENCES1. SHERARD J.L. Hydraulic fracturing inembankment dams. Journal of GeotechnicalEngineering, ASCE, October 1986, vol 112,no 10, 905-927.

Fig. 4. Stresses at elevation 145m onSection 1

46

Discussion

J.A. CHARLES (B.R.E.)In a session concerned with the State of theArt of the Embankment Dam it may be of interestto have a brief account of a recent NATOAdvanced Study Institute (ASI) on "Advances inRockfill Structures". The ASI was held at, andorganised by, LNEC Lisbon. There were 70participants and lectures came from Australia,Brazil, France, Germany, Japan, Portugal, Spainand U.K. Although other types of rockfillstructure were considered, the ASI wasprincipally concerned with rockfill dams. Someof the issues which emerged during the twoweeks of the ASI are as follows:

Col compression when saturated forthe first time rockfill may reduce involume. Poorly compacted rockfill with alow moisture content is most vulnerable.It can be questioned whether heavycompaction is sufficient to prevent theoccurrence of this phenomenon or whetherin many cases it is also necessary towater the rockfill. Collapse compressionis probably the least well understoodfacet of rockfill behaviour, yet in somesituations it is the most important.

Index properties of parent material Itwould he nelpful if these properties(e.g. particle strength) could be used topredict fill behaviour. They are simpleand cheap to determine whereas tests onthe fill require large expensive samples.However it is questionable whether thereis much correlation between indexproperties and fill behaviour; thedensity of the fill tends to have adominating influence on fill performanceand obscures the effect of factors suchas particle strength.

Theoretical soil models These continueto attract the attention of soilanalysts. It does not appear that anincrease in sophistication leads to anybetter prediction of constructionalbehaviour of a rockfill dam that thesimple BRE one dimensional compressionapproach. Reservoir impounding is muchmore complex; there would be collapsecompression in the upstream fill and thisis difficult to model.

4. Importance of details For rockfill damswith thin membrane (concrete, asphalticor geomembrane) details of joints etc.can be of major significance. Forexample with concrete faced rockfilldams, problems are often associated withthe perimetric joint.

5. Lew grade rockfill There is increasinginterest in the use of low graderockfill. This raises the question ofwhen does rockfill become so low gradethat it is effectively earthfill.

6. Seismic behaviour There is also growinginterest in seismic behaviour, althoughheavily compacted rockfill seems tobehave very well. The lecture on thistopic was illustrated with manyphotographs of liquefaction failures, butthey were all of sands, none were ofrockfill.

It is intended that the lectures will beproduced as a book in the NATO ASI series.

P. TEDD (B.R.E.)Referring to Paper 3 and in response to theChairman's request to have discussion oninstrumentation, I would like to describe anovel instrumentation system that was installedat Roadford dam. An important feature of thedesign and construction of the dam is thejunction of the asphaltic membrane and theconcrete cut-off structure at the upstream toeof the dam. Relatively incompressiblesandwaste has been placed close to the cut-offstructure (see Fig. 1) to limit the movement ofthe membrane close to the cut-off. TheBuilding Research Establishment (BRE) wasrequested by Babtie Shaw and Morton, consultingengineers to South West Water to design andinstall an instrumentation system that wouldmeasure the deflection of the upstreamasphaltic membrane close to the concretecut-off structure. The instrumentation madeuse of the E-L (electro-level) system developedat BRE during the last 20 years (ref. 1).

The E-L is a gravity sensing electrolytictransducer that provides an output voltageproportional to the tilt angle. It consists ofa small glass sealed tube, partially filledwith an electrolytic fluid and with metalelectrodes in contact with the electrolyte.Those used at Roadford were about 30mm long and6ram diameter. Their range is approximately +

The embankment dam. Thomas Telford, London, 1991 47

STATE OF THE ART

Extent of E-L instrumentation

Inspection gallery andcut-off structure

18 box sections housing E-Ls8 at 0.5m long10 at 1.0m long

Drainage layer

Sandwaste

Fig. 1 Section of Roadford dam showing extentand location of electro-levelinstrumentation

48

DISCUSSION

3 (+ 52mm displacement over a metre length)and the long term accuracy is better than 50seconds of arc (0.25mm displacement over ametre length).

Electro-levels were fixed to the inside of aseries of stainless steel box sections, 50mm by100mm (see Fig. 2 and 3) joined together by asliding pin joint. Up to 18 sections werejoined together extending some 14m up the slopefrom the cut-off structure (see Fig. 1). Theywere placed on the upstream slope immediatelybeneath the asphaltic membrane with the lowestbox section being fixed to the cut-offstructure. Cables from the instruments weretaken into the inspection gallery and read witha portable readout. The E-L measures theangular rotation of each box section in thevertical plane such that displacementsperpendicular to the line of the box sectionand therefore the upstream slope of the dam canbe obtained. Summation of the displacementsfrom each box section gives a displacementprofile of the membrane relative to the cut-offstructure.

The E-L system installed at Roadford survivedthe placing and rolling of the asphalticmembrane, and has provided satisfactory resultsduring the partial impounding of the reservoir.Full details of the system installed atRoadford will be presented in Ref • 2.

References

1. Pri ce, G., and Wardle, I.F. TheBuilding Research EstablishmentElectro-level monitoring system.Technical Consultancy paper. BuildingResearch Establishment, Garston.

2. Tedd, P., Price, G., Wilson, A.C. andEvans, J.D. Use of the BRE electro-levelsystem to measure deflections of theupstream asphaltic membrane of Roadforddam. 3rd In. Symp. on Field Measurementsin Geomechanics. Oslo, Norway. 1991.

Fig. 3 Completed length of box sectionsbedded in place

Fig. 2 Box sections bedded in place duringinstallation

49

STATE OF THE ART

S.G. TOMBS (Binnie & Partners)The Authors of Paper 3 referred to "collapsesettlement". A dramatic example of 'collapsesettlement' of rockfill occurred at CogswellDam in California. The dam was constructedfrom granite rockfill placed in 7.6m lifts.The rockfill was mainly Class A specified as "awell graded mixture, 40% of which to vary inweight from quarry chips to 1000 pounds, 30%from 1000 to 3000 pounds and the remaining 30%from 3000 to 14000 pounds and the mixture wasnot to contain more than 3% of its total weightin quarry dust and the maximum dimension of anypiece was not to exceed 3 times its minimumdimension. All rock was to be sound, hard,durable, angular quarried rock weighing notless than 160 pounds per cubic foot and to beunaffected by air and moisture and of suchtoughness as to withstand dumping without undueshattering or breakdown and to have a minimumcompressive strength of 5000 pounds per squareinch." * ' During placing a layer of muchfiner rockfill was produced on the surface ofeach lift by the action of caterpillar tracks.This layer was scarified prior to placing thenext lift. No sluicing was carried out becauseof a lack of water. Part of the dam wascompleted to the full height of 85m by thebeginning of December 1933. On the last day ofDecember a major storm swept in from thePacific Ocean and by noon on 1st January 1934about 380mm of rain had fallen on the dam. Thecrest at station 4 + 80 K ' settled by 1.8movernight and by 13 June 1934 the totalsettlement recorded (since the beginning ofDecember 1933) had increased to 4.1m.Subsequently large quantities of water wereadded to the rockfill and by 8 August 1935total settlement had increased to 5.3m.

In the late 1960s we carried out a number oftests on samples of rockfill in oedometers andcompared the results with tests on samples ofchert. The rockfill materials were Siluriancleaved mudstone similar to that subsequentlyused for the shoulders of Brianne Dam andgranitic gneiss from Venemo Dam. Samples wereprepared dry and compacted to a dense state.Loading was applied and the initial settlementwas measured. The samples were then flooded byallowing water to enter through porous media atthe base of the oedometer. In the case of theSilurian cleaved mudstone and the graniticgneiss the settlement increased immediately byabout 95% and 13% respectively of the initialsettlement. The settlement of the wet Siluriancleaved mudstone was therefore nearly doublethat of the dry material for the same appliedloading. The contact points in an angularrockfill will deform and fracture until theproduct of contact area and rock strength equalthe contact load. Wetting of the rock surfacereduces the strength of the rock by a varyingamount depending on rock type. Therefore afterwetting further fracturing takes place at thecontact faces so that the product of wetstrength and new contact area again equalsapplied load. Wetting of the chert producedlittle additional settlement. However wherechert particles were crushed under load failurewas by shattering producing razor sharp needleswhich could not be handled.

50

It should be noted that when the Siluriancleaved mudstone and granitic gneiss sampleswere removed from the oedometer and dried andthe tests repeated exactly the same resultswere obtained showing that the dry strength isonly temporarily reduced while the material iswet. This phenomenon of temporary reduction ofstrength in certain environments applies toother engineering materials and the propertyhas been used to advantage in some machiningoperations.

In order to assess the minimum amount of waterneeded when placing and compacting rockfill, soas to achieve the maximum immediate settlement,oedometer tests may be carried out on a seriesof appropriately compacted samples atincreasing moisture contents until noadditional settlement occurs at greatermoisture content.

References:

1. Baumann, P., Rockfill dams: Cogswell andSan Gabriel dams. Trans ASCE Vol 1251960 Part 2 p29ff.

2. Spielman, J.V., Discussion. Trans ASCEVol 125 1960 Part 2 p60.

J.H. McKENNAI would like first of all to praise Dr. Chinand his co-authors for publishing a mostvaluable paper. Case histories are always ofinterest but one where the pre-constructiondesign assumptions were found to needsignificant alteration during construction, isof particular interest.

Secondly, the paper concerns the Coulee Eastdam. However, 10 km to the west is the CouleeWest dam with the same height (28m). There thefoundations "performed generally as expected".It appears from the text that the foundationconditions at both dam sites were the sameexcept that the West dam foundation clays werethinner being up to 35m thick compared with theEast dam foundation clays which were up to 60mthick. I would ask the authors to comment onthe reasons why they got it right for the Westdam foundations but did not get it right forthe East dam foundations.

Thirdly, the pre-construction pore pressurecalculations using "linear elastic" stresschanges gave pore pressures up to 35% lowerthan those measured in the field. However, thepore pressure calculations using "finiteelement" calculated stress changes, done during(or after?) construction gave pore pressureswhich are in remarkable agreement with thefield measurements. Would the authors confirmtherefore that Henkel's and Law & Bozozuk'sequations give the correct predictions as longas the stress changes are correctly predicted?Would they use the same methods for their nextdam on similar foundations?

Fourthly, the parameters used in the finiteelement analyses were derived from laboratorytest data supplemented with data from thetechnical literature and that to get

DISCUSSION

deformation agreement "some fine-timing of themodel was necessary". It would be of interestto the profession if the authors wouldtabulate;

1. the original parameters derived from thelaboratory data and

2. the final parameters that were used

so that it will be possible to assess theextent to which the pre-construction laboratorydata had to be corrected in order to obtain theobserved field performance.i

Finally, it is my experience from myinvolvement on international dam review boardsthat a finite element or finite differencemethod analysis is now standard design practiceon all major dams to be built on thick softclay foundations. Deformation and porepressure predictions are essential elements ofmodern dam design methods.

References

McKenna J.M. (1989). Properties of corematerials, the downstream filter and design.Clay Barriers for Embankment Dams, 63-72.Institution of Civil Engineers. ThomasTelford, London.

T.A. JOHNSTON (Babtie Shaw & Norton)The integration of the design and constructionprocesses, as described in Paper 1, hasoccurred on dams in the United Kingdom.Kielder Dam is one example of the cross sectionbeing modified in response to construction porepressures. It is interesting to learn that inCanada this approach is called "theobservational method".

A potential draw-back of the observationalmethod is that the rapid construction possibleon modern earth-moving projects is at odds withthe time constraints imposed by the need toread instruments, collate results, assess theirsignificance and initiate changes on site. Howdid Dr. Chin and his colleagues resolve thepotential conflict between the designer's needfor time to observe construction and thebuilder's wish to construct the dam as quicklyas possible.

The effective stress method was used in thedesign of the dam. Could Dr. Chin provide someinformation on the factors of safety adopted inthe initial design and for the modifiedcross-section. It would be particularlyinteresting to know the calculated factor ofsafety at the time the designer decided toincrease the size of the toe berms.

References

Paper 8575 (a)

Paper 8703 (b)

The Kielder Headworks - ProcICE Part 1, 1982, 72, May.Geotechnical Aspects of theKielder Dam - Proc ICE Part 1,1983, 74, Nov

A. CAMPBELL (Laboratory Manager, SABCCN)

Comments on "Instrumentation of MRICA Dam"Paper 6by Burton and Ferguson :Clause 16 G I Technical only visited to

commission instrument houses,i.e. 3 visits plus one otherbefore impounding. SABCONLaboratory installed all theinstruments, prior to thosevisits.

Clause 19 Silt in tubes is consideredhighly unlikely. All breaks inline were reconnected afterremoving approx. 3 metres oftubing on either side of break.Lines were checked beforejointing by flushing line withNitrogen Gas.

Breakages/malfunctions mostprobably caused by long tubinglengths, and differentialsettlement between core, filterand rockfill. Also, tubing forPneumatic Piezometers was only3mm ID.

See; "Field Instrumentation -Accuracy performance, Automationand Procurement" by S.D. Wilsonfrom International Symposium onField Measurement inGeoraechanics. Zurich September 5- 8 1983.

Clause 25 Readout units were unreliable,not really suitable for climateand initially were of digitaltype. At least 1 No. unit forpiezometers was replaced withordinary Bourdon Gauge type.

Clause 29-32 Pipe was stored horizontally onfloor of container (don't thinkit was a contributory factor).Extrusion of Aluminium casing maycause a spiral of 1 degree/3mlength. Extruded and machinegrooved plastic casing (as usedat MRICA), has been known tospiral 18 degree in a 24m length.Exposure to hot sunlight can alsocause spiral to develop onoccasion. Up to 3m of pipe (whennew connection made) would besticking out above fill andbecomes impractical to shade.

There is often the discussion ondurability properties comparingaluminium relative to plasticcasing and we believe plastic wasadopted for this reason.However, from a contractor'sviewpoint we consider that abetter alignment can bemaintained in utilising analuminium casing.

Clause 39 SABCON had the only laboratory onsite, which was freely availablefor use by PLN.

51

7. The safety of tailings dams and lagoons in Britain

A. D. PENMAN, Harpenden, UK, and J. A. CHARLES, BRE, Watford, UK

Tailings arise from many mining, quarrying and industrial processes. The need for tailings damsand lagoons has grown with increased output of tailings and restriction of discharge intorivers. There are now a large number of tailings lagoons in Britain, with over 1500 associatedwith mines and quarries. Although the hazards posed by tailings dams can be similar to thosefrom dams which impound water, tailings dams and lagoons are not subject to reservoir safetylegislation. Many tailings dams come within the scope of the Mines and Quarries (Tips) Act of1969, but not all do so. There have been a number of incidents affecting safety and the hazardposed by the sudden release of tailings is examined in the context of the safety of Britishtailings dams.

INTRODUCTION1. Large quantities of fine grained waste

materials arise in many industrial, mining anddredging activities. The term "tailings"derives from their production at the tail-end ofprocessing plants as waste. Because tailingsoriginate in a wet process, it is usuallycheapest to transport them from the planthydraulically, allowing them to flow by gravityin open channels and flumes, or to pump themthrough pipelines. The most common form ofstorage is in lagoons retained by embankments.The cheapest form of construction material forthe embankments is usually the tailingsthemselves.

2. When volumes of tailings were not large,they were discharged into streams and rivers ordirectly into the sea. Small amounts of wastefrom the chemical industry and dredgings fromcanals have been stored in lagoons since the1920's. The main need for tailings lagoons hasarisen with the increased output of tailings andrestriction of discharge into rivers resultingfrom the various Clean Water Acts passed duringthe 1950's. Information is presented in Table 1about major producers of tailings in Britain,but it should be recognised that there are manyother processes and industries which producesimilar types of waste materials.

NUMBERS AND SIZES OF TAILINGS DAMS3. There is no central register which

includes all British tailings dams and thereforethere is no accurate knowledge of the totalnumber of tailings dams and very limitedinformation about sizes, shapes, design,foundation conditions, construction methods andoperation of lagoons. In a preliminary surveyof the safety of tailings dams and lagoons inBritain (ref.4), three approaches were adoptedto obtain more information.4. Inspectorate of Mines and Quarries. Under

the Mines and Quarries (Tips) Act 1969 (ref. 5)and the subsequent Regulations, 1971 (ref. 6),notification of all tips connected with mines

The embankment dam. Thomas Telford, London, 1991

Table 1. Major producers of tailings in UK

Source

Coal mining

Thermalpowerstations

China claymining

Sand andgravel pits

Salt basedchemicalindustry

Dredging

Material

Fine discardfrom washery

Pulverisedfuel ash

Micaceousresidues andother fines

Silt and claywaste fromscreeningplantHot liquidsand sludges

Clay, siltand fine sand

Examples

Lagoons retained bydams of coarsediscard; Taylor,1984 (ref. 1)Gale Common lagoon,154 ha, 51 m high;Haws et al, 1990(ref. 2)Kernick Dam, 90 mhigh when complete;Illsley et al,1976(ref. 3)Low bundssurrounding workedout areas

Lagoons retained bydams up to 17 mhigh of limestabilised boilerash etcLagoons adjacent toManchester ShipCanal

and quarries must be made to the inspectorate,including tailings dams and their lagoons. Thisinformation has been used by the inspectorate tocompile a National Tips Register which was firstcompleted in 1975. The report of the ChiefInspector of Mines and Quarries for 1975 refersto this and gives the following information. Atthe end of the year, there were 2244 tips ofsolid refuse and 1478 tailings lagoons. By theend of 1976, the number of tips had decreased to2136, but the number of lagoons had increased to1547. The decrease in number of tips wasattributed mainly to the merging of groups oftips into single, large tips for inspecting and

53

TAILINGS DAMS

reporting purposes. More recent figures,released in July 1987, have shown that there arenov almost the same number of lagoons as in1976, ie 1550 made up of 770 active classifiedlagoons, 479 active unclassified, 215 closedclassified and 86 closed unclassified lagoons.The definition of "classified" is given later.An approach to HM Principal Inspector of Minesand Quarries in March 1987 confirmed that whilehe could identify those tips classed as"liquid", he was unable to help with heights andslopes, etc.5. British Section of International

Commission on Large Dams. When the WorldRegister of Tailings Dams (ref. 7) waspublished, only 7 tailings dams were recordedfor the whole of Britain. A request was made toall members of British Section of theInternational Commission on Large Dams forinformation on tailings dams for which they haddetails which they could release, to provideadditions to the British entry in the WorldRegister. This request brought in a further 10examples, most of which were already knownthrough published literature.6. Planning Authorities. A request for

information was made to 72 Planning Authorities.It was hoped that since all major lagoonsrequire planning permission, these authoritieswould be able to supply details of tailing dams.Replies have shown that Planning Authorities areunable to give details of height and shape oftailings dams, but many have been able to locatethe positions of lagoons and it is clear thatthere are many lagoons in addition to thosecoming within the control of the Mines andQuarries (Tips) Act 1969. Within the county ofCheshire, during the year 1979/80, waste fromthe heavy chemical industry amounted to 21.53 x106 tonnes, from the Ship Canal 2.4 x 106

tonnes, and from power stations 0.66 x 106

tonnes: most of this waste went to lagoons. Atotal of 19.32 x 106 tonnes was deposited inlagoons in five districts within the county.None of these lagoons had to be reported to theMines and Quarries Inspectorate.

LEGISLATION AND INSPECTION7. The failure of a colliery spoil tip at

Aberfan in 1966 led to new legislation tocontrol waste heaps and tailings lagoons. Atthe time of the enquiry by the Tribunal into thecause of the Aberfan disaster, there appeared tobe few regulations controlling tips and lagoonsin any part of the world. In Britain, theInspectorate of Mines and Quarries had noresponsibility for tip safety. The report of thegeotechnical investigation into the causes ofthe Aberfan failure (ref. 8) recommended thattips should be regarded as engineeringstructures to be designed by civil engineers andthat a Tip Safety Committee should be set up toadvise government. The Tribunal of Inquiryendorsed these recommendations with thefollowing results:(a) a Tip Safety Committee was set up andadvised government on the drafting oflegislation,(b) the Mines and Quarries Inspectorate wasstrengthened by the creation of a civilengineering branch.

Mines and Quarries (Tips) Act 19698. A Mines and Quarries (Tips) Bill was

drafted to supplement the existing Mines andQuarries Act, 1954, and became the Mines andQuarries (Tips) Act 1969 (ref.5). It wasdescribed as an Act to make further provision inrelation to tips associated with mines andquarries; to prevent disused tips constituting adanger to members of the public; and forpurposes connected with these matters.9. In the Act, the expression "tip" means an

accumulation or deposit of refuse from a mine orquarry (whether in a solid state or in solutionor suspension) other than an accumulation ordeposit situated underground, and where any wallor other structure retains or confines a tipthen, whether or not that wall or structure isitself composed of refuse, it shall be deemed tcform part of the tip for the purposes of theAct.

10. The operation of the Act was laid down byStatutory Instrument No. 1377 (1971): the Minesand Quarries (Tips) Regulations (ref.6), whichcame into effect on 1 October 1971. It gave aclearer separation between a tip of coarsediscard and a tailings lagoon by stating that a"classified tip" was:(a) a tip consisting of refuse accumulated ordeposited wholly or mainly in a solid state andnot in solution or suspension and(i) covering a superficial area of landexceeding 10 000 nr, or(ii) of height exceeding 15m, or(iii) standing on ground at a slope exceeding 1on 12.(b) a tip consisting of refuse accumulated ordeposited wholly or mainly in a solution orsuspension and(i) with lagoon level more than 4m above thelevel of any part of the neighbouring landwithin 50m of its perimeter, or(ii) with lagoon volume exceeding 10 000 m3

11. The tips were further separated by theAct, into "active classified tips", i.e. thosebeing added to during working of the mine orquarry, and "closed classified tips" meaningthose no longer used for disposal of waste. Aninspector may exempt any tip from the /

application of any of the provisions of theregulations, if he is satisfied that they wouldbe inappropriate. Tipping operations have to besupervised by a competent person, who has toreport any defects in a book, also the personresponsible for the tip is required to recordaction taken to remedy the defect in a book.

12. Before construction of a tip can begin,notice must be given to the inspector for thedistrict, who must be satisfied that thecondition of the foundation soil and proposedmethod of construction are adequate to ensuresafety. There is no special provision fortailings dams, other than that a tailings damand its lagoon are regarded as a classified tipif the lagoon is to be more than 4m aboveadjoining land or exceed 10 000 m3.

Technical guidance and inspection13. Following the formation of the civil

engineering branch of the Inspectorate, morethan a hundred inspectors received specialtraining in soil mechanics and the factors

64

PAPER 7: PENMAN AND CHARLES

affecting tip stability. The National Coal Boardprepared a "Technical Handbook on Spoil Heapsand Lagoons" and "Codes and Rules for Tips"(ref. 9). Following publication of the Minesand Quarries (Tips) Act 1969, they prepared aproduction department instruction entitled"Management of Tips". The main emphasis in allthese documents related to coarse discard tips:there was little reference to lagoons and nospecial instruction concerning the inherentdanger of tailings dams and the fact thatfailure is likely to lead to a flow of liquefiedtailings over a considerable distance. By 1971the China Clay Association had produced anexcellent handbook on the disposal of wastematerials. The principles of design for tipsand lagoons were explained in practical termsfor those persons who have to carry out theday-to-day operations. The British Quarryingand Slag Federation have issued a useful bookletwhich will assist quarry owners to prepare thenecessary documentation.

Reservoirs Act 197514. Water retaining dams are controlled by

the Reservoirs Act 1975 (ref. 10) which appliesto reservoirs which are capable of holding morethan 25 000 m3 of water above the natural levelof any part of the land adjoining the reservoir.Important features of this reservoir safetylegislation include:a) Local authorities are designated asenforcement authorities with specific dutieswhich include keeping a register of reservoirsand demanding full documentation from owners andengineers for dams in their area.b) Construction and inspection of dams can onlybe undertaken by qualified civil engineers whoare members of specially constituted panels setup by the Secretary of State for theEnvironment.c) Owners are required to appoint a namedsupervising engineer from the appropriate panelto keep the reservoir under continualsupervision between inspections.

15. The Reservoirs Act 1975 specifies that"reservoir" means a reservoir for water as suchand accordingly does not include a mine orquarry lagoon, which is a tip within the meaningof the Mines and Quarries (Tips) Act 1969.Despite this, some owners of tailings lagoonshave had them designed and their constructionsupervised by a panel engineer. Thus thespecialist knowledge and expertise of civilengineers authorised to design and supervise theconstruction of water retaining dams has beenutilised in the design and construction of sometailings dams.

16. The requirements of the the Mines andQuarries (Tips) Act 1969 can be compared withthe requirements of the Reservoirs Act 1975. Theformer only requires 30 days notice to be givento the district inspector prior to commencementof tipping. Method of construction must bespecified by tipping rules made by the minemanager or quarry owner, but there is nostipulation for an overall designer for the tip.Prior to commencement of tipping (orconstruction of a tailings dam) the owner mustobtain a report from a competent person on themethod of carrying out the intended tippingoperation and on every other matter likely to

affect the security of the tip, eg siteconditions, the total volume to be tipped. TheRegulations 1971 require every active tip to besupervised by a competent person. There is noindication as to how the competence of thepersons is to be assessed.

FAILURES AND SERIOUS EVENTS17. In other parts of the world, failures of

tailings dams have caused major loss of life.The failure at Buffalo Creek in West Virginia on26 February 1972, resulted in 125 deaths. Thefailure at Stava in northern Italy on 19 July1989 resulted in 269 deaths when two 20 m hightailings dams collapsed and the resulting250 000 m3 mudflow engulfed Stava and destroyedpart of Tesero. Fortunately no incidents ofsuch severity have occurred in Britain, butthere have been a number of uncontrolledreleases of tailings. Initially lack ofexperience in the construction and management oftailings lagoons, when they were first requiredin Britain, resulted in some failures.

18. In South Wales, at Ty Mawr colliery, alagoon was formed on the hillside above thecolliery, retained by a bund of coarse discard.Sufficient consideration had not been given tothe water balance and the lagoon overtopped thebund in December 1961, sending a small flow oftailings down the hillside, which reached theaerial ropeway that carried the waste from thecolliery. A second lagoon was constructed onthe other side of the ropeway, so that anyrelease would not harm it, but on 25 March 1965,the bank failed, releasing a flow of liquefiedtailings which reached the colliery car park,damaging some cars, and almost reaching theshafts. The seriousness of this incident wasappreciated, and the Coal Board requested alltheir colliery managers in South Wales area tocheck on their tipping arrangements,particularly in relation to tailings, andreport. The Aberfan slip, which occurred duringthe morning of 21 October 1966, was not directlyconnected with tailings disposal. In thisdisaster 114 school children were killed whenone of the tips failed, causing a flow slidewhich engulfed their school.

19. Another incident, similar to those at TyMawr, occurred on 24 March 1966 at Williamthorpecolliery. A slurry pond which had been builtinto the Old Dirt Tip collapsed, sending a flowof tailings over an adjacent road which wascovered to a depth of 3m and remained closed for10 days. At Stoney Middleton, in Derbyshire,the retaining dam of a settling pond burst andthere was a damage to property and roads, butfortunately there was no loss of life. Thisfailure occurred on 8 February 1968.

20. The annual reports of HM Chief Inspectorof Mines and Quarries have shown that dangerousoccurrences related to tips and lagoonsassociated with mines and quarries havecontinued since they began to be reportedfollowing Aberfan. Although the numbersreported under the provisions of the Mines andQuarries (Tips) Act 1969 have fluctuated fromyear to year, they have shown no consistentdownward trend, which might have been expectedto follow the Regulations controlling tip andlagoon construction. Over 60 dangerousoccurrences involving unstable or potentially

65

TAILINGS DAMS

unstable lagoons were reported between 1968 and1986.

21. Concern about lagoons was expressed bythe Chief Inspector in his report for 1968 whenhe said that it was evident that lagoon bundstability was a matter requiring specialattention. He pointed out that difficultieshave resulted from inadequate compaction of theearthwork which allowed seepage, saturation andin due course, failure. Occasionally additionalmaterial was placed directly on top of asaturated bund, only to become water-logged andso take part in an even larger slip.

22. A large retaining bund of a 4 hectaretailings lagoon at a fluorspar processing planthad been progressively heightened withoutincreasing the width of the base, in order toincrease the capacity of the lagoon. Followingheavy rains, which reduced the freeboard at thecrest of the bund, strong winds caused waves togo over at one point, eventually causing abreach 6m wide and 2m deep. More than 14 000 m3

of liquid flowed through a nearby village, someparts of which were flooded to a depth of lm.It was evident that the lagoon had not beenproperly designed for the capacity it wasultimately required to hold and that, inparticular, the upper stages were notconstructed on sound engineering principles.

23. A major sand tip failure was caused bytipping sand residues over an old mica tailingslagoon. The initial slip triggered a flow slidewhich caused the sand to run some 200m acrossopen country. It was an example of thedangerous instability which can occur when tipsare not constructed on sound principles.

24. The 10 lagoon retaining bund failuresreported for the period 1969/70 were almost alldue to excessive seepage caused by inadequatecross section, or failure to incorporateadequate internal drainage in the bund orallowing an excessive depth of free water tostand over the settled tailings. A major breachat one bund was caused by running tailings intothe lagoon at such a rate that it flowed overthe crest.

25. In 1971 a lagoon bank collapsed when thesilty sand foundation became saturated and thebank moved forward towards a freshly excavatedarea. Seepage had been apparent on the surfaceover a period and should have served as awarning.

26. One of the three lagoons reported in 1972had been overfilled and the warning was given bythe Chief Inspector that lagoons formed on coalmine refuse tips, particularly those at highlevel, are undesirable because seepage pathsthrough the colliery refuse are difficult topredict with accuracy. The failure of the othertwo was attributed to excessive seepage throughthe bunds. It was pointed out that lagoon damsare often built with readily available butunsuitable material which does not have therequired engineering properties.

27. One of the lagoons which failed during1973 discharged tailings on to a public road andat another lagoon, which had been brought backinto use after a period of inactivity, anextensive fissure appeared along the crest ofthe lagoon bank accompanied by movement of thetoe. Part of the bank and its foundationconsisted of clayey material and no provision

56

had been made for the drainage of seepage water,although this is important in lagoon banks madeof impervious refuse.

28. Coal recovery from an old colliery refusetip was using a washery process dischargingtailings into a lagoon formed in the refuse.During 1974 the lagoon overflowed, causing rapiderosion of the bank and releasing 20 m3 of waterand slurry. A potentially more dangeroussituation arose at a sand and gravel quarrywhere a 6m length of lagoon bund collapsedadjacent to a draw-off pipe and released 9000 m3

of tailings. This flowed into a series of lowerlagoons causing overflows on to adjacent land.This originated in a piping failure and it wassaid that the onset of piping had been concealedby silt which had built up around the draw-off.

29. Two lagoons gave trouble in 1975. A bankseparating two sections of a large lagoonfissured along its centre-line, thereby makingit insecure. The bank had not been maintainedat its designed height and slope, which resultedin overloading of the foundations. The bank ofanother lagoon in a sand quarry collapsed,resulting in tailings cascading through a seriesof lagoons, through the quarry and on to apublic road.

30. The very dry summer of 1976 was thoughtto have contributed to the failure of one lagoondam due to clay shrinkage. This released about7000 m3 of tailings. When the rains came, theywere very heavy and were blamed for the failureof a tailings dam at a sandstone quarry. Thetailings flowed 500m and flooded the quarrydrying plant, an occupied cottage and a minorroad.

31. At another lagoon, the rockfill dam faileddue to piping through the clay seal on the innerface of the dam. About 7000 m3 of water floodedthe quarry floor and submerged a dieselexcavator. At another site, refuse was beingplaced over an old lagoon when it slipped due,it was thought, to build up of pore pressureunder the increasing weight of material.

32. A fatal accident occurred in 1985 at anactive spoil tip and lagoon where the reshapingof the site required the location of the tippingoperations to be changed. An articulated dumptruck driver, employed by earth movingcontractors, was drowned when he reversed thetruck towards uncompacted ground at the edge ofthe lagoon. A slip occurred over a length ofabout 24m and about 2500 tonnes of recentlyplaced colliery discard disappeared into thelagoon together with the dump truck.

33. In another area where landscaping wasbeing undertaken, some lagoons had been infilledduring freezing weather. A dam failed,releasing 3000 m3 of highly viscous tailingswhich flowed into a second lagoon. Thisdisplaced wetter material blocked the drainageculverts and consequently diverted through atunnel on to adjacent land. Failure wasattributed to tipping on to deeply frozen slurrywhich subsequently thawed.

34. At a quarry, a disused lagoon ofsolidified slurry had, over the years, beenovertipped with the refuse. As a final layer ofsoil-forming material 1 to 1.5m thick was beingplaced, the tailings dam failed. It was saidthat this activity was not covered by the Minesand Quarries Act.

PAPER 7: PENMAN AND CHARLES

HAZARD EVALUATION35. Tailings have a water content high enough

to permit flow through flumes or pipelines. In alagoon the particles settle into a meta-stablestructure, developing an effective stress underself-weight with corresponding shear strength.When the surface of a lagoon is allowed to dryby evaporation, the pore water suctionsdeveloped in the surface layers may produce asurface crust, which may be strong enough tocarry the weight of people walking on it andlow ground-pressure earthmoving machines.Disturbance of the surface structure can destroythe bond between particles, throwing allself-weight on the pore water and reducing shearstrength to zero. Such a phenomenon occurslocally when a machine breaks through a driedsurface, producing quick conditions, allowingthe machine to sink uncontrollably into the massof tailings.*

36. If the retaining embankment is breached,tailings will be released from the lagoon. Dueto their high moisture content, the tailings mayliquefy and flow. The storage of hydraulicallyplaced fine grained waste materials above thelevel of the surrounding ground thus forms apotential hazard to people and property at thelower level.

Hazard analysis and risk assessment37. The magnitude of the hazard is related

to the height of the dam above the surroundingground, the volume of tailings in the lagoon,the distance of the threatened area from thelagoon and the flow properties of the tailings.Hazard analysis involves not only an evaluationof the magnitude of the hazard, but also anassessment of the probability of failure of thetailings dam and the consequences of such afailure. The hazard may be assessedsubjectively or some method of calculation maybe adopted. A study of the use of probabilisticrisk assessment for a water retaining earthembankment dam has been described by Parr andCullen, 1988 (ref.ll). Despite the difficultiesin establishing the probability of many of thebase events on which the analysis rested, Parrand Cullen recommended continuing development ofthe approach.

Comparison of tailings lagoons with tips38. The hazard posed by tailings dams and

lagoons can be compared with those due to tipsof coarse waste materials in a solid condition.The Aberfan tip disaster has already beendescribed. In July 1987 the 150 m Little Johntip, near the village of Roche in Cornwall,collapsed (ref. 12). Debris flowed 100m down acountry lane and an eyewitness said that thefailure occurred very quickly. Failure of atailings dam could release a much largerquantity of tailings from a lagoon than wouldusually be involved in a tip failure.

Comparison of tailings lagoons with waterreservoirs

39. Parr and Cullen, 1988 (ref. 11)recommended that in line with practice in otherindustries which offer major hazards, the waterindustry should consider developing emergencyplans for dam failure and a scheme for reporting

"near misses" and making this available to panelengineers. If this approach were to be adoptedfor water retaining structures, it may bequestioned whether it should also be consideredfor tailings dams.

40. Failure of a water retaining dam maycause an escape of water that represents therelease of a considerable amount of energy.Damage may be done by the force of the waterflow as well as the damage caused bysubmergence. Tailings are more dense than waterand can therefore exert a greater pressure onobstacles to their flow. Tailings may push overwalls, crush cars and wreck services. Thetailings released from the lagoon at Ty Mawrcolliery in March 1965 reached the colliery carpark where it caused damage to several cars andalmost reached the shafts. It is probable thata release of the same volume of water, even ifit had flowed through the car park, would havecaused relatively little damage because it wouldnot have risen to the same depth as tailings andwould have exerted less force. The water woulddrain away and the incident might not beconsidered worthy of report. The tailings, onthe other hand, remained as a more permanentreminder of the escape which had occurred.

41. The power of tailings flow wasdemonstrated after the collapse of a tailingsdam on 11 November 1974 at the Bafokeng platinummine in the western Transvaal. "The heavy fluidliterally cut away half the winder house andcarried with it heavy steel girders, brickbuildings, vehicles, reservoirs and heavystores. Although erosion of the original groundwas minimal, all surface installations in thepath of the flood were destroyed and a largequantity of slimes entered the shaft, trappingworkmen underground and dragging the shaftequipment to the bottom" (ref. 13).

Dam break analysis42. Estimates of flood damage following the

failure of water retaining dams are increasinglybeing based on dam break analysis. Suchanalyses permit assessment of damage caused notonly by inundation but also by the velocity offlowing water. These calculations can be usedas part of a hazard analysis or in thedevelopment of emergency planning procedures.The major uncertainty relates to the extent ofthe breach and the speed with which it develops.A full examination of the hazard presented by atailings dam can only be made if the potentialflow of the tailings released from the lagoon isstudied. A realistic assessment of damage wouldrequire knowledge of the volume of tailingswhich would be released, the distance thetailings would flow and the velocity of flow.Application of dam break analysis to tailingsdams requires a better understanding of thecharacteristics of the liquefaction and flow oftailings.

Seismic risk43. In many parts of the world, earthquakes

have caused uncontrolled release of tailings(refs. 14 and 15). Although Britain isconsidered to lie in a relatively non-seismiczone, Britain is not free of earthquakes. Outof more than 2000 recorded earthquakes which

57

TAILINGS DAMS

have occurred during the last 700 years, themaximum had a magnitude of M = 5.5 (Richterscale). Dams constructed of tailings by theupstream method, with steep downstream slopes,are particularly susceptible to earthquakeshock. In general, in Britain, no allowance forearthquake has been made during the design oftailings dams and it would be of value toidentify those most likely to be affected. Itwould be of benefit to measure the dynamiccharacteristics of tailings dams when they havereached various heights in order to check designassumptions. Assessment of in situ propertiesand dam behaviour by controlled excitation couldassist safe design. Little is known about thedegree of damping and the low bulkcompressibility of the saturated material andinteraction with the sensitive structuredeveloped within the lagoon mass may lead tounexpected results.

CONCLUSIONS44. There are a large number of tailings

lagoons in Britain. Over 1500 are associatedwith mines and quarries alone. They form aninexpensive method of waste disposal for a widevariety of industries and there is no reason toexpect that the volume of waste materialsdisposed by such methods will decrease.

45. Storage of tailings is usually aboveexisting ground level and requires theconstruction of a tailings dam. Such storagepresents potential hazards to life and propertywhich are similar to those occasioned by damswhich impound water reservoirs.

46. Incidents affecting the stability oftailings dams and involving the release oftailings continue to occur in Britain.Overseas, there have been failures involvingmajor loss of life and property.

47. Although the hazards posed by tailingsdams are similar to those from dams whichimpound water reservoirs, tailings dams are notsubject to reservoir safety legislation. Manytailings dams come within the scope of the Minesand Quarries (Tips) Act of 1969 but not all doso. The Mines and Quarries (Tips) legislationdoes not provide publicly available registers oftailings dams and does not clearly define howthe competence of a person appointed to build orinspect tailings dams is to be assessed.

48. The continuing number of dangerousoccurrences involving unstable tailings dams andthe consequent uncontrolled release of tailingsemphasise that it is essential that theprinciples of geotechnical engineering areapplied in the design, construction andinspection of all tailings dams irrespective ofsize and whether or not they are associated withmines and quarries.

49. The high standard of safety establishedfor embankment dams which impound waterreservoirs owes much to detailed studies of thebehaviour of particular dams and the opendiscussion of problems and hazards among civiland geotechnical engineers. More could be doneto study the performance of British tailingsdams and lagoons and to evaluate the hazards towhich they are subject.

58

The work described in this paper forms part ofthe research programme of the Building ResearchEstablishment and is published by permission ofthe Director. Crown copyright

REFERENCES1. TAYLOR R K (1984). Composition andengineering properties of British collierydiscards. National Coal Board, London.2. HAWS E T, MARTIN P L AND ORANGE-BROMEHEAD R(1990). Gale Common ash disposal scheme -concept, design and construction. BNCOLDConference on The Embankment Dam, Nottingham.3. ILLSLEY A E, WAKELING T R M AND HUMPHREYS JD (1976). The Kernick and Portworthy micaresidue disposal schemes for ECLP in Cornwalland South Devon. Transactions of 12thInternational Congress on Large Dams, Mexico,vol 1, pp 437-461.4. PENMAN A D M (1987). Preliminary study ofsafety of tailings dams and lagoons in Britain.Report prepared for Building ResearchEstablishment, Garston, Watford.5. HER MAJESTY'S STATIONERY OFFICE (1969).Mines and Quarries (Tips) Act 1969 Chapter 10.An Act to make further provision in relation totips associated with mines and quarries: toprevent disused tips constituting a danger tomembers of the public: and for purposesconnected with these matters. HMSO, London.6. HER MAJESTY'S STATIONERY OFFICE (1971).Statutory Instrument (1971) No.1377. Mines andQuarries. The Mines and Quarries (Tips)Regulations 1971. HMSO, London,7. INTERNATIONAL COMMISSION ON LARGE DAMS(1982). World Register of Mine and IndustrialTailings Dams, 1982-Edition 1. ICOLD, Paris.8. BISHOP A W, HUTCHINSON J N, PENMAN A D M andEVANS H E (1969). Geotechnical investigationinto the causes and circumstances of thedisaster of 21 October 1966. A selection ofTechnical Reports submitted to the AberfanTribunal. HMSO, London, pp 1 - 80.9. NATIONAL COAL BOARD (1970). Technicalhandbook on spoil heaps and lagoons. Also Codesand rules for tips. NCB, London.10. HER MAJESTY'S STATIONERY OFFICE (1975).Reservoirs Act 1975. An Act to make furtherprovision against escapes of water from largereservoirs or from lakes or lochs artificiallycreated or enlarged. HMSO, London.11. PARR N M AND CULLEN N (1988). Riskmanagement and reservoir maintenance. Journalof Institution of Water and EnvironmentalManagement, vol 2, no 6, December, pp 587-593.12. SMITH G (1987). Village in fear after claytip collapses. Observer, 26 July 1987 p.3.13. RUDD R T (1979). The failure of a slimesdam at Bafokeng - The Bafokeng disaster and itslegal implications. The Civil Engineer in SouthAfrica, vol 21, no 6, pp 146 - 150.14. DOBRY R AND ALVAREZ L (1979). Seismicfailures in Chilean tailings dams. ASCE Journalof Soil Mechanics and Foundation EngineeringDivision, vol 93, no SM6, pp 237-260.15. MARCUSON W F, BALLARD R F AND LEDBETTER R H(1979). Liquefaction failure of tailings damsresulting from the Near Izu Oshima earthquake,14 and 15 January 1978. Proceedings of 6thPanamerican Conference on Soil Mechanics andFoundation Engineering, Lima.

8. Tailings dams of the copper mining plant Elatzite after eight yearsof operation

C. B. ABADJIEV, Higher Institute of Architecture and Civil Engineering, Sofia, Bulgaria,and A. A. KARADIMOV, Nonferrous Metallurgy Corp., Sofia, Bulgaria

SYNOPSISThe paper gives a detailed description of the structure and the method of execution of the

tailings dam at the Elatzite Copper Mining Plant, built after the downstream construction methodby hydrocycloning. The tailings dam was designed to reach an ultimate height of 145 m, and in June1989 it was 95 m high. An analysis is made of the design solution and of the alterations duringthe construction period of the structure and of the physical properties of the deposited cyclonedsand and slime. Lessons and conclusions are drawn on the basis of 8 years of operation, concerningthe design, construction and operation of this type of tailings dams.

DESCRIPTION OF THE TAILINGS DAMThe tailings dam of the Elatzite Copper

Mining Plant is being constructed after thedownstream method by hydrocycloning. It is lo-cated about 70 km east of Sofia in the upperpart of the hilly catchment area of a smallravine (Fig.l) at an altitude of 600 to 700 m.The area of the tailings dam at the final stageis to be 268 ha, and of the catchment area-330ha. The average water discharge of the naturalrunoff in the ravine is 15 1/s. The tailingsdam is located in an area of seismicity with apredicted 1000 year return acceleration of 10%of gravity.

The plant has a maximum annual tailings

10

Fig. 1

1,2,3 COLLECTOR4 UPPER MAIN DAM5 LOWER MAIN DAM6 CREST OF FINAL HEIGHT

7 DRAINAGES8 PUMP STATION

9 SLURRY PIPELINE10 SLURRY TROVGHS11 WATER PIPELINE

output of 10 mill.t. The tailings dam isdesigned to use to the maximum the capacity ofthis site to store the tailings with an ulti-mate volume of 130 mill.m^. The outflowing tail-ings have a mean diameter of 0.068 mm, and thequantity of the particles smaller than 0.074mm(200 mesh) is 70% on the average.

With a short, but wide reservoir site andthis particle size distribution of the tailings,most suitable and economical in comparison withthe several alternatives discussed, proved tobe the downstream construction method by hydro-cycloning. In order to be able to make maximumuse of the capacity of this site the tailingsdam has to have a final height of 130 m abovestream bed at the downstream toe, with a possi-bility to raise it to 145 m. The projectedcross section of the tailings dam is shown inFig.2,

The upstream starter dam was erected ofheavily weathered moscovite-biotite gneiss bor-rowed from a quarry in the reservoir site,sincethe open-cast mine was too far off to allow theuse of pit waste rock material. The height ofthe starter dam was determined on the basis ofthe expected relation of cycloned sand to slime,presuming a 1:3.5 downstream slope of the sandbody (by analogy with the Brenda tailings damin Canada, ref.l) and a 1:2.0 upstream slope.The percentage of the cycloned sand during thefirst 5 years was expected to be 50% of thetotal tailings quantity and further on 30%. The1:1.5 upstream slope of the starter dam wasscreened with clay up to half its height.

The downstream toe dam is 30 m high and isfilled of fresh granite as a permeable body,allowing draining of the sand. Two drains, each10 m wide, are laid along the entire length ofthe tailings dam under the sand part betweenthe two dams parallel to their axes. They aredesigned, just as the toe dam, to drain mainlvthe processing water carrying the cycloned sandover the temporary downstream slope off thesand part.

The slurry is discharged from the plant with

The embankment dam. Thomas Telford, London, 1991 59

TAILINGS DAMS

POND BEACH

UPPER MAIN DAM

Fig. 2 LOWER MAIN DAM

density of 1.16 to 1.19 t/m3, which means thatone weight unit of tailings corresDonds to 3 to3.5 weight units of water. The specific gravityof the tailings is 2.70 g/cm3. The slurry flowsfreely from the plant along the right bank ofthe tailings dam in concrete troughs, and onlyin one steep section - in two pipelines. Fromthe high right bank to the starter dam it flowsdown under pressure in two pipelines of 500 mmdiameter with inside basalt coating. The distri-buting steel pipeline along the starter damcrest, which feeds the hydrocyclones, in 1989had the same length as the starter dam, but atthe final stage it is to be 2000 m lone. Thework proceeded consecutively in sections alongthe pipeline. At the start two sections wereset UD. and in 1988 they were already four. Thediameter of the pipeline in the first sectionand the first half of the second one is 700 mm.then it becomes 630 mm. At the beginning (in1981) 75 hydrocyclones were installed on thestarter dam crest. In the eighth year of opera-tion (1989) 130 hydrocyclones covered thelength of 1600 m. The diameter of the hydrocyc-lones is 500 mm tapering under an angle of 20°,the diameter of the sand nozzle being 46 mm,They are coated with rubber lining. At full ca-pacity of the plant up to 35 hydrocyclonesoperate simultaneously. The average water pres-sure of the cyclones was projected to be 2 bars.

The secondary dams are deaigned to be 5 to 6m high,of cycloned sand. The distribution pipe-line is moved on each secondary dam, and thehydrocyclones are raised at each 1.5 m (Fig.3).First the second part of the pipeline is raised

Fig. 3

1 DISTRIBUTION PIPELINE

2 HYDROCYCLONE

3 SLIME HOSE

and then the first one.In the final 15 m of the tailings dam height

the relation of the necessary sand with respectto the slime will be higher to the extent thatit will be possible not to cyclone any more,butto spigot the entire quantity of the tailingsafter the upstream construction method. The ba-lance will have begun to be possitive severalyears before that. In order that the sand partdoes not overtake inadmissibly the slime part,cycloning is to be stopped periodically withspigotting continuing with the entire quantityof slurry. Thus a layer of coarser material isto be deposited in the slime part, which willdrain the slime and accelerate its consolida-tion. As a result stability will be improved.

Stand-pipe piezometers of PVC pipes dia.63 mmare provided only in the sand part to monitorthe phreatic surface.After settlement of the hard particles the

clear water is discharged over three slope col-lectors located in the ground folds, into theopposite bank of the pond (Fig.1).They consistof concrete troughs and are covered with con-crete slabs, thus raising the spillway sill andcontrolling the pond level.Under the starter dam, the sand part and the

toe dam, the flow continues into steel pipesdia.800 mm in reinforced concrete casing.Thesedischarge into the reservoir of the recircula-ting pump section, located downstream of the toedam. At their end they are provided with slidevalves to stop the flow when necessary (forconcreting the slabs or in case of emergency).The seepage water caught in the drains is alsodischarged into the pump station reservoir.Water is discharged always from the spillway

most distant from the site of slime discharge.

OPERATION OF THE TAILINGS DAMThe tailings dam was put into operation at

the beginning of 1981. Changes in the structureand the technology had to be introduced duringthe operation since not everything could beimplemented as planned.The plant was commissioned in stages. The

first year it begun to work with 50% of its ca-pacity, in the second - with 75%, and only inthe third - with 100%. During the first 2.5years the plant functioned under different andvarying conditions, which hampered the work ofthe hydrocyclones. Besides that, experiencehad to be acquired during the first year. Theexpected separation of 50% as coarse fractioncould not be achieved. During the first yearthe hydrocyclones separated only 42% of the

60

PAPER 8: ABADJIEV AND KARADIMOV

SECONDARY DAMS DESIGNED DAM SECTION

/ ^ULTIMATE DAM SECTION

SAND LEVEL 1.07.1989

.1.01.1089

J.01.19881.01.1984

SUPPORTING DAMS

UPPER MAIN DAM

SUPPORTING DAMS

Fig. 4

tailings as coarse fraction. Since they didnot operate regularly, the separated cyclonedsand amounted to only 35% of the entire tail-ings quantity. The remaining 65% were dischargedupstream of the starter dam as the slime. Besi-des that the cycloned sand was deposited with aflatter slope. Fig.4 shows the slopes of thesand deposited in the course of the years. Inthe upper part the slopes are about 1:3. whilein the lower part they reach up to 1:10. Figure4 reveals that the sand part was not raiseduniformly every year, although the annual tail-ings quantities were almost equal. This was sofor each specific profile, since operation pro-ceeded consecutively on separate sections, andthe changing of the sections did not coincidewith the sequence of the years. In order to se-cure that the rise of the sand part overtookthe slime, it became necessary to use rockfillfor the 6 m high first secondary dam. It waslaid on the starter dam and on the slime sincethe sand part had remained lower. For the samereason the second and the third secondary damswere built of cycloned tailings only up to aheight of 3-4 m, while their upper parts, 2 mhigh, were made of rockfill borrowed from aquarry.

In the second section the cycloned sand partlagged more behind because of the topographicconditions and for organizational reasons.Therefore it was necessary there to constructtwo suppprting dykes of borrowed rockfill,3 to3.5 m high, laid on the temporary sand slope,ata distance of 45 m from the starter dam (Fig.4).With their help the level of the sand part wastemporarily raised.

During the first year the cyclones separated42% of the total tailings quantity in order tomaintain the sand part higher than the slime.During the following years the percentage ofseparation by the hydrocyclones was graduallyreduced, in the sixth year being 36%, and inthe eighth - 32%. The actually separated sandin the first year was 35%, in the sixth - 31.7%and in the eighth - 28%. During the first yearthere exhisted the danger of the sand remainingunder the slime level. For that reason the se-condary dam had to be made of borrowed rockfill.Gradually the balance was improved, during thesecond year the sand body overtook the slime by2.30 m and in the sixth year the sand was al-ready higher by 8 m. In order not to increasethe level of the sand above the slime by morethan 8 m, which endangered the upstream slopestability of the sand part, from the seventh

year onwards spigotting was conducted with theentire tailings quantity over the slime. Thisdirect spigotting with the entire tailings wasincreased every year.Since the 1:3.5 downstream sand slope could

not be acheived, it was decided to build up-stream of the toe dam 3 secondary rockfilldams each 5 m high (Fig.4).The following tests and measurements were

carried out on the tailings dam. The averagediameter of the sand particles ranged from 0.20to 0.26 mm, the particles smaller than 0.1 mmbeing 25 to 35% and the coefficient of uniformi-ty varying at about 6. The slime particles hadan average diameter of about 0.04 mm.The pressure in the first operating hydrocyc-

lone was maintained within 2.0 to 2.3 bars. Thepressure in the last hydrocyclone was 1.6 to1.9 bars. The first hydrocyclone separated sandof an average diameter of 0.25 to 0.26 mm, andthe last - 0.20 to 0.23 mm.The hydrocycloned sand flew out of the cyc-

lones as slurry with high density; one sandweight unit of sand corresponding to one weightunit of water. With this density and with thehigh permeability of the sand, the water wasquickly drained away and no substantial segre-gation of the sand occured. This is evident inFig.5. The slime was segregated along towardsthe pond, since after the first year a non-sub-merged beach was always maintained, which in1986 was 60 m long, and in 1989 - 150 to 200 m.The decrease of the average slime diameter to-wards and in the pond at the end of 1986 isshown in Fig.5.The cycloned sand was deposited with a low

density. The samples taken from the surface ofthe sand slope indicated an average bulk densi-ty of about p - 1.4 g/cm3. The void ratio wasabout e = 0.95 to 1.05 and the average relativedensity Id = 0.41 at a bulk density of f max =

1.81 g/cm3 and pm±n - 1.21 g/cm3. The samplestaken from the sand surface had an angle of in-ternal friction of iff = 26° to 32°, cohesioncf = 0.1 to 0.2 x lO^Pa and permeability k =a x 10-3 to a x 10-4 cm/s. The compressiontests give a void ratio of e = 0.9 and permea-bility of k = a x 10~5 cm/s for a vertical load-of 5 x 105Pa, and e = 0.7 and k = a x 10~6 Cm/sfor a vertical load of 14 x lO^Pa. Thereforeit could be assumed that in depth the bulk den-sity and the relative density are higher.The slime had a permeability of k = a x 10"~5

cm/s at the beginning of the beach, and furtherinside the beach k = a x 10~6 to a x 10""? cm/s,

61

TAILINGS DAMS

Fig. 5

respectively angle of internal friction {p? =22o to 15°, and cohesion c = 0.2 x 105 to0.35 x lO^Pa.The cycloned sand settled mainly in the upper

part, while in the lower part it remained con-siderably below the ultimate dam section. Forthat reason after the seventh year the nozleswere adjusted to a somewhat wider opening, toalow the sand to flow out with more water, soas to be conveved by gravity into the lowersand part. This leads inevitably to an encreaseof the fine fractions in the cycloned tailings.However this compromise was necessary to shapethe required profile of the tailings dam.The stand-pipe piezometers indicated a water

level not higher than 1 to 2 m above the ground.This phreatic surface was maintained mainly bythe water flowing with the sand from the cyclo-nes. The phreatic surface in the section whichwas being cycloned was by 2 to 3 m higher. Thislow phreatic surface ensured the seismic stabi-lity of the tailings dam.Since the upper layers of the sand part cree-

ped downwards over the downstream slope, thepiezometers tilted and got broken. This made itnecessary to replace them with new steel pipes,which however also tilted.The drains under the sand part functioned

well. They lowered the phreatic surface andconveyed a total discharge of up to 80 1/s ofdrained water.In the third year of operation a permanent

rotation sprinkler system of a range 45 m wasset up to reduce dusting from the temporarydownstream slope. The system was laid on thissame temporary downstream slope. In the processof building up of the sand body, the horizontalfeeding pipes were deformed and broken due tosettlement of the deeper sand layers, and thevertical pipes - due to creeping of the sand.Rotation sprinkling line proved to be ineffec-tive during continuous strong wind. The systemwas changed to a semipermanent one with sprink-ler laterals shaping vertical water screens.

The purpose of the screen was to catch the dustand simultaneously to moisten the slope. How-ever it also proved to be inadequate because ofthe large area of the dusting surface - 63 ha.Now a new method for chemical and chemical-bio-logical stabilization of the Temporary down-stream slope is being developed.

CONCLUSIONThe downstream method of tailings dam

construction by hydrocycloning is highly relia-ble and economical. In this case it was themost appropriate. Inspite of its merits, itraised the problem,that in the initial stagesthe sand fraction was not sufficient to supportthe slime. Dusting was more intensive and thedesired final downstream slope was difficult toshape.Not everything could be predicted in the

design. In the course of operation changes hadto be introduced consisting in building addi-tional supporting dykes, changing the methodof shaping the ultimate downstream slope, aswell as its draining and stabilization toresist the effect of water and wind, both duringoperation and after completion of the tailingsdam.

REFERENCES1. Klohn,E.J. and Maartman,C.H. Construction

of Sound Tailings Dams by Cycloning andSpigotting, in ,Tailings Disposal Today\ Proc.of the First International Tailings Symposium,Tuscon,Arizona, 1972.

62

9. Waste retention embankments on soft clay

A. R. GRIFFIN, Binnie and Partners, Redhill, UK, C. K. PENG, Syed Muhammed Hooi danBinnie, Kuala Lumpur, Malaysia, and D. L GUDGEON, Binnie and Partners, Redhill, UK

SYNOPSIS

To evaluate the preliminary design for embankments on soft clay which was basedon parameters from insitu and laboratory tests a trial embankment was constructedand monitored. The results of these trials enabled a re-evaluation of the designwith considerable savings on fill material.

INTRODUCTION1. The Tioxide Group pic is a leadingmanufacturer of titanium dioxide, a white pigmentused in paint, rubber, plastics, textiles,toothpaste and many other products. The Companyis preparing to open a new plant at an industrialsite on the east coast of Malaysia in Terengganustate.2. The major waste by-product of themanufacturing process is gypsum and ironhydroxide initially produced as a slurry. Thisslurry is to be dewatered by a series ofhydraulic presses producing a *cake' of gypsumand iron hydroxide, and a generally clearfiltrate of water which contains mainly gypsum(calcium sulphate) in solution.3. Before approval for the construction of thefactory could be given the entire process wassubject to a strict environmental impact study.An outline scheme proposed for the handling ofthe gypsum and wastewater was approved by theMalaysian Department of the Environment. Thisscheme provided for the containment of the solidgypsum within a controlled landfill area,retained by impermeable embankments. The waste -water is to be directed through a series oflagoons to enable monitoring of the water and toallow any remaining suspended solids to settlebefore final discharge to the environment.Because of the high rainfall in the area(3000mm/annum) the lagoons were also designed toreceive run-off from the landfill site whichwould contain some resuspend gypsum material.THE SITE4. The factory site is located just north of thetown of Chukai in Terengganu State (see Fig. 1.) .After visits and investigations at severalpotential storage areas the wastewater treatmentplant was chosen to be sited in the area adjacentto the factory (Fig. 2) because it was known tobe underlain by clays. This would provide animpermeable base to the lagoons and landfillarea, avoiding the need to install an artificialand expensive impermeable lining. This area washowever flat, low lying and swampy with anapproximate elevation of 3.0 m above mean sea

The embankment dam. Thomas Telford, London, 1991

STRAITSOF

MALACCA

I0i*E 102* 103*

Fig.l - Site location

level and subject to flooding. Drainage from theswamp westwards was via a recently excavatedchannel leading to the small Sungai Ruang and themain Kemaman estuary.

Geological setting5. The factory site was known to be underlain byrecent alluvial deposits. The flat swampy areawas surrounded by forested hills consisting ofLower Carboniferous metasediments (phyllites,slates, quartzites and schists) with intrudedgranites of Upper Carboniferous/Lower Permianage. These rocks also form the bedrockunderlying the alluvial deposits. The alluvialdeposits are predominently soft black clays(believed to have been deposited within the last

63

TAILINGS DAMSAdditional boreholes in valley

I

53(21) j

54T15)|

46(14) | 48(16) 50(12)

. D .__ Out|jne Qf sjte - 1 D

45(8) 47(4) 49(16) 51(14)

Steel works

45 Borehole (m)100m ^ thjCkness of clay (m)

Fig.2 - Waste water treatment site

10000 years), underlain by sands and stifferclays representing an older deposit.6. These soft clays are of marine origin and arewidespread throughout the Far East but areparticularly common in the Mekong Delta (SouthVietnam and Cambodia), Chao Phraya delta ofThailand and on the coastal plains of peninsulaMalaysia, Sumatra and Java. They can reach amaximum thickness of 30 m at the coastline butthin inland.7. They have been studied in recent yearsparticularly by Cox1 among others, mostly inrelation to the construction of highwayembankments.

Site Investigation8. A detailed site investigation was implementedin stages. Initially boring was carried out toassess the thickness of alluvial materials withsampling and testing to evaluate soil properties.Use was also made of a large volume of data froman earlier investigation at the actual factorysite which included boring and Dutch Cone

, SCTTLEMENT PONDS

testing. Second stage investigations includeddynamic probing aimed purely at confirming thecontinuity of the soft marine clay across thesite. In the final investigation stage, 4boreholes were put down at the location of thetrial embankment (see below). In 2 of the holescontinuous 100 mm diameter piston sampling wascarried out through the soft clay and in theremaining 2 holes field vane test at 0.3 mintervals using the Geonor penetration vaneapparatus. A simplified section of the site isshown in Fig. 3.

Marine Clay Properties9. At the proposed waste retention and lagoonsite the general soil profile consisted of 8 to20 m of very soft to soft greenish grey siltyclay with traces of sea shell and fine sand. Theupper one to two metres generally consisted of adark brown peaty clay or occasionally a clayeypeat with decayed vegetation, roots and woodfragments.10. Typical soil profiles and index propertiesare shown in Fig. 4 together with the results ofthe corrected in-situ vane shear tests.11. The clays are of high to extremely highplasticity with high liquidity indices. Particlesize distributions indicate between 20 and 50% ofclay size material and 5 to 15% fine sand. Thein-situ vane shear strengths were corrected usingthe method of Bjerrum2. The corrected vanestrengths are generally consistent for the top7 m at about 10 kN/m2 below which they increasedlinearly with depth with a Cu/Po' ratio of about0.35 to 0.40. The shear strength profileindicates a weathered crust with a depth ofweathering of about 5 to 6 m similar to theprofiles for recent clays in Bangkok2 but deeperthan might be expected for clays adjacent to thecoastline.

12. The results of oedometer tests andconsolidation properties of the soil are shown inFig. 5.13. The pre-consolidation pressures (or criticalpressure) appear to increase with depth althoughthere is considerable scatter of points with theover consolidation ratio varying from 0.3 to 4.5The lower values undoubtably reflect sampledisturbance.

tCWWY LAOOOM*

EA3T WEST

KEY

:••• ;.;.; Sand

C i , Soft marine c l a y ^

• X \ Silt, clayey silt +:e* Sandy clay, clayey sand

••stiff clay

Granite and weathered granite Shale and mudstone and hard clayand/or silt (weathered rock)

Fig.3 - Simplified geological profile

64

0

2

4

6

8Ef 10Q

12

14

16

18

20

0 20 40 60 80 100 0

PAPER 9: GRIFFIN, PENG AND GUDGEON

4 8 12 16 0 4 8 12 16 0 8 16 24 32 40

MCx MCa PL• LL Liquidity index

X

X X

iX*

Denstiy: mg/m3

x Bulk densitya Dry density

V

x 8 '

Vane corrected: KN/m2

Very soft dark brownclayey peat with roots

Very soft olive greensility clay with traces of s

Very soft to soft olive greslightly sandy silty clay

Soft greenish grey togrey silty clay

Very soft dark brown peaV 'silty clay

Slightly sandy silty clay

Fig.4 - Index properties

oc

2

4

6

8

10

12

14

16

18

?n

) 0-51

o

o

° o- a oo

O 0-

a *&o

o oO O Q G

-

o ° ,o

o

-

-

1 1 - 5 2 2 - 5 0 1 2 3 4 Ci i a ' '

a

o

oo

a o

op

P^ o a

o o

o ao

Cc

o Swamp areaa Trial embankment

1 ° x •

x *a x °x

°* x *X

• . ' • • ;X

** a *

« a a

X X

X

X

Coefficient of volume compressibility(x10"?m2/kN)

x Swamp areaa Trial embankment

) 1

° aa

arj

Q « a

a a

aa

aa aa

2 3 4 5

a

oa a

a

a a

a 9

\ a a

a

a

Overconsolidation ratio

a Swamp area• Trail embankment

Fig.5 - Consolidation properties

PRELIMINARY DESIGN14. The scheme called for embankments for thelagoons with a minimum crest height of 2.0 m andfor the embankments around the proposed gypsumlandfill a height of 2.5m. Behind the latterembankments the gypsum is to be stored incompacted layers rising to a height of about12 m. The purpose of these retaining bunds wasto contain run-off from the gypsum landfill anddirect flow to the settlement and monitoringlagoons. The embankments are to be constructedwith completely to highly weathered granite andmetasediments with a central core zone taken downbelow root zone. A drainage blanket 0.5 m thickis to be provided under the shoulders of theembankment.15. The stability of the embankment at variousstages of design was analysed using Sarma's3

method for both circular and non circular failure

surfaces. The strength profile adopted for theanalysis is shown in Fig. 6. At the time of theanalysis not all soils data was available andstrength points showed some scatter. Thestrength profile adopted was generallyconservative.16. The following assumptions were used in theanalysis:1) The marine clay layer was taken to be 16 m

thick.2) Soil profile as Fig. 6.3) Embankment to be built in stages due to soft

sub-soil conditions, stage 1 to 2.0mfollowed by a *rest' period of 1.75 yearswith subsequent raising to 2.5 m.

4) Foundation was assumed to improve with timedue to consolidation.

5) Minimum factor of safety adopted to be 1.40.

65

TAILINGS DAMS

UNDRAINEO STRENGTH ( Cu) (kn /m 2 )

10 20 30 tO 50

Fig.6 - Design Strength profile

17. The embankment profiles adopted forpreliminary design are shown in Fig. 7. This canbe compared with profiles of the surroundinghighway, an example from Bangkok, and thesubsequent trial embankment dimensions.18. The total final settlement profile of theembankment due to consolidation of an underlying16 m thick layer of soft clay was calculated tobe 0.70 m under the crest of 2 m high bank. Therate of settlement was extremely difficult tocalculate since assumptions were based on limiteddata. The coefficient of consolidation (Cv) wasbased on a few field permeability test resultsand oedometer tests. A lower than average valueof field permeability was used in the calculationof Cv to allow for the reduction in permeability,especially in the upper parts of the marine clay,due to the increase in effective vertical stressand subsequent decrease in void ratio. It wastherefore expected that the initial short termsettlement would be faster than calculated. Atthis preliminary design stage the effect ofoverconsolidation of the crust or weathered zonewas not taken into account. Using a value of Cvof 2m2/yr it was estimated that the degree ofconsolidation of the marine clay after 2 yearswould be 27%.

J HIGHWAY AT SITE

TRIAL EMBANKMENT CONSTRUCTION19. Stability analysis showed that the minimumfactor of safety was very sensitive to the shearstrength values assumed for the underlying marineclays. This was particularly true at shallowdepths where the critical slip surfaces arelocated. It was also felt that the strengthvalues adopted were conservative and mayunderestimate the in-situ material strength ofthe clays. In addition due to sample disturbancethere was not great confidence in the degree andrate of settlement calculated for the embankmentprofile. It was therefore decided to confirmdesign assumptions by constructing and monitoringa trial embankment.20. The trial embankment was located close to theproposed area for the settlement lagoons. Fourboreholes were put down at the site for detailedin-situ testing and sampling as described above.The average ground elevation was approximately3.2m above mean sea level.

Site preparation21. Clearing and cutting of the site was donemanually and trees hand felled. Roots were leftin the ground providing an undisturbed foundationfor the embankment.

Instrumentation22. The following instruments were installed inthe foundations of the embankment to monitor thebehaviour of the foundation during and afterconstruction of the embankment:

12 pneumatic piezometers (P)4 surface settlement plates (SP)4 deep settlement plates (DP)20 deformation markers (DM)10 theodolite stations (TS)

23. The layout and sections were shown in Figs.8 to 10

24. The pneumatic piezometers used were suitablefor low and medium pressure range accurate to ±0.2 m head. Prior to installation they weresoaked and tested for leaks on site under apressure of 20 m head. They were installed bylowering into boreholes using a PVC placing tubeand pushed in for the final 300 mm to therequired tip level. The boreholes weresubsequently backfilled with 3 :1 bentonite/cementgrout. The tubing from the 12 piezometers waslaid in a trench and terminated at a terminalpanel and read-out gauge housed in timberbuilding 15 m from the embankment.25. Surface settlement plates were constructed of50 mm dia galvanized steel pipe welded to 600 mmx 600 mm steel plate (5 mm thick). Each lengthof pipe is 1 m long with screwed couplings forextension and is provided with a 100 mm diameter

20 40 60 m

TRIAL EMBANKMENTBANGNA-BANGPAKONG HIGHWAY

PRELIMINARY DESIGN

Fig.7 - Embankment Profiles

66

PAPER 9: GRIFFIN, PENG AND GUDGEON

2-5

®

Slope 1 in 3

Slope'1 in 4 '

®'|# 13® 18®1|

Crest _ DP1B j | J/Slope 1 in 3 DP2 SP21"Z I H Q|I

/Barn/ Slope 1 in 3

49

5®

9®

10®

Q Surface settlement8 Deep settlement plate® Deformation markerA Theodolite station

• Boreholes

7-5m,7-5m ,

16®

176&

Slope

_,SP4

1 in 4

_ \14® 19®^

15® 20®

fis.

j

M-

- Trench for piezometers leadsPiezometers installed on Earth

Temporary building

Fig.8 - Tr ia l embankment plan

I Deep settlement1 plates

T Deformationplates

(ormatioimarker

0 Piezometer1 Surface settlement

plates

Fig.9 - Trial embankment section

as necessary/other length)

50 mm steel pipe (extended as necessary)

20mm steel rod (extended 10mm deep notchas necessary by welding on cut into top of rodothr lenth) /

20 mmSteel rod-

Deformation markerSurface settlement plate Theodolite station Deep settlement plate

Fig.10 - Instruments

steel sleeve. Deep settlement plates wereconstructed of 25 mm diameter high tensile steelrods with a 300 mm diameter steel screw at thebase (Fig. 10). Monitoring of settlement was byprecise levelling and referenced to two temporarybench marks outside the trial embankment area.26. Deformation markers were driven beyond thetoe of the embankment and on the embankment slopeto measure lateral deformation. Measurement of

the lateral movement/displacement of the markerswas by precise sighting with a theodolitereferenced to stations outside the embankmentarea.

Fill Material27. The fill material was well graded verygravelly sand (predominently coarse sand and finegravel) with less than 5% fines. Duringconstruction in-situ density tests were carriedon each fill layer. Bulk density ranged fromabout 1.64 Mg/m3 to 1.96 Mg/m3 with a mean valueof 1.80 Mg/m3. Moisture content was about 5%.

Construction of Embankment28. The embankment was to be constructed inlayers. The first two layers were 0.5 m thickand the remaining 6 layers each of 0.25 mbringing the embankment to its design height of2.5 m. Due to time constraints it was notpossible to allow long periods of time formonitoring between each layer and on average only3-4 days was available between placing layers.The embankment was brought to its final height in35 days.29. Fill was brought onto the embankment by 6-wheeler trucks along an access road made up of0.5 m thick sand. Fill was tipped at the edge ofthe embankment and was spread out using a backpusher and bulldozer of self weight 1 tonne and4 tonnes respectively.

MONITORING OF TRIAL EMBANKMENT30. Generally the reading of all instruments wascarried out on a daily basis. With the exceptionof piezometer P4 and deep settlement plate DP3all instruments functioned satisfactorily.Piezometer P4 was later replaced with a new setof piezometer and leads. Unfortunately betweenthe placing of the 4th and 7th layer thepiezometer read out unit malfunctioned and anumber of readings were lost.

Piezometric levels31. Piezometric levels are shown plotted asexcess pore pressure in Figs. 11 and 12 .Piezometers P5 and P6 placed under the zone ofthickest fill responded during the placing of thesecond layer (0.5 m - 1.0 m) with a maximumexcess pore pressure of 2.5 m - 2.7 m.Piezometer P4 (1 m below original ground level)responded but due to apparent rapid dissipationof pore pressure maximum excess head obtained was0.5 m. A similar effect was noted forpiezometers P7-P9 (Fig.11). The piezometersplaced beyond the toe of the fill (P1-P3 and P10-P12) showed negligible response.

Settlement32. The settlements of the surface (SP1 - SP4)and deep plates (DPI - DP4) are shown in Fig. 13As with the piezometers negligible response wasnoted until the placing of the second layercommenced. Maximum settlements, by day 64 underthe thickest fill section was 0.58 m and 0.64 mrepresenting about 23 to 26% settlement. At adepth of 10 m below the bank settlement was about5%.

67

TAILINGS DAMS

10 20 30Day

40 50 60

Fig.11 - Piezometric levels

3

2-5

# 1-5Q.

•

© 0-5

-0 -510 20 30

Day40 50

Fig.12 - Piezometric levels

Fig.13 - Embankment settlements

Deformation33. Movements of the deformation markers areshown in Figs. 14 t o ig Measurements were takenon movement of the top of the marker and base.During the early filling stages there was sometilting of the markers but after the filling of

the third layer movements were generally uniform.34. Deformation was surprisingly small at the toeof the main slope (DM 2, 7, 12 and 17) withinitial movement towards the bank after placingthe third layer (10-30 mm) followed bydeformation away from the toe. The line ofmarkers 5 m away from the toe (DM1, 6, 11 and 16)showed very little movement after an initialdisplacement of 20-30 mm after placement of thefirst layer except Dll which showed a suddenmovement during the placing of the third layer(maximum lateral displacement of 80 mm).35. Deformation markers at the toe of the lower

-100

Fig.14 - Embankment deformation

100fin

6040200

-20-40-60-80

-100

X

V

Q

O-

-

-

-

-

-

3T8T

13T18T

r1

3B8B

13B18B

>n? A/ o vydwj

V M KMcSPc

, i

// — '

-

fa, V

2-5

2 E

1-5 j

I-0-5

10 20 30Day

40 50 60

Fig.15 - Embankment deformation

slope all showed a positive movement away fromthe bank with a maximum displacement ranging from36 to 78 mm. Five metres from the toedisplacements were negligible after placement ofthe first layer.

DESIGN REVIEW36. The preliminary embankment design was drawnup on the basis of the shear strength profileshown in Fig. 6. This resulted in very flatslopes. The trial embankment was constructed tovery much steeper slopes, 1 on 3 one side and 1

68

PAPER 9: GRIFFIN, PENG AND GUDGEON

-100

f—1

I/ 1 1 1

ftx

V

aoi

4T

9T14T

19T

4B

9B14B

19Bi i

2-5

2

-1-5

- 1

-0-5

10 20 30 40 50 60Day

Fig.16 - Embankment deformation

100

80

60

E 40

^ 20

1 0I 20Q -40

- 6 0

- 8 0

-100

i- x 2Tv 7Ta 12To 17T

-

_ X" " y 0

/ 1

2B7B

12B r~17B I—'

r s£

/-

-1 1 1

2-5

2 E

1-51

- 1

10 20 30 40 50 60

0-5

0

Day

Fig.17 - Embankment deformation

100

80

60

40

20

0

-20- 4 0

- 6 0

- 8 0

mn

X

V

oo

k

I

5T 5B10T 10B15T 15B 120T 20B /

r1

1i i i

/

J

j\ ^

-

1 1 i

2-5

-2 E

1 I- 0-5

10 20 30 40Day

50 60

Fig.18 - Embankment deformation

on 3 with a 2.5 m berm on the other. It wastherefore expected that failure would be reachedbefore the final height of 2.5 m was obtained.37. In the event the embankment did not fail soit could therefore be assumed that the factor ofsafety against shear failure was greater thanunity. The margin of safety was, of course, notknown. A back analysis showed that a shear

strength profile developed from the vane testsdone on the site gave a factor of safety of justover one. However, the embankment did not showsigns of distress in the form of excessivehorizontal deformation that would be associatedwith embankments built on soft clay and close tofailure.38. Previous experience4 with trial embankmentson soft clays demonstrated that a factor ofsafety of at least 1.4 was necessary foracceptable performance. Below this margin,delayed failure was possible. It was thereforeconsidered likely that the trial embankment hada similar margin of safety. For this to be thecase, the strength of the foundation clay wouldhave to show an increase in strength over themeasured vane strengths. Confirmation of thiswas found in the piezometer readings. PiezometerP4, at a depth of 1 m, and under the highest partof the embankment, recorded a piezometric levelequivalent to a pore pressure ratio (ru) of 0.21when the embankment was completed. Since thetheoretical (ru) without consolidation would havebeen expected to approach 0.7 or 0.8, it wasclear that considerable consolidation hadoccurred. A similar behaviour was observed atpiezometer P7, the other piezometer at shallowdepth under the embankment. The piezometers ata depth of 3.5 m also show some consolidation butnon was apparent at the piezometers 6 m deep.This is consistent with the length of therespective drainage paths.39. It may be noted the consolidation describedabove showed as a maximum settlement of thefoundation of about 600 mm.40. The above conclusions were applied to thepreliminary embankment design by increasing thestrength adopted for the top 1 m thick layer ofthe foundation from 5kN/m2 to 7kN/m2 to reflectthe increase in effective stress in this layer.Strengths below this layer were not changed.This modification enabled the widths of the bermsto be decreased by 5 m with a saving of about50,000m3 of fill , justifying and demonstratingthe value of trial embankments.

ACKNOWLEDGEMENTS: The authors are grateful to MrJohn Graham, Project Manager of Tioxide Group picfor permission to publish this paper. We wouldalso like to thank Mr Stevens and Mr Davies forcarrying out the stability analyses and finallyDr Walbancke for reviewing the paper.

REFERENCES:1. COX.J.B. A review of the engineeringcharacteristics of recent marine clays in S EAsia. Research Report No.6, Asian Institute ofTechnology, Bangkok, 1968.2. BJERRUM.L. Problems of soil mechanics andconstruction on soft clays. 8th ICSMFE, Moscow,1973, Vo.3., p 111-159.3. SARMA.S.K. Stability analysis of embankmentsand slopes. Geotechnique, Vo.26. No.3., p423-433, 1973.4. BINNIE AND PARTNERS. Report on foundationtests for Anglian Water Authority (unpublished)1976.

69

10. Tailings deposition predictive computer modelling

P. WATERMEYER and N. A. THOMPSON, WLPU, Ashford, UK

SYNOPSIS

The construction and infilling of tailings depositories are dynamic processes. Operationalefficiency depends on the form, method and control of impoundment. Accurate prediction ofinfilling allows designers to assess: life of depository; reservoir area; spillway requirements;rate of wall construction required; and effect of different deposition modes. Development, testingand application of the WLPU tailings deposition model (TADAM) is described. The program is basedon over fifty years' experience of tailings dams around the world and is designed to maximiseflexibility and minimise data manipulation. Studies at two mines illustrate the macro and microscale modelling potential of the program.

INTRODUCTION1. The confining wall for a tailings

depository is a unique structure, in that it isusually built from the same material that it isdesigned to impound. Finely processed minewaste is mixed into a slurry and hydraulicallytransported to an overland storage site.Discharge of material takes place from discretelocations around the depository perimeter.Material flows away from the point of dischargeand is deposited. The rate of filling andstructural development of the dam wall is thuscontrolled by the operation of the dischargepoints. Tragic failures of tailings structuresin recent years have focused attention on theirsafe design and management.

2. Tailings impoundments are undercontinual construction throughout the period oftheir service lives, typically 30 or 40 yearsor more. Impoundment is therefore a dynamicprocess and the efficiency of deposition willbe influenced by the form that the impoundmenttakes. Prediction of the life of a tailingsdepository can be of great importance to theeconomic viability of a mining project. Tooshort a life of depository will necessitatefuture major capital expenditure to provideadditional storage. Too long a life ofdepository implies inefficient use of capitalduring the cash intensive pre-production phase.

3. Control of the tailings dam pond sizeand location, as well as the freeboard betweenpond surface and dam wall are essentialelements of depository management. Appropriatefilling patterns must be specified by theengineer responsible for technical managementof the tailings structure. In reality, anylocalised encroachment of the pond will bedealt with by the mine operators, who will openor close the appropriate valves, or move theappropriate discharge points, based upon theirindividual experience. This is, however, a

short term solution; long term mis-managementof filling could result in inefficient usage ofavailable storage, or even structuralinstability. The efficient use of theavailable storage is of major importance to theoverall mining operation, and therefore longerterm waste disposal management approaches needto be adopted.

4. The authors have developed a softwarepackage called tfTADAM" to model the hydraulicdisposal of tailings within overland storagebasins. This program provides a method ofpredicting long and short term disposal effectsand is an efficient and cost effective tool forthe design of tailings dams.

DESCRIPTION OF TADAM PROGRAM5. TADAM uses an incremental generation

procedure to generate successive stages in thedevelopment of a tailings depository. Theprogram is written in FORTRAN and runs on a VAX8200 Computer. Run times are normally one totwo minutes.

Input data6. TADAM utilises a topographical database

containing elevations of the "original" groundsurface on a rectangular grid, typically at 25,50 or 100 metre spacings. This database isproduced by digitising the results oftopographical surveys and is stored as a binaryfile named LOWFILE.

7. The position of any confining wall isstored in a separate binary file namedWALLFILE. This minimises the amount of datamanipulation involved in modification of thewall location.

8. The extent of the area available fortailings deposition is defined in a thirdbinary file, called VALIDFILE.

The embankment dam. Thomas Telford, London, 1991 71

TAILINGS DAMS

Output files9. Deposition (or dredging) generates a

new tailings surface which is stored in a newbinary file called TOPFILE. This file isinterchangeable with LOWFILE and will probablybe used as the new base surface for future runsof the program.

10. ZONEFILE contains the zones ofdeposition for each deposition point. The mainoutput listing is stored in the ANSWERFILE,this includes a summary of the volumes andtonnages of deposition classified by depositionpoint and full listings of LOWFILE and TOPFILE.The output listing is cumbersome and is rarelyreferred to; a summary of total tonnage andvolume of deposition is displayed on the screenat the end of each run and y^reen/hard copygraphics facilities are available.

Parameter file11. Operation of the program is controlled

by a simple parameter file which contains theminimum possible amount of data and thereforeallows an experienced operator to run theprogram, assimilate the results and edit theinput data ready for the next run within acycle time that should not exceed five minutes.The parameter file contains: wall and pondelevations; density and slopes of tailings(dredged, subaqueously deposited andsubaerially deposited); and names of the inputand output binary data files.

Slope of deposited tailings12. The behaviour of mine tailings during

hydraulic deposition has been discussed byBlight and Bentel (ref. 1) and other authors.The slope of the tailings beach is influencedby many factors including: specific gravity;coefficient of uniformity; moisture content ofdelivered tailings; depth of phreatic surface;rate of discharge per unit length ofembankment; rates of rainfall and evaporation;and even wind speed. Blight and Bentel comparedimensionless beach profiles for four differenttypes of tailings, these indicate that theprofile of copper tailings beaches is onlyslightly concave and can be approximated aslinear. Diamond, platinum and gold tailingsproduce increasingly concave non-linear beachprofiles. WLPU's database of tailings slopes,accumulated from over fifty years ofexperience, supports these findings.

13. TADAM incorporates flexible andsophisticated facilities for the specificationof tailing beach slopes. At the conceptualdesign stage, linear approximations are oftenused, typically 1:200 or 1:100 for sub-aerialdeposition, 1:30 for subaqueous deposition and1:15 for dredged slopes. Detailed design andproject monitoring normally require the use ofnon-linear profiles, these may be based onlaboratory trials or surveys of existingdeposition or may be selected from an extensivedatabase of beach profiles that have been usedon similar assignments. TADAM's slope databasecomprises surface elevations and tailingsdensities at 50 m increments from the point ofdeposition, intermediate points areinterpolated using cubic curve fitting.

Modus Operandi14. In the first iteration, TADAM examines

each deposition point in turn by evaluatingadjacent grid points to identify file sectors,with that deposition point as origin, wheredeposition is possible. The grid pointsdefining the edges of these sector(s) areentered into the deposition table for the nextiteration as potential deposition points. Thecumulative distance from the real originaldeposition point is also entered into thetable; this distance accurately simulates thereal flow path and is the key to the successfulimplementation of the program. At the end ofthe iteration the deposition table is examinedand points from which no deposition occurredare eliminated. Each subsequent iterationexamines the points one grid step further out.

15. The elevation of the deposited tailingsat each grid point is determined by referringthe distance from the deposition point to thestored beach profile. Deposition can continue,in a straight line, until the elevation of thedeposited tailings would be below the groundsurface or below the pond level. Grid pointswhere these conditions occur are identified as"hillpoints" or "waterpoints" respectively andthese points become virtual deposition pointsfor further deposition.

16. Hillpoints allow the tailingsdeposition to follow the topography of theimpoundment and are probably the mostsignificant improvement over the various"straight line" deposition models.

17. Waterpoints mark the transition fromsubaerial to subaqueous deposition. They areunlikely to lie exactly on the edge of the pondand the program makes an adjustment to allowfor this lying between the grid points.

18. Volumes of deposition are calculated byre-examining the valid area and multiplying thedifference between the new tailings surface andthe original ground surface at each grid pointby the grid spacings. Volumes are apportionedto the real deposition points using ZONEFILE.

TESTING OF THE TADAM PROGRAM19. Durrant (ref. 2) has described an

earlier program, written by the senior author,called "DUMPS" (ref. 3) and his own researchwork using the Medusa Geographical InformationSystem. The Medusa Geographical InformationSystem is a relatively expensive package whichis unlikely to be economically viable formodelling tailings deposition.

20. Durrant has described a series oftests, designed to demonstrate the limitationsof DUMPS and the advantages of a moresophisticated modelling system. His five testcases have been repeated using TADAM, theresults for a single deposition point with are-entrant corner are presented in Figure 1.These tests show that DUMPS (and any othersimple deposition model) is unable to modelhills and re-entrant corners with acceptableaccuracy. TADAM can model these features andis a practical alternative to the MedusaGeographical Information System.

72

PAPER 10: WATERMEYER AND THOMPSON

DEPOSITION POINT

DUMPS

MEDUSA

TADAM

Fig. 1 Contour plots of DUMPS, Medusa andTADAM tailings deposition models fordischarge around a re-entrant corner.

APPLICATION OF THE TADAM PROGRAM TO SOHARCOPPER PROJECT, SULTANATE OF OMAN.

21. The Sohar Copper Project is situated27 km inland from the coast in an area ofvery limited water resources. Its locationis shown in Figure 2. As a result the OmanMining Company LLC have to rely to a largeextent on seawater make-up for the processplant.

22. The mine is required to operate itstailings depository in the Wadi Suq in amanner which will prevent the pollution ofadjacent fresh water aquifers.

23. In 1984 WLPU was appointed to adviseon methods of maximising the quantity oftailings stored in the dam and thedevelopment of a system for recycling thetailings pond water and minimising both theseepage from the reservoir and theimportation of seawater (ref. 4).

24. The Sohar Concentrator produces aboutone million tonnes of tailings each year.Estimates of the size of the existing orebodies indicate that the plant will operatefor about eleven years. At present thetailings dam impounds about 5 million cubicmetres of tailings and a maximum of400 000 cubic metres of free water.

25. TADAM has been used for computermodelling of the filling of the dam with theaims of achieving the optimum depositionpattern and of restricting the passage ofseepage water, high in dissolved solids,throughout the underlying gravels. Waterbalance studies were made in parallel withthe filling trials to determine the scope forrecycling water from the pond and reducingacid generation therein.

26. In 1984 the pond volume was about230 000 m3, the pond level was 245.3 m and theedge of the pond was about 200 metres fromthe crest of the embankment. Computer aidedfilling trials indicated that continuation ofthe filling regime then in use would resultin a pond volume of 1.3 million m3 by December1989 and a pond level of 256.1 m. WLPUrecommended the implementation of pond waterrecycling which was forecast to result in apond volume of 25 000 m3 and a pond level of251.9 m by December 1989.

27. The anticipated advantages of pondwater recycling were:

i) The small pond, distant from theembankment, would result in a large beachwith maximum sub-aerial deposition, and hencea higher average density of tailings.

ii) The small pond would result in alarger volume of the depository beingavailable for the storage of tailings ratherthan water.

iii) The small size of the pond and itsdistance from the embankment would reduce thevolume of seepage loss under the dam wall andhence reduce the pollution risks.

iv) The small pond would result in areduction in the acidity of the pond watermaking it more suitable for recycling.

28. In December 1989 the actual pondvolume was estimated to be 50 000 m3 and itslevel was 250.81 m (ref. 5). The tailingsdam had performed just as predicted by the1984 computer modelling. The pond wasconfined to the north west extremity of thetailings dam and the edge of the pond wasabout 600 m from the crest of the embankmentas shown in Figure 3. The beneficial effectof the smaller pond on the rate of seepage isdemonstrated by the reduction in the rate ofpumping of water from the interception trench(Figure 4.)

73

TAILINGS DAMS

SOUTHYEMEN \

ARABIAN

SEA

n.ooo

10000-

9000"

8000

7000-

6000

5000'

4000.

3000-

2000-

1000

Fig. 2. Location of Sohar Copper Project.

Fig. 4.Sohar Copper Project, volume of waterpumped from interception trench to plant;February 1986 - December 1989.

MAIN EMBANKMENT

NLPU TAILINGS DAH MODEL

Original ground SurfaceHim BUBMaxima

MininMaxim

:ontour elevation 230 «:ontour elevation 261 •at 1.00 • intervalsf deposited tailings:ontour elevation 230 a:ontour elevation 261 •

Contours at 0.50 n intervalsEnbankment —Dredge Point DSubaqueous deposition Point NSubaerial depositionLinear Point LNon-Linear Point N

ScaleHorizontal gridline spacing 25.Vertical gridline spacing 25.

OMAN MINING COMPANY LLC.SOHAR TAILINGS DAM.Predicted f i l l i n g pat tern.Assumed Pond Level 249.9 molume Deposited 253975. cu.m. G. No. 1728/

Fig. 3. Sohar Copper Project TADAM results for December 1989.

74

PAPER 10: WATERMEYER AND THOMPSON

APPLICATION OF TADAM PROGRAM AT ANDINA MINE,CHILE

29. The Andina Division of CODELCO, Chile'sstate copper mining corporation is situated inthe Andes, close to Aconcagua, the highest peakin the Americas. Its location is shown inFigure 5. The division mines and concentratessome 40 000 tonnes per day and has two existingtailings disposal dams in rugged mountainvalleys at altitudes above 2000 m, overlookingthe township and mine support facilities.Disposal is currently to Los Leones dam, aconventional earth and rockfill structure whichrequires to be raised periodically to createadditional tailings storage capacity. Atpresent the dam stands some 160 m high and itsheightening by conventional methods is becomingincreasingly costly.

Coquimbo

Valparaiso

) Aconagua

•Mendoza

SOUTHPACfFICOCEAN

Val d i v i a "v

Cordoba

MINE

ARGENTINA

Fig. 5. Location of Andina Mine.

30. WLPU was commissioned to carry out apreliminary study of alternative tailingsdisposal technologies which might be applied tothe existing sites to reduce costs without lossof security (ref. 6). The sites areseismically highly active and there isextensive agricultural, residential andinfrastructural development downstream.

31. Five unconventional techniques wereexamined. These, and various combinations ofthem, were applied to the two valley sites tocreate nine disposal options, each of whichwould provide an additional 30 years storagecapacity, amounting to some 430 million tonnesof solids. The nine options were developed topre-feasibility level and subjected tocomparative economic analysis. Single-siteschemes emerged more favourably than multipledevelopments and the further raising of LosLeones dam, by the centre-line technique usingcompacted hydrocyclone underflow as fill, wasidentified as the preferred option on bothtechnical and economic grounds.

32 4 The TADAM program was used to model allthe options for tailings deposition in the LosLeones and Rio Blanco valleys. It alsoprovided depth-capacity information for floodrouting studies.

33. Original ground levels were digitisedfrom 1:5 000 and 1:10 000 maps onto four50 x 50 m grids, representing the existing LosLeones and Piuquenes depositories and twoadditional areas. The model was calibrated bycomparing the volume of tailings as "predicted"by TADAM with the actual volume determined froma bathymetric survey. A slope of 1:350 forsubaerially deposited tailings resulted in amodelled volume of 35.18 x 106 m3 whichcorresponded to a measured volume of37.2 x 106 m3.

34. The favoured option is shownschematically in Figure 6 and the TADAM resultsare shown in Figure 7. This option wouldinvolve, for the 30 year capacity adopted asthe basis of the study, the gradual raising ofthe existing Los Leones dam to a final crestlevel of 2200 m, an increase of 90 m over itscrest in Autumn 1989. Raising of theembankment would be by the centreline methodusing compacted cyclone underflow. This optionwould also involve the extension of thefoundation downstream to a new minimum level of1900 m giving a final overall height of 300 m.This would place the Los Leones embankmentamongst the world's highest tailings dams. Thetotal cost of the preferred scheme over the30 year period was estimated to beUS$ 197 million at end of 1988 prices,amounting to US$ 0.456 per tonne of tailings.

EXISTINGDEPOSITION

Fig. 6. Schematic Diagram of Favoured Option.

35. TADAM was also used to determinestorage capacities for an unusual extension ofthis study. This involved an examination ofthe viability of using the raised Los Leonestailings dam for the seasonal storage of40 x 106 m3 of water for controlled release toaugment seasonal low flows to a hydro-electricpower station downstream. The additional costof this option was estimated at US$ 34 million,which was thought unlikely to be matched by thepotential economic benefits.

75

TAILINGS DAMS

DEPOSITION POINTS

POND

DECANT TOWERAND TUNNEL

KLPU TAILIN6S DAM MODEL

Original ground SurfaceHiniaua contour elevation 2000 aMaxima contour elevation 2210 aContours at 10.00 a intervalsSurface of deposited tailingsMiniaua contour elevation 2000 aMaxiaua contour elevation 2205 aContours at 0.50 a intervalsEabankaent •Dredge Point [Subaqueous deposition Point ISubaerial depositionLinear Point INon-Linear Point *

ScaleHorizontal gridline spacing 50.Vertical gridline spacing 50.

DEPOSITION POINT

H—U > Uhl—.

ANDINA MINE.LOS LEONES TAILINGS DAM: OPTION NUMBER 1.Predicted filling pattern on completionAssumed Pond Level 2190.0 mVolume Depo3ited 329.220 Million cu.m. bn6. No. 1830/

Fig. 7. TADAM Results for Favoured Option.

CONCLUSIONS36. The dynamic processes involved in the

construction and operation of tailingsdepositories have been outlined. Therequirement for an efficient and accurate modelof the deposition process has beendemonstrated. Development and testing of sucha model has been described.

37. The tailings dam model utilises linearor non linear beach profiles, digitised basetopography and an unlimited number ofdeposition points, allowing flexibility duringimpoundment trials. Confining wall positionsare. specified in separate data files allowingmaximum flexibility, minimising the need forlaborious alterations to data and enabling avariety of wall construction methods to beassessed. The program uses virtual depositionpoints to model subaqueous deposition andtopographical effects, it is thereforeconsiderably more accurate than the previousgeneration of deposition modelling programs,including DUMPS.

38. Applications of TADAM have includedtailings deposition and flood routing studiesin Africa, Asia, Europe and South America. Theaccuracy and value of the program have beenproven by its success in modelling tailingsdeposition on numerous projects including theSohar Copper Project in the Sultanate of Omanand, on a larger scale, at Andina Mine inChile.

AcknowledgementsThe Sohar Copper Project is owned and

operated by the Oman Mining Company LLC.Andina Mine is owned and operated by the AndinaDivision of Codelco-Chile. The permission ofboth Clients to publish this paper isgratefully acknowledged.

References1. BLIGHT, G.E. and BENTEL G.M. The

behaviour of mine tailings during hydraulicdeposition, Journal of South AfricanInstitution of Mining and Metallurgy, April1983, pp84.

2. DURRANT, A.M. A computer-aidedsimulation of hydraulic tailings disposal.Ph.D. Thesis, Loughborough University ofTechnology, 1988, 134-136.

3. WLPU CONSULTANTS. DUMPS -Tailings DamFilling Trials Program. In-house user manual.1978.

4. WLPU CONSULTANTS. Report to OmanMining Company on the Raising of the SoharCopper Project Tailings Dam. October 1985.

5. WLPU CONSULTANTS. Report No. 7 to OmanMining Company on the Monitoring Programme atSohar Tailings Dam. March 1990.

6. WLPU CONSULTANTS. Inform final delestudio preliminar de la disposicion noconvencional de relaves. Report to Codelco-Chile, Division Andina. March 1989.

76

11. Geotechnical aspects of the construction of tailings dams- two European studies

M. CAMBRIDGE and R. H. COULTON, WLPU, Ashford, UK

The disposal of mine waste behind embankment dams has been practised for many years, but has beenpoorly served in the technical press. Well publicised failures and environmental concern have ledto increasing use of specialist geotechnical engineers in the design process. The safe, efficientand environmentally acceptable deposition of mine tailings relies on engineering, not only of theconfining wall, but of disposal methods. Two case studies are reported which indicate therequirement for an integrated multidisciplinary approach to design and stress the need forflexibility to suit mine development and the adoption of technology appropriate for the disposalenvironment.

INTRODUCTION1. In the initial planning stage of a new

mine .attention is focused on proving themineral reserve and on developing the oreextraction technology necessary to optimiserecovery. During the economic and technicalevaluation of the project therefore tailingsdisposal is invariably a secondaryconsideration. Moreover, the problemsassociated with the disposal of the wasteproduct are often over simplified, andtechnology and design principles from similaroperations adopted. This approach, oncecommitted to paper, invariably leads toimplementation and construction. It is onlyduring full production that the deficiencies ofthe tailings disposal facility become apparent,often necessitating a review of the adopteddisposal method. Major and costlymodifications to the disposal system may thenbe required.

2. The development of a successfultailings disposal scheme involves a multi-disciplinary appraisal of the project in orderto provide the mine with a safe, efficient,economic and environmentally acceptable form ofstorage.Historical Perspective

3. To many mine operators tailings, thecrushed fine waste rock emanating from theprocessing of the ore, have low priority andare of no commercial value. The tailingsproduct for disposal will, as a result of theconcentrating process, invariably be in slurryform often including a suspension of clay sizedparticles. These waste products havehistorically been deposited behind crudeconfining embankments where sedimentation ofthe product takes place enabling 'clean water'to be decanted into the river system orreturned to the plant.

4. For many years, disposal was anunsophisticated process, the demands ofproduction being insufficient to require anybut the most simplistic of structures and, as a

result, geotechnical design input, inspectionand monitoring were minimal. Increased demand,coupled with the processing of lower grademineral deposits, has led to progressivelylarger structures with an inherently greaterhazard from failure. Much has been gained fromthe successful techniques and failures ofearlier tailings disposal projects, and manylessons remain to be learned, particularly withrespect to economic disposal methods, pollutioncontrol and the transfer of technology betweendifferent climatic zones. In recent years moreattention has been paid to the geotechnicalaspects of tailings dam construction which, incombination with increased environmentalconstraints, has led to the employment ofspecialists at an early stage of projectplanning. In many cases such technical inputhas only resulted following well publicisedfailures, and, due to increased publicawareness and environmental concern.

5. The most positive aspect has been theintroduction of legislation controlling designrequirements, inspection and monitoringroutines. This legislation was instigated inthe UK following the Aberfan disaster and hasformed the model legislation adopted in othercountries, e.g. Zambia. All aspects oftailings disposal in the UK now fall under thecontrol of the Mines and Quarries Inspectorate.It is impractical for the Inspectorate to fullyappreciate all geotechnical aspects of tailingsdam construction and it is necessary,therefore, that great reliance be placed on theadvice of specialist geotechnical engineers.However, much of the available information hasyet to be disseminated throughout the industry,and consequently the design of tailings damsstill remains poorly understood by manypractising dam engineers.

6. Many specialist engineers engaged ontailings disposal projects concentrate solelyon the design of the retaining wall and neglectserious consideration of the mechanism oftailings disposition. This philosophy, whilst

The embankment dam. Thomas Telford, London, 1991 77

TAILINGS DAMS

providing a potentially conservative design,often does not provide the client with the mosteconomic method of disposal and fails toconsider the implications on restoration. Theaim of this paper is to demonstrate, by meansof two examples, the need to adopt anintegrated approach to the design of a tailingsdisposal facility.

PRINCIPAL ASPECTS OF TAILINGS DAM DESIGN7. The principal factors to be considered

in the design of any tailings facility are:-(i) Ore Characteristics-geology, mining

method, previous mining on the site(ii) Mill Product - concentration method,

geotechnical characteristics of thetailings, slurry pulp density,throughput, plant operating hours andany future changes in the processingoperation.

(iii) Disposal site - climate, geology,topography, availability ofconstruction materials, hydrology,

concern was the impact of a large tailingsdisposal facility on the Cornish landscape.

(iv)seismicity.Environmentpopulation,constraints

- existing habitats/effluent discharge

(including dust and gasemissions), restoration and both longand short-term environmental impact.

8. In addition, the mine operator willrequire that sufficient storage be provided forthe disposal of both proven and projected orereserves in a safe manner and at a cost whichis not detrimental to the viability of theproject. During the planning stages,particular emphasis should be placed uponproving the environmental acceptability of theproject to local inhabitants and to statutoryauthorities. Finally, the depository shouldhave sufficient design flexibility to enablechanges in grade, ore type, through-put andtotal storage volume to be accommodated.

9. Such design requirements areillustrated in the following case studies. Thefirst demonstrates the need for flexibility,whilst the second highlights the problemsinherent in adopting techniques fromclimatically different regions.

CASE STUDY NO. 1 - Wheal Jane Tin Mine,Cornwall

10. Cornwall has a long history of tinmining stretching back at least 2000 years and,as a result, old workings have a significantimpact on the landscape. The mines to the westof Truro, the area of interest (Fig. 1), wereworked from the 16th Century. 19th Centurymining activity in particular resulted in theproduction of large quantities of waste, muchof which has been deposited in the rivers andestuaries. The resurgence of mining in the20th Century in response to rising metalprices, led to a revival of mining in Cornwalland to a reappraisal of tailings disposalpolicy. The construction of a tailingsconfinement structure in a tourism orientatedCounty necessitated the design of anenvironmentally acceptable structure. Hence,when Consolidated Goldfields proposed thereopening of the Wheal Jane Mine on the site ofthe former Nangiles workings, an initial major

Fig. 1 Clemows Valley Tailings Dam

Design Review11. WLPU first became involved in the

project in 1968, at which stage it was apparentthat planning consent for the mining projectwas dependent on the provision of asatisfactory tailings disposal facility in theadjacent Clemows Valley.

12. An initial appraisal indicated the maindesign criteria to be:-

(a) economic design utilising, as far aspossible, the coarse fraction of thewaste product in the confining walls

(b) landscaping to blend with the surroundsas far as practical

(c) effluent control by the design of asuitable settlement area to enable thesedimentation of the finest particlesand to meet consent limits imposed bythe Water Authority

(d) geology with respect to the provisionof suitable dam foundations, thelocation of old workings and tominimise any possible risk of abreakthrough of mine tailings into thenearby workings

(e) hydrology, including the provision ofappropriate river diversion works andspillways, and the disposal of minewaste (underground) water.

78

PAPER 11: CAMBRIDGE AND COULTON

Particle Size Distribution Chart

BRITISH STANDARD SIEVE SIZESISO 212 300 425 600 f i t 2urn iim itm urn um mm mm fnrn nwii fnrn ftnn% nfiwi

PRIMARYOVERFLOW

GRAOEO UNIT" SLUMPEDSTRUCTURES WITH CLAYEYLENSES. GRAM SIZE MCREASESOOWHVOPOS.PYRITE CONTENTINCREASES DOWNWARDS.

SANDY LATER WITH A LITTLE

SUS-HORIZONTAL LAYER WOOEPMEO BY SUOHT GRAINSIZE ANO COLOUR CHANCE.

MEDIUM TOUOHTGREY. HORIZONTAUYU Y E R E a SLIGHTLYCUYEY SILT. WITHCLAYEY ANO SANDYLAYER. TRACES OfPYRITE ARECONCENTRATED INSANDY LAYERS.

SECONDARYUNDERFLOW

HORODNTAL LAYEMN0 OEFMED8Y SLIGHT COLOUR CHANGES

LOG OF TA IL INGS BEACH

0 05 01

PARTICLE SIZE0 5 10

MILLIMETRES

CLAYFINE | MEDIUM | COARSE

SILT FRACTION

FINE [_ MEDIUM |_ COARSE

SAND FRACTION

FINE J MEDIUM [ COARSEGRAVEL FRACTION

Fig. 2 Typical Tailings Gradings and Structure

Clemows Valley Tailings Dam13. The initial design located the dam

between the surface outcrops of the lodes inthe Clemows Valley, the eventual position ofthe upstream wall being arranged to preventdeposition of tailings above the hanging wallof the productive ore body. The pre-depositionworks included a starter wall constructed fromcompacted earthfill derived from the valleysides, a rockfill toe comprising mine waste, animported crushed granite filter and appropriatedecanting and emergency spillway facilities*The initial design included extensive use ofthe coarse fraction of the tailings product inthe staged construction of the retaining wall.The classified tailings were to provide boththe main structural portion and the freedraining zone within the embankment, Seepagecontrol through the dam section was to beprovided by the coarse tailings, the filtersystem, and finally by the coarse rock toe laidon the valley floor.

14. Two waste materials (Fig 2.) areproduced by the mine comprising a fine overflowand a coarse underflow and are delivered to thetailings dam from the primary hydro-cyclonefacility located in the processing plant. Thefine product, which predominantly consists ofsilt and clay sized particles, exhibits poorconsolidation characteristics and due to theconsequently low effective stress is of limitedshear strength. This material is thereforedeposited at the northern end of the depositoryat a location remote from the confining wall.

15. The underflow from the primary cycloneis pumped to a series of secondary hydro-cyclones located on the dam wall. Here the

coarse fraction is separated for wallconstruction whilst the finer, intermediatesized material is subaerially deposited to forma tailings beach. As the slimes fraction hasbeen removed by the primary cyclones, both thewall and beach materials rapidly consolidateand develop a high shear strength, <t>' values ofup to 43° have been measured in the laboratory.The deposition method results in a laminatedbeach material in which, due to anisotropyhorizontal drainage into the higherpermeability coarse tailings zone is encouragedand, as a result, consolidation is furtherenhanced. The tailings beach material isconsequently predominantly unsaturated and ofsufficient shear strength to enable theupstream method of embankment construction tobe employed with the confining wall partiallyfounded on the tailings beach.

16. Tailings deposited sub-aerially fromthe embankment walls form a gently slopingbeach which consolidates rapidly to a densityof up to 1.5 t/m3, allowing access on foot tothe edge of the pond. In contrast, the finetailings deposited sub-aqueously from thenorthern periphery of the depository, settle toa much lower density of the order of 1.0 t/m3

and possess insufficient shear strength toallow access.

Design History17. This initial design was based on early

predictions of ore grade, processingrequirements, and mining methods. However,within two years of start-up in 1970, majorprocess plant modifications necessitated there-assessment of the structure. As a result ofthe smaller proportion of sand sized material

79

TAILINGS DAMS

available, considerable field testing ot tnehydro-cyclones was undertaken in order toincrease the efficiency of wall sectionconstruction. However, the finer grindproduced by the mill reduced the volume of wallbuilding material available and necessitatedthe revision of the dam section to incorporatea thicker downstream earthfill zone than theoriginal facing proposed for landscapingpurposes.

18. Since the first design review, furthermajor modifications of the embankment cross-section and of construction methods have beennecessary to accommodate subsequent changes inboth mining and milling processes. Thesechanges, which predominantly reflect thevulnerability of the mining operation tofluctuations in world metal prices, areillustrated in Fig. 3 and are summarised asfollows:-

1976- Change in mining method withsubsequent replacement of coarse tailings bya wide earthfill zone to form the mainstructural component, thereby allowing theuse of the coarse tailings fraction forunderground backfill. Inclusion of achimney drain to replace the embankmentdrainage zone previously provided by thecoarse tailings.1977-78 Mine closure resulting in cessationof deposition during change in ownership,continuation of monitoring and preparationof abandonment proposals.1979- Mine reopened and productionrecommenced.1980- Major design review to enable fulluse of the available coarse tailings as aresult of increased production, resulting inthe reduction in the width of the earthfillzone. Reinstallation of the hydro-cycloneson the dam for coarse tailings wall buildingand embankment drainage, leading totermination of chimney drain construction.1988- Purchase of South Crofty Mineleading to increase in mill throughput andthe proposed increase in the final crestheight by 6m to meet long term storagerequirements from both undergroundoperations. The increased coarse tailingsavailable enabled optimisation of earthfillvolumes and a more economic cross-section.

Summary19. To date some 2.5 x 106m3 of storage has

been provided by construction of a 1.4 km longconfining wall to a maximum height of 40m.Each change in mining or milling practise hasbeen accommodated and the Clemows ValleyTailings Dam has provided an efficient disposalsite for some 20 years. On completion ofdeposition to the proposed final crest level of76m, the main embankment will be approximately53m high, providing 5.8 x 106m3 of storage.

20. At the time of writing, due to therecent downturn in the world tin price,underground mining at the site will shortlycease, although processing of South Crofty orewill continue in the short term. Unless metalprices substantially increase, tailingsdeposition at the site may cease. Restorationproposals may, therefore, be required to

CYCLONEDTAILINGS

F I N E :••:[•:•::'•;::: Z.l;:.T A I L I N G S •'':::':•:•'•:.•'

Fig. 3 ChronologicalSection

Development of Dam

provide a suitable long-term scheme oflandscaping and reclamation in keeping with theoriginal permission.

21. The flexibility incorporated into thedesign of the Clemows Valley tailings dam hasallowed the structure to be modified throughoutits life. The use of the waste products in theembankment cross section has been continuouslyreviewed to ensure the economic construction ofthe confining structure and efficient disposalof the mine waste products. The disposalmethod adopted will also allow the relativelyrapid rehabilitation of the site shouldtailings deposition cease.

CASE STUDY NO. 2 - Neves Corvo Copper Mine.Portugal

22. The complex copper ore body at NevesCorvo mine located near Castro Verde inSouthern Portugal is an extension of the"pyrite belt" which runs across the Iberianpeninsular. The Neves Corvo mine wasidentified in the 1970s and was commissionedsome ten years later following establishment ofa partnership between the Portuguese Governmentand RTZ (Somincor). The development of thismine, one of the richest copper ore bodies tobe discovered in recent times, was undertakenon a design and construct basis by acombination of British, American and Portuguesecompanies with WLPU involved during the laterstages of the design process as ReviewConsultant.

80

PAPER 11: CAMBRIDGE AND COULTON

Fig. 4 Cerro do Lobbo Tailings Deposition

Cerro da Lobo Tailings Dam23. The initial design of the tailings

disposal facility was essentially dictated bydiscussions with the environmental authoritiesin Portugal who were concerned at the high acidgeneration potential of the predominantlypyritic tailings. Canadian tailings disposaltechnology was therefore adopted as mostappropriate as the problems associated with theoxidation of similar ores, and the ready supplyof water, have led to the adoption offacilities involving total inundation, ie. sub-aqueously placed tailings remain submerged.

24. The stage 1 Cerro do Lobo tailings dam(Fig. 4) was therefore designed as aconventional 28m high rockfill dam providingsome 3 x lO^3 of storage, an additional7 x lO^3 of storage was to be provided byincreasing the dam height to 35m.

25. Initially, the dam core was designed asupstream sloping but following review of thegeotechnical characteristics and theavailability of the low plasticity clay, a re-design involving a central core was undertaken.For environmental reasons, a zero dischargeoption was selected and only an emergencyspillway constructed. Construction of the damwas undertaken during 1987 and was completed byDecember in time to receive the first tailings.

Tailings Disposal Review26. The WLPU review of the tailings

disposal facility identified a number of majordesign items for further detailed study.However, of prime importance was theimplication on both the deposited tailingsdensity and final restoration proposal ofadopting subaqueous disposal methods. Inconsultation with Somincor a major study wastherefore put in hand to investigatealternative more efficient tailings depositionmethods which would remove the onerous longterm water demand on abandonment, improvedensities and provide an environmentallyacceptable method of operation. WLPUinstigated a programme of field and laboratorytesting to assess the potential improvements intailings disposal density which could beachieved without significant detriment to theenvironment.

27. Initial laboratory studies indicatedthat the subaqueous deposition of tailingswould result in the formation of a low densitydeposit requiring in excess of 300 years toconsolidate under self weight. Littleconsolidation would have occurred at the end ofthe 20 year depository life and capping torestrict oxidation and permit satisfactoryrestoration would be impractical for manyyears. The Company were therefore faced withthe prospect of maintaining the depositoryflooded in perpetuity with the associatedprovision of long term make-up water to replaceevaporation and seepage losses.

28. The period of exposure required toallow access for restoration and capping couldbe substantially reduced by adopting subaerialdeposition methods. Deposition of the tailingsabove pond level improves density, increasesstorage efficiency and accelerates the rate ofconsolidation as a result of both improveddrainage and increased effective self weight.In climatically favourable areas the depositeddensity is also enhanced by evaporation of freewater from the tailings surface.

29. The principal disadvantage in usingsubaerial deposition for the disposal ofpyritic tailings is the risk of acid generationdue to the oxidation of iron sulphide.Research in Canada (Ref 2 and 3) has shown thatpyritic tailings can be satisfactorilydeposited subaerially providing the tailingssurface is only exposed to atmospheric dryingfor a limited period. Appropriate disposalmethods can limit exposure of the tailings

FALL FALL

OPEN 150 DIA.END PIPE

TEE INVERTPIECE HOLES

Fig. 5 Spray Bar Apparatus

surface by undertaking systematic deposition ofthe slurry by means of spraybars or similartechniques.

30. A typical spray bar is shown in Fig. 5and comprises a 40 m long length of 150 mmdiameter tubing with holes drilled at regularintervals along the invert. Tailings are fedinto the spray bar via a valved tee connectionlocated in the centre of the perforated pipe.The apparatus is mounted above the tailingsbeach by means of either timber posts orscaffold tubes. Flow into the spray bar andthe gradient of the spray bar limbs arecarefully adjusted to achieve a near uniformdistribution of slurry throughout the length ofthe apparatus. Operation will then result inthe uniform distribution of tailings slurryover the beach, effectively producing sheetflow within which the velocity is sufficientlylow to allow sedimentation of both coarse andfine particles, thereby minimising the amountof particulate matter reporting to the pond.

81

TAILINGS DAMS

31. Depending on the tailings productionrate, a series of 5 to 10 spray bars would beoperated for a fixed period on alternatesections of the beach allowing each area ofdeposited material to consolidate as a resultof both drainage and surface evaporation. Aseach layer is of the order of 50 to 100 mmthick, consolidation rapidly occurs and by thetime the next layer of slurry is depositedconsiderable densification has occurred.

32. Although the technique is principallyused to enhance the deposited density, themethod is also advantageous in controllingoxidation of pyritic waste. For the depositionof the Neves Corvo tailings, the controllingfactor is, therefore, the control of acidgeneration, improvement of density being afurther benefit.

33. To demonstrate the applicability of thetechnique in the semi-arid environment atSomincor, a series of both laboratory and fieldtrials were undertakenLaboratory Studies

34. The subaerial consolidation of atailings layer was simulated in the laboratoryby measurement of moisture content, density,shear strength and pH in a series of 50 mmthick slurry samples, comparative evaporationrates were derived from a control containerfilled to the same depth with water. The mainconclusions drawn from this ntudy were:-

35. The laboratory results indicated thatsignificant enhancement of tailings densitycould be achieved without the generation of anacidic tailings mass. To confirm these resultsa series of field trials were undertaken in atributary valley of the tailings dam. Siteevaporation rates indicated that, depending onthe season, between 2 and 8 mm of evaporationoccur each day. Allowing a suitable margin foroperating error, a deposition cycle of between2 and 5 days was proposed. However, toaccommodate monitoring and operationalrequirements, a 7 day cycle was adopted.

Spray Bar Trials36. A series of field tests were undertaken

between August and October 1989 at which timethe test programme was prematurely curtailed byabnormally high rainfall flooding the trialarea. Measurement of pH and moisture contentwere carried out, where feasible, on a dailybasis as summarised in Fig. 7, which showsaverage pH remaining above 7 throughout.Accurate measurement of dry density in thesemi-solid tailings was extremely difficult,and an indication of density was inferred fromthe moisture content density relationship,further testing is in hand to confirm the fieldvalues.

(i) The pH of the initially alkalinetailings samples remained above pH 7until the samples cracked (Fig. 6), atwhich point the pH rapidly decreased.The onset of cracking thereforeprovided a good indicator forsubsequent deposition.

(ii) Densities of the order of 1.85t/m3 wereachieved at the commencement ofcracking, some 85% higher than theinitial slurry density of 1.0t/m3.

(iii) Approximately 20 mm of evaporationtook place from the water filled panprior to the commencement of cracking.

2 1

2 0 -

1-8 -

16 -

1 4 -

, 0 .

0 8 -

0 6 -

0 4 -

0 2 -

0 -0

BLEEDPOINT

\ 1

//

10

C R A C K P O I N T ^

i t - " ' ' ^^DRY

X

X

20 30PAN EVAPORATION I

DENSITY

H

40 50mm)

6C

•11

10

• 9

• 8

• 7

- 6

• 5

• 4

• 3

2

1

40 , RAINFALL mm

2o]

n 7 PH7 U 21 28 4 11 18 25 2 9 16 23

VlV^ ..A-V -- ^ ^..r7 U 21 28 4 II 18 25 2 9 16 23

40-j MOISTURE CONTENT • /•

30

20-

7SPRAY

U 21 28

BAR IN OPERATION

II 18 25 2

U 21 28 . 4 II 18 25 . 2 9 16 23AUGUST H-« SEPTEMBER ++•— OCTOBER -

Fig. 7 Spray Bar Field Trial Results

37. The results of the field test clearlyindicate that controlled subaerial deposition,by means of spray bars, could be undertaken,without the generation of an acidic tailingsmass. In addition, the estimated tailingsdensity for the study period which variedbetween 1.55t/m3 and 1.95t/m3 provided anaverage value of some 1.8t/m3.

Fig. 6 Laboratory Evaporation Results

82

PAPER 11: CAMBRIDGE AND COULTON

Summary38. The attainment of densities of this

magnitude throughout the life of the depositorywould allow the storage of an additional 6 x10 t of tailings over that by subaqueousdeposition and, more importantly, should enablethe company to adopt an alternative restorationscheme for the depository.

39. The tailings deposition techniqueadopted at Somincor, although adequate to meetshort term disposal commitments, will inhibitrestoration. The subaqueous method ofdeposition selected is well suited to theclimatic conditions associated with most ofCanada, where long term water supplies forrestoration by inundation are freely available.The method is, however, less suitable for usein a semi-arid climate in which evaporationexceeds precipitation and water supplyinvolves costly pumping from remote reservoirs.The field trails indicated that subaerialdisposal may provide a more appropriate methodof tailings deposition for this mine.

CONCLUSION40. The design of a successful tailings dam

not only involves the consideration oftechnical aspects of the retaining embankmentbut also the implication of the selected

disposal method on both final restoration andoperational flexibility. To achieve both ofthese objectives the mechanism of tailingsdeposition must be understood, carefullysimulated by laboratory testing and, whenfeasible, by field trials and continuouslyreviewed during the life of the disposalfacility.

ACKNOWLEDGEMENTS41. The Wheal Jane Mine is operated by

Carnon Holdings Ltd and the Neves Corvo Mine bySociedade Mineira de Neves Corvo S.A(Somincor). The permission of both Clients topublish this paper is gratefully acknowledged.

REFERENCES1. Forbes P.J. and Cambridge M.: The designand construction of tailings dams in Cornwall,Presented to Geological Society EngineeringGroup Regional Conference, September 1980.2. Haile J.P. and Kerr T.F. Design andOperation of the Myra Falls Tailings DisposalFacility, Vancouver Geotechnical Society's 4thAnnual Symposium on Geotechnical Aspects ofTailings Disposal and Acid Mine Drainage, 1989.3. Ferguson K (1989) Peronal Communication.

83

12. Spillway systems for tailings dams

P. J. FORBES, S. A. CALE, and L. F. CLELLAND, WLPU, Ashford, UK

During its operating life economics normally dictate that a tailings dam is constructedprogressively, just keeping ahead, with an appropriate freeboard, of the deposited waste andsupernatant water. The dam thus normally requires a spillway with a crest level which is at thecurrent reservoir surface and can be varied as an operational procedure to suit depositionprogress. The role of this structure, often termed a "decant" is crucial to successful disposal,safety and environmental protection. Some of the commonly used decanting structures are described,with reference to design points and pitfalls which have been experienced with each.

INTRODUCTION1. The main object of tailings dam

construction is the storage of solid wastebrought in from outside - and not, as in waterstorage dams, to catch as much natural runoffas possible. The target in siting a tailingsdam is therefore usually to reduce naturalrunoff into the reservoir to a minimum. Asecondary objective, for economic reasons, isto provide a ratio of storage capacity toembankment volume that is as large as possible;a site with the main embankment located in avalley, and therefore across a water course,normally gives a high ratio. In very flatterrain both objectives can be satisfied bychoosing a site remote from a valley, resultingin a non-impounding reservoir with embankmentsaround the full perimeter. Occasionally a non-impounding reservoir can be created in a valleysite by diversion of the river around or underthe reservoir. Usually, however, the planningresults in a cross-valley structure and acertain amount of natural runoff into thereservoir has to be accepted. Where permittedby the location of the concentrator, a sitecommanded by a small catchment area is selectedto keep to a minimum the spillway dimensionsand provisions for flood absorption and tolimit the amount of water which might becontaminated by the waste product.

2. The spillway facilities provided for atailings dam have the twin duties of evacuatingfrom the reservoir the surplus liquordischarged into the reservoir with the tailingsand the natural rainfall runoff. For theformer duty, even with the addition of waterfrom direct rainfall, the dimensions of thestructure, if the reservoir is non-impounding,may be quite modest and are often determined bythe practicalities of construction andmaintenance, rather than by the designdischarge. In some projects in Canada theprincipal method of decanting is through afilter drainage blanket specially located onthe embankment or natural surface for thispurpose. In impounding reservoirs the design

The embarkment dam. Thomas Telforcl, London, 1991

discharge originating from natural runoff maybe two or more orders greater than thatoriginating from the tailings liquor.

3. As with any spillway, those servingtailings dams must be of such a capacity thatthe water rise does not encroach on thefreeboard provision. This is particularlycrucial where the embankment is constructedusing part of the tailings product itself.Deposited tailings is highly erodible and anyovertopping may lead rapidly to complete damfailure and be followed by liquefaction of thecontained solid material. Additionally it isnecessary to avoid the possibility of the waterrise resulting in the establishment of atemporary subsurface flow regime such that thepore pressures exceed the limitations imposedby considerations of stability and piping.

4. Spillway facilities for tailings damsoften take the form of a service spillway fornormal discharge and an auxiliary or"emergency" spillway for the more severe andless frequent flood conditions. The formerfacility is usually termed a decanting systemor "decant" - a reflection of its duty to allowfree passage of the surface water, whilepreventing the passage of the stored solids.

5. In this paper some of the features ofthe arrangements of spillways for tailingsdams, particularly those relating to decantingsystems, are discussed. Examples are includedof typical structures, together with some ofthe problems encountered.

CONCEPTUAL ARRANGEMENTS6. One of the major differences between

reservoirs used for water retention and thoseused for tailings storage lies in the rate ofconstruction. The dam for a water retainingreservoir is usually constructed to its fullheight in a relatively short period and theduty of the structure remains relativelyconstant thereafter. The spillway arrangementscan consequently be designed as permanentstructures remaining in place and operableduring the life of the reservoir with a minimum

85

TAILINGS DAMS

of operational control and with appropriatemonitoring and maintenance. The constructionof both the dam itself and the spillway isnormally fully supervised by engineers.

7. Some tailings dams are similarlyconstructed to their full height at the startof the mining operation, particularly where thecatchment area and therefore natural runoff arelarge in relation to the reservoir capacity.For the majority of tailings dams, however,this approach would be quite uneconomic andwould result in front end capital expenditureon construction which might only benefit theoperation in 10 or 20 years. Since the life ofthe mining operation itself may be uncertainand controlled by market forces, suchexpenditure could be entirely wasted.

8. The normal approach is to construct theretaining embankment of the reservoir either instages of modest height increment or even,particularly where the construction makes useof the disposable product, as a graduallyrising wall, keeping ahead of the depositionlevel within the reservoir by only the marginneeded for freeboard, flood storage and anoperational safety allowance. This approachaffects decanting system design in two ways -the sill level has to be capable of beingraised in very small height increments and themethod of raising it has to be relativelysimple and capable of being managed byoperatives with minimal supervision.

9. The siting of the decanting system andthe level of the sill affect the pattern ofdeposition profoundly (ref. 1). If it is tooclose to the embankment the less permeablebarrier created by the deposited fine tailingsis reduced in effectiveness. If it isincorrectly located in relation to thedeposition points, or if the sill level israised excessively, the deposition level mayrise too fast in relation to the embankmentconstruction along a portion of its length andthe density of the deposited material may bereduced, depleting the future storage capacity.If the level is set too low the pond of waterabove the deposited tailings may be reduced toa size too small for the sedimentation of thesolids to take place effectively.

10. The following types of decantingsystems are those most frequently used to caterfor the needs of tailings reservoirs which arenon-impounding or subject to modest naturalinflows:

Towers11. This type of system, illustrated in

Fig. 1, is probably the most commonly used. Itconsists essentially of a vertical towerdesigned to be raised within the reservoir anda sub-horizontal conduit passing under theembankment at or below foundation level, eitherat the bottom of the valley, for systemsinserted at the start of deposition, or higherup on the flank, for systems inserted later inthe life of the depository.

12. The tower may be constructed to itsfull height at inception, with ports in thesides, closed sequentially as the level ofdeposition in the reservoir rises to preventingress of solids. More often, however, it is

86

constructed of steel or reinforced concreteannuli of modest depth, added progressively toperform the same function. The rings aretransported to the tower either by raft orusing a pontoon arrangement and a floatingplatform is normally constructed around thetower to facilitate their erection. The systemhas been employed traditionally in miningoperations since the early days of miningactivity, when civil engineering played noactive role in the operations. The integrityof the structure has varied widely, with manyearly tailings dams having failed as a resultof the collapse of the structure.

13. The conduit portion is similar to thebottom outlet of water retaining dams and hastraditionally been constructed of steel piping,concrete piping or, more recently, of either ofthese encased in a structural concretesurround. Corrugated metal conduits have alsobeen used. The reliability of the conduit hasagain been variable and many failures haveoccurred involving piping adjacent to the lineor migration of solids into the conduit, as aresult of collapse or joint separation,followed by chimney development at criticallocations in the embankment.

D A M PARTLY CONSTRUCTEDFROM C L A S S I F I E D T A I L I N G S

TOWER RAISED ASTAILINGS ELEVATION RISES

\TAILINGSBEACH

P O N D

7 1 7 TT"7 \

OUTLET CONDUIT

Fig. 1 Tower System

Chutes.14. This type of decanting structure

differs from the tower arrangement in that thereceiving portion takes the form of a chute setinto a natural slope of the reservoir oroccasionally into the embankment. Asillustrated in Fig. 2, slabs are used to closethe overt of the chute progressively as thetailings deposition level rises. The conduitportion of the arrangement may be connected tothe bottom of the chute, as in the towerarrangement, but the opportunity is alsoafforded in this case for additional conduitsto be connected at upper levels. This may beconvenient for high pressures to be avoided inthe chute itself or in the conduit. The lowerportion of the chute and the lower conduit maythen be sealed off and filled.

15. Where there is a conveniently steep andregular natural slope this arrangement hasadvantages over the tower arrangement inrespect of its ease of operation. The robustsection and thickness of slabs required toresist the pressures and the greater lengthinvolved, however, tend to make it moreexpensive.

16. A variation on the chute system is the"piccolo" decant which normally comprises asteel pipe with numerous tee-pieces. This islaid on the ground and is used where steep

PAPER 12: FORBES, CALE AND CLELLAND

sided valleys permit access to the line at alltimes and in a convenient position with respectto the proposed filling pattern and poollocation. Ports are sealed with blank flangesas the tailings level rises. This is one ofthe earliest forms of decanting systems andemploys simple pipe technology which is usuallyreadily available at most mines.

SETTLEDTAILINGS

PONDSURFACE

R.CCHUTE

R.COUTLETCONDUIT

Fig. 2 Chute System

Stoploas17. Where saddles are available or the

natural slopes under the embankment areshallow, small channel-type spillways, fittedwith, grooves and stoplogs, may be used fordecanting (Fig. 3). These have the merit ofpermitting avoidance of the uncertainties ofthe security of structures below the depositedtailings and embankment. For economy, however,it may be necessary to construct a large numberof these to serve the full range of depositionlevel in the reservoir, since the cost of eachrises exponentially with its height. Inaddition it is necessary to ensure thatadjacent portions of the embankment aredesigned to full water retaining standards tocater for the presence of the pool, which forthe other systems may be located remote fromthe embankment. This may substantiallyincrease the cost of the embankment,particularly where tailings construction isbeing employed, and this type is often reservedfor the final arrangement constructed at thetime of decommissioning of the tailingsdisposal facility.

CHANNEL PASSING THROUGHLOW PORTION OFEMBANKMENT OR THROUGHNATURAL GROUND ON THEFLANK

SETTLEDTAILINGS

| I(R.C.OUTLET CONDUIT| ALTERNATIVE FOR

1 | HIGHER EMBANKMENTS)

I LL./

WEIRBOARDS

Fig. 3 Stoplog System

Pumps18. While somewhat stretching the

definition of "decanting systems", the use ofpumps as an alternative to the provision offixed free flow structures is frequently thebest, and at times the only possible, method ofwater evacuation. Pumps may be mounted on araft (Fig. 4), with the discharge and accesssupported by a ' floating causeway, or on araisable framework attached to the reservoirflank.

FLOATINGOUTLETP I P E

S E T T L E D T A I L I N G S

Fig. 4 Pump System

19. The advantages lie in the ease ofrelocation to suit the prevailing depositionconditions; the clear water pond may be movedin this way further from the embankment or fromside to side to raise or lower local depositionlevels or to increase the width of theimpermeable deposition zone. Where theembankment is located in a steep, narrow-sidedvalley or where the foundations areinaccessible, as in the case of a site wheredeposition has already started or where anexisting decanting system is no longerfunctional, a pump arrangement may be the onlyoption, either as a temporary expedient or asthe permanent solution.

20. The disadvantages of the system includethe cost of the energy to run the pump, theinconvenience of having to have operationalcontrol at a possibly remote site, thelimitations imposed by the capacity of the pumpand the possibility of pump or power failure atcritical periods.

21. The prevention of suspended solids fromemerging with the discharge is often less easyto control since the intake pipe draws fromdeeper in the water than do the sills of free-flow decanting arrangements. Small auxiliaryreservoirs may be needed to control this if thewater pond is to be kept small and, where thepump is located on the flank of the reservoir,earthfill training walls may be needed toachieve a similar effect.

22. Pumps have been used most successfullyin large non-impounding reservoirs with hightailings discharge rates such as that at theEast Rand Gold Operation in South Africa and innon-impounding reservoirs where the dischargedwater requires treatment before discharge tothe river systems.

23. The increasing need to return waterfrom tailings dams back to the processing plantmeans that pumps have frequently to be sited atthe dams in addition to or instead of otherdecanting systems. Floating pumps make themaximum use of available head in the reservoir,

87

TAILINGS DAMS

a percentage of which will be lost if water isdischarged to a fixed pump station sited belowthe dam.

ENGINEERING ASPECTS24. Normal principles have been applied to

the design of decanting structures,particularly where their engineering has beenentrusted to civil engineers. Some features ofthe design, however, are worthy of specialattention and among these are the following:

Hydrology25. Where the reservoir ie large in

relation to the catchment area the shape of theinflow hydrograph becomes of lesser importanceand the criterion has often been to ensure thatthe reservoir is able to absorb the largestanticipated storm over a given period, withoutan outflow allowance, whatever the timedistribution of the precipitation. Theessential criterion then is to ensure that thedischarge is evacuated adequately before thenext storm and precipitation/duration/frequencycurves are employed.

26. The question of the allocation of anacceptable risk cannot follow exactly thepattern often used for the construction periodof water retaining dams. At each incrementalheight and at each incremental change in thereservoir storage characteristics a new floodassessment must be made. The risk cannot,however, be related to the time in which thesenew conditions exist but rather to thesummation of the different sets of conditionsover the entire life of the project, which maybe 30 years or more. A conservative selectionof design precipitation conditions thus resultswhen the frequency method of analysis is used.In most cases, however, the criterion is basedon probable maximum precipitation. The highlyerodible nature of the tailings material andthe disastrous consequences of overtoppingsupport the need for this conservatism.

27. In many of the tailings dams in whichthe embankment is constructed of tailings thesurface of the less permeable zone is in theform of a beach extending out from thestructural portion of the embankment. Thefreeboard must be measured downward from thetop of the beach at the lowest portion alongthe periphery of the embankment. Encroachmentabove this level could set up interim untowardflow patterns within the pervious outer shell,even if the exposure is short lived. Where theouter wail is composed of the total tailingsproject, in which segregation and layering hasoccurred, the result can be equally serious andthe catastrophic failure of the Bafokeng Dam inSouth Africa is reported to have resulted fromsuch a condition.

28. Where the natural inflow is large orwhere the reservoir flood absorption isinadequate to prevent encroachment on thefreeboard, the decanting system is normallybacked by an emergency spillway cut into theflanks of the valley. Minimal erosionprotection is usually required for thesestructures, since each is provided usually foronly one season, and it is accepted that therewould be some erosion damage, if they were to

come into operation, provided the erosion couldnot extend to the embankment. A precipitationcondition assessed to have a 100 year returnperiod may be used as the threshold above whichan emergency spillway would come intooperation.

Hydraulics29. Since the decanting level must be

continually altered as the elevation oftailings in a depository increases, thehydraulic performance of the decanting systemis also continually changing. It is usuallyimpractical to provide a hydraulicallyefficient structure, such as a tower with abellmouth entry; instead attention is normallygiven to ease of construction and operation.However, a number of useful features have beendeveloped for incorporation in these systems toimprove performance and a selection aredescribed in the following paragraphs.

30. The hydraulic performance of a typicaltower and conduit system is illustrated inFig. 5 and is seen to be divided into threeflow regimes with, respectively, weir, orificeand pipe control.

31. Weir flow occurs at low heads and theflow is controlled by the inlet arrangements.In this flow region the highest discharge perunit head is achieved and it is therefore theregion in which the dam should operate. Thelow head over the crest of the tower permitsadditional tower sections, or rings, to befitted relatively easily.

32. With a circular tower, should theheight of water over the crest rise above aboutone quarter of the internal diameter, the nappewill meet in the centre to form a boil. Whenthe height of water reaches about half theinternal diameter weir flow will have ceased.Orifice flow then occurs and can be unstabledue to air entrainment. Gulping and surgingcan occur, depending on the configuration ofthe tower and conduit. This flow region istherefore best avoided.

33. If a further increase in water leveloccurs the tower and conduit become full ofwater and pipe flow takes place. In thisregion control switches to the outlet and the

INCREASINGCREST ELEVATION

PIPE CONTROL

ERRATICFLOW RANGES

DlSCHARGE

Fig. 5 Discharge Characteristics in TowerSystems

88

PAPER 12: FORBES, CALE AND CLELLAND

increased frictional forces . result in arelatively small increase in flow for acorrespondingly large increase in head. Largeflood events can therefore cause a large risein water level against which it must be ensuredthat the dam embankments are secure. Thisemphasises the need for a conservativeassessment of the likely floods.

34. Consideration must also be given to thehydraulic characteristics of the horizontalconduit and care must be taken that this hasadequate discharge capacity to cope with theweir and orifice flow regimes.

35. A typical inflow/outflow hydrograph isgiven in Fig. 6 and illustrates the lowdischarge capabilities of a small decant andthe need for adequate flood storage or anemergency discharge spillway. The practicaldifficulties and cost of raising an emergencyspillway, contiguous with a decanting system,usually means that attenuation is adopted asthe normal method of flood control. Adequatefreeboard at all times is therefore required.

36. The construction of towers from eitherreinforced concrete rings or flanged steel pipesections has been described. The diameter ofthe overflow weir is equal to the diameter ofthe tower and thus limits the length of theweir crest for weir controlled discharge. Oneway of improving this is to attach a largerdiameter pipe section to the top of thedecanting tower, effectively increasing theflow into the structure for a given height ofwater above the sill. The larger section canbe moved upwards progressively as subsequentrings are added. This larger diameter pipesection can also provide a useful secondaryfunction; by raising it some distance abovethe pond level, the new rings may be attachedto the tower without the disturbance of flowingwater•

INFLOW 1 IN 100 YEAR STORM

5 10 15 20 25 30 35 40 45 50HOURS

Fig. 6 Typical Hydrographs

Tower Sections37. The rings used to form the tower

structure are normally constructed from precastconcrete or flanged steel pipe sections. Theseshould be light enough to be manoeuvrable butstrong enough to resist the stresses imposed bythe tailings.

38. The tower is located in a medium whichis subject to an increase in thickness withtime and therefore continues to consolidatethroughout the active life of the depositoryand generally for some considerable time afterdecommissioning. It is thus necessary for thetower to resist the downdrag to the extent ofthe limiting sheath friction or to be able tobe compress to accommodate the downdrag once

the tailings has settled into a solid state.Fine tailings has a high compression index butthe coarser material has a moderate value.Some success has been achieved by the use ofrubber annular inserts between the concrete orsteel sections. The estimated compression ofthe contiguous tailings is matched with thecompression in the rubber such that thevertical stress imparted by the rubber to theconcrete or steel ring results in a load notgreater than could be sustained by the section.An essential feature of the arrangement is thatthe rubber has to be restrained from inwardmovement•

39. At the Wheal Jane Mine in Cornwall theinitial decanting arrangement, set up sometwenty years ago, comprised a tower formed ofprecast concrete rings with 45 mm rubberinserts and a conduit section passing under theembankment (Fig. 7). The rings were boltedtogether through the inserts. Regularmonitoring of the compression (Fig. 8) wasundertaken and it was found that thecompression agreed reasonably well with thatestimated, although the rate of deflectiontended to be irregular with some suggestion ofintermittent sudden relaxation of stress.

PONDSURFACE

\ / \ 7 \Fig. 7 Wheal Jane Decanting Tower

40. The deformed shape of the inserts wasrather different from that anticipated andthere was pronounced bulging of the rubbertoward the inner face of the tower (Fig. 9).The inserts were glued to the concrete on oneside (to the new ring being added); it was ofinterest to note that the shear stress exceededthe glue bond and on this face the insert slidtowards the centre of the tower to the extentthat the bolt restraint would permit. On theother, unglued, face the friction resisted theshear and there was considerably less sliding,illustrating the superior bond between theuntreated faces of rubber and concrete.

41. These towers are gripped in thetailings downward from about 2 m below the

89

TAILINGS DAMS

HORIZONTAL DEFLECTION VERTICAL D E F L E C T I O NOF RUBBER INSERTS DEFLECTION OF TOWER

OF INSERTS

10 20 30 40 10 20 0 200 £00 600 800

DEFLECTION mm

Fig. 8 Monitored Tower Deflection

CONCRETE DECANT

RINGSFORCE

o NO LOAD APPLIED

FORCE

b. PARABOLIC DEFORMATION

FORCE

c.BOND FAILURE CAUSING"LIP'TYPE EXTRUSION

d.TOP ANDBOTTOM BONDFAILURE RESULTING IN

"MUSHROOM'TYPEEXTRUSIONS

Fig. 9 Rubber Insert Deformation

surface of the tailings and their deflection inthe horizontal direction is thus controlled bythe strains in the mass of material.

42. In the latter stages of its life theWheal Jane tower started to show a marked tilt(Fig. 8) which brought forward the date of itsreplacement with a new structure. The reasonfor the movement was not determined, since thecost of detailed examination could not bejustified, but it was considered that a likelycause was the collapse of ancient shallow mineworkings a short distance from the tower,resulting in the horizontal movement of themass, to fill the void.

43. In other projects, where the careneeded to emplace the rubber inserts was notlikely to be achieved, the tendency has beentoward structures of shorter life. Steelsections have been used with thin rubbergaskets and the height of the tower has been

90

limited, generally to less than 20 m. A newtower and conduit is then started.

44. An alternative arrangement has beendevised to cope with the continuingconsolidation without the need for the rubberinserts. The tower is constructed helically toa spiral shape as illustrated in Fig. 10. Thissystem has not yet been tested but would besuited best to a rapidly rising deposit ofconsiderable final height and small waterdischarge requirements, where the additionalcosts of steel in the structure would notreduce its competitiveness with other systems.

Fig. 10 Helical/Spiral Tower

Sub-horizontal Outlet Pipes45. The design of the conduit associated

with the tower and chute systems involvesdealing with the same phenomena as areencountered in the design of bottom outlets forwater retaining dams. The tensile stress setup by the horizontal strain generated by theembankment construction has to be overcome bythe conduit being made suitably ductile or ableto slide relative to the surrounding ground;the variable compression of the foundation hasto be allowed for; consideration has to begiven to the positive projection conditiongenerated by the different stiffnesses of theconduit and the adjacent ground; and attentionhas to be paid to the durability of thestructure with respect to its projected life.

46. The question of the durability of theconduit is perhaps less emphasised in decantingsystems than in water retaining dam bottomoutlets since the conduit constructed for theformer purpose frequently has a limited usefullife, at times of only a few years duration,after which the conduit is sealed and filled.The other design considerations, however, areemphasised particularly strongly in tailingsdams, especially in those using tailings forconstruction, since the material is highlyerodible and the smallest breach in thewatertightness of the conduit can lead rapidlyto internal erosion and failure.

47. In many projects the target has been tofound the conduit well below the surface of the

PAPER 12: FORBES, CALE AND CLELLAND

natural ground and, where possible, to seat iton sound rock. The positive projectioncondition can thereby be countered and both theelongation and differential settlement reducedor eliminated. An additional safeguard toeliminate joint opening has been the encasementof the pipe structures in reinforced concrete,generally quite heavily reinforced in thehorizontal direction.

48. A section used in a project in PapuaNew Guinea, where the bore of the conduit waslarge and the structure was required towithstand the load imposed by some 80 m oftailings, is illustrated in Fig. 11.

T A I L I N G S .

Fig. 11 Conduit under Heavy Loading

49. One of the methods used to reduce theeffect of the elongation has involved theprovision of a coating on the outside of theconduit, as illustrated for a project in Zambiain Fig. 12.

BITUMEN SLIP LAYERS(MINIMUM THICKNESS 10 mm)

mwmmm

.STEEL PIPE(610mm DIA)

/ I 7 \ 7 \ 7 T 7 T

Fig. 12 Friction Reduction around Conduit

EE"" 0-

s so3 100UJ

-• 150

CONDUIT INVERT

1100

CHAINA6E (m)ISO

Fig. 13 Conduit Displacement

50. The conduit portion of the arrangementin Cornwall, referred to earlier, was foundedon partly weathered rock and considerable carewas taken in ensuring its integrity inaccommodating the variable settlementanticipated. At the end of the life of thestructure the vertical displacements, asdetermined by closed circuit TV monitoring,were found to be as illustrated in Fig. 13.

Rehabilitation51. There is an ever greater emphasis on

the provisions for tailings dam decommissioningand statutory authorities now require proposals

to be formulated in considerable detail. Thespillway arrangement is a particularlyimportant feature of the decommissioning and isexpected to be trouble-free thereafter, with aminimum of maintenance and monitoring.Arrangements involving conduits and towers arenot generally favoured to form the finalspillway and it is normally necessary for themto be replaced by a conventional side channelarrangement or for the last stages ofdeposition to be served by a stop-logarrangement, which can be left in a sound anddurable condition with a minimum of capitalexpenditure at the decommissioning stage.

REFERENCESl.WATERMEYER P. and THOMPSON N.A. TailingsDeposition: Predictive Computer Modelling.BNCOLD 6th Conference on the Embankment Dam1990

91

13. Clay mining waste disposal problems - central and peripheral

C. D. ROUTH, A. STREET, and C. F. J. GRIGG, MRM Partnership, Bristol, UK

SYNOPSIS The planning and design of residue disposal schemes for the mineral processing industryraises a variety of problems for the designer, and operator. This paper identifies some of theeconomic, engineering and physical constraints which face those working in the mineral industry,with special reference to the disposal of china clay tailings.

INTRODUCTIONl.For the purpose of this Paper, Mining

Waste Products are divided into three broadgroups: Coarse; Intermediate; Residues. Theproblems and solutions discussed in this Paperare based largely on experience of thedisposal of waste materials produced by themethod of mining china clay employing highpressure water jetting and wet processing.However, a number of the problems associatedwith the safe disposal of the waste products -particularly the residues - and the solutionsto these problems described in the paper applyalso to residue disposal associated with otherproducts.

2.The Paper concentrates on the disposal ofthe finer waste products, or residues, behindembankment dams, because it is in the handlingand safe disposal of these materials that mostof the problems arise. Some mention is madeof the disposal of intermediate and coarsewaste materials (particle sizes ranging fromsay medium sand through to fine or mediumgravel, and including overburden). In civilengineering terms, the disposal of coarsewaste products (ranging in size from coarsegravel to up boulders) is, from the miningengineer's point of view, a matter ofselecting the most appropriate, economical andconvenient plant, and from the designer'spoint of view a matter of making the best useof such material in the design andconstruction of retaining structures for thefine residues.3.Because the proportions in which the three

groups of waste occur and arise from themining operation is determined by the geologyof the mine area, and because theseproportions are likely to vary as the mine isdeveloped, the design of the most economicaloverall disposal scheme has to take account ofthe mine development programme. Because thisin turn may be conditioned by unforeseeablemarket changes, the overall design has to beas flexible as possible in order toaccommodate changes in the proportions ofconstruction materials available from themining operation.4.From the foregoing general observations,

it follows logically that the design problem -

The embankment dam. Thomas Telford, London, 1991

as far as the storage of solid materials isconcerned - commences with, and dependsentirely upon, the safe storage of fineresidues. The availability of intermediateand coarse by-products may be seen as animportant parameter in this basic designproblem; and the disposal of any excessintermediate and coarse materials, over andabove that required for fine residue disposal,can be superimposed upon the residue disposalscheme, conditioned by purely physicalparameters such as landscaping, optimal landuse and, occasionally, subsequent sale orother purposes.

COARSE WASTE PRODUCTS5.Depending upon the geology of the site, it

may sometimes not be economically worthwhileseparating the coarse from the intermediatewastes. In other cases, particularly in theearly stages of development of a mine or pit,the preoccupation of the mine manager with allthe problems of opening a new site -development of access roads, siting of officesand works, and problems of incomplete landacquisition, can lead to the construction ofall-in heterogeneous spoil tips comprising amixture of overburden, coarse and intermediatewaste materials stripped from the claydeposit, in such a manner as to deprive thedesigner of the fine residue disposal schemeof the opportunity to make the best use ofthese individual ingredients in the residueretaining structures. In either of thesesituations, there is a clear case for planningthe fine residue disposal scheme as anintegral part of the overall long-termdisposal scheme, at the outset, as distinctfrom the time when mineral processing hasactually commenced, perhaps several monthsafter the initial clearance of topsoil,overburden and any rock overlying in thedeposit.6.The disposal of coarse waste materials

alone, as loosely defined earlier in terms ofparticle size, presents few civil engineeringproblems. As these materials compriseessentially rock particles having a minimumsize roughly equivalent to coarse gravel,

93

TAILINGS DAMS

there may not be the need for elaborateadvance testing in order to obtain reasonableparameters for a central design and it is safe

to assume that such materials will have aminimum value of ' of about 35° to 40°. Thepossible need to crush some material in orderto produce drainage or filter materials forincorporation into the fine residue disposalscheme should be borne in mind and theirdurability under load or in contact withchemical laden process water should beconsidered, although the double-handling andprocessing of the drainage materials for suchuse has been largely obviated - in the UK atany rate - by the use of geotextiles, with orwithout built-in collecting drainage systems.

INTERMEDIATE WASTE PRODUCTS7.As defined earlier, these include the

sand/medium gravel components of the wastesabstracted in the first of the clay refiningprocesses. By contrast with the subsequentrefining and handling processes, carried outby hydraulic methods, the intermediatematerials are usually abstracted, andsubsequently handled, by purely mechanicalmeans.

8.The designer of the fine residue disposalscheme will probably wish to utilise the sandelement in order to form part of the retainingworks for the fine residues. He willtherefore require

(a) that it be kept separate from theoverburden

(b) possibly some advanced testing of itsengineering qualities for designpurposes

9.As regards sand tips, which may (or maynot) form part of, or abut, the residueembankment in a composite scheme there may bea tendency to assume that the material is bothnon-cohesive and free draining; and to deducefrom this that the material may safely bedumped or tipped to a more or less indefiniteheight, at any angle less than or approaching'. This is by no means a safe assumption,

not only because imperfections in the refiningprocess may result in a significant claycontent, or solely because the actualpermeability after placement may prove lowerthan expected, but also - and perhaps mostimportantly - because the possibility ofpartial or total liquefaction in certaincircumstances cannot be casually disregarded.Both research, (refs. 1 and 2) and experiencehave indicated that 'pure1 sand tips candevelop flow slides which can only beattributable to some form or other ofliquefaction process.

10.Empirically it may be deduced frompublished accounts of flow slides that theywill not (or, more precisely, may not) occurwhere the exposed slope is at a tolerablyshallow angle, perhaps in the order of 20° orso. It is suggested here, however, havingregard to the quite frightening properties offlow slides (which may commonly come to restat an angle little more than 2 or 3°) thatempirical rules of thumb are only adequate if:

(a) the designer had considerableexperience of the material he is

94

dealing with; and(b) not only is the design height of the

proposed tip limited to a maximumvalue well within his experience, buthe is also absolutely certain thatthe tip will not be raised for sometime after his departure.

11.There has been in recent times at leastone significant failure of a sand tip duringor immediately following heavy rain and theremay be other examples. This has prompted theAuthors to ponder another aspect of the designof intermediate waste disposal works, namelythe cumulative effect of heavy rainfall oninternal stability of sand tips. They do notclaim to have pursued this question in anydepth, beyond wondering whether the laminarstructure of a compacted sand tip, coupledwith the comparative readiness with which itmight be expected to admit rainwater from itssurface, might not in certain circumstancespermit the build up of surprisingly highporewater pressures within the lower portionsof the tip body.

12.In composite schemes where co-disposal ofintermediate wastes and fine residues takesplace it is, therefore, absolutely vital thatstability analyses should be carried out onthe complete cross section, as eventuallyconceived, as well as on intermediate stages.At the risk of stating the obvious, what maybe quite reasonably perceived as a relativelythick outer layer in the early stages willdwindle in relative thickness, in relation tothe height, as the scheme develops, renderingthe outer layer relatively insignificant, andits strength irrelevant to the stability ofthe whole mass on completion. However obviousthis may appear, it has probably constitutedone of the principal causes of general failurein bygone years as generations of site foremenhave conscientiously and confidently compactedwhat appears to be adequate retaining bundswhose relative size progressively shrank aslift succeeded lift (and foreman succeededforeman).

FINE RESIDUE DISPOSAL13.We turn now to the principal topic of

this Paper, namely the residue disposal works,or 'tailings dam' although quite frequently,the design engineer will be satisfied if theretaining structure acts as a filter ratherthan as a dam. It may be helpful at thisstage to contrast some of the problems arisingin the early stages of development between afine residue disposal structure and anorthodox dam constructed for the purpose ofretaining water, for whatever purpose.

Planning and site constraints14.Having 'prospected' the site for a water

storage reservoir, the legal framework existsin which statutory powers can usually beinvoked to obtain all the land necessary tocomplete the scheme, subject of course topublic consultation. By contrast, a miningcompany has no such powers and the search andidentification of mineral resources, as wellas the acquisition of the land necessary inorder to mine and process the mineral and on

PAPER 13: ROUTH, STREET AND GRIGG

which to dispose of waste products, must beachieved against a background of planningcontrol, environmental studies, land ownershipconstraints and public goodwill generally.15.The disposal of waste products earns no

money for the mining company. The economicsof mining dictates that disposal must beachieved as economically a possible, usuallymaking use only of the materials availablefrom the mine and as far as possible,expenditure on retaining structures deferred,for obvious economic reasons, as long aspossible. This last point represents one ofthe most powerful reasons for selecting a•step by step1 method of construction, bycontrast with most water storage reservoirs(granted that examples do exist of the latteralso being developed in more than one stage asdemand grows).

16.The foundations for the disposal schememay very well not be ideal, because the verysoil which is attractive as a mineral ore maywell have second rate geotechnicalcharacteristics, simply because it is a claybearing material.

Hydrological considerations17.It has been suggested in an earlier

paragraph that the design engineer may welloften be satisfied if his so-called 'tailingsdam1 simply acts as a safe and stable filter.This is not always the case and it issometimes necessary for the tailings dam toact, to a greater or lesser extent, as areservoir for the reuse of water for thepurpose of mineral processing. A water-balance study , taking account of whateverhydrological data may exist for the site, isthus an integral part of the project planningas a whole, but in overseas countries, thelack of adequately documented hydrologicaldata may be a problem, not only in relation tothe overall water balance for the miningprocess, but also in the design of adequatemeans of coping with flood inflows.

Water control structures18.Since the cheapest type of site for fine

residue disposal (as with water storagereservoirs) is typically a natural valley, itfollows that the optimal tailings disposalsite will be situated in a natural watercoursehaving its own catchment area. The design ofsuitable spillway arrangements may be to someextent simplified by permanent streamdiversions or catchwaters; on the other handthe problem is in certain respects morecomplicated that the design of spillways forwater storage reservoirs by the fact thatoverall economics of the mining developmentnormally dictate progressive development ofthe retaining bunds in preference to theinitial construction of a dam to the fulleventual height. The engineer must thereforedevise a series of spillways, or other meansof safely releasing floodwater, at asuccession of increasing levels as thedisposal scheme develops. Since both tailingsand, often, the materials forming theretaining bunds, are highly erodible, theprevention of overtopping is every bit as

vital as in the case of a traditional earthembankment dam impounding a water storagereservoir. At each stage in the tailings damdevelopment, the spillway works must thereforebe adequate and yet, each having a limiteduseful working life, must be designed with aview to reasonable economy.19.It is fairly typical to make use, as far

as possible, of a common wastewater channeldownstream of a series of spillways. This canincorporate the necessary drop-structures atthe downstream end of the perimetercatchwater, designed to minimise flood flowsonto the surface of the tailings dam.

20.The design of stream diversions orcatchwaters and, to some extent, ofintermediate-level spillway structures,presents the design engineer with an almostphilosophical problem concerning the selectionof appropriate parameters for the 'designstorm1. Both instinctively and logically itwould seem uneconomical to design temporarywater control works, having a useful life ofperhaps two to five years at most, to copewith the same long-term flood risk as theeventual completed scheme.

Abrasive suspended solids21.The fine residues themselves, carried to

their disposal lagoon in suspension in water,are almost certain to be highly abrasive. Itis the experience of the Authors that themining company, with its many years ofexperience, need hardly look to itsconsultants for advice or expertise in thedesign or manufacture of the necessarypipelines. However, in the design ofancillary works, such as catchwaters,diversions and drop structures, the designengineer must be keenly aware of the abrasivenature of even the relatively smallconcentrates of fine material still carried insuspension in overflows and possible 'natural1

surface flows draining toward the lagoon.

22.Indeed, in one case in South Devon, wherea public road diversion incorporated areinforced concrete culvert carrying surfacewater which had been partly contaminated bytailings upstream, considerable abrasiondamage was suffered in the culvert invertconcrete, notwithstanding the fact that theflow velocities were not particularly high.23.In the case of some diversion and

catchwater channels, the channel cross sectionmay be designed with a relatively small invertchannel incorporating extra, sacrificial,concrete designed to take normal or dryweather flows with a substantially wider crosssection above, with more sparing use ofconcrete, to take the larger, obviously moretransient, design flows.

Physical design of the trailings 'dam*24,Using the traditional nomenclature, the

design of tailings dams has often beenclassified in three groups or types, viz (i)upstream method; (ii) centraline method; (iii)downstream method. Although each of thesenames may be something of an oversimplification, they are acceptable as a basisfor the comparison of three approaches to the

95

TAILINGS DAMS

step-by-step development of tailings lagoons.Although perhaps almost too familiar to thoseconcerned with tailings dam design, they arereproduced for ready reference in Figure 1.Because of the many and varied site-specificconstraints affecting the progressivedevelopment, both of the mine and of thedisposal area - some of which have beenbriefly referred to earlier - any generalisedcomparison between the three types ofconstruction should be read with some cautionand may well be invalid for particular sites.

Downstream method25.An embankment constructed by the

downstream method encroaches progressively onthe land downstream of the original starterbund and full account must be taken of this inthe layout of any stream diversions oroverflow works discharging into any originalwatercourse downstream.

26.Since the downstream slopes aresuccessively covered by fresh lifts, thesystem clearly does not lend itself toprogressive planting.

Tailings DischargeBeach Formation

CINTRKLINI METHODTailings Discharge

B«och Formation

Beoch Formation

UPSTREAM METHODF i g u r e 1 - SCHEMATIC EMBANKMENT CONSTRUCTION

27.Because virtually the whole of thecompleted dam section will have beenconstructed of either intermediate or coarsewaste fill, the downstream slope isconditioned by the shear strength parametersof the coarser wastes and can therefore ingeneral be steeper than in the case of theupstream method.

Centreline method28.Centreline construction may be

advantageous, for example if the retainingbund or dam has to be constructed betweenabutments comprising two well-defined narrowspurs of land, such that a shift upstream ordownstream would increase significantly thelength, and hence the bulk, of the retainingstructure.29.As with the downstream method, this

system depends upon the ready availability ofrelatively large quantities of intermediate orcoarse waste fill for the downstream shoulderof the completed cross-section; moreover, forpractical reasons either the rate of yield ofcoarser wastes, as the mining processdevelops, must closely match the rate ofrequirement for successive lifts of fill

material, or there has to be a convenientsource of excess waste material readilyavailable within a fairly short haul distance.30.Also in common with the downstream method

any catchwaters or overflow works must becarried clear of the ultimate positioning ofthe downstream toe of the completed section.Whilst this point may appear obvious, it issometimes less obvious that the full extent ofthe disposal scheme as originally envisagedmay be subject to subsequent alteration byreason, say, of production changes or economicfactors, in which case the existence of apossible expensive spillway channel and dropstructure immediately downstream of theoriginally proposed profile may be anembarrassing and unwelcome restraint on theadditional development of the dam.

Upstream method31.Upstream construction as currently

practised by the Authors and their colleagues,is essentially based upon fairly commonplacemethods of stability analysis using measuredvalues of shear strength parameters, forecast(and monitored) values of porewater pressureand incorporating reasonably tight control ofthe method of introducing the residue into thelagoon from the upstream face of the temporaryretaining bunds.32.Typically, the finished overall slope of

such a development will be in the order of 11or 12 degrees. The retaining bunds or atleast most of the upper ones, can besatisfactorily constructed utilising thepartly consolidated fine residue from thebeach near the perimeter (consolidated bydesiccation) although the first one or tworetaining bunds have to be constructed usingintermediate or coarse wastes, or acombination of the two. To prevent migrationof the fine residue particles through theseearly starter bunds, geotextiles or gradedfilters are frequently incorporated on theupstream face of the starter bunds and somedegree of under drainage, also utilising man-made fibres, is occasionally required.33.As remarked earlier, there is empirical

evidence suggesting that failures byliquefaction, or flow slides, may not beexpected in cases where the overall downstreamslope is substantially less than, say, 20°.Nevertheless, modern developments may bethought to come close to, or to exceed, thelimit of practical experience of upstream-method construction, and, following a study ofresearch work published elsewhere onliquefaction generally, it was considered thatfurther research work was justified,specifically related to china clay tailings,but of probable application elsewhere, toestablish, if possible, the essentialparameters confirming the susceptibility ofthe stored residues to liquefaction.34.In the upstream method of construction,

overall stability obviously depends upon theshear strength of the deposited fine residueand upon maintaining as low a phreatic surfaceas possible. For both reasons, it is highlydesirable or virtually essential, thatresidues should be introduced to the disposal

96

lagoon from the retaining bunds in acontrolled manner and the supernatant waterkept as far from the perimeter bunds aspossible.

35.Similarly in general terms, the natureand permeability of the foundations of a newscheme over which the designer may have littlecontrol are of course highly relevant to thelikely eventual flow net set up.

Environmental considerations36.There is a far greater need these days to

consider the environmental impact of anydisposal scheme than previously, with everyindication of stricter restraints in thefuture. These restraints include, forexample:

(a) visual impact - the need to develop alandform sympathetic to thesurrounding landscape

(b) progressive restoration, includingearly development of vegetation andtree cover

(c) Quality of supernatant water (wherenot required for recycling andprocessing) discharged to naturalwatercourses. Hitherto, control haslargely been restricted to limitingsuspended solids and pH control;analyses and control of otherelements may well be required infuture

(d) dust control(e) control of surface erosion and the

contamination of surface drainage

PAPER 13: ROUTH, STREET AND GRIGG

flows which may themselves havealready been contaminated in passingthrough or near the process plant.

37.From design considerations, thedesirability of the formation of beaches nearthe perimeter bunds, encouraging partialconsolidation by desiccation and the lowestpossible phreatic surface has already beenemphasised. Perversely, this can give rise tothe considerable environmental disadvantage ofdusting in dry windy weather, not only on thebeach itself but also the exposed surfaces ofthe upper retaining bunds before theestablishment of grass or vegetation cover.System planning and budgeting, and theprovision of access, should take account ofthe periodic need for surface watering bybowsers.

REFERENCES1. HIRD, C C and HUMPHREYS, J D. 1977.An experimental scheme or the disposal ofmucaceous residues from the China ClayIndustry. Qv Jonl Eng. Geology, Vol. 10, pp.177-194.

2. HIRD, C C and HASSONA, F A K. 1989. Somefactors affecting the liquefaction and flow ofsaturated sands in laboratory tests.Engineering Geology, 28 (1990) pp. 149-170.

97

15. Gale common ash disposal scheme - concept, design andconstruction

E. T. HAWS, P. L MARTIN, and R. A. N. ORANGE-BROMEHEAD, Rendel, Palmer andTritton, London, UK

General aspects of ash disposal are followed by particular application to the major Gale Commonscheme. The design of the 51m high embankments is presented in the context of the geology,materials available and environmental protection. Construction methods are described.

INTRODUCTION1. About 20% of the content of British coal

burned in thermal power stations remains asash. A 2000 MW power station is capable ofburning up to 100,000 tonnes of coal in a weekand records show that in practical terms thesepower stations produce around 600,000 tonnesof ash per annum.

2. Modern thermal power stations are sitednear the coal fields, and in the mining of thecoal to fuel the stations, around 60% of thematerial brought from the face is discarded.For a large productive mine, there can be morethan 1,000,000 tonnes of waste per annum.3. The ash from the stations is of two

types, -pulverised fuel ash - (pfa) andfurnace bottom ash - (fba).

Pfa is generally composed of uniformly gradedspherical silt sized particles containingwithin it around 2% hollow glass balls(cenospheres) which appear as 'floaters' whenthe pfa is disposed of into lagoons.Fba is generally composed of well graded sandand gravel particles and is in demand forconstruction drainage purposes.

Principles and Philosophy4. Ash disposal is wholly an overhead for a

power station and consequently its cost mustbe minimised. Where disposal is adjacent tothe station the ash can be disposed of inconditioned form by truck or conveyor.However, for more distant sites the cheapestform of transportation is generally as aslurry in a pipeline. This means lagooning atthe disposal site, and dam structures.

5. Economics dictate that the pfa and fbabe used for dam building and while trucking ofconditioned material is possible it may not beenvironmentally acceptable or economical. Thedewatering of the slurried ash at the site isthe alternative, achieved by vacuum filtrationto near optimum moisture content. Fba isproduced separately and in relatively smallquantities and is normally transported byroad. It should be noted that the excavationof lagooned material for embankmentconstruction is not practicable

The embankment dam. Thomas Telford, London, 1991

with modern stations due to the separationof the fba from the pfa. With separated pfare-excavation is difficult as the loosesaturated silty material tends to behavethixotropically when disturbed.

6. Materials of lower permeability thanpfa may be required in ash dam constructionfor seals. Because of the relationship ofmodern large thermal stations with locallydedicated pits, colliery waste (shale) isoften the cheapest available material.

7. The pit/power station relationship alsogives rise to the common occurrence ofmining subsidence under ash dams andlagoons. Additional factors thereby imposedon design are:

- dealing with deformations,strains, and cracking

- allowing for 'earthquake' shockunder vigorous subsidence, egliquefaction of hydraulically placedpfa.

8. It is the authors' practice to arrangedisposal schemes so that lagooning can besuspended in a lagoon subject to activesubsidence as defined by an upper limit oftensile strain.

9. The CEGB and successor companies havemaintained a policy of having their ash damsdesigned, constructed and certifiedaccording to the requirements of theReservoirs Act.

10. Environmental protection is of greatimportance in ash disposal and proceduresare described in the accompanying paper,(Dennis et al).

11. Because of the pressure on economics,some features may be adopted in excess oftheir use in conventional dams. Examplesare:

- geotextiles and membranes- "fir tree" construction, where the

upstream shoulder is founded onearlier deposits of slurry placedpfa (triangular instead oftrapezoidal dam section; liquefactionmust always be guarded against),flexible pipelines laid directly in

99

TAILINGS DAMS

fill of internal bunds, connectingdecant towers.

12. Further features distinguishing ash damsfrom conventional water retaining structuresare as follows:-

- The phreatic surface under steady seepagethrough an homogeneous embankmentretaining saturated slurry, is concaveupwards instead of the normal convex.This is because a vertical through the"reservoir" is a flow line, not anequipotential, head loss occurringthrough the lagoon ash.

- The construction period is veryprolonged. This requires great carewith maintaining setting-out systemsdesign manuals etc, but normally permitsexclusion of the 'end of construction'

f analytical case.f- The retained pfa is sensitive to piping! under high exit gradients, and is notl1 self-filtering.

Furthermore ash dam factors of safety tend tobe at the lower limits of acceptable norms.

APPLICATION TO GALE COMMON13. The primary purpose of the scheme is the

disposal of pfa from two 2000 MW powerstations, Eggborough and Ferrybridge C,located some 5 and 7km distant. Additionally,up to lm tonnes p.a. of shale (mainly coarsediscard with some pressings) from a nearbycolliery at Kellingley is disposed of on thesite. The ash is pumped to site as slurry andthe shale is delivered by lorry. Stage I,started in 1967, is nearing completion, andnow contains more than 30m cu.m. of material.

Preparatory works for Stage II are welladvanced (see Fig.l of Dennis et al).

14. The scheme disposes of the pfa by thefilling of lagoons with slurried ash and bythe use of conditioned pfa as the mainconstruction material for the lagoonembankments, part of the slurry being driedto near optimum moisture content in a vacuumfiltration plant.

15. The prime use of the shale is to formby fir tree methods, the division bundbetween the pair of lagoons forming eachstage of the development together with the'water' faces of the enclosing embankments.However, it is available in such largequantities that in Stage 1 it was used toform most of the embankment abutting thesucceeding phases of the scheme, and inStage II there is to be a core of shaledownstream of the chimney drain where itsgreater density enhances embankmentstability and improves safety againstprogressive failure.

16. The lower levels of the Stage IIembankments will be built in full widthhorizontal lifts, each 3.3m in depth. Thewater face will be inclined at 1:2 at lowerlevels and will be nearly vertical at upperlevels comprising a series of fir tree liftsbuilt out on to the previous lifts of ashsettled in the lagoon.

17. The profile of the Stage II embankmentis shown in Fig.l. The embankment materialsare placed according to normal earth dammethods and standards. Lagoons are filledalternately.

KEY TO GEOLOGY1.UPPER SANDIwhere present)"silty fine SAND*

2AJPPER CLAY'laminated siUy CLAY*

3. LOWER SANO and BASALSAND cr.d GRAVEL

TYPICAL STAGE H SECTION4.LOWER CLAY/StLT

(where present)5. BEDROCK-UPPER PERMIAN MARL

and SHERWOOD SANDSTONEWATEH PRESSURE ASSUMPTIONS SHOWNIN BOXES.

| r _ _ =3

| I I PIPE FINGER DRAIN

atJ f

n «£ I j _ STONE FINGER DRAIN I

5i l IL .

I | I i - D R A I N COLLECTICU\r\/ & MEASUREMENT

\ SANO ^ -UUUI IL tTION * . j IRON PIPE

PFA > N . PFA

IMRERMEABL.£ / X \ l^t

- t

TYPICAL HALF SECTION ON

PIPE FINGER DRAIN

TYPICAL HALF SECTION ON

STONE FINGER DRAIN

PLAN

F i g . l . Typical Stage I I embankment detai ls

100

PAPER 15: HAWS, MARTIN AND ORANGE-BROMEHEAD

18. For full protection of the underlyingaquifer against leachate, thfe wholi lagoon bedof Stage II is covered by an impermeablemembrane. This protection also extends underthe chimney and finger drains. In order toexpedite draining down of the lagoons uponcompletion of slurry placing, a dendriticpattern of drains on top of the membrane leadsto the outlet culvert, controlled by valves.Some 4 or 5 lifts of the embankments will bebuilt before slurry placing and the lagoonsare flooded with water initially to protectthe membrane.

THE SITE19.Gale Common lies on the south western

margin of the Vale of York at National GridReference SE 535 215. To the west there is alow NW-SE trending escarpment rising to over+30m O.D., passing through Cridling Stubbsnorthwards towards Knottingley. The RiverAire passes within approximately 4km of thesite to the north, with the Aire and CalderNavigation and the M62 Motorway situated inbetween, on low lying ground which extendseastwards away from the escarpment towards theHumber estuary. A Yorkshire Water Authorityaqueduct crosses the site on the west side ofthe main lagoons.

20. The site is generally flat, lyingbetween approximately +6 and +8m AOD with aslight fall from NW to SE. Especially in thearea of Stage I, it is naturally marshy withpatches of peat deposits. The main sitesurface drainage consists of a perimeterditch.

Geology21. The site lies well to the south of the

limit of Devensian tills in the Vale of York.It does, however, lie within the area occupiedby Lake Humber in Devensian times andinvestigations have shown that below a thincover of Flandrian Stage peat or clay theentire site is underlain by lacustrine siltyclay and interbedded sands. Betweenthese deposits and bedrock are sands andgravels of probable fluvial origin.

22. As shown in Fig.l the superficialdeposits across the site typically consist ofthe following sequence:-

Deposit

Upper SandUpper ClayLower SandLower Clay/SiltBasal Sand and Gravel

Thickness

0-1.5m0-3.7m0-2.7m0-1.7m2-6.0m

23. The solid geology comprisesCarboniferous Middle Coal Measures at depth,overlain by Upper and Lower MagnesianLimestone and Permian Marl and partly by theSherwood Sandstone Group. The boundarybetween the Sherwood Sandstone Group andPermian Marl subcrops diagonally across GaleCommon striking approximately NW-SE. ThePermo-Triassic strata dip gently eastwardswith the Magnesian Limestone forming theescarpment to the west of the site where

easterly dips of up to 5° are recorded.East of the escarpment bedrock is concealedby the superficial deposits. Major faultsfollow a WSW-ENE trend with thePontefract-Knottingley trough fault passingnorthwest of the site. A secondary faulttrending WNW-ESE underlies the southern partof the site.

24. The Magnesian Limestone, SherwoodSandstone Group and the basal sand andgravel are all important aquifers as takeninto account in the design of the works.

25. The site lies within the YorkshireCoalfield and is underlain by several seamsof coal. Coal has been extracted from a1.5m thick seam at a depth of some 700mbeneath Lagoons A and B (Stage I) and miningof a 2.3m thick seam has recently beencarried out at about 660m beneath theemergency lagoons where over 2m ofsubsidence has been observed. It isanticipated that mining beneath the Stage IIlagoons will eventually take place.

26. Groundwater is generally within 1 to2m of ground level, ^lery shallowgroundwater gradients have been observedwith flow generally to the south east in thebasal sand and gravel aquifer. There isevidence for a small upward hydraulicgradient and slightly artesian groundwaterconditions in the sand, gravel and bedrockdying out from west to east across the site.

27. Gale Common lies within a region ofmoderate seismic activity for the UnitedKingdom where Lilwall (1976) (Ref. 1)calculated Intensity 5-6 ground motionshaving a return period of 200 years.

GROUND INVESTIGATION28. Extensive ground investigation has

been carried out. The Stage I investigationincluded 24 percussion borings and 28shallow auger holes. In 1974 excavation anddetailed inspection of pits and trenches wasalso carried out in Stage I. (Taylor et alRef. 2.)

29. The 1982 investigation for Stages IIand III was carried out in two phases. Phase1 comprised 38 boreholes, 6 trial pits, 2trial trenches and 10 static conepenetration tests. The boreholes were sunkusing cable tool percussion methods orflight augers through the superficialdeposits. 3 boreholes were extended intobedrock by rotary core drilling. Soilsamples were obtained for logging andlaboratory testing. In selected holes theboring was progressed without the additionof water so that high quality samples ofgroundwater could be obtained and field andlaboratory chemical analyses were carriedout. 19 standpipe piezometers wereinstalled.30. Phase 2 included the sinking of some 117

probes to establish the thickness of the upperclay across the site.and the installation of afurther 4 standpipe piezometers forhydrogeological monitoring purposes.

101

TAILINGS DAMS

GEOTECHNICAL DESIGN31. A typical profile of the Stage II

embankment is shown in Fig. 1.32. The geotechnical design is governed

principally by the Upper Clay layer which hasa relatively low strength compared with theother foundation soils (peaty material havingbeen removed in preparing the site) and withthe embankment materials. This deposit is ofintermediate to high plasticity and is oftenlaminated, with tree and plant roots.Near-vertical discontinuities noted in trialpit and trench inspections are considered toresult from desiccation following the drainingof Lake Humber. Tests indicate that thedeposit is slightly overconsolidated,considered to be due to the effects ofvegetation and desiccation.

33. From consideration of the geology andtopography of the site and detailed inspectionthe clay deposits are not believed to containpre-existing shear surfaces. Based on theresults of the laboratory tests designparameters given in Table 1 are believed to bereasonable lower bound peak parameters forshearing along the laminations.

Table 1. Properties and parameters assumed indesign for the Upper Clay.

Natural moisture contentLiquid limitPlastic limitBulk densityDrained shear strength0' peak (design value)J3' residual (min value)c'peak (design value)c'residual (min value)Coefficient ofcompressibility mvCoefficient ofconsolidation cv

24 to40 to 75%22 to 30%1.7to2.0 Mg/m3

17 degrees11 degrees3 kN/m2

00.16 to 0.30mz/MN2.0 to 4.0mz/yr

34. The parameters determined fromlaboratory tests for other constructionmaterials and for the lagoon pfa are given inTable 2.

Table 2. Engineering properties assumed indesign for the embankment constructionmaterials and lagoon pfa.

Compacted Compacted LagoonColliery pfa pfaShale

BulkDensity Mg/nr3 2.07 1.60 1.63

MaximumDry Density Mg/nr3 1.78 1.46

OptimumMoisture Content % 8.5 17.5

Permeability m/s < l.OxlO7 1.5xlO"6

0' degrees

c' kN/m2

29

0

29

0

22

0

35. For drainage zones use has been madeof local limestone and fba. Typicalparticle size curves for these materialstogether with those of pfa and shale areshown in Fig.2 .

36. Extensive embankment stabilityanalyses have been carried out for Stage Iand in the design of Stage II. Because ofthe soil types and the slow rate ofconstruction it was determined thatconstruction excess pore pressures would notoccur and so only drained conditions needconsidering. Effective stress analysesusing pore pressures predicted from seepagetheory were therefore carried out.

37. Typical piezometric design assumptionsare shown in Fig. 1 together with an exampleof a trial failure surface analysis.Analyses used computer programs of the Sarmaand Morgenstern and Price methods and handwedges.

38. The static factor of safety requiredagainst both overall and local toe failureis 1.40. This was judged to be the minimumallowable commensurate with the assumptions,particularly those for the Upper Clay Layer.Following Carsington, the design wasreviewed to assess any potential forprogressive failure particularly in view ofthe relatively brittle nature and thinnessof the Upper Clay layer which would tend toconcentrate strain. It was concluded thatthe assumptions made concerning the drainedshear strength of the clay and the factor ofsafety criterion in the conventional stabilityanalyses were such that progressive failurewas unlikely to be a problem. However, it wasconsidered prudent to carry out some finiteelement analyses to check the potential forprogressive failure.

39. The embankment has also been checked toensure a factor of safety greater than unityfor seismic loading or shock loading due tomining induced movements. For these cases itwas assumed that the lagooned material hadzero strength. (Haws et al Ref.3).

40. Foundation and embankment pore pressuresare monitored. Most of the original 22No twintube hydraulic piezometers continue tofunction satisfactorily 25 years after theywere installed. As Stage I developedadditional piezometers were installed, andthere is currently a total of 46.

-

1/

/

1

_—

f

• — "

f

y

i

ro

—

- -

0002 0-006 002 O006

T60mm

Fig. 2. Typical particle size curves

102

PAPER 15: HAWS, MARTIN AND ORANGE-BROMEHEAD

41. For Stage II 33 hydraulic piezometershave been installed in the foundation soilsand additional ones will be installed in theembankment during construction to monitor theseepage condition between the lagoon and thechimney drain.

42. The piezometer monitoring to date hasnot indicated any cause for concern and hasconfirmed that there is no build-up ofconstruction pore pressures.

CONSTRUCTIONMaterials supply, characteristics, laving, &testing

43. £fa The pfa is pumped from the twopower stations via 530mm pipelines. These areof asbestos cement which, although cheap, haveproved to be susceptible to erosion. Thepipeline from Ferrybridge is now beingreplaced by one of high density polyethylenewhich trials have shown to be highly resistantto pfa erosion.

44. At rest the pipeline is kept filled withwater so that when ashing out a power stationan initial slug of water of low slurry contententers the lagoon.

45. When slurry of solids content 20 - 40%reaches the site it is diverted to the vacuumfiltration plant. Here as much as possible isdewatered to about 20% moisture (optimum beingaround 17.5%) and the rest is disposed of intothe lagoons as unsuitable. About 55% of thepfa reaching the site is thus processed andused in the embankments.46 As pfa is made up of spherical particles

it is easily compacted and has a flat standardcompaction curve. This makes it possible tolay throughout the year in layers up to 500mmthick, achieving compactions in excess of 98%maximum standard compaction with 4 passes of aT182B Vibrating roller.

47. The in situ density test uses thelarger 150mm diameter core cutter to reducethe effect of dimensional error on a verysensitive result. The top 150mm layer ofmaterial is discarded because ofoverstressing or disturbance by traffic.Three cores are then taken going down in180mm units so that the bottom of the lowestcore is between 600 and 700mm below thesurface covering the re-worked top of theprevious layer.

48. Shale Although the site is owned andoperated by National Power, it is also theprincipal disposal area for discard fromKellingley Colliery. Included in thediscard are press tailings, the amount beinglimited by the need to stockpile thematerial in a tip throughout the winter.

49. The material arrives saturated fromthe colliery washer, so when placed directinto the works it is spread in 100mm layersto dry for a day to lower the moisture tothe 8.5% optimum before being rolled by twopasses of a 10 tonne grid roller. Thisleaves a high surface area to aid drying andabsorb surplus moisture from the succeedinglayer of wet shale. Laying can only becarried out when drying conditions prevail,which experience has shown to be dry weatherwith a day temperature in excess of 15°C.

50. Testing is by the normal sandreplacement method ignoring the 'top 150mm'of temporarily saturated material.

Construction cycle51. At all times construction must provide

new capacity before existing lagoons arefull. In the case of 'A' lagoon (Dennis etal,Fig.l) crest raising is carried out when ~~the lagoon is out of commission. Howeverowing to the filtration plant effluent onlypassing into B lagoon that lagoon nevercomes out of service so that the placing ofa nib of shale into the lagoon followed bycrest raising has to be done with thelagoon live. The amount of shale 'lost'into the lagoon is clearly minimised bycarrying out this operation towards the endof the lagoon filling cycle.

52. Where the outfall has to be raised forthe live lagoon the method adopted has beento isolate the area by forming a shale bundaround the outfall tower, having firstraised the tower concrete, if necessary overwater. Cleaning out, infilling the towerand placing damboards follow as well as theplacing of stone filled gabions 2m high soas to form a barrier to prevent the layer of'floaters' and vegetation from reaching thetower. The 'floaters' rise with each lagoonfilling and are now between 1 and 2m thick.The lagoon being live means that pumps arerequired to be on hand when men are working inthe pond area around the tower. In operationeach tower has a collar boom around it as asecondary protection to prevent floaters beingcarried back to the river in the return waterwhich must not exceed 40 ppm.

53. The Stage I construction was dictated bythe initial shortage of embankment material,so that for each raising of the embankmentcrest the form most economic in use ofmaterial was adopted. This led to acontinuous operation of placing a minimum ofmaterial on the temporary outside face up to alevel from which the crest raising for a newlagoon could be executed in a comparativelyshort time in the summer (necessary because ofthe use of shale at the water face). Repeatedtemporarytreatment by hydroseeding was thus necessaryto the outer face of the embankment for dustprevention and environmental reasons.

54. For Stage II, construction will be bycomplete horizontal layers across theembankment permitting outerface sowing as theembankment rises, thus presenting a betterimage.

Survey controls55. For Stage I, survey stations were

established on the toe bund wherever crosssection checks were desired. This arrangementwas suitable for the tacheometric surveymethod applied until the advent of the EDM.The stations also provide control for thelandscaped outer face. Here the practice isto raise the embankment in 2m stages, surveyin pegs on the correct outerface and then trimand soil to this line. This not only enablesirregular patterns to be followed but also

103

TAILINGS DAMS

ensures that the soil is placed on a preparedsurface uncompacted by plant. It also leavesa degree of unevenness which helps to avoidthe "engineered slope" normally found in earthdams.

Other works56. On a scheme of this size the ancillary

works are substantial,examples are:-the provision and maintenance of 5km of7.9m HRA road for 17 tonne axle loads,-the provision of a monitored drainagesystem isolated from surroundings, thewater being returned to the river withlagoon outfall water,-the provision and replacement of slurrypipelines within the site both on and offthe embankment,-embankment, drainage and surveywork arising from mining subsidence,-the provision and operation ofinstrumentation.

Initial development of Stage II57. Apart from an initial contract to

establish the road system, the outfallarrangement, and some of the embankmentformation, the work of placing thefoundations of the Stage II embankment hasbeen carried out as an extension to thenormal work of operating the site.

58. Both stripped topsoil and subsoil havebeen stored in two separate groups accordingto a plan prepared by the LandscapeConsultant and MAFF.

59. This has been followed by the initialembankment construction of shale with amembrane lining of black POLYTARP laid onthe floor of the lagoons and up the innerslopes for the first 3 m above ground level.

60. Between the dendritic drains (para.18) were laid dumplings of press tailingsfrom the colliery to hold down the membraneagainst wind action. In spite of this,considerable amounts of air were trappedunder the membrane and it was necessarycarefully to vent the "whales" which aroseas the lagoon filled.

61. Finger drains leading outwards fromthe chimney drain encircling the embankmentwere first shaped, lined with membrane up to+10mA0D and then filled with a stone drainand a piped drain at alternate locations.At their outer ends these drains are led viaindividual manholes into measuring chambersso that embankment seepage can be examinedin sectors.

62. Two blocks of instrumentation, one inthe south side and one in the east, recordfoundation piezometric levels usinghydraulic instruments.

Outfall system63. The effluent from the lagoons is

handled identically to that for Stage I withdischarge via concrete outfall towers and a900mm diameter pipeline laid through theembankment in a reinforced concrete tunnel.The towers will be raised in 6.6m stageswith continuity of reinforcement maintainedby the use of couplers. The height of thesestages is dictated by the need to place anaccess platform on the top of the tower.Lagoon water level is controlled bydamboards placed from the platform. Theseare of reinforced concrete with their facesrecessed so that they can be removed with ascissor grab.

64. On completion of each 3.3m lift of thelagoons reinforced concrete is placed tofill the damboard slots and complete theoctagonal tower, the damboards beingretained as a back shutter for safety.65. As developed for Stage I, shale is

placed around each tower to form a platform tocarry stone filled gabions to hold back thefloaters.

CONTRACTS66. Construction has been by a series ofcontracts of two years duration, the newsucceeding the old without any time gap.Early contracts were undertaken by Lehane,McKenzie & Shand and the more recent ones byTaylor Woodrow (Northern), with a Stage IIpreparation contract going to Henry Boot.

ACKNOWLEDGEMENTAcknowledgement is made to National Power Co.for permission to publish this paper.

REFERENCES1. LILWALL R . C Seismicity and seismic

hazard in Britain. Seismol. Bull. Inst.Geol. Sci., 1976 No 4.

2. TAYLOR R.K., BARTON R., MITCHELL J.E.,and COBB, A.E. The engineering geology ofDevensian Deposits underlying PFA lagoonsat Gale Common, Yorkshire. Q.J1 EngngGeol 1976 vol. 9. 195-216.

3. HAWS E.T., PEDLEY W.A. AND NISBET R.M.Brotherton Ings Ash Lagoons. 12th CongInt. Comm. on Large Dams, Mexico 1976.

104

16. Gale common ash disposal scheme - planning, environment,operation and restoration

J. A. DENNIS and D. J. HILLIER, Rendel, Palmer and Tritton, London, UK,and H.T. MOGGRIDGE, Colvin and Moggridge, UK

The scheme is for the disposal of waste materials from two power stations and a colliery,by forming circumferential embankments of these materials and filling the central voidswith slurried ash. In view of its prominence, great attention has been paid to theprofile and restoration of the 'hill' so formed. The life of the project is some 50 yearsand it is being developed in 3 stages. Work on the second stage has recently commenced.

PLANNING AND ENVIRONMENT

Introduction1. The disposal of ash from a major

coal-fired power station can be a considerableproblem, not least environmentally. WhenEggborough Power Station (2000 MW) was beingplanned, a number of schemes for the disposalof the ash were considered, but finally, withthe nearby Ferrybridge X C Power Station (also2000 MW) being built at about the same time,the concept of a joint disposal site hadobvious advantages. A new colliery was alsobeing opened at Kellingley and the waste fromhere was considered to be a useful materialfor embankment construction. This opened upthe possibility of dealing, on one site, withyet a third disposal problem.

2. The general area in the vicinity of GaleCommon is flat and featureless, and the scaleof disposal - some 65 million cubic metresover the full life of the scheme - meant thatan above-ground scheme would lead to thecreation of a man-made hill on a scalerivalling Brayton Barff near Selby; offeringthe prospect of adding a pleasant feature tothe local landscape.

3. From the outset great attention was paidto finished profile and final landscapetreatment. Lagooning was planned to a totalheight of about 50 metres, making theembankments - which in effect are peripheraldams - among the highest earth dams in theUnited Kingdom. The scale of the scheme issuch that the horizontal embankment crestswould, from certain viewpoints, stretch forover a kilometre across the skyline without asignificant change in level. Therefore,notwithstanding that the outline of the schemein plan is shaped so as to offer interestingchanges of silhouette and light pattern, thecrest, if left without further treatment wouldhave been a disastrous culmination to anotherwise well-planned scheme. Hence it wasdecided to place dry material on the finishedlagoons to form a crown, thus presenting amore pleasant and natural final appearance;and by forming this material to a general

south westerly slope gaining someagricultural advantage.

4. The lagoons are planned to be finishedto different levels to follow as closely aspracticable the overall final profile andhence reduce the volume of material whichhas subsequently to be placed dry.

5. Originally the site was of only poor tomedium quality agriculturally. The worstareas were in the northern part and plannedcoal mining would eventually reduce yetfurther its agricultural value.

6. In view of the considerable life-spanof the Power Stations (which could beextended by re-planting) it was clear that adisposal scheme based on a stageddevelopment would be the most appropriate.

7. A 3-stage scheme was developed, withthe first stage at the northern end of thesite (for the reasons explained above), asecond stage to the south east, and a finalstage to the south west. Each stagecomprises two lagoons, so as to permit theirbeing raised and filled alternately. Inaddition, a separate pair of low-levellagoons, for emergency purposes, was to beconstructed in the north west area of thesite. (See figure 1).8. Figure 1 also shows the Yorkshire Water

aqueduct reservation which traverses thesite. This reservation is kept clear of theworks and is crossed by "bridges" so thatheavy loads will not damage the pipelines.

9. The whole arrangement was discussedwith the local planning authority who gaveapproval to the outline scheme.

Stage 110. The main constructional materials for

the "hill" are pulverised fuel ash (pfa) andcolliery waste. (Conditioned pfa andcolliery waste for the embankments, slurriedpfa for filling the lagoons.) The initialembankment construction had to be wholly ofcolliery waste since pfa would not be

The embankment dam. Thomas Telford, London, 1991 105

TAILINGS DAMS

available economically until the stations werecommissioned. The preliminary site works forStage I therefore included the construction ofinitial bunds (in colliery waste), separatelyenclosing the main and emergency lagoons, andallowed for continued construction in bothcolliery waste and pfa as the latter becameavailable.

11. These preliminary works also includedthe construction of the vacuum filtrationplant-house and associated tanks, etc., theslurry pipelines from the power stations, theon-site pipelines, roadworks, site drainage,drains for the main embankments, effluentsettlement ponds and a booster and returnwater pumphouse. The latter, is firstly topump the slurry into the lagoons when thesereach a level beyond the capability of thepower station pumps, and secondly to pump theeffluent back to the River Aire, from whichriver the power stations drew the water in thefirst place.

12. Eleven feet, the height of the initialembankments, (translated more recently to 3.3metres) has remained the lift height by whichthe embankment and lagoons have subsequentlybeen raised.

13. Because of the great width of the mainembankments at their base and the limitedsupply of colliery waste and conditioned pfa,it was not possible until comparativelyrecently, to construct these embankments infull-width horizontal layers. It wasessential to build up the inside of theembankments to keep pace with lagooning, butlittle more material was available than thatrequired to ensure the stability of the risinginner face. This led to a scar face beingvisible from outside the site, and (later) tostricter requirements governing theconstruction of Stage II (see para 18).

14. This scar face was open for most of thelife of the first stage of the work. It wasnot until 1989 that the outside faceeverywhere caught up with the lagoon face andthe soiling and seeding of the outer face,which followed construction as soon aspossible, was able to provide anenvironmentally satisfactory appearanceoverall.

15. Dust nuisance is always an environmentalproblem when constructing earthworks in pfa.Although a thin crust does form on finishedsurfaces, this is fairly easily disturbed. Inorder to reinforce this crust, temporarysurfaces, where of short duration, weresprayed with vinyl, and where of longerduration, were hydroseeded.

16. Floaters, which form a thick crust onthe tops of the lagoons, are very easilydisturbed by wind. However, the surface canbe bound together by a covering of vegetation,

which also breaks up the air flowimmediately above the vulnerable surface.At Gale Common,rushes, reeds, long grass andeven small trees have flourished on thefloater crust, providing cover and a habitatfor an interesting variety of wild life.

17. A variant to the originally plannedconstructional sequence has recently beeninvestigated and will probably be followed.The total volume of material in the "crown"on Stage I is about 2,750,000 cubic metres.It has now been shown that by stepping eachsubsequent lagoon well within the outline ofthe one below, further lagooning above theoriginally envisaged levels will bepossible, so that a large part of the"crown" can be formed from settled ash, thusoffering a considerable constructionaleconomy. The scheme would be incorporatedin the next contract which is due tocommence in the autumn of 1991.

Stage II18. It is estimated that lagooning in

Stage I will have been completed to theoriginally planned levels by early in 1992.However, there has already been a surfeit ofbank-building material over and above thatrequired for Stage I and it has beenpossible to commence the construction of theStage II embankments well before lagooning(in Stage II) has had to start. Thisenables the embankments to be built to fullwidth from the outset (a requirement of thePlanning Authority) and soiling and sowingof the outer faces to commence as well.

19. Another constructional difference fromStage I, again of environmentalsignificance, is the fact that for Stage IIan impermeable membrane has been laid acrossthe floors of the lagoons and below thechimney and finger drains. This is toensure that the contaminated leachate fromthe lagoons is prevented from seepingdownwards and entering the underlyingaquifer. So far, there has been noindication of such being a problem in thecase of Stage I, but it was neverthelessconsidered a worthwhile precaution in thisenvironmentally sensitive era.

20 Although the bank building of Stage IIstarts from a more favourable position thandid Stage I, construction is stillcontrolled by the rate of material supply.The first target - which must be achievedbefore Stage II becomes operational - is tocomplete the banks to such a height that thelagoons provide sufficient head to drive thereturn water to the settling ponds, whichare located near the return water pumphouse,some 2500 metres from the outfall towers.

21. The second key stage occurs when the

106

PAPER 16: DENNIS, MOGGRIDGE AND HILLIER

banks have reached a level, such that withtheir decreasing width (with increasingheight), the rate of supply of bank-buildingmaterial can support construction at the samerate as the lagoons are being filled.Clearly, from this stage onwards sparematerial will be available for completing the"crown" on Stage I. In fact, this criticallevel (+19.7m) should be reached well beforethe corresponding depth of ash has beendeposited in the lagoons and therefore, forsome considerable period before the slurriedash reaches this level, all the "dry" materialshould be available to be placed in the"crown" to Stage I.

22. The requirement to build to a level of+19.7 in advance of lagooning explains theinner profile of the Stage II embankments,which have to be free-standing (trapezoidal)up to this level, but can adopt a "fir tree"configuration at higher levels. (See fig 1 ofHaws et al)

23. Stage II lagooning should continue untilabout the year 2007, when it will also becrowned with dry material. However, by thattime, the Stage III embankments will have beencompleted to their initial critical level andthese lagoons will be ready for operation.Stage III will not be complete until about theyear 2020.

OPERATION

Owner Operations24. The Booster and Return Water Pumphouse

and the Vacuum Filtation Plant (VFP) at GaleCommon are manned and operated by NationalPower on a 24-hour basis. National Power alsoorganises the delivery of furnace bottom ash(fba) to the site by road.25. Pumping around and off the site is

controlled at the pumphouse; but the bulk ofthe controls are at the filtration plant wheredensity meters determine when slurry is to betaken into the plant for filtering.

26. Once in the filtration plant, the slurryis thickened by transferring slurry from thebottom of a secondary tank while the surpluswater and thin slurry are decanted off via aperimeter trough in the primary tank. Thethickened slurry is kept in motion byrecirculation and a rake arm while it isprogressively drawn off for filtration. Aftera time the grading of the remaining slurrystarts to blind the filters and at this stagefiltration ceases and the slurry is pumpedaway to XB' lagoon while the filters arecleaned.

Contractual Arrangements27. The initial construction works for Stage

I of the Gale Common scheme were mainlycarried out prior to 1970 and were completedduring the first operational contract,1970/71. Since then, the operationalcontracts have normally been for two-yearterms.

28 Originally, each two-year contract wasnot subject to variation of price and thissituation existed until 1975. However, by1976, two-year contracts were embracingvariation of price clauses. Consequently itwas considered that a long contract periodmight be financially advantageous to theEmployer. Accordingly, in 1976, tenderswere sought for both 2-year and 5-yearterms, each incorporating variation ofprice. In the event, however, the "5-year"tenders did not offer obvious financialadvantages and an alternative 2-year "fixedprice" offer by one of the tenderers wasaccepted. Thereafter, all normal contractshave been subject to variation of price.

29. Until 1984 the Contractor was Lehane,Mackenzie and Shand Ltd, and since then ithas been Taylor Woodrow ConstructionNorthern Ltd. The current contract hasanother year to run.

30. In addition to the contracts describedabove, there have been two others, namely aSite Investigation Contract for the StageII/III area of the Site (by Soil MechanicsLtd in 1982) and the Stage II Initial WorksContract in 1986/7 carried out by Henry BootNorthern Ltd.

31. On the Employer's instruction,contracts are based on the CEGB Generalconditions of Contract for constructionalWorks, 1971 Edition (which document issimilar to the old ICE 4th Edition) and onthe old ICE Standard Method of Measurement,1974 Edition.

32. Over the years the details of thecontracts have been refined and adapted, butessentially they have been contracts tooperate the Site for a specific period oftime, including any incidental works thatmight become necessary (although the lasttwo contracts have also included furtherpreparatory works for Stage II).Consequently the Bills of Quantities havetended to include a fairly large P.C. andProvisional Sums element to coveroperational and maintenance requirementsunknown in detail at the time of tender.

33. Since 1980 the contracts have includedintermediate completion date requirements.Such Key Dates relate to items of work forwhich there is a definite programmerequirement, such as clearance of shalestockpiles by the end of the summer period(so that the maximum storage capacity is

107

TAILINGS DAMS

available at the onset of winter, during whichlittle or no shale can be placed in theWorks). Other Key Dates have applied to suchitems as emergency lagoon embankment raising,road resurfacing, pipeline replacement etc.Monthly valuations are checked on Site andpassed to the Engineer's Head Office where apayment certificate is prepared and issued tothe Employer, for payment within 42 days ofreceipt of the valuation on site.The minimumamount for interim certificates is currentlystipulated as £50,000, but the volume of workis seasonal and actual monthly payments varyconsiderably.

Site Organisation34. The current contractor, Taylor Woodrow

Construction Northern, operates the site as amanagement contractor with all work beingcarried out by the principle subcontractor,V.H.E. The latter keeps a small permanentstaff on site while maintaining a flexibleposition with the bulk of his plant andlabour.

35. Pfa is handled by five Moxy articulateddumptrucks, a loading shovel, a bulldozer and

ro a vibrating roller. Normally shale can onlybe placed into the Works between April andOctober and involves a loading shovel,lorries, two or three bulldozers and acompactor. At least one bulldozer, forstockpiling, and one backactor, for all thetrimming and trenching associated with thejob, remain on site full time. The totalmanpower varies from 20 - 35 men, winter tosummer.

36. The Consulting Engineer provides on sitean ER, inspector and clerk, with appropriateadditional seasonal and specialist back-upfrom head office as required. The ER hasresponsibility for supervising the Works andfor managing the site on behalf of the StationManager of Eggborough Power Station, who hasresponsibility for the overall operation ofthe lagoons.

Communication37. The scheme is monitored by both the

Employer and the Consulting Engineer atregular monthly meetings with the Contractor.In addition, the site staff, Contractor andEmployer meet British Coal respresentatives ona monthly basis at "pit" level to resolve anyproblems in connection with the shale supply.

38. At a higher level there are meetings atleast annually with the appropriateauthorities to discuss mining subsidence,landscaping, overall management, and planningand the environment.

Checking39. Routine records are maintained with

reference to:- weather (the site is a Met. office

rainfall station)- piezometric levels (6 locations in theembankments and borehole dip readingselsewhere)

- mining subsidence (levels and strains)- shale and pfa densities, and furnacebottom ash (fba) grading

- daily lagoon levels and monthly seepagereadings

- daily pipeline and embankmentinspections.

Environmental Protection40. The day to day control of dust on the

site is covered by the inclusion of twowater bowsers in the contract - one onpermanent duty and the second on standby.The machines are alternated weekly toensure, as far as possible, that the standbywill operate effectively when required.

41. The extensive interim embankmentsurfaces, which are beyond reach of thebowsers, are vinyl sprayed or hydroseeded(see para 15).

Certification of Lagoons42. Originally the Stage I main lagoons

and the emergency lagoons were inspected andcertified under the Reservoirs (SafetyProvisions) Act, 1930 and although it hasnot been necessary to register them underthe Reservoirs Act, 1975, National Power, asa responsible operator, treats them asthough they were so registered.Accordingly, they are inspected andcertified by Mr E T Haws as ConstructionEngineer, and preliminary certificates areissued by him, to the Employer, each time alagoon embankment is raised. Ultimately afinal certificate will be issued for eachlagoon.

43. Because the Stage II lagoons willstand filled with water (to protect themembrane lining) for some time before beingused for slurry, they have been registeredunder the Act. They, too, have beeninspected and certified by the ConstructionEngineer and no doubt this procedure willcontinue, although the present intention isto de-register the lagoons when they becomeoperational (say 1991/92).

44. The preliminary certificates issued todate for the various lagoons have sometimescontained conditional clauses specific toindividual lagoons and to situationspertaining at the time, in addition tonormal clauses stipulating requirements fordaily inspections, keeping of appropriaterecords, maximum permissible slurry inputs,water and spillway levels, etc.

108

PAPER 16: DENNIS, MOGGRIDGE AND HILLIER

1. Layout of Scheme

Fig 2. South-east Cornerof Stage II(model)

Fig 3. West Side ofStage I (model)

109

TAILINGS DAMS

Examples of additional conditions are thosepertaining to pipelines on the embankments,the required action when settled ash builds uplocal to the points of input, action to betaken in the event of outfall valvemalfunction and clauses relating to miningsubsidence.

Costs45. The disposal of ash is an inescapable

facet of coal-fired electricity generation.Some ash can be solid, but overall thedisposal is effectively an operationaloverhead.There is generally a ready market for any fbaand Eggborough also sells some of its pfa makeas a pozzolan, but the Authors have noinformation regarding the receipts from suchsales. Neither have they any details ofNational Power's own costs relating to ashdisposal.

46. Therefore the following approximatecosts are based solely on the civilengineering contract sums at the date of thosecontracts and on the total quantities of pfasent to Gale Common.

47. For the whole period, to the end of1989, the cost (including initial constructionworks amounts to approximately 80p/tonne ofdry ash; for the more recent period 1987/9 itis approximately £1.75/tonne (including theconcurrent initial works for Stage II). Thelatter figure reflects the more up-to-datecosts of the preparatory works for Stage II,compared with the Stage I pre-1971 valuesincluded in the former figure. The mostrecent "disposal only" contract (1986/7) givesa value of approximately 80p/tonne.

LANDSCAPE AND RESTORATION

Concept48. The design strategy for restoration of

Gale Common hill was established soon afterthe engineering concept. The late BrendaColvin CBE PPILA produced the first design inthe early 1960s, the CEGB at that timepioneering the idea that a restoration schemeshould be prepared before site work started.The design approach was a poetic response to anew problem; the preliminary report of 1962for instance cited "the objective of creatinga completely new landscape feature, ofdistinguished and pleasing form, contributingto the interest of the landscape, as doexisting hills" "High contours, thusindicated, will be visible from greatdistances in this flat area. Crops on thesecontours exposed to higher winds than existinglevels will require shelter planting in theform of tree belts. High priority should,however, be given to the visual effects oftree belts."

49. The top of the hill was designed as anarable plateau sloping towards thesouth-west. The sides of the hill, pastureor wood because of their steepness, wereimagined as terraces spiralling upwards, thecorners of the hill being elongated intoterraced silhouettes. The report explainsthe ideas behind this design: "In Britiainman-made terraces were common in neolithicand iron age periods both for agricultureand defence. If this new hill can be madein terraced form it would be of comparablescale and character to those early ones; anabstract sculptural group. No attempt atreproduction of other earthworks isproposed. A frank artefact may bepreferable to a naturalistic hill in thisflat area where it can scarcely be seen aspart of the surrounding geologicalstructure".

50. These themes have guided developmentof the project over the succeeding two and ahalf decades, latterly under the guidance ofthe Author who, as partner of Brenda Colvinfrom 1969 until her death, is still retainedas Consultant.

Engineering Restraints51. The cross section of the bulk of the

hill is determined by the need for steepside slopes to maximise the proportion ofinterior volume to dry-placed banks. Theaverage slope is 1:3, with engineeringrestrictions on the acceptable length ofsteeper slopes and on deviation from theengineering section line. However, at outercorners it has proved possible to increasethe distance between crest and toe withoutplacing much extra material, by takingadvantage of the horizontal differencebetween the tightest lagoon shape possibleon plan and the angular corners of thesite. This has given the opportunity todesign an interesting and less steep profilewhere the hill is most visible insilhouette. The photograph (figure 2.) of amodel made in the early 1980s illustrateshow the south-east corner of Stage II, forinstance, takes advantage of this concept.

52. The plan form of the hill, alreadycurvaceous due to exigencies of the site,has been further enriched by such devices astaking advantage of re-entrants which occurwhere a later stage is butted against anearlier. Rather than smooth out the form,the hollow is emphasised. Tracks and berms,to prevent run-off water accelerating toofast down the steep sides, spiral up theside of the hill, not parallel but in aslightly variable relationship to eachother. "Though the separation between theberms never reaches a distance beyondacceptable engineering criteria, the

110

PAPER 16: DENNIS, MOGGRIDGE AND HILLIER

possibility of variability has made itpossible to design a rich superficial form.This flexibility also makes it possible toarrange tracks to leave the base and arrive atthe top in convenient locations.

53 The concept of dealing with berms in anirregular spiralling manner derived fromfollowing what at first seemed yeryconstricting engineering criteria to solve allthe demanding structural problems. Yet thevery strictness of the engineering disciplinesimposed - minimum and maximum berm widths,minimum and maximum gradients, maximumdeviations from structural cross sections,strict rules at the rnost fragile positions,maximum distances between berms - all thesegive an underlying unity of form to theapparent casual ness of layout in the finallandscape details. In just such a way a treewhich looks haphazard in form is constructedby the hidden geometry of natural forces.

Woodlands and Pastures54. The photograph (figure 3.) of the

model's west side, giving some idea of thisunderlying land form design, shows theprinciples of woodland layout adopted. On thetop plateau, simple edge and intermediatebelts shelter potential arable fields; smallclumps pick out high points to enrich thesilhouette of the hill. On the sides of thehill trees are blocked into elongated woodsfrequently continued from existing woodlandbelow, picking up the spiral lines and leavinglarge areas of hillside under pasture. Duringthe layout of the Stage I hillside woods onsite, long sweeping lines have been emphasisedto show more clearly the land form below, asthis was perceived to be more attractive inappearance than too great an artifice ofcasual informality.

55. Since the model in the photographs wasmade, studies of pasture field patterns forthe hillside have resulted in fragmentation ofopen land into enclosures 1-2 hectares insize. These have been grouped according toMAFF recommendations for sheep farming. Thedetailed shapes of the fields was determinedby trial and error on a working model. At therequest of the planning authority a largerxoutbye' pasture is proposed for thesouth-east corner of Stage II to show the landform more clearly at the point.

56. At original ground levels, several oldwoods have been kept, and are being managed toenrich wildlife; a ground cover of bluebellsand anemones has been given renewed vigour bygentle management improvements. An attempthas been made to move the floor of a wood,lost under Stage II, onto the woodland sidesof Stage I; it is too soon to know with what

success. One ancient wood is receiving aspecial management regime followingecological advice. New screen woods havealso been established along the edges of thesite at ground level to hide low levelvehicular movements and stockpiles ofcolliery waste from local roads andvillages.

57. The chemicals in pfa and its verysmall particle size limit the diversity oftree species which can be grown. The CEGBhas long been carrying out scientificstudies of what will grow. Site trials werealso undertaken as soon as an ash face wasavailable; these revealed that minorvariations in restoration technique wereless important than the correct choice ofspecies and certain basic ameliorationmeasures. A soil layer on top, ripped intothe compacted ash surface at the interface,induced root growth in tolerant species.Root pits were dug after five years whichshowed good penetration into the pfa, sothat both stability and tolerance of the dryYorkshire plain climate can be hoped for infuture.

Soils and Maintenance58. Great care has been taken at Stage II

with soil stripping. Four separatelong-term soil heaps cater for top soil intwo grades and subsoil in two grades. Thetop grade is the light sandy soil whichforms the limited areas of top classfarmland beneath Stage II; this will be usedto restore the top of the hill for arableuse. The sides of the hill use second gradesubsoils for restoration, which is oftenheavier in texture, or a mixture of soilsfrom Stage I stripping.

59. Vegetation establishment at GaleCommon has been excellent. This is partlydue to thorough trials and demandingspecifications. Much of the credit howevermust go to excellent standards of plantingand aftercare achieved by the MobileResources Unit of National Power. In thecreation of landscapes all efforts ofimagination and technical understanding arein the end dependant upon good husbandry ofthe living skin.

60. The aim of restoration is to create anew living skin across the surface of thehill to profiles and plan shapes which bothplease the eye and satisfy foreseeablefunctional requirements.

ACKNOWLEDGEMENT

Acknowledgement is made to National PowerLtd for permission to publish this paper.

Ill

Discussion

In view of the large number of papers in thissession the chairman for the session,Dr A D M Penman, introduced the subject andreviewed the papers.

A D M PENMANThis subject attracted 35% of the paperssubmitted to the Conference. The nine papers inthis session (papers 7-16) can be considered interms of the organisations that have submittedthem (Table 1), or in terms of the heights ofthe tailings dams (Table 2).

The Bulgarian dam described in paper 8 was theonly one built almost entirely from tailings.During the work, the effect of delivery pressureon the sizes of coarse grains from the cycloneswas measured (2.0 to 2.3 bars producedparticles of 0.25 to 0.26mm, while 1.6 to 1.9bars gave 0.20 to 0.23mm). Slurry density interms of solids to water by weight was 1 to 3.0- 3.5 while by volume this was 1 to 7.6 - 9.6.In the early stages when, because of the valleyshape, it was difficult to keep dam heightsafely above lagoon level with the coarsefraction available, design might have beenhelped if they had had access to the TADAMprogramme described in paper 10.

The large pulverised fuel ash dams and lagoonsat Gale Common have been described in papers 15and 16. While tailings dams do not come underthe Reservoirs Act, CEGB decided to have Phase 1I designed and constructed in accordance with theAct. With the formation of National Power inpreparation for the privatisation of CEGB, Phase2 will not be put under the Act when it is beingbuilt with pfa in the lagoon. Initially,however, a layer of water has had to be put overthe impervious plastic sheeting covering thefoundation to hold it down. The volume of waterinvolved brings it into the Reservoirs Act, butapparently the intention is to bring it outagain once fly ash is pumped into the lagoon.The earthquake intensity mentioned in paper 15is apparently Modified Mercalli and the"dendritic" pattern of drains means simply thatthe drains were laid out like the roots of atree or system of tributaries going into ariver.

Tailings from the Neves Corvo copper mine inPortugal is rich in pyrites and to avoid theacid caused by oxidation, the tailings wasdischarged into the lagoon below water and kept

The embankment dam. Thomas Telford, London, 1991

below the water level. This produced very lowdensities. To get a greater weight of tailingsinto the storage volume, spray bars were used toplace the tailings above water but keeping itcontinuously wet (paper 11).

Paper 9 is a geotechnical paper on the art ofbuilding low banks on very soft clay. It canusefully be compared with Paper 1. The softclay clearly had a dried crust and wasreinforced with tree-roots. It is not clear howgypsum cakes are to be stored to a height of 12mwithout instability.

The problems of waste disposal from china claymining are discussed by paper 13. Paper 12gives a valuable assessment of things that cango wrong with overflow systems: a vital factorfor the safety of tailings dams which in generalare particularly sensitive to overtopping. Itis suggested that most failures initiate fromfaults in the overflow arrangements.

The safety of British tailings dams is discussedin paper 7. It points out that there is noregister of all tailings dams and so it is notknown how many exist. A comparison is drawnbetween the requirements of the Reservoirs Actand those of the Mines and Quarries Act thatrelate to tailings dams. The high standard ofsafety of embankment dams owes much to thedetailed studies of behaviour that have beenmade. We need to apply this approach totailings dams.

Various names have been given in the papers tothe same thing, eg in South Africa theimpoundment is often called the dam and theretaining dam is called a wall. The terminologygiven in Table 3 is therefore proposed for usein describing tailings dams.

Table 1. Organisations submitting papers

Organisation Paper No

WLPU (now Knight Piesold) 10 11 12Rendel Palmer & Tritton 15 16MRM Partnership 13Binnie & Partners 9Building Research Establishment 7Higher Instit Arch & Civ Eng, Sofia 8

113

TAILINGS DAMS

Table 2. Heights of tailings dams

PaperNo

10

8

15,16

11

11

9

Tailingsdam

LosLeonesElatziteMineGaleCommonWhealJaneCerro daLoboTioxideGroup pic

Materialstored

CoppertailingsCoppertailingsPFA

TintailingsCoppertailingsGypsumwaste

Country

Chile

Bulgaria

UK

UK

Portugal

Malaysia

HeightPrest

160

95

51

40

28

(m)Propd

300

145

51

53

35

2.5

- There is an increasing use of sand filling inunderground operations and this tends todecrease the sand content of tailings availablefor dam construction.- Tailings are not always released from the damcrest. In cases where it is more economic,tailings are released at the head of a valleyand retained by a traditional earthfill damwhich, initially may simply retain water, beforethe tailings in the valley extend to it.- There should be a register of all thetailings dams in the UK.- There is no doubt that earthquake risk mustbe taken seriously.- Instrumentation and monitoring should besimilar to that used for embankment dams butbecause of the much longer construction period,redundancy needs to be built in to cope withlosses of equipment that may occur.

Table 3. Proposed terminology

Preferred term Non-preferred term

Dam WallLagoon Reservoir, DamWall drain Chimney drainContinuous wall drainTailings - singular Tailings - plural

C B ABADJIEV (Higher Institute of Architectureand Civil Engineering, Sofia)I should like to express my gratitude to theBritish Dam Society for the opportunity to shareexperience with British colleagues. I wish togive a brief overview of paper 8 with particularattention to three problems:- The area of the tailings dam was the same asthat of Nottingham University (where theConference was held), about 300 hectares. Thedam was built by the downstream method using 75cyclones on the starter dam crest, the numberincreasing to 130 during this last year, the 8thyear of construction.- It was difficult to achieve the design slopefrom the coarse discharge from the cyclones.This was overcome by enlarging the dischargeaperture of the cyclones to give a wetter mixthat spread more readily.- Dust control. Rotational sprinklers with arange of 45m, then vertical wetted screens wereused but with various problems. Currently twonew methods using chemical and chemical/biological stabilisation are being used. Thisinvolves waste from the timber paper celluloseindustry and seeding grass, which has to be keptwatered.

S A CALE (Knight Piesold)Following Dr Penman's remarks, I should like tomake several points:

N THOMPSON (Knight Piesold)Dr Penman asked about the value of computersoftware such as TADAM (paper 10).Tailings dams involve retaining structures(referred to as the "wall" in the ICOLD manualon tailings dams and dumps) and retainedmaterial (referred to as the "stockpiledmaterial"). The ratio of volume of stockpiledmaterial to volume of wall varies widelydepending on the topography. One of the aims ofthe design process is to maximise this storageratio within the constraints of the site. TADAMis a design tool. It models tailings depositionand calculates the capacity of an irregularlyshaped impoundment filled with a material thathas a sloping upper surface. This calculationcan be carried out by hand using contour plansand a planimeter. The value of TADAM lies inthe automation of an otherwise tedious andtime-consuming design process.

Paper 10 describes the use of TADAM for thedesign of two tailings dams. The Los Leones damat Andina is constructed in extremely ruggedterrain and would have a final overall height of300 m. The Sohar dam in Oman is constructed ina relatively flat area and will have a finaloverall height of about 33 m. The Andina minestudy involved the investigation of four damsites with approximately one hundred possiblefilling patterns for each site. Each fillingpattern would take about one man-day tocalculate using contour plots and a planimeterso the whole volumetric study would take about400 man-days. The actual study using TADAM tookabout 40 man-days. The value of this softwarein saving time and reducing tedium is obvious.

The effort involved in optimising the layout ofa tailings dam decreases as the originaltopography becomes flatter. TADAM would nothave saved a significant amount of time if ithad been used for the design of the Gale Commonash disposal scheme. For this reason thesoftware is not used by Knight Piesold's officein Sydney, Australia. However, the value hasbeen proven by use on many schemes at theirAshford, Derby, Denver and Vancouver offices.

114

DISCUSSION

S HAWES (Independent Consultant)For some twenty years, my practice has beenretained by a major agricultural produceprocessor to design, construct, maintain andrepair settlement lagoons and effluentreservoirs. The ten million tonnes of produceprocessed each year results in 75% beingdischarged as water, and 7% as topsoil. Thus,with rainfall and plant foul water run-off,storage lagoons are required for nearly tenmillion tonnes of water and seven hundredthousand tonnes of soil. This output is storedin lagoons awaiting purification beforedischarge to river, or excavation to dryingbeds. These lagoons vary in depth from two toten metres, and have a capacity of betweenthirty thousand and two hundred thousand cubicmetres.

In some cases the mud, which varies from finesand to clay, is dropped as "tailings", thewater passing on into storage in reservoirswhich the client accepts as such under the 1930and 1975 Acts. More recently there has been achange to combining settlement with storage,thus a lagoon will contain about one tenthsettled mud, nine tenths water with a BOD of4000 and a fair amount of rotting organicmatter.

Failures during some thirty years have variedfrom breaches due to overfilling, land drainsbelow embankments blowing, causing water supplyintakes in the rivers below to be closed, banksslipping due to their having been cut away atthe base when mud was being dug out and crackingdue to inadequate compaction. Old mild steelpipes have rotted, and in one case the pipere-activated itself and sent the contents of onelagoon down to the town sewage works which itclosed for several days. Frequently ponds areemptied of mud by cutting the banks away, andthen re-building without compaction.

Whilst Section 1 of the Act refers to "water assuch" it immediately goes on to say thataccordingly it does not include a mine or quarrylagoon which is a tip within the meaning of theMines and Quarries (Tips) Act 1969. What itdoes not say, is that any lagoon which is not atip within that Act is automatically outside theReservoirs Act 1975. It then says quiteunequivocally that it covers any reservoir"designed to hold" or "capable of holding" morethan 25 000 cubic metres of "water". Note thatit here refers to "water", and not "water assuch".

Whilst the bulk of the water which my clients'lagoons hold comes as "water as such", it iswithin a short time dirty water with a BOD of3000 to 4000, and, as it enters the lagoon, itcarries topsoil which has been washed off theproduce. I feel I must now refuse to advise onsuch lagoons until they have been registered aslarge raised reservoirs, since whilst they areunregistered they are, in my opinion, illegalstructures, and since I would have no legalpower to fix a top water level, if failureoccurred I could be an "accessory before orafter the fact". Many of them have not beendesigned to hold water, simply because they have

not been designed. They just grew. But theyare capable of holding water as defined inSection 1 of the Act, and I cannot see thatlegally or logically they do not hold "water assuch".

Discussion amongst engineers is often biased bythose who had a hand in fashioning the Act, whoknow what was intended. But now we have tointerpret the Act as it is written, and not asit may have been intended. What is ratheralarming about the present situation is thatthose with experience of cost effectivesolutions in this area, particularly those withresponsibilities for pollution prevention, areto be excluded from practising by edict of theReservoirs Committee. They refuse to recommendthe Minister to create a suitable Panel forthose engineers whose experience is in the fieldof industrial and agricultural medium sizedwater and effluent holding structures. This is afield where more, not less, engineering input isnecessary, and where the very largeconsultancies have little experience of thetechniques which produce the safe and costeffective solutions so necessary to thecompetitiveness of British industry.

A D M PENMANDuring the drafting of the 1975 Reservoirs Actan endeavour was made by the committee of theInstitution of Civil Engineers to bring in thestorage of liquid other than water with a viewto tailings dams and the fact that in Americathey had been storing oil behind embankmentdams. It was turned down as something whichwould put additional complications into therulings of the Act.

The greater density of tailings makes itsdestructive power much greater than that ofwater. Escaping tailings can crush cars anddemolish buildings which, under the same depthof escaping water, would only be damaged bybeing wetted. As Mr Hawes has indicated, manytailings lagoons could hold more than 25 000 m3

of clean water above the tailings if they becomeflooded and it might be argued that they shouldcome within the existing Act, although clearlyit would be preferable to have special provisionfor tailings dams.

A STREET (MRM Partnership)In his introductory presentation Dr Penmansuggested that the upstream method of tailingsdam construction tended not to be the preferredapproach, indeed it was frowned upon. There areof course a wide variety of mining and quarryingprocesses, each producing different types andquantities of solid waste (overburden, wasterock, crushed fines, etc) and tailings. Inaddressing the design of mining and quarryingwaste disposal schemes it is necessary to workwithin certain constraints which include:

the need to utilise the waste materials beingproduced; the rate of production may vary andthis may be a significant controlling factor inthe choice of tailings dam construction method

115

TAILINGS DAMS

- the need to ensure the stability of thestructure, both during construction and longterm

the need to develop a scheme which hasminimal visual impact; this is becoming a keyconcern, particularly in the UKFor certain schemes the upstream method oftailings dam construction may well prove to bethe preferred approach. There are a number ofgood examples of such dams currently underconstruction in the UK which provide an economicand visually unobtrusive solution, as well asbeing technically sound.

With regard to the need for a register oftailings dams, it is my opinion that such aregister is required. A number of tailings damsfigure among the largest "embankment" structuresin the UK. Under the Mines and Quarrieslegislation the "competent person" is requiredto inspect and report on such structures atprescribed intervals. Unfortunately thequalifications and experience required for the"competent person" are not defined. Thereclearly would seem to be some merit in extendingthe existing 1975 Reservoirs Act to covertailings dams since this would bring the"Supervising Engineer' and "Inspecting Engineer"into play. If this is considered to be goingtoo far then a formal register of dams would atleast be an improvement on the current, illdefined position.

E T HAWS (Rendel-Parkman)There has been considerable discussion onanomalies within the Reservoirs Act. One suchitem concerns the exclusion of ash lagoons fromthe legal requirements of the Act. However, theresponsible attitude of the CEGB and itssuccessor companies has been to operate suchlagoons as if they were within the purview ofthe Act. The final legal step of registrationhas naturally been avoided. As an indication ofthe effects of the legal distinction it isworthy of note that at Gale Common the newlagoons for Stage II are firmly within the Actwhilst they contain only water as temporaryprotection to a sealing membrane, whereas theywill come out of the Act directly ash starts tobe placed. Regarding the cost of applying theregulations of the Act to such lagoons, the onlycost item is the fees for certification whichare very small indeed related to project cost.

Studies are currently underway into thepossibility of achieving part of the domedfinish required for landscape architecture onStage I by lagoons within embankments set infrom the main crest. This implies founding theadditional bunds on lagooned ash from which itis hoped that floaters have been displaced.Nevertheless allowance will have to be made forthe residual presence of floaters which pose aninteresting soil mechanics problem as they areof less density than water. Floaters orcenospheres are hollow glass balls of very smalldimensions. They are of considerable value asfiller or base for such items as paint and facepowder if they can be retrieved economically.Such retrieval is currently underway but is a

difficult operation as handling is troublesomeand ash blow must be avoided at all costs,particularly with the adjacent M62. Removal ofthe floaters also requires stripping off thesubstantial vegetation which has floated up thusleading to a chance of wave action and alsoremoving the habitat currently enjoyed bywildlife.

The remark has been made that dams have been putout of operation because of mining subsidence.However, it is the case with power stations onthe coalfields that, although the power stationsthemselves are on a pillar sterilised fromextraction, the ash lagoons are not soprotected. Movements of a metre and above havebeen experienced and the lagoons have been putback into service satisfactorily after repairingcracks, drainage systems etc. For this reasonthe lagoon arrangements commonly includealternative containers so that active slurryplacing can avoid lagoons currently sufferingsubsidence.

The culvert containing the outlet pipework fromthe Gale Common outfall towers has been subjectto foundation settlement and stretch under 50 mof fill and a sag of 600 mm has beenaccommodated safely.

Finally on the subject of membranes, acomparatively cheap polythene sheet has beenplaced over the whole of the base area of theStage II lagoons to ensure protection of theunderlying aquifer from leachate. It was notfeasible economically to underdrain or surchargethis membrane and consequently "whales" appearedwhen the lagoon was flooded. This phenomenon isapparently common in these circumstances and webelieve was entirely due to trapped air with noknown presence of methane. In view of thelimited purpose of the membrane the procedureadopted was to make a small incision in each"whale", thus releasing the trapped air. It isbelieved that any residual seepage through thesesmall cuts will be quite negligible in terms ofany possible effect on the aquifer.

R ORANGE-BROMEHEAD (Rendel, Palmer and Tritton)The new ash lagoons at Gale Common have beenlined with a plastic membrane to minimise thepossible pollution of the underlying aquifer.When laying the membrane, material was placed onit only sufficient to keep it down against theeffects of wind. Upon filling the lagoon it wasexpected that the water would drive the air outfrom under the membrane. In the event, airbecame trapped in bubbles which appeared abovethe rising water giving the impression of a bayof whales. The solution finally adopted was tolower the lagoon level so that a man in waderscould puncture each 'whale' with a smallincision. The loss of impermeability to thefuture ash filled lagoons is not seen assignificant.

116

DISCUSSION

H T MOGGRIDGE (Colvin & Moggridge)I wish to refer to some landscape aspects of theGale Common scheme:- The plants on the top of the lagoon, growingon the floaters, were not an accident, but weredesigned as a means of suppressing dust.- The landscape consultant can make acontribution to the edge of the site in terms ofsaving existing woods, planting etc. This isimportant for good public relations.- Some trees have been planted in the ash-hilland have 6 years growth. Trial holes have shownroots entering the ash layer beneath 150mm oftopsoil. There is a limited number of speciesthat grow successfully on ash.- The after use of the site will be mainly foragriculture including fields for sheep and somefor arable with some areas of woodland.- The cost of this landscaping work is about 1%of the total. We were landscape consultants atRoadford reservoir and there the cost was about2%.

A R GRIFFIN (Binnie and Partners)Further to the comments made by Dr Penman duringhis summarising of paper 9:- Gap in the piezometric data. It should beexplained that data was obtained during thisperiod. Readings showed a rapid increase inpore pressures suggesting that failure of thetrial embankment was imminent. However, visualobservations and readings of deformation andsettlement did not indicate that the embankmentwas under undue stress. Subsequently thesupervising engineer obtained a second pneumaticpiezometer readout unit to check results andfound that the original unit was malfunctioning.As a result it was decided to remove theerroneous or suspect data from the graph.

- Vegetation under the trial embankment was cutand rolled to form a mattress under theembankment fill. Larger tree roots were grubbedout. This is normal practice in south east Asiaparticularly under road embankments crossingsoft alluvial clays. The mattress acts like areinforcing fabric and supports the fill butmakes stability analysis difficult.

- The original design was based on limited vanestrength data obtained from boreholes put downby others nearby for factory construction. Thedata was insufficient to enable a detaileddesign to be carried out and subsequently a moreextensive investigation included closely spacedin situ vane tests and piston sampling.However, the results of laboratory tests on thepiston samples suggested that the material wasdisturbed and therefore could not be relied on.It was necessary before this data becameavailable to produce a preliminary design withestimates of quantities for tender purposes.The preliminary design was thereforeconservative and the aim of the trial embankmentwas to define a more realistic but stablestructure.

The trial embankment enabled a reassessment ofthis design with savings on materials. Also therevised design will enable a review of the

process of placing the waste gypsum materialbehind the retaining embankments and the heightsto which the gypsum can be stacked. Thepreliminary design allowed the waste material tobe stacked in layers 1 m thick with 20 to 35 mwide benches between layers raising the wholestructure to about 12 m high in the centre.This design is being reviewed as a result of thetrial embankment test with a view to reducingthe bench widths and increasing the storagecapacity of the tip.

- Gypsum waste or tailings. This materialconsists of fine coarse silt size crystals(generally needle or lozenge shaped) mixed withiron hydroxides. It can be placed in thelandfill area and rolled and compacted.Laboratory compaction tests (2.5 kg) indicate amaximum dry density of about 1.27 Mg/m3 andoptimum moisture content of 24%. The moisturecontent of the waste on production is about 23%but 'ages' rapidly, the iron hydroxide rapidlydehydrating to iron oxides which in turn act asa cementing agent between the gypsum crystals.Consequently its properties are time dependent.Quick undrained tests on compacted samples withmoisture contents ranging from 38% to 13% gavestrengths of 119 kN/m? to 317 kN/m2

respectively. Effective stress tests gave <(>'values in excess of 50° at low effective cellpressures.

Tests and observations therefore indicated thatthe properties of the soft marine clays woulddictate the final design of the embankment andtailings retention area. Also since the factorof safety of the embankment was very sensitiveto small changes in the adopted strengthparameters of the soft marine clay it wasconsidered that a trial embankment was essentialto enable a sensible economic design to beproduced.

B G CHIN (Klohn Leonoff)Dr. Penman has drawn a comparison between papers1 and 9. A difference is that a total stressmethod was used in paper 9, whereas we usedeffective stress methods in paper 1. When usingundrained strengths, you must predict how fastthey will increase during construction, whereaswith effective stresses you take pore pressuredissipation into account. In our Forty MileEast Dam we observed virtually no dissipation,while at the trial embankment in Malaysia ruwas of the order 0.5, indicating quite a lot ofdissipation during construction, giving a gainof strength in the foundation.

C G GREGORY (Rofe Kennard and Lapworth)With reference to the question of whether ashlagoons came under the provisions of the Minesand Quarries (Tips) Act 1969 or the ReservoirsAct 1975 and its predecessor the Reservoirs(Safety Provisions) Act 1930. My firm has beenresponsible for the detailed design andsupervision of construction of several large ashlagoons and the most recent lagoon at FiddlersFerry Power Station stored water to a depth of

117

TAILINGS DAMS

approximately 10 metres above the natural levelof Cuerdley Marsh on which it was constructed.The enclosing lagoon embankment wasapproximately 3 kilometres in length and henceit could be seen that a very considerable volumeof water was stored. Such a structure surelycame within the Reservoirs Act 1975 and the Actwas followed in that an appropriate panelengineer was appointed to undertake the designand supervision of construction of this andother similar ash lagoons elsewhere.

G N JONES (National Power)There have been a number of references in thepapers presented today and ensuing discussionsto ash lagoons operated by the CEGB and one ofits successor companies, National Power.Firstly, I would like to offer some reassuranceto everybody that National Power, as aresponsible company with operational safety asone of its highest priorities, intends tocontinue the policy of the former CEGB and totreat large ash lagoons as if they are

reservoirs under the Reservoirs Act. This willeven apply when they are not registered as suchwith the local authority.

In the discussions about extending the scope ofthe Reservoirs Act to cover tailings dams, Iquery whether this could be done withoutconsiderable thought and probably re-drafting tocover extra provisions particular to this typeof structure. A couple of points come to mind:

Tailings type dams typically have very longconstruction periods and the Reservoirs Actrequirement for a single Construction Engineermay be difficult to meet. It would be nice tothink that Panel Engineers were immortal but, inany case, consultancy agreements of the requiredlength would not be palatable to clients such asourselves. A greater emphasis on supervision ofconstruction is, I believe, required.- The definition of capacity and therefore thelower bound definition of reservoirs under theAct needs careful thought in a tailings dam. Atwhat point does the material contained cease tobe liquid or to potentially liquefy during afailure?

118

17. Evaluation of dam safety at a series of hydropower damsincluding risk assessment

D. S. BOWLES, L R. ANDERSON, and T. F. GLOVER, Utah State University, USA, G. S.TARBOX, Consultant Engineer, Wanconda, IL, USA, R. B. WAITE, Utah Power andLight Co./Pacific Corp. USA, and P. E. YIN AU-YEUNG, ECI, Denver, USA

Utah Power and Light Company (UP&L) owns a series of dams on the Bear River in Utah and Idaho,U.S.A. These dams are regulated by the Federal Energy Regulatory Commission (FERC). Not all thedams currently meet the FERC's standards for flood and earthquake loading. A dam safety evaluationstudy was performed using an incremental consequence assessment and a risk assessment. Each damwas evaluated considering its potential for complete or partial failure due to floods, earthquakes,internal causes, or upstream dam failure. The safety evaluation procedure and results arepresented. Also the dam owner's perspective on the role of the dam safety evaluation results inthe selection of remedial measures which were accepted by the FERC is presented.

INTRODUCTION1. Several Utah Power and Light Company (UP&L)

dams on the Bear River in Utah and Idaho, U.S.A.were found inadequate based on the currentprobable maximum flood (PMF) and the recentlyrevised maximum credible earthquake (MCE). Thisfinding was surprising to UP&L management becausethe dams had a satisfactory record of performanceand because the highest floods of record were lessthan 10% of the new PMF's. Also the verycompetitive market for electrical power did notjustify the investments that the FERC appeared tobe requiring. According to UP&L's Waite (1989a)the estimated cost of remedial upgrades was $20to $25 million. This translated to $155 to $195per installed kilowatt or an increase in theaverage cost of generation of three to four millsper kilowatt-hour.

2. UP&L decided to commission a comprehensiverisk-based evaluation of their dams in order toprovide them with a thorough understanding uponwhich to base their proposals to the FERC. Theevaluation was costly to perform, but the savingsin remedial action costs made possible as a resultof the study more than justified the costs (seePara. 36). The study was conducted by ECI, Denver,Colorado, and RAC Engineers and Economists, Logan,Utah. A detailed description of this study ispresented in ECI/RAC (1988).3. The Bear River UP&L dam safety evaluation

was performed through an incremental consequenceassessment (ICA) and a risk assessment (RA). ICA,which is recognized by FERC regulations, providesestimates of increases in economic damages or lifeloss for postulated scenarios of dam failurecompared with cases which consider no dams on theBear River (i.e., natural flows). While ICA dealswith incremental consequences due to a dam beingadded to a natural river system, it does notinclude consideration of the chance of thesefailure scenarios actually occurring; therefore,it is a "what if" type of assessment. To add theperspective of the chance of occurrence, riskassessment was performed for the UP&L Bear Riverdams. The risk assessment approach is

The embankment dam. Thomas Telford, London, 1991

specifically mentioned in FERC EngineeringGuidelines, and it is currently used by the U.S.Bureau of Reclamation (Von Thun 1987) forevaluating the safety of existing dams. Also,its use has been recommended for evaluation ofexisting dams in reports by the American Societyof Civil Engineers (1988) and the NationalResearch Council (1983, 1985).4. The remainder of the paper is divided into

four sections. The next section summarizes theprocedure followed in the Bear River study. Thefour study dams are briefly described in thefollowing section, together with their hydrologicand seismic setting. In the fourth section resultsof the evaluation are summarized for one of thedams and the use of these results in the damsafety decision-making process is discussed. Thefinal section contains some conclusions on thevalue of information obtained from incrementalconsequence assessment and risk assessment in damsafety decision-making.

PROCEDURE5. The Bear River study was conducted in

accordance with the procedures described byBowles, Anderson and Glover (1987) and by the U.S.Bureau of Reclamation (1986). The overallframework for risk assessment in dam safetyevaluation is summarized in Fig. 1. The figure isdivided vertically into the four major steps ofthe risk assessment procedure. These steps aredescribed in the next subsection. The developmentof a risk model, upon which the risk assessmentis based, is described in the followingsubsection. Fig. 1 is divided horizontally intothe sequence of events represented by the model.

Risk Assessment Steps6. In this section the four major steps in a

dam safety evaluation risk assessment aredescribed (see row headings in Fig. 1). Riskidentification involves recognizing and listingthe various factors which could contribute to therisk of dam failure and organizing these intological event sequences which cover all reasonably

119

RISK, HAZARD AND SAFETY

IDENTIFICATION

ESTIMATION

INITIATINGEVENT

EXTERNAL

SEVERE STORMEARTHOUATC

UPSTREAM DAM

FAILUREINTERNAL

PIPING

SYSTEMRESPONSE

OVERTOPPINGDEFORMATION

SLOPE INSTABILITY

OUTCOME(PARTIAUCOM°LETE)

BREACHNOSREACH

EXPOSURE

TIME OF DAY

SEASONWARNING TIME

CONSEQUENCE

ECONOMC DAMAGELOSSOFUFE

EONVIRONMENTAL

SOCIAL

UPSTREAM

WATERSHED CHANCESUPSTREAM DAM

S T R U C T i m .

MODIFICATIONS

SAFETY INSPECTIONS

WARNING SYSTEMSROOD PROOFING

FVEFCENCY

RELOCATIONSLAND USE

ZDNNG

EMENTS INSTRUMENTATIONOPERATING

RESTRICTIONS

SELECTION OF RISK

AVERSION MEASURES

ACCEPTANCE

Fig. 1. Risk-Based Method for Assessing DamSafety Improvements (Adapted from Anderson et al,1987)

probable failure modes. Such an organization isreferred to as an event tree. It serves as therisk model for evaluation of existing dam safety,or the effectiveness of proposed rehabilitation(risk aversion) alternatives. The second step isrisk estimation which involves assigningprobabilities and consequences to each failuremode, represented by a branch in the event treemodel.

7. The product of the second step is anestimate of the probability of failure, and lifeloss or economic consequences associated with eachfailure mode, or combination of failure modes, forthe existing dam (i.e., the do nothingalternative). If these risks are unacceptable,the assessment proceeds to the third step of riskaversion. This involves formulation andevaluation of remedial action (rehabilitation)alternatives. Risk aversion can be achieved byreducing the probabilities associated with anevent tree branch, or by reducing theconsequences. In both cases structural andnonstructural measures should be considered. Fig.1 lists examples of aversion measures in theaversion step part of the diagram. These examplesare linked by arrows to the probability orconsequence that would be expected to be reducedby their implementation. The product of theaversion step is an estimate of these reductionsfor each aversion measure.8. The final step in the risk assessment

process is the decision on what degree of safety,or equivalently what residual risk, is acceptable.Although the engineer can supply information andrecommendations for the risk acceptance decision,the decision is usually made by the dam owner,operator or regulator. Information available froma risk assessment can be categorized into severaltypes: probabilistic, economic, safety, legalliability, and insurance. Bowles (1990) discussesthe role of risk acceptance criteria in dam safetydecision-making.

Risk Model Development9. Risk model development commences with the

identification of a sequence of events (see columnheadings in Fig. 1), beginning with events thatcan initiate dam failure, and ending with the

120

consequences of failure. Initiating events canbe classified as external or internal. Externalevents include earthquakes, floods, and upstreamdam failure. Internal events includechemical/physical changes in soil or concreteproperties or latent construction defects. At lowlevels these events would not normally lead to damfailure. However, at high inflow rates a rapidrise in pool level could lead to overtopping, ora severe earthquake could result in structuraldeformation or liquefaction. These and otherdam-foundation-spi1lway-reservoir system responsescan lead to the outcome of the sudden release ofthe reservoir contents. The magnitudes of theresulting life loss and property or environmentaldamage will depend on various exposure factors.These can be defined by flood routing to determinethe path of the flood wave, area of inundation,and travel time; the time of the day and seasonof the year; and the effectiveness of any warningsystems and evacuation plans. Consequences areclassified as life loss and economic loss whichincludes property damage, cost of dislocations,and lost project benefits. Environmental andsocial consequences also can be considered.

10. During the identification step,professional judgment and experience, review ofavailable information, and site visits are usedto develop a list of the types of initiatingevents, system responses, outcomes, exposurefactors, and consequences which apply to aparticular dam-foundation-spillway-reservoirsystem. Using this information an event tree isdeveloped. Each branch in the event treerepresents a failure mode.

11. To implement the risk model, requires theestimation of probabilities and consequences foreach event tree branch. Several cases should beconsidered: 1) natural flow (i.e., no dam); 2a)existing dam without failure; 2b) existing damunder various failure modes; and 3) variousstructural and non-structural rehabilitationalternatives (including different levels of eachalternative, e.g., various spillway capacities).

STUDY AREA AND DAMS12. In downstream order the study dams are:

Soda Point, Grace, Oneida and Cutler. UP&Lregulates flows in the Bear River by divertingwater to and from Bear Lake which is locatedupstream of Soda Point Dam. The hydrologic regimeis dominated by the spring snowmelt event. Ageneral storm of regional extent, combined withsnowmelt, defines the PMF. However, the riskassessment also considers local summerthunderstorms since these independent events canoccur at high enough magnitudes to threaten thesafety of the Bear River dams. Procedures used forestimating flood frequency relationships andextending them to the PMF are described in Au-Yeung and Anderson (1989).

13. The seismic hazard in the region isstrongly related to the presence of the 370-kmlong Wasatch Fault zone which extends fromGunnison, Utah on the south to Malad City, Idahoon the north. The Wasatch Fault zone is an activewestward-dipping, normal fault. Geologic evidenceindicates that it has experienced many largemagnitude earthquakes during the last 10,000years. Future activity is expected to produceearthquakes with a maximum Richter magnitude in

PAPER 17: BOWLES, ANDERSON, GLOVER, TARBOX, WAITE AND AU-YEUNG

the range 6.5 to 7.5. A uniformly-distributedbackground seismicity was used to represent otherfaults in the region that are capable ofgenerating small to moderate earthquakes, whichwould not rupture the ground surface. Thehistoric record was used to establish the annualfrequency of exceeding given levels ofacceleration (Au-Yeung and Anderson 1989).Maximum accelerations ranged from 0.45g at SodaPoint, Grace, and Oneida Dams to 0.66g at CutlerDam.

UP&L Dams14. The four UP&L study dams are briefly

described in this section. For the Oneida Dampotential failure modes and remedial actionalternatives are also described as background forthe presentation of results in the next section.

15. Soda Point Dam is a concrete gravity damwith a short embankment section at the leftabutment. It is 103-feet high and 490-feet longwith a 14,000 kW generating station integratedinto the concrete section. It has a spillwaycapacity of 63,000 cfs. and a PMF of 72,100 cfs.The dam was built in the mid-1920's.

16. The 52-foot high Grace Dam is a timber cribstructure, which serves a 33,000 kW powerplantthrough a 5.5 mile penstock. It has a long but lowembankment section extending to the rightabutment. The spillway capacity is about 14,000cfs. and the is 63,700 cfs. Grace Dam was builtin 1910 and modified in 1951.17. The Oneida Dam, built shortly after 1910,

is a concrete gravity structure, 110-feet high.An earth embankment dam, separated from the maindam by a ridge, closes off a low saddle to theleft of the main dam and has a crest elevationsix-feet higher than the concrete dam. The totalspillway capacity is about 12,000 cfs. and the PMFis 74,700 cfs. The hydropower generating stationhas a capacity of 29,000 kW. The potentialhydrologic failure modes are overstress of theconcrete dam and overtopping of the embankment.Overstress failure of the concrete dam and slopeinstability of the embankment were considered asthe earthquake failure modes. Failure modes ofthe concrete dam was postulated to result in totalfailure of the dam and failure of the earthembankment was postulated to range from partialfailure to total failure. Several structural andnon-structural remedial actions were developedto the conceptual level for the Oneida Dam site.They were: 1) no action, 2) decommission the dam,3) anchor the main concrete dam to allowovertopping and raise the embankment dam toprevent overtopping, and 4) install a floodwarning system.

18. The 112-foot high Cutler Dam is a concretegravity arch structure that was built in the late1920's. The hydropower generating station has acapacity of 30,000 kW. The spillway capacity is22,000 cfs. and the PMF is 195,500 cfs.

SUMMARY OF RESULTS19. Each of the four UP&L dams was assessed

individually and for its serial interactions withother dams. To illustrate the type of informationobtained from the study results from Oneida Damare used. The event tree risk model for OneidaDam is presented in Fig. 2. A summary of resultsfor the Oneida Dam is presented in Table 1. The

Table 1. Summary of Findings - Oneida Dam

FERCguidelines

Economic damage

Prob. (net life loss)

Cost-to-save-a-life

Prob. (breach failure)

Prob. (partial failure)

Benefit-Cost ratio

Total annual cost

Hydrologic

General:Non $ 0 5 mUP&L 27-5 I

Thunder:Non $ 0-3 nUP&L 25-7

1 in 17 500/y

Earthquake

Non $ 0 8 mUP&L 26 2

8 lives1 in 43 500 y

Concrete Embankment8 lives 6 lives1 in 18 OOO y 1 in 110 OOO y

Decommission $ 1 6 B

Anchors $ 2 7 B|

Warning system $ 107 m

1 in 43 500 y 1 in 15 500 y

1 in 6500/y

Anchors < 1 %

Do nothing 5 2 400 (min)Decommission $ 249 000Anchors $ 420 OOO

table is divided vertically into sections forconventional dam safety assessment following theFERC guidelines and sections for ICA and RA.Results are summarized for hydrologic, earthquakeand internal initiating event types. Referenceis made to the following remedial actionalternatives: do-nothing, decommissioning thedam, anchoring the concrete dam and raising theembankment, and implementing a dam failure warningsystem.

Incremental Consequence Assessment20. Incremental Hazard to Human Life: Failure

of the existing Oneida Dam is not expected toresult in additional life loss above thatprojected due to the effects of a natural floodwithout the dam in place. This finding alsoapplies to cases where Oneida Dam fails due toflood-caused failure of the upstream Soda PointDam. Therefore, upgrading of the dam to safelypass the PMF (by installing anchors in theconcrete dam and raising the embankment) is notprojected to reduce hazard of life loss.

21. For an earthquake-caused failure of OneidaDam, life loss is predicted to be about eightlives. Upgrading the dam to withstand the maximumcredible earthquake could be achieved by addinganchors to the concrete dam which would reducepredicted life loss to zero.

22. Incremental Economic Damages: Increasesin economic damages due to dam failure vary withthe flow rate at which dam failure is postulated(see Fig. 3). The maximum increase for theexisting Oneida Dam for a general storm flood isprojected to be $28 million, with only $0.5million of non-UP&L losses. For thunderstormfloods, the maximum increase is projected to be$26 million with only $0.3 million of non-UP&Llosses. These levels of UP&L damages, while notsmall, are according to UP&L representatives,within insurance coverages that UP&L carries.

23. Damages for earthquake failure of theexisting Oneida Dam are estimated to be up to $27million for an overstress failure of the concretedam with only $0.8 million of non-UP&L losses.No earthquake failure damages are predicted forthe embankment dam because it is considered tomeet the MCE standard.

Risk Assessment24. Risk of Incremental Life Loss: No chance

121

Load

ing

In

itiat

ing

type

ev

ent

Sys

tem

resp

onse

Ou

tco

me

Hyd

rolo

gic

even

ts

Sei

smic

_ev

ents

Inte

rnal

_ev

ents

Failu

reu

pst

ream

—

dam

Gen

eral

sto

rm

Thun

ders

torm

Ear

thqu

ake

0 to

750

0 cf

s

7500

to

14

500

cfs

74 7

00 t

o 1

01 7

00 c

fs-

0 to

750

0 cf

s

7500

to

14

500

cfs

68 5

00 c

fs

0 to

00

5 g

00

5 to

0-1

0 g

> 0-

45 g

Con

cret

e ov

erst

ress

».

Foun

datio

n a

bu

tmen

t J]

>M

isce

llane

ous

—'

Pip

ing

em

ban

kmen

t d

amS

lope

sta

bilit

y em

ban

kmen

t d

amP

ipin

g a

t ou

tletw

orks

Foun

datio

n s

tabi

lity

emb

ankm

ent

dam

-

Thun

ders

torm

abov

e S

od

aP

oint

Dam

Ear

thq

uak

eat

So

da

Poi

nt D

am

Inte

rnal

at S

od

aP

oint

Dam

0 to

750

0 cf

s

7500

to

14

500

cfs

> 68

600

cfs

22

500

cfs

22 5

00 c

fs

- N

o f

ailu

re-

Con

cret

e ov

erst

ress

- E

mba

nkm

ent

over

topp

ing

- N

o f

ailu

re-

Con

cret

e ov

erst

ress

- E

mba

nkm

ent

over

topp

ing

- N

o f

ailu

re-

Con

cret

e ov

erst

ress

- E

mba

nkm

ent

over

topp

ing

- N

o f

ailu

re-

Con

cret

e ov

erst

ress

- E

mba

nkm

ent

over

topp

ing

- N

o f

ailu

re-

Con

cret

e ov

erst

ress

- E

mba

nkm

ent

over

topp

ing

- N

o f

ailu

re-

Con

cret

e ov

erst

ress

- E

mba

nkm

ent

over

topp

ing

- N

o f

ailu

re

- C

oncr

ete

over

stre

ss

- N

o f

ailu

re

- C

oncr

ete

over

stre

ss

- N

o f

ailu

re

- C

oncr

ete

over

stre

ss

- N

o f

ailu

re

- C

oncr

ete

over

stre

ss

• N

o f

ailu

re

• S

truc

tura

l fa

ilure

- N

o f

ailu

re•

Con

cret

e ov

erst

ress

- E

mba

nkm

ent

over

topp

ing

- N

o f

ailu

re-

Con

cret

e ov

erst

ress

- E

mba

nkm

ent

over

topp

ing

- N

o f

ailu

re-

Con

cret

e ov

erst

ress

- E

mba

nkm

ent

over

topp

ing

- N

o f

ailu

re-

Con

cret

e ov

erst

ress

- E

mba

nkm

ent

over

topp

ing

- N

o f

ailu

re-

Con

cret

e ov

erst

ress

- E

mba

nkm

ent

over

topp

ing

Pas

s flo

odB

reac

h c

oncr

ete

dam

Bre

ach

em

ban

kmen

t

Pas

s flo

odB

reac

h c

oncr

ete

dam

Bre

ach

em

ban

kmen

t

Pas

s flo

odB

reac

h c

oncr

ete

dam

Bre

ach

em

ban

kmen

t

Pas

s flo

odB

reac

h c

oncr

ete

dam

Bre

ach

em

ban

kmen

t

Pas

s flo

odB

reac

h c

oncr

ete

dam

Bre

ach

em

ban

kmen

t

Pas

s flo

odB

reac

h c

oncr

ete

dam

Bre

ach

em

ban

kmen

t

No

rel

ease

of

rese

rvoi

r co

nte

nts

Bre

ach

of

conc

rete

dam

No

rel

ease

of

rese

rvoi

r co

nte

nts

Bre

ach

of

conc

rete

dam

No

rel

ease

of

rese

rvoi

r co

nte

nts

Bre

ach

of

conc

rete

dam

No

rel

ease

of

rese

rvoi

r co

nte

nts

Bre

ach

of

conc

rete

dam

No

rel

ease

of

rese

rvoi

r co

nte

nts

Bre

ach

of

conc

rete

dam

Pas

s flo

odB

reac

h c

on

cret

e d

amB

reac

h e

mb

ankm

ent

Pas

s flo

odB

reac

h c

oncr

ete

dam

Bre

ach

em

ban

kmen

t

Pas

s flo

odB

reac

h c

oncr

ete

dam

Bre

ach

em

ban

kmen

t

Pas

s flo

odB

reac

h co

ncr

ete

dam

Bre

ach

emb

ankm

ent

Pas

s flo

odB

reac

h co

ncr

ete

da

mB

reac

h em

ban

kmen

t

c/)

<e

3.0

2.8

-

2.4

2.2 2

1.8

1.8

1.4

1.2 1

0.8

_

0.6

-

0.4

-

0.2

-

0

INC

RE

ME

NT

AL

D

AM

AG

ES

($

)O

MC

I0A

DA

M -

DCM

CRAL

STO

CM

2.8

2.4

-

2.2

-

2 -

1.8

-

1.8

-

1.4

-

1.2

-

1 -

0.8

"

0.6

"

0.4

"

0.2 0

CO

NC

RET

E D

AM

OV

ER

STR

ES

S

FAIL

UR

E-

TO

TA

LE

MB

AN

KM

EN

TO

VER

TOPP

ING

FA

ILU

RE

- TO

TAL

.SE

E

EX

PA

ND

ED

SC

ALE

DE

TA

IL

BEL

OW

INF

LO

W (

kofs

)Q

TO

TAL

+

NO

N-U

P&

L

DE

TA

IL-E

XP

AN

DE

D

SC

AL

E

CO

NC

RE

TE

DA

MO

VER

STR

ESS

FA

ILU

RE

- N

ON

- U

P 8

L

NO

FA

ILU

RE

AN

D N

ATU

RA

L F

LOW

- TO

TAL

EM

BA

NK

ME

NT

OVE

RTO

PPIN

G F

AIL

UR

E-N

ON

-UP

a

L

NO

F

AIL

UR

E-N

ON

- U

P 8

L

NA

TUR

AL

-NO

N-

UP

a L

INF

LO

W (

kofs

)D

TOTA

L

\ N

ON

-UP

&L

Fig. 2.

Event

tree for O

neida

Dam

(ECI/RAC 1988)

Fig. 3.

Incremental

Dama

ges

for

Oneida Dam

General

Storm

(ECI/RAC 19

88)

PAPER 17: BOWLES, ANDERSON, GLOVER, TARBOX, WAITE AND AU-YEUNG

of incremental or increased life loss is projecteddue to flood-caused failure of Oneida Dam whencompared with the case of no dam.

25. It is predicted that if an earthquakefailure of the concrete dam occurred, then about8 lives could be lost. The chance of failure ofthe Oneida concrete dam is estimated to be 1 in43,500 per year for earthquake failure, and 1 in18,000 per year for an internal failure. Ifinternal failure of the embankment occurred, thenit is estimated that about 6 lives could be lost.The chance of such a failure occurring isestimated to be about 1 in 110,000 per year.These chances of life loss resulting from thefailure of Oneida Dam are much lower than thehistorical probability of life loss from damfailures in the United States due to all causes(i.e. flood, earthquake and internal failures),which is 1 in 5,000 per year. However, theexisting concrete and embankment sections of theOneida Dam have been found to satisfy FERCcriteria with respect to their internal condition.Also, the embankment section is considered to meetthe MCE standard.

26. Cost-to-Save-a-Life: The cost ofincreasing human safety can be expressed on a "perstatistical life saved" basis (i.e.cost-to-save-a-life). This is the cost ofproviding safety and is not in any sense a valuefor human life. Since no life loss could beattributed to the Oneida Dam under flood loading,it follows that upgrading of the dam would not bepredicted to save any lives. Therefore, thecost-to-save-a-life for remedial upgrading of theflood performance of the dam is infinitely large.

27. The cost-to-save-a-life for installinganchors in the concrete section of the Oneida Damto withstand the maximum credible earthquake iscalculated to be approximately $2.7 billion perlife saved. A dam break/flood warning system wasconsidered for reducing the hazard to human lifein the event of an earthquake or internal failureof Oneida Dam. It was calculated that thecost-to-save-a-life for this system isapproximately $107 million per life saved.However, this system is not expected to reducelife loss at the Oneida Hydro Facility itself,since it is located immediately below the dam.If the Oneida Dam is decommissioned, thecost-to-save-a-life is calculated to beapproximately $1.6 billion per life saved.

28. These costs can be compared withcosts-to-save-a-life calculated for regulatedareas such as nuclear power plant design ($4 - $10million), environmental protection ($4 million)and occupational health and safety ($4.5 millionand $300 million for OSHA Benzene regulations).

29. Probability of Dam Failure: The chance ofa breach failure of Oneida Dam, from floods,earthquakes, internal causes and upstream damfailure (see Fig. 4), is estimated to be 1 in6,500 (1.6 x 10 ) per year. This is lower thanthe historical probability of dam failure in theUnited States due to all causes. The chances offailure of either the concrete or embankment damare estimated to be approximately equal (see Fig.

30. Information on annual failure probabilityis combined with the incremental consequenceassessment results (incremental economic damages)in histograms in which incremental damages are

UPSTREAM DAM (7.7%)

INTERNAL (41.0%)

GENERAL (22.4%)

THUNDER CU.1%)

EARTHQUAKE (14.7%)

Fig. 4. Probability of Failure by InitiatingEvent for Oneida Dam (ECI/RAC 1988)

INTERNAL - EMBANKMENT (5.8%)

INTERNAL - CONCRETE WALL (35.3%)

CONCRETE OAM OVERSTRESS (18.8%)

EMBANKMENT OVERTOPPING (38.1%)

Fig. 5. Probability of Failure by System Responsefor Oneida Dam (ECI/RAC 1988)

shown separately for general and thunderstorminitiating events, and for damages occurring tothe owner (UP&L) and to other parties (non-UP&L).Figs. 6 and 7 are examples of histograms of netdamages to UP&L and to other parties for thegeneral storm. Failure probabilities are dividedbetween components attributed to the concrete damand the embankment dam hydrologic failure modes.

31. Benefit-Cost Ratio: Economic benefits arepredicted to be less than one percent of theestimated cost for installing anchors in theconcrete dam and raising the embankment. Nostructural alternatives were considered forinternal failure modes since the Oneida Dam hasbeen found to meet FERC standards for these cases.

O.OOOO2-

0.OOOOI -

0-1 1-3 3-3D 30-35 35-37 >3?DAMAGES (MILLION t)

V / I CONCRETE DAM »\M EMBANKMENTOVERSTRESS OVERTOPPING

Fig. 6. Histogram of Net Non-UP&L Damages forOneida Dam - General Storm (ECI/RAC 1988)

123

RISK, HAZARD AND SAFETY

0.00003 -

0.00002-

a: 0.00001-

2-30 20-25 25-2?DAMAGES (MILLION i)

CONCRETE DAMOVERSTRESS

E 3 EMBANKMENTOVERTOPPING

Fig. 7. Histogram of UP&L Damages for Oneida Dam- General Storm (ECI/RAC 1988)

32. Total Annual Cost: The sum of thepredicted annualized damages (net risk costs) andestimated annualized costs is $420,000 forinstalling anchors in the concrete dam and raisingthe embankment, $2,400 for the do-nothingalternative (i.e. maintain the existing dam) and$249,000 for decommissioning the facility. Thus,the existing dam alternative was found to have thelowest total annual cost of these threealternatives.

33. Environmental Impacts: A reconnaissance-level environmental evaluation of dam failureimpacts was performed. For flood-caused failurescenarios, the additional area of environmentalimpact is predicted to be small when compared tothe natural flooding case. The probability of dambreach impacts occurring was found to beapproximately 1 in 6,500 per year.

Serial Dam Failure Modes34. The potential for serial dam failure was

assessed for all reasonably probable failure modesat upstream dams. This involved many floodroutings and the evaluation of downstreamconsequences. Under the general storm theprobability of dam failure was slightly increasedby the presence of upstream dams for all damsexcept Grace Dam which is protected atintermediate flood magnitudes by the upstream SodaPoint Reservoir. For thunderstorm loading it isonly in the case of an overstress failure of SodaPoint Dam that the probability of downstream damfailure was predicted to increase slightly.Seismic or internal initiating events are notexpected to lead to serial dam failure except atGrace Dam as the result of this type of failureat Soda Point Dam.

Study Outcome35. The Bear River Risk Assessment was

presented to FERC by UP&L, to justify theirproposals for upgrading the dams. These proposalswere accepted by the FERC. The views expressedby the owners representative (Waite 1989b) aretestimony to the value of the risk assessment ofthe four Bear River dams and another on Hams Forkin Wyoming:

36. "Was the Study Worth It?: Including the

internal utility costs the risk assessment studyfor the five dams cost close to $500,000. Whatdid we get for our money besides FERC'sdiscomfort? First and foremost, we developed anin-depth understanding of these dams' potentialfor failure, and we internally justified thenecessary remedial activities. Without thisthorough review, we would probably have had badfeelings about any of the work for a considerabletime to come, and we may have otherwise soughtvery costly legal remedies. Second, the studydeveloped alternatives that would probably havebeen missed or bypassed without this penetratingscrutiny. Based on our initial estimates andcontingency plans, we feel the study came veryclose to saving us $10 million in current remedialcosts, about 40% to 50% of the money we hadanticipated spending. Third, we felt that FERCwas better able to appreciate the benefit ofavoiding some of the work we would have otherwisedone, and we were better able to appreciate someof their concerns ... Fourth, some of the work,such as the incremental flood studies would havebeen needed in any case, and they were a materialportion of the study cost, perhaps 20% of thetotal ... On the whole, it was very well worththe effort."

CONCLUSIONS37. The Bear River study showed that estimated

dam failure probabilities were low. Predictedincremental damages were low, and in most casesdamages would affect the owner to a far greaterextent then other parties. The probability oflife-loss was estimated to be low, and the cost-to-save-a-life was calculated to be high for allstructural and nonstructural alternatives. Noeconomic justification for alternative fixes couldbe shown. Evaluation of potential serial failuremodes did not show large increases in failureprobabilities from this type of initiating event.In the case of Grace Dam, a decrease in failureprobability can be attributed to the protectionprovided by the upstream Oneida Dam. The approachto risk assessment used in the Bear River studydid not involve placing a value on human life, nordid it involve using a specific decisioncriterion, such as minimum total annual cost. Theselection of remedial actions was made by the damowner and regulator using study results and givingweight to any case where incremental life loss wasidentified (however, unlikely) and where socialimpacts of dam failure would be severe.38. The results summarized in this paper

illustrate some types of information which can beprovided by incremental consequence and riskassessment dam safety evaluation studies. Otheruses of information obtainable from theseapproaches include: the assessment of liabilityexposure for dam owners and operators, the choiceof interim measures for improving safety at damswhich are awaiting permanent rehabilitation, theefficient allocation of effort for dam safetystudies, the sequencing of remedial actions at agroup of dams which cannot be budgeted orscheduled to be performed simultaneously, and theprovision of a basis for insurance coverage ofdams. Bowles (1990) describes several riskassessment studies performed for U.S. dam ownerswhich illustrate some of these applications andtheir value in dam safety decision-making. In each

124

PAPER 17: BOWLES, ANDERSON, GLOVER, TARBOX, WAITE AND AU-YEUNG

case the information obtained has proven usefulin dam safety decision-making process.

REFERENCES1. American Society of Civil Engineers (1988)Evaluation procedures for hydrologic safety ofdams. Report prepared by the Task Committee onSpillway Design Flood Selection of the Committeeon Surface Water Hydrology of the HydraulicsDivision. 95 p.2. Anderson, L. R. and D. S. Bowles (1987)Seismic response component in the risk assessmentof dams. Proceedings of the Session on SeismicConsiderations in Risk Analysis of Dams,Structures Division, Specialty Conference,American Society of Civil Engineers, pp. 241-260.3. Au-Yeung, Y. and L. R. Anderson (1989) Damsafety evaluation for a series of Utah Power andLight hydropower dams, including risk assessment:Seismic and hydrologic consideration, andpotential for serial failure. Proceedings of the6th Annual Conference of the Association of StateDam Safety Officials, Albuquerque, New Mexico.4. Bowles, D. S. (1987) A comparison of methodsfor integrated risk assessment of dams. In"Engineering Reliability and Risk in WaterResources", L. Duckstein and E. Plate (Eds.), M.Nijhoff, Dordrecht, The Netherlands.5. Bowles, D. S., L. R. Anderson, and T. F.Glover (1987) Design level risk assessment fordams. Proceedings of the Session on SeismicConsiderations in Risk Analysis of Dams,Structures Division, Specialty Conference,American Society of Civil Engineers, pp. 210-225.6. Bowles, D. S. (1990) Risk assessment in dam

safety decision making. In: Risk-Based DecisionMaking in Water Resources (Ed. Y. Y. Haimes andE. Z. Stakhiv), Proceedings of an EngineeringFoundation Conference, American Society of CivilEngineers. 29 p.7. ECI/RAC (1988) Bear River FERC Part 12DDecision Analysis Study. Technical ConsultingReport to Utah Power and Light Company.8. National Research Council (1983) Safety ofexisting dams: Evaluation and improvement.National Academy Press, Washtington D.C. 354 p.9. National Research Council (1985) Safety ofDams: Flood and earthquake criteria. NationalAcademy Press, Washington, D.C, 321 p.10. U.S. Bureau of Reclamation (1986) Guidelinesto decision analysis, ACER Technical MemorandumNo. 7, Denver, Colorado.11. Von Thun, L.( (1987) Use of risk-basedanalysis in making decisions on dam safety. In"Engineering Reliability and Risk in WaterResources", L. Duckstein and E. Plate (Eds.), M.Nijhoff, Dordrecht, The Netherlands.12. Waite, R. B. (1989a) Dam safety evaluationfor a series of Utah Power and Light hydropowerdams, including risk assessment: Ownerperspectives on the issues. Proceedings of the6th Annual Conference of the Association of StateDam Safety Officials, Albuquerque, New Mexico.13. Waite, R. B. (1989b) Dam safety evaluationfor a series of Utah Power and Light hydropowerdams, including risk assessment: Ownerperspectives on the role of the evaluation in theselection of remedial measures. Proceedings ofthe 6th Annual Conference of the Association ofState Dam Safety Officials, Albuquerque, NewMexico.

125

18. Safety considerations with existing embankment dams and intheir raising

D. J. KNIGHT, E. A. JACKSON, J. HALCRO-JOHNSTON, and C. MAKINSON, Sir AlexanderGibb and Partners, Reading, UK

SYNOPSIS. Experiences with a variety of embankment dams of greatly differing ages in variousparts of the world are described in relation to safety, both in their existing state and in theirraising. The dams involved contain: a concrete core wall; homogeneous sections; an upstream facecore; a central core with rockfill shells; a deteriorating draw-off culvert and ancient sluices.Safety lessons are summarised in respect of the inspection and investigation of, and the designand construction of works on, operational dams.

INTRODUCTION1. Dams, by their nature, are generally

amongst the longest functioning of thestructures bui l t by man, and often s t i l lpresent much of their general o r i g i n a lappearance to the present observer. They areseldom discontinued or removed, because thewater they retain is increasingly valued asboth a basic and a precious resource.

2. With the increased emphasis quiteproperly being given to their continued useand safety, dams-particularly embankments -are being regularly inspected, and wherenecessary investigated, repaired or improved.Because of their increasing age, these actionsinvolve safety considerations at each step ofthat process. Some of those considerationsare described for two cases of existing dams,the first relating to the 66-year-old Coedtydam in North Wales and the second to ancientembankment dams in Sri Lanka.

3. The need to maximize water resources cansometimes require the raising of exist ingdams, which involves additional safetyproblems during the design and constructionstages. Two examples follow, in which theraising of the Gaborone dam in Botswana andHinze dam in Australia are described.

SAFETY OF EMBANKMENT DAMS IN THEIR EXISTINGSTATECoedtv dam - North Wales

4 * H i s t o r y . Coedty dam i s s i t u a t e d in theConway Valley, and was bui l t in 1924 toaugment the hydro-electric capacity of theCowlyd and Eigiau dams conveying water by as e r i e s of tunnels, pipel ines and l ee t s(surface channels) to the Dolgarrog powerstation.

5. The dam (Fig.l) is a 250 m long, 11 mhigh earth embankment with upstream anddownstream slopes of lv:2h and l v : 2 . 5 hrespectively. The upstream slope is protectedagainst wave action by stone pitching, and thedownstream slope is grassed. The embankmentincorporates a thin reinforced concrete cutoffwall placed centrally on the dam axis, and

The embankment dam. Thomas Telford, London, 1991

extending to 3.7 m below natural ground level.6. Bedrock at the dam site is overlain by

some 6-10 m of glacial till drift deposits.The drift typically comprises poorly sorteddeposits of generally granular material, butincluding variable amounts of silts and clays.The embankment is founded on, and constructedof, the till.

7. The draw-off works comprise a 1.8 mdiameter reinforced concrete culvert feedingthe penstock which leads to Dolgarrog powerstation 2 km away and 250 m below. A linevalve and automatic self-closing butterflyvalve are installed on the draw-off pipeimmediately downstream of the embankment.8. In 1925, the completed embankment was

overtopped following failure of the Eigiau damupstream, forming a breach 60 m wide at thetop and 18 m at the bottom. Sixteen liveswere lost in Dolgarrog village. This event,recorded by a bronze plaque adjacent to theroad through the village, led directly toenactment of the 1930 reservoir safetylegislation. The dam was rebuilt, since whena number of other repair works have beencarried out in attempts to reduce the quantityof seepage passing through the dam, andincluding the installation of a riveted steelliner to the draw-off pipe.9. An attempt to reduce seepage by grouting

through the core was made in 1972; grout takeswere high and grout appeared on the surface.In 1986 further wet areas appeared on thedownstream face, and it was then decided thatthe reservoir should be maintained some 1.5 mbelow full storage level. A full safetyassessment of the embankment was initiated.

10. A ground investigation was undertakenin 1988 to establish the soil properties andseepage conditions within the embankment, aswell as to obtain data for stability analysesand the design of any necessary remedialworks. Standpipe piezometers were installedin a number of boreholes and a monitoringprogramme established.

11. The problem identified. The safetyassessment yielded the following conclusions:

127

RISK, HAZARD AND SAFETY

( l )

( 2 )

( 3 )

( 4 )

The concrete core wall was in a poorcondition, and was expected to continueto deteriorate due to soft water attack.In the area where the embankment wasreconstructed following the breaching in1925 the water level in the downstreamshoulder was low, being apparentlydrained by the permeable strata in thefoundations. Elsewhere the phreaticsurface was high.The stability of the downstream shoulderwith the reduced reservoir level wassatisfactory. However, if the reservoirlevel was restored to its original levelfurther deterioration of the core wallwould result in increased seepage. Ifthe phreatic surface rose to a levelcons i s t en t with no core wall, thestability would give cause for concern.Although the upstream slope had shown nosigns of d i s t r e s s as a r e s u l t ofoperating under conditions of continuousand rapidly fluctuating water levels formany years stability analyses for therapid draw-down condition showed a factorof safety of near unity.The draw-off conduit was in poorcondition. The steel lining was corrodedand a short unlined concrete section atthe downstream end showed signs ofleakage. Also, the draw-off conduit wasnot provided with an upstream guardvalve.

12. A number of V-notch weirs wereconstructed to monitor seepage flows.However, because of the permeable strataunderlying the embankment i t was consideredthat flow measurements would be unlikely toindicate actual seepage losses realistically.Nevertheless, measurements made beforelowering the reservoir level had shown asubstantial increase in flow as the reservoirapproached the full storage level . Theelevation at which such increases in flowoccurred was consistent with the areas ofdefective concrete encountered during thedrilling. The highest measured seepage losses

( 5 )

amounted to some 3 1/s, as compared with acalculated value based on flow rates of some20 1/s at maximum water level.

13. The solution adopted. In view of thedamfs history of problems, and its location ina narrow valley perched high above Dolgarrogvillage, i t was considered prudent to designremedial works on pessimistic assumptionsabout future behaviour, viz:-

* The core wall being totally ineffectivein reducing seepage

* Severe rapid draw-down conditions.

Because of fa i lure to reduce seepage throughrepair of the concrete cutoff wall i t wasdecided that provision would be made only tocontrol rather than reduce seepage.

14. Remedial works (F ig .2 ) to theembankment therefore comprised:-

(1) Widening the c r e s t and p lac ingstabilizing fill incorporating a drainagelayer on the downstream slope, dividedinto five separate seepage collectionpanels in plan (Fig.l).

(2) Placing stabilizing f i l l to about halfembankment height on the upstream slope.

15. Similarly pessimistic assumptions weremade on possible future deterioration of thedraw-off culvert, and the following works wereundertaken:-

(3) Lining the existing steel-lined culvertwith a 1.5 m dia. glass-reinforcedplastic liner.

(4) Demolishing the existing 1.8 m dia.unlined concrete cu lve r t sectionimmediately downstream of the embankmentand reconstructing, with fabricated steelliner pipes, and including an accessbranch for inspection.

16. The absence of a shut-off guard valveon the upstream end of the culvert representeda se r ious shortcoming, both from anoperational point of view and also for the

Reservoir

Berm El. 26500

Toe of originol embankment

Wove wallEl. 271-25

(T) = Impervious (till)

(3) - Droinage zone

f Downstream surface reinstated by0 15 m of tcpsoil and seeded

0Original profile

Fig. 1. Plan of Coedty dam

128

Fig. 2. Modifications to Coedty embankment

PAPER 18: KNIGHT JACKSON, HALCRO-JOHNSTON AND MAKINSON

safety of the embankment should leaks developin the buried section of the culvert. Thenormal safety standards called for any pipeinto which people can enter for inspectionpurposes to be protected by two guard valves.In this instance, however, it was agreed thatfor various reasons a single valve would beacceptable. Accordingly, the remedial works(Fig.3) included the demolition of theexisting intake screen supporting structureand i t s reconstruction incorporating abut ter f ly valve and air vent pipe. Anextension of the existing steel access bridgeto accommodate the valve-operating headstockwas also necessary. A number of other worksunrelated to safety were also undertaken.

Footbridge |

Geared headstock

El.270 57

Valve operating shaft

Valve operating shaftguides,attached to air pipe

New 300 NB flanged pipe(air entry/exit)

Original screensbuilt into newheadwall structure

1500NBGRPbellmouth

150 mm dia PVC pipe

El. 259-45

1500 NB butterfly valve—'

• -1500 to 1A00 N B flanged andplain enaed steel pipe 1500long with puddle flange

1—1500 NB flanged steel pipe 1000 longwith 300 NB flanged branch

Fig . 3 . S e c t i o n a l e l e v a t i o n showingmodifications to Coedty draw-off intake

17. Construction. Construction was fairlystraightforward. The main problem was how todeal with the river inflows while working onthe draw-off culvert, as this was the soleoutlet from the reservoir. The only solutionwas to pump. It was decided that the pumpsshould be able to deal with low flows only,and the possibility of the works being floodedoccasionally was accepted by the client as arisk to be paid for if i t occurred. In fact,because of the long dry summer i t was onlytowards the end of the work that anysignificant flooding occurred, but no damagewas caused.

18. The work was carried out between Juneand December 1989. In January 1990 thespillway gates were raised and the originalmaximum r e s e r v o i r operating l eve l wasrestored. Seepage flow measurements andpiezometr ic l e v e l s i n d i c a t e that theembankment is functioning satisfactorily.

Ancient embankment dams in Sri Lanka19. Safety assessment of embankment dams in

Sri Lanka presents a number of uniqueproblems. There are some 300 or so majorr e s e r v o i r s (those supplying 80 or morehectares of irrigation) and tens of thousandsof smaller ones. These are all impounded bylow earth dams or bunds designed to serve thecommercial needs of the country's denselypopulated rural areas. Many of the reservoirs(referred to as 'tanks1 in Sri Lanka) havebeen in existence since the last century andrecords, show that some of them have been inalmost continuous use since first constructedby Sri Lanka's ancient kings (Fig.4), someover 1600 years ago. The tanks form anintegral part of the rural environment and thecommunities depend on them for much more thanjust their irrigation needs. They are part ofthe social fabric of the population so thattheir safety is taken as a matter of course.The failure of the Kantalai Tank bund in April1986 (ref . 1) with the loss of 127 l ivestherefore came as a devastating shock to theSri Lankan people.

Fig. 4. Typical ancient Sri Lankan embankmentdam

20. The need for a universal system forhazard rating for Sri Lankan reservoirs, onethat can be accepted and applied by a l lagencies responsible for reservoir operationin the country, has now been recognised. Thecase in Sri Lanka for adopting a hazard ratingsystem is the same as for other countries witha large stock of dams, namely:-

* to provide the basis for a uni f iedapproach to assessing flood criteria fordams

* To formalise the need for more intensiveinspection procedures and maintenanceresources targeted to high hazardstructures.

21. In order to promote the concept ofhazard rating within the country's engineeringcommunity (ref .2) , proposals were formulatedfor a hazard rating table based on thatadopted by the Australian National Committeeon Large Dams1 dam safety sub-committee.

129

RISK, HAZARD AND SAFETY

Table 1. Hazard rating for reservoirs in Sri Lanka

Very high High Significant Low Very low

Excessivenumbers oflives at risk

Loss of identifiable lifeexpected because ofcommunity or other signifi-cant developmentsdownstream

No loss of life expected,but the possibilityrecognized. No urbandevelopment and no morethan a small number ofhabitable structuresdownstream

No loss of lifeexpected

Reservoirs in remotejungle or sea coastlocations presenting nomeasureable risk topersons or property

Excessive economic losssuch as serious damage tocommunities, industrial,commercial or agriculturalfacilities, importantutilities, the dam itselfor other storagesdownstream

Appreciable economicloss such as damageto secondary roadsrminor railways,relatively importantpublic utilities,the dam itself orother storagesdownstream

Minimal economic loss,such as farm buildings;limited damage toagricultural land, minorroads, etc

Dam essential for services,and repairs not practicable

Repairs to dampracticable oralternative sourcesof water/power supplyavailable

Repairs to dam practicable.Indirect losses notsignificant

Their table was used as being the mostappropriate for Sri Lankan conditions, butwith an extension to the number of categoriesto reflect the wider range. The benefit ofadopting a system which is compatible withpractice in other countries, so that commonexperience can be shared, is clear. Theproposed hazard rating table has now beenapplied in a preliminary exercise to rate allof the major i rr iga t ion tanks, and i sreproduced here as Table 1.

22. In conjunction with hazard rating, thestandards for flood design generally adoptedin Sri Lanka have been reviewed. Table 2 setsout flood standards which are believedgenerally to reflect current practice fordesign of new dams in Sri Lanka, and maytherefore be used to check the adequacy ofexisting dams.

23. In order to improve the evaluation ofthe safety of major irrigation dams, modelforms for dam safety inspection were preparedand applied to twelve representat iveinspections. The dams selected for suchinspection were a l l major structures, butdisplaying particular symptoms which had beencausing concern. The model form for damsafety report ing provided for makingrecommendations for measures to be taken inthe interests of safety.

24. The following summarises the main orrecurring features of those measures inrespect of the twelve dams inspected.

25. Reservoirs. The water level was to bekept down in some cases until remedial workshad been completed.

26. Bunds. Further slope erosion was to bearrested by remedial works; a berm at one damwas to be completed to ensure stability; crestlow points were to be raised to restorefreeboard; drainage measures were to beimplemented.

27. Sluices. Ancient dry stone barrelswere to be demolished, as posing a severe riskto safety (being similar to the Kantalai typewhich collapsed after piping, ref .3); gates

Table 2. Minimum flood safety standards asgenerally represented by current designpractice in Sri Lanka. (Return periods of floodsin years)

Hazardrating

Very high

High

Significant

Low

Very low

Reservoir size factor

Greater 100 to 10than 100

PMF

10 000

1000

100

no minimum no

PMF

5000

500

50

minimum

- V x H

Lessthan 10

PMF

2000

200

20

no minimum

V = Volume of water in million m retained bydam with reservoir level at highest floodlevel

H = Height of dam measured in metres asvertical distance between crest level andlevel of lowest adjacent natural ground

PMF = Probable Maximum Flood as derived byrainfall maximisation procedures

were to be checked for higher heads at one damand replaced at another by larger ones toallow faster emergency drawdown; barrels wereto be strengthened where the need was shown byleakage test ing; s t i l l ing basins were to berepa ired and a new s l u i c e was t o beconstructed to allow more rapid draw down inan emergency.

28. Spillways. Aparc from usual mattersrelating to capacity and gate mechanisms,

130

PAPER 18: KNIGHT JACKSON, HALCRO-JOHNSTON AND MAKINSON

radio equipment was being considered for oneremote dam for warning of emergency use of thespillway.

29. Downstream flood corridors. Floodwarning systems were being considered as wellas arrangements with local authorit ies tosecure l i fe and property in the event of majorrelease from two spillways located in closeproximity to built-up areas.

30. The above points are thought to typifythe dam s a f e t y problems l i k e l y to beencountered at most of Sri Lanka's ancientdams •

SAFETY ASPECTS IN RAISING OF EMBANKMENT DAMSGaborone dam raising, Botswana

31. The design for the ra i sed dam,constructed 1983-84, is based on an 8 m risein the full supply level for the reservoir,and includes flood discharge arrangementscapable^ of passing a probable maximum flood of7000 m / s . The original zoned embankment hasan upstream sloping s i l ty clay core and adownstream random zone separated by a chimneydrain. Investigations were limited by theneed to respect the integrity of the core andavoid deep penetration i n t o i t , whi leconstruction records from the early 1960s wereunavailable. The upstream face of the corewas subject to a degree of softening and thedownstream random zone was l a r g e l y ofimpermeable material. The permeability of thechimney drain was in doubt unt i l exposedduring the new construction work, when i t wasshown to be a stiff gravelly clay rather thana modern permeable chimney.

32. The sandy alluvium of the originalfoundation was cut off by a contiguous piledwall linked to a grout cur ta in in the

underlying weathered granite, and historicallysignificant seepage water existed at thedownstream toe. The only suitable damalignment, through a granite outcrop offeringa spillway s i te , necessitated incorporatingthe existing dam into a 25 m final heightembankment (Fig. 5).

33. It was not safe to drill through theexis t ing core and improve the cutoffprovisions, so the design re l i e s on thelengthened seepage path maintaining similarhydraulic gradients in the foundationalluvium. Internally the raised dam has a newsand chimney zone and a high capacityhorizontal drainage blanket. The safety ofthe foundation against piping failure wasfurther enhanced by the addit ion of atrapezoidal collector drain of graded filtermaterial, running along the downstream toe.The whole of the existing dam and the raisedcore were thus treated as an integral upstreamcore zone, supported by a massive newdownstream random f i l l shell zone.

34. The crucial stage of construction forupstream slope stability occurred when thereservoir was at an unusually low level andcomprised removal of the upper 4 m of theexisting dam core, while the advanced heightof new core f i l l downstream maintained therequired freeboard of the existing reservoir.

35. The f i l l had to be placed as rapidly aspossible to reach crest level, consistent withs t a b i l i t y of the upstream slope. Porepressures both within the existing core andabove the core junction level were thereforemonitored with hydraulic piezometers andcontrol surveys of the upstream slope werecarried out. In the event pore pressure riseswere not significantly high (ref. 4).

Raised dam crest

XTemporary fil l levelprior to construction of core junction

2-25

Impervious fil l

Raised full supply level

Existing dam crest

©Riprap and 1

f i l t ersImpervious

Weatheredgranitebedrock

Sand / gravel drainExisting contiquouspiled cutoff andgrout curtain

5. Raised section of Gaborone dam

Raised fuse crest

Full supply level-i

Filters©Riprap

Relief wells

985-El.m

7 r r " " \Existing dam Weathered granite

bedrock

J ) Impervious coreGeneral f i l l •-SandGravel

(5) Fine rockfilf

Medium rockfilf^Rockfilled gabions © RjpRap

Concrete toe

Fig. 6. Fuse section of Gaborone dam

131

RISK, HAZARD AND SAFETY

36. In line with advances on flood riskanalysis a spillway design based on a 1000year flood was upgraded to PMF. The dome ofsound granite offering a foundation for theungated spillway weir was of strictly limitedlength. To avoid the excessive cost ofincreasing crest height along the f u l lembankment length i t was decided to follow acommon precedent in Africa and provide a fusesection within the embankment. If the floodexceeds 0.5 PMF the fuse is overtopped and itsdesign (Fig. 6) ensures rapid erosion, down toa protected base level to retain a residualreservoir. The lack of data from actualoperation of such fuses, and the conjecturalform of the PMF flood hydrograph, led to theadoption of a wavewall to increase theavailable emergency freeboard. Standby plantwas also provided to i n i t i a t e erosion i fnecessary. The risk of occurrence of a floodthat would activate the fuse provision is lessthan 1:1000 years, and the fuse merelyconcentrates the breach at a s e l e c t e ddischarge channel rather than damaging theentire dam crest and downstream toe.

37. The existing concrete draw-off towerand culvert were incorporated into the newscheme. The heightened tower was tied toanchorages in the bedrock by prestressing barsto avoid flotation problems. The existinglength of culvert, set in a trench through thebedrock below the embankment, was providedwith a reinforced concrete liner to take theextra vert ica l loading. A new extensionlength and downstream valve house were thenadded, and grouting and drainage provisionsincorporated to reduce seepage along theoutside of the culvert. The draw-off waterpasses through a s tee l pipe within theconcrete culvert and township supplies werenot interrupted for more than 24 hours duringconstruction.

Hinze dam raising, Australia38. Hinze dam, located 70 km south-south

east of Brisbane, Queensland, was built 1974-76 as a 48 m high rockfill embankment with acentral clay core. It was designed with theintention of future raising by 22 m as thenext stage in augmenting water supplies to theGold Coast and adjoining areas of AlbertShire. Since then, however, the forecastwater demand almost doubled for the period1985-2000, with the existing yield expectingto be reached by 1990. These considerations,together with optimisation studies, formed the

20 L

See crest detail

Stage HI crest El. 10600

- Stage II crest El. 93-50

F.S.L. stage II El. 82

Stage I crest El. 75

E.S.L. stage I El. 64.47

Reference line

Fine filter

utureStage I Stage II stage HIprofile profile profile

Fig. 7. Hinze dam two-stage raising

132

basis for a two-stage raising of the dam by atotal of 30 m, Fig.7. The first stage raisingwas designed and constructed, with completionin 1989, to facilitate subsequent raising tothe ultimate height.

39. Two other factors s ign i f i cant lyinfluenced both design and construction of theraising, namely: (a) changes in procedures forestimating probable maximum f loods inAustralia since the original design, resultingin considerably increased peak outflows; thesecould not be accommodated by the earlierenvisaged raised dam and spillway arrangement;(b) a d e c i s i o n at the beginning ofconstruction to improve the flood mitigationaspec t s of the dam by incorporating arectangular slotted spillway in the raisingproject: a lower 24.5 m wide part reduced thepeak outflow for floods up to the design 100year flood, whilst greater floods passed alsoover the adjacent shoulders of the spillway.This required a combined surcharge andfreeboard of 11.1 m, resulting in a revisedStage II dam crest level at El.93.5 m, inorder to discharge the PMF of 2420 m / s . Thisrepresented an additional 3.5 m to theor ig inal ra i sed c r e s t l e v e l to whichconstruction had already begun.

40. The Stage II crest arrangement (Fig. 7,inset) features a flat-inclined filter barrierto water above El.88.0 m, which represents a25 year return period flood, whilst El.89.2 mrepresents the 100 year design flood andEl.92.5 m the PMF. None of the floods wouldbe of long duration and the PMF would resultin water levels above El.88.0 m for a periodno greater than 3 or 4 days. During such aperiod the top 5.5 m of the Stage II dam wouldact as a permeable bund. The quantity ofs e e p a g e would be c o n t r o l l e d by thepermeability of the flat-sloping one metrethick fine f i l t e r zone acting as a core,protected by the outer coarse f i l t e r anddischarging into a special general drainagezone.

41 . The design concept i s simply anextension of the existing (Stage I) clay coreand rockfill shel ls , with the core slightlyinclined downstream to minimise the amount ofr o c k f i l l (zone 3W) to be placed in thereservoir. Such rockfill was restricted tofresh or slightly weathered only, to minimisepollution of the water. To avoid damage tothe exposed right hand side of the drawoffculvert, 3W material over the culvert wasrequired to be placed from the left towardsthe right abutment. Fine, clean rockfill ofmaximum size 200 mm was specified to be placedaround the intake tower to minimise damage tothe concrete. The upstream rockfill berm 3Wwould ensure that there will be no rockfillplaced through water for the ultimate StageIII embankment.

42. To safeguard freeboard and to minimisedistortion of the Stage I core, new rockfillwas to be constructed to at least Stage Icrest level before new core constructionbegan, and maintained at as similar a level aspossible upstream and downstream.

43. A minimum thickness of 2 m of corematerial was required to be stripped from the

PAPER 18: KNIGHT, JACKSON, HALCRO-JOHNSTON AND MAKINSON

TOTAL STRESS NORMAL TO CONCRETE-CLAYINTERFACE FROM HYDRAULIC FRACTURE TESTS.WITH RANGE OF MEASURED SHUT-INPRESSURES SHOWN AS EXTENSIONS.-

METRES2

CONCRETEOUTLET CULVERT(LOOKINGDOWNSTREAM) 50mm DIA.TEST HOLES FOR HYDRAULIC FRACTURE

AND PORE PRESSURE MEASUREMENT TESTS

Fig. 8. Hinze dam core/draw-off culvert areastress conditions

Stage I embankment to eliminate a slightly dryand stiff zone from the core at its geometricchange point, and thus reduce the risk ofarching and cracking. At 2 m depth the 10year old existing core had a moisture contentof about 20%, which was within the range ofthe original placement moisture contents. Thenew core was benched into the existing core ona lv to 1.5h slope, with a 3 m working benchdownstream. To avoid further drying out andto minimise risks due to temporarily reducedcore freeboard, cutting down of the Stage Icrest and replacement with Stage II f i l l wasto proceed in sections of a minimum lengthnecessary for construction purposes only.

44. Low stress conditions within a portionof the Stage I core, between a rock abutmentand the side of the nearby concrete draw-offculvert (Fig. 8), and a state of hydraulicfracture were suspected and proved (ref .4) .Although the hydraulic fracture was safelycontained by the downstream f i l t er , i t wasdecided to enhance existing stresses in thecore zone. This was achieved by a programmeof squeeze grouting from within the culvert.The first three series of injections over theupstream two thirds of the core width took onemonth, and succeeded in raising the minortotal principal stress by about 150 kPa, witht h e p i e z o m e t r i c p r o f i l e r i s i n gcorresponding ly . A fourth s e r i e s ofinjections was made after a seven month wait,when testing showed that the stress levelsachieved by the previous work had beenm a i n t a i n e d , and a modest a d d i t i o n a lenhancement was then achieved. Overall stresslevels on completion of the dam raising werefound to be acceptable.

45. In summary therefore, the safety issuesinvolved with this raising were related to:( i ) a reappra i sa l of procedures forestimating the PMF, ( i i ) incorporating floodmitigation requirements within the projectafter start of construction, involving apermeable barrier, ( i i i ) constructionsequences and rockfill sizing, and (iv) stressenhancement of the core adjacent to the draw-off culvert. First spil l ing for the raiseddam occurred in April 1989 after rapidfilling.

CONCLUDING REMARKS ON SAFETY LESSONS DERIVEDFROM CASE HISTORIES DESCRIBEDRegarding the inspect ion and i n v e s t i g a t i o n ofoperat ional dams

46. Investigations can at best onlypartially discover the internal condition ofan old dam. The older the dam the more likelyit is that ful l design and constructiondetails will not be available. In such casesexternal inspection will provide the onlydirect knowledge of a dam's condition, unlessinternal investigations are made. Theplanning and doing of investigations inoperational dams regardless of age, however,must proceed with a high level of caution. Atal l times they must be capable of beingquickly stopped, and procedures modified wherethey are suspected of causing a local butsignificant deterioration in an embankment'sintegrity. This particularly applies to al lwork within a dam core, and at interfaces withadjacent structures such as draw-off culverts.

Regarding the design and construction of workson operational dams

47. Safety must be the primary concern inthe design and construction of works onoperational dams and wil l include: themaintenance of freeboard and the ability topass floods; control of temporary and long-term seepage by adequate geometry anddrainage; careful construction sequences toavoid damage to existing work; r e a l i s t i cassumptions about the possible long-termdeterioration of significant elements such asdraw-off culverts and their surrounds, andconcrete core walls in old dams; continuedmonitoring and assessment of a dam's behaviourafter completion of remedial works or raising.

ACKNOWLEDGEMENTSThanks are expressed to the following

organisations for permission to refer to thework described: National Power; OverseasDevelopment Administration and Sri LankaIrrigation Department; Water U t i l i t i e sCorporation, Botswana; Gold Coast CityCouncil, Queensland, Australia; Damcorp; GibbAustralia (Pty) Ltd; Sir Alexander Gibb &Partners Ltd.

REFERENCES1. BYRD T. Speedy s tudy p o i n t s t o K a n t a l a ip i p i n g . New C i v i l Engineer, 1986, 29 May, 26-29.2 . HALCRO-JOHNSTON J . Hazard r a t i n g ofreservo irs in S r i Lanka. SLNCOLD B u l l e t i n ,v o l . 1 . , No.4 , Sr i Lanka National Committee onLarge Dams, July 1989, 21-26.3 . KNIGHT D.J. Discussion on the measurementand s i g n i f i c a n c e of horizontal earth pressuresi n t h e puddle c l a y c o r e s of o l d earth dams.Proceedings, Institution of Civil Engineers,Part 1, 1987, vol.82, December, 1261-1265.4. KNIGHT D.J. The proven usefulness ofinstrumentation systems on varied damprojects. Geotechnical instrumentation inpractice. Institution of c ivi l Engineers,1990, 401-414.

133

20. Woodhead Reservoir - remedial works

R. W. CHALMERS, Babtie Shaw and Morton, Glasgow, UK

Remedial works to Woodhead Reservoir were required to alter the overflow provisions to handle aprobable maximum flood (PMF) inflow while restricting the outflow to lower reservoirs in theLongdendale Valley cascade. Woodhead Dam has had a noteworthy history with severalmodifications to the original design since construction first started in 1847. Severalrestrictions were placed on the remedial works as a result of the dam being situated in thePeak National Park. The risk to the Works, as a result of various factors, are considered forboth the temporary remedial works phase and the permanent works.

INTRODUCTION1. Following an inspection in 1986 under

the Reservoirs Act 1975 of North West Waterfsfive reservoirs in the Longdendale Valley, 25km east of Manchester, the AR Panel Engineerrecommended that the safety provision of allreservoirs in the valley be increased towithstand a probable maximum flood (PMF)condition.

2. This recommendation was based onthe method of predicting severe floodsdeveloped by the Institute of Hydrology,"Flood Study Report", NERC, 1975 (Refl).

3. Several alternative methods ofachieving the required degree of floodprotection were examined. Nine of thesealternatives were the subject of anenvironmental appraisal includingconsultation with over 50 interested parties.The preferred method, selected on the basisof environmental grounds, involvedconcentrating the major works at WoodheadReservoir, with only minor work required atthe lower reservoirs.

4. At the time of the AR Panel Engineer1sinspection the outlet capacity of thespillway at Woodhead was assessed as 170m /s. This corresponded with the reservoirat crest level, less an allowance for waverun-up, and a flood with a return period ofthe order of 1 in 3,000 years. The spillweirwas not the control on the discharge capacitybut rather the relatively narrow channelleading from the weir. The dam had beenclassified in accordance with the ICE Notesfor Guidance (Ref 2) as being a category Adam and therefore should be able toaccommodate PMF inflows. By increasing thedischarge capacity of the overflow atWoodhead the discharge capacity of thesubsequent reservoirs in the cascadewould have proved to be insufficient and

increased provisions at all five reservoirswould be required. However, by restrictingthe flow and holding the flood water atWoodhead, only minor work in the four lowerreservoirs is required.

5. From the analysis of the outlet worksof Torside Reservoir, the^next reservoir inthe cascade, and its time lag effect, it wasestablished that the inflow to Torsidereservoir required to be limited to a maximumof 128 m /sec to ensure that no major workswere required. By restricting the flow fromWoodhead reservoir to Torside to 128 m /s,flood inflows to Woodhead in excess of thisoutflow had to be stored. Therefore a raisedembankment and throttled outlet was requiredto increase the flood storage volumeavailable.

HISTORY OF THE DAM6. The storage and use of the Etherow

River in the Longdendale Valley was coveredby an Act of Parliament which received RoyalAssent on 9 July 1847. The Longdendalescheme, and its contemporary, the Rivingtonscheme for Liverpool, were the first majorworks of this extent in the country. Veryfew engineers were experienced in this sizeof project. Manchester City Council, thepromoters for the scheme, selected JohnFrederick Latrobe Bateman as the Engineer.

7. Construction of the Woodheadembankment commenced in August 18^7. Theembankment was formed where the Manchester toSheffield turnpike road, on the north side ofthe valley, and the Manchester, Sheffield andLincolnshire Railway, on the south sideapproached one another. The embankmentcomprised of a central puddle clay coresupported by granular shoulders of locallywon weak and weathered sandstone. The puddleclay extended 3.3 m (10 ft) into the

The embankment dam. Thomas Telford, London, 1991 135

RISK, HAZARD AND SAFETY

Existing upstream slope

Disusedrailwaytrack

Figure 1: Plan of raised embankment

foundation (Ref 3 and 4).8. This original embankment had side

slopes of 1 to 3 for the upstream slope and 1to 2 for the downstream slope. The upstreamslope was pitched with stone. The embankmentwas within 7 m (20 ft) of its final 29 m (90ft) height when the impounded water reached adepth of 9 m (30 ft). Seepage at this pointhad reached a balance with the inflow and nofurther water was being stored. Leakage hadbeen observed at the downstream toe. Manyyears were spent trying to stem the leak by avariety of methods, with several causes beingpostulated.

9. In 1864 Bateman decided that the onlyway to fully impound was to construct a newembankment with a deep concrete cut-offextended through the under-lying landslipdeposits into the almost impermeable shales.A study of the valley downstream of the firstembankment indicated to Bateman that the mostsuitable location would be immediatelydownstream and slightly at an angle to theoriginal embankment. Construction of thisadditional core, cut-off and shoulder wasquickly started and was completed in 1877, afull 30 years since commencement ofconstruction.

10. The embankment has remainedvirtually unchanged to the present. However,since 1887 there has been doubt about thecapacity of the flood control works and fromthat date until 1945 the reservoir was helddown 1.52 m (5 ft) to provide flood storage

capacity. Several flood studies have beencarried out throughout its life (Ref 5 and6). In the late 1930fs an additional lengthof spillweir, deepened watercourse and a newspillway chute were constructed to meet therequirements of the 1933 Interim Report ofthe Institution of Civil Engineers FloodsCommittee.

GENERAL DESCRIPTION OF THE WORKS11. Following a detailed study, by Binnie

& Partners, of the whole of the LongdendaleValley and the flood safety provisions ateach of the five reservoirs in cascade (Ref7), it was determined that all reservoirs inthe valley could not accommodate a PMF event.Remedial works were therefore required forall five reservoirs to meet the requirementof the 1975 Reservoirs Act. The study alsoconsidered the necessary actions to meet therequirements and it concluded that the mostfavourable means of providing additionalsafeguards for the whole of the valley was tocarry out all the major works at Woodhead.

12. To maximise the benefits of the works,while minimising the environmental impact,flood routing and increasing reservoirretention time was considered the mostapplicable. Woodhead being at the head ofthe cascade and having almost 50/5 of thecatchment area of the valley has particularinfluence on the whole of the Longdendalesystem. Therefore, the majority of the workwas concentrated at Woodhead.

136

PAPER 20: CHALMERS

t *of new of raised

public road embankment

I 'Revetment surface |protection,topsoited \and grassed. | 2*. 6 . 5 m A.O.D.

Existing Originalmasonary road 1.2 m minimum |

Figure 2: Typical section of raised embankment

13. The time lag of flood flows throughthe reservoir was relatively short and inorder to increase this it was necessary torestrict, or throttle the outflow. From thestudy of the lower reservoirs it wasnecessary to limit the outflow to around 128m /s during PMF conditions. This could beachieved by forming a throttle 4.5 m wide by3.5 m high. This, however, results in theinflowing flood water being retained in thereservoir and the reservoir rising byapproximately 6.0 m. The flood storagerequirement is therefore to increase theheight of the embankment of the order of 7.0m.

14. The raised embankment was designed asa simple section (Fig 2) with two fill zones,an inclined upstream rolled clay zone to formthe water barrier, supported by a zone ofgranular fill. The upstream clay membranewas connected to the original upstream puddleclay core (Core No 1). However, it was knownthat this core was not effective, as Batemanhad demonstrated. Piezometers installed,during the site investigations, in the infillbetween the two puddle clay cores reflectedthe reservoir level. Study of the monitoringresults indicated that, if the reservoirlevel rose as a result of storing the PMFinflow, the phreatic level in the infill zonewould rise above the maximum height of thedownstream puddle clay core (Core No 2).However, there was a sufficient time lag tolimit the maximum rise possible, during asingle flood event significantly below themaximum reservoir level. This would causesevere leakage and instability problems withthe raised embankment. The phreatic levelwas therefore held below the level of thedownstream core by the inclusion of adrainage blanket which intercepts the risingwater in the infill and conveys it todischarge into the spillway channel. Thequantity of water loss is anticipated to be

of the order of 150 1/sec.15. This large loss of water is only

acceptable as it occurs during severe floodconditions, and would not be stored. Thedrainage blanket is a graded filter sandwichwhich will prevent loss of material from theinfill. The grading was designed using therules proposed by Sherard (Ref 8). Otherdetails of the design are indicated onfigures 1 and 2.

RESTRICTIONS ON THE CONSTRUCTION16. During the consultation stage

restrictions were placed on the constructionphase to reduce its impact on theenvironment. In particular, as the workswere located in the Peak National Parkwinning of fill material in the valley wasnot allowed. It was anticipated that allmaterial necessary for the works would beimported into the valley. However, therewere also concerns about the traffic impacton the already heavily used Woodhead Pass(A628). During the investigation stageseveral sources of fill were examined. TheManchester to Sheffield Railway, which runsalong the south side of the valley andthrough the Woodhead Tunnel at the head ofthe reservoir, was closed in 1982. However,the track was still ballasted and was beingconsidered as a haul route to the dam, toreduce the load on the public highways.Examination of the track revealed that thequantity of ballast was apparently in excessof the required granular fill quantity forbhe works. This was investigated and foundto be suitable for granular fill. There wastherefore only a requirement for clay, toform the water barrier, and granular filtermaterial to be imported from out of thevalley.

17. During a previous investigation intothe operation of the discharge arrangementsthe reservoir was emptied. The silt in the

137

RISK, HAZARD AND SAFETY

reservoir basin was observed to be severalmetres thick. The stream flow eroded thesilt leaving near vertical walls. During aperiod of heavy rain following a dry spell alarge volume of silt slipped and liquefiedcausing the draw-off tunnel to be blocked bysilt. To ensure that a similar situation didnot recur North West Water required that theremedial works be carried out without takingthe reservoir out of commission.

18. The reconstruction of the spillwaychannel and throttle was required to be onthe line of the old channel. Physicalhydraulic model studies were carried out atSalford University in 1986 to establish theoptimum alignment. The decision to locatethe new channel on the line of the oldchannel and with the reservoir not out ofcommission, has posed problems in themaintenance of an adequate discharge capacitythrough the Works during the constructionperiod.

RISK ASSESSMENT19. The construction of an orifice

controlled discharge channel for an existingreservoir previously controlled by a larger

0

1 -

— 2 —

" 3 -

J 5 -

1 6 -

« 8"x:Q.«lo

S

1

/

/

/

torogt

2

^ II

7/

Volume

31

//

/

( x

1

/

10* ^

5

y.

h6

Availflood

7

(Tbl.storog

Figure 3a: Reservoir capacitybelow sill level

0 5 5 10 25 SO ii)0 200- 500 10003^500010000

capacity side-spillweir introduced severalfactors which required to be considered incomparison to the normal spillweir wideningproject. These additional factors all had abearing on the risks imposed on the Works.

i 6-

= 5

1,-

All valvesclosed

Figure 3b: Flood volume againstreturn period(Modified spillway)

2 5 10 25 SO ISO 200 500 1000 10000

Safe Return Period ( Years)

Figure 4: Overtopping risk

20. During the design, the hydrologicalanalysis indicated that the inflow for PMFconditions would more than fill the reservoirbasin even if it had initially beencompletely empty (Fig 3). As previouslystated, problems had been experienced withthe draw-off arrangements during a periodwith the reservoir empty and therefore it wasdecided that the reservoir could not becompletely emptied during the constructionperiod.

21. For the construction period it wouldbe necessary to accept a risk of the spillwaychannel works being inundated due to floodwater discharging through the channel. Itwas also necessary to accept a risk of theembankment being overtopped during the periodof the works. The acceptable risk is a verydifficult concept for most engineers toembrace and fully quantify. It was acceptedthat the relatively short duration of theworks could allow a theoretical risk factorto be calculated. However, this onlyfactorised the long term acceptable risk.The question as to what is acceptable stillremained. It was considered that as theexisting works were lower than the finalacceptable risk, but at a risk level that hadbeen carried for over forty years, this levelof provision during the Works would beappropriate.

22. It was accepted that the reservoirwould have to be held down to avoid floods ofshort return period inundating the spillwaychannel works. The risk of inundation ofthese works was based on precedence and abalance between the economic value of theloss of storage and the cost to the worksfrom damage as a result of the inundation.

138

PAPER 20: CHALMERS

245- PM.F

239-00 1 in IQOCytors238-28 Sill lev

Figure 5: Section showing flood lift

This risk element was set as a flood eventwith a return period of 1 in 50 years whichwould cause the reservoir to increase inlevel to that of the weir level, any greaterinflow not being fully stored. To achievethis the reservoir had to be lowered and heldat maximum level of 4.5 metres below the weirsill level (see Figures 3a and 3b).

23. With the lowered reservoir level, therisk of overtopping the original embankmentcrest level, if no water could pass throughthe spillway channel was that withapproximately a 1 in 200 year return period(Figure 4). This was significantly below thepreviously defined acceptable risk.

24. The winters in the Peak District canbe very severe and it was the Employer's viewthat the limited draw-off facilities couldnot maintain the reservoir at the reducedlevel below the overflow weir during thewinter period due to the high run-off with afrozen catchment or with snow melt. It wastherefore required that no work be carriedout in the spillway channel during the winterperiod, October to March inclusive.

25. With the orifice fully constructed aflood with a return period of approximately 1in 1,000 years would cause the reservoir torise from a water level at the sill level toa level in excess of the original crestlevel. It was therefore stipulated in thecontract that provison be made at all timesfor facilities to handle flood inflow with areturn period equal to or in excess of thatwhich the original spillway couldaccommodate. This could be achieved by usingeither, or a combination of, a purpose builtchannel, the original channel or the newpermanent works channel.

26. A further requirement in the contractwas that if the culvert roof was constructedprior to the winter closedown then theembankment had to be raised to a level of242.5m OD over the full length of the dam.

Therefore, even with the reservoir leveluncontrolled, a combination of flood storageand discharge would ensure that the risk ofovertopping would not exceed that previouslydefined as acceptable.

27. Following normal tendering proceduresthe contract was let to Alfred McAlpineConstruction Limited and constructioncommenced in early April 1989 on an eighteenmonth contract. An auxiliary spillwaychannel was provided during the initial phaseof the culvert construction to maintain therequired discharge capacity. Once the newculvert was functional the auxiliary channelwas removed and the fill was raised to abovean elevation of 242.5m OD. This work wascarried out during the first constructionseason.

28. Weather conditions during the lastquarter of I989 were very good and arelaxation was granted for the contractor tocontinue working in the spillway channeluntil the end of October.

29. To improve on the accepted risk, NorthWest Water attempted, and generallysucceeded, in holding the reservoir leveldown 4.5m below the sill level. Only duringperiods of flood inflow was this not achievedand at no time over the 1989/90 winter didthe reservoir level exceed 2.0m below silllevel.

30. Construction continues through 1990,with the contract completion programmed forearly October 1990. The risk of overtoppingthe embankment will continue to reducethroughout 1990 until when the embankment isat final level and final surface protectionis complete at which point the works will beable to accommodate full PMF conditions.

139

RISK, HAZARD AND SAFETY

CONCLUSIONS31. Risks are imposed on the Works from

various quarters. These must be assessed toensure that they are being fully catered forand that no unacceptable risk is beingcarried by any of the interested parties. Bydetailed analysis the level of risks beingcarried during remedial works can bedetermined and a level of risk determinedwhich is acceptable to all parties.

ACKNOWLEDGEMENTS32. The Author is grateful to North West

Water for their permission to publish thispaper and to my colleagues and the Partnersof Babtie Shaw & Morton for their assistanceand encouragement in its preparation.

BIBLIOGRAPHY1. NERC, Flood Study Report. Institute ofHydrology, 1975.2. Floods and reservoir safety: an

engineering guide. Institution of CivilEngineers, London, 1978.3. BATEMAN J.F.L.T., History andDescription of the Manchester Waterworks,1884.4. QUAIL T., Reservoirs in the Hills,North West Water, 1972.5. WILDSMITH A.C., Flood Relief Works,Longdendale Catchment Reservoirs. Paperpresented to the NW Association of Instutionof Civil Engineers, 1945.6. CLERKE R.W.G. et al, LongdendaleHeadworks. Flood Channel Improvements.Paper presented to the Northern Section ofInst. of Water Engineers, 1972.7. BINNIE & PARTNERS, Modifications toLongdendale Reservoirs, Unpublished, 1984.8. SHERARD J.L. & DUNNIGAN L.P., Filtersand leakage control in embankment dams. ASCESymposium on Seepage and Leakage from Damsand Impoundments, Denver, 1985.

140

21. The Bureau of Reclamations new downstream hazardclassifications guidelines

D. J. TRIESTE, US Department of the Interior, Denver, USA

SYNOPSISA new Bureau of Reclamation (Reclamation) document, "Downstream Hazard Classification Guidelines"(ref. 1), has been published for guidance in determining a dam's downstream hazard classification.This document reflects Reclamation policy and philosophy regarding flood danger to people, combinedwith new quantifiable depth-velocity flood danger level criteria. This document includes guidanceand criteria for performing dambreak/inundation studies suitable for downstream hazardclassification purposes. This paper summarizes key portions of the "Downstream HazardClassification Guidelines".

INTRODUCTION

1. A downstream hazard is defined as thepotential loss of life or property damagedownstream from a dam and/or associatedfacility (e.g. dike) due to floodwatersreleased at the structure of waters released bypartial or complete failure of the structure(ref. 1). Downstream hazard classification isnot associated with the existing condition of adam and its appurtenant structures or theanticipated performance or operation of a dam.Rather, downstream hazard classification is astatement of potential adverse impact on humanlife and downstream developments if adesignated dam failed.

2. Although the Guidelines are intended tobe used for all dams, they are especiallyuseful for dams whose failure flood wouldaffect only a small population (e.g. smalldams). The purpose of the Guidelines is : (1)to define the Safety Evaluation of ExistingDams (SEED) method for assigning a dam'sdownstream hazard classification; (2) toprovide guidance and present methods, for thepurpose of downstream hazard classification,for estimating the downstream area susceptibleto flooding due to a dam failure; (3) toprovide guidance and criteria foridentification of downstream hazards; and (4)to bring objectivity and consistency intodownstream hazard classification.3. Reclamation's "Downstream Hazard

Classification Guidelines" can be obtained bysending a request to: Bureau of Reclamation,Denver Office, P.O. Box 25007, Denver, Colorado80225, Attention: D-7923A. The cost is $2.00per copy (Foreign handing: Surface mail $0.50,Air Mail $1.00).

PURPOSE OF DOWNSTREAM HAZARD CLASSIFICATION4. Dams within the U.S. Department of the

Interior (DOI) are given a downstream hazardclassification for two reasons:

1. The DOI Manual, part 753 (ref. 2),

The embarkment dam. Thomas Telford, London, 1991

established that a downstream hazardclassification is to be assigned to every DOIdam.2. Downstream hazard classification serves asa management tool for determining which damsare to undergo the full SEED (SafetyEvaluation of Existing Dams) process. Damshaving a low downstream hazard classificationare excluded, whereas those having asignificant or high downstream hazardclassification are included.

DOWNSTREAM HAZARD CLASSIFICATION SCHEME5. The system presented in table 1 is used

by the SEED program for classifying Reclamationand other DOI dams.

Table 1.scheme

Downstream hazard classification

Classification

Low

Significant

High

Lives-in-jeopardy

0

1-6

More than6

Economic loss

Minimal(undevelopedagriculture,occasionaluninhabitedstructures, orminimal outstandingnatural resources)

Appreciable (ruralarea with notableagriculture,industry, orworksites, oroutstanding naturalresources)

Excessive (urbanarea includingextensive

141

RISK, HAZARD AND SAFETY

community,industry,agriculture, oroutstanding naturalresources)

Lives-in-jeopardy6. Lives-in-jeopardy is defined as all

individuals within the inundation boundarieswho, if they took no action to evacuate, wouldbe subject to a dangerous situationcommensurate with the depth-velocity flooddanger level criteria explained later inparagraphs 11 through 14.

7. Lives-in-jeopardy is divided intopermanent and temporary use. Permanent useincludes permanently inhabited dwellings,worksite areas, and industrial areas, whereastemporary use includes roads, campgrounds, andother recreational facilities.

IDENTIFICATION OF DOWNSTREAM HAZARDS8. Identification of downstream hazards is

one of the most important topics of theGuidelines. Therefore, it will be given extraconsideration in this paper.9. Sometimes downstream hazards

classification is obvious. That is, ananalysis is not necessary because lives-in-jeopardy and/or property damage could bedetermined with little doubt. If a downstreamhazard is not obvious, and/or "possibledownstream hazards" have been identified, thena dam-break/inundation study is performed forthe purpose of determining the impact of a damfailure flood on the possible downstreamhazards. A possible downstream hazard is onethat has been identified as having thepossibility to constitute a downstream hazard,but field work and/or analyses need to beperformed for confirmation. Possibledownstream hazards are identified fromtopographic maps, photographs, field surveys,and information from "locals". Downstreamhazards include any situation that is suspectedof having a potential for lives-in-jeopardy oreconomic loss due to a dam failure.

10. Analysis does not always prove apossible downstream hazard to be a confirmeddownstream hazard; many "gray areas" exist indownstream hazard classification. Analysis mayindicate that a residence may be flooded by 1foot (0.3 m) of water, but will this result inloss of life? If a failure flood overtops ahighway bridge, will the bridge be destroyed?If not, will a vehicle be carried by thefloodwater or go out of control due thehydroplaning? Or, will a vehicle crash due toa damaged road or bridge after the flood haspassed? Questions and gray areas such as theseare the underlying reasons for creation of theGuidelines.

11. Figures 1-5 contain curves of depthversus velocity that are indicative ofdangerous floodflows for various possibledownstream hazards. These curves assist theanalyst in making decisions regarding questionssuch as those addressed in paragraph 10. Thecurves are presented for the following

HIGH DANGER ZONE - Occupants of most houses are in dangerfrom floodwater.

JUDGEMENT ZONE - Danger level is oased uDon engineeringjudgement.

LOW DANGER ZONE - Occupants of most houses are notseriously in danger from flood water

3 0 V e l o c i t y ( m / 3 ) 5 0

-\1.0 7.0

3.0

10 15

Velocity (ft/8)

Figure 1. - Depth-Velocity flood danger level relationship for houses built on foundations.

HIGH DANGER ZONE - Occupants of almost any s.ze moo e nome are mdanger from flood water

JUDGEMENT ZONE - Danger level is based upon engineering ludqement.LOW DANGER ZONE - Occupants of almost any size motwie nome are not seriously

0 , 5 40 Velocity (m/s)

Figure 2. - Depth-velocity flood danger level relationship for mobile homes.

Velocity (m/s)

2.0 3.0 4 . 0

HiGH DANGER ZONE - Occupants of aimost any size passenger rvehicle are >n danger from flood water.

JUDGEMENT ZONE - Danger ieve< is based upon engineeringjudgement

LOW DANGER ZONE - Occupants of almost any size passengerly in danger from

10 12 14

Figure 3. - Depth-velocity flood danger level relationship for vehicles.

situations: (1) houses on foundations, (2)mobile homes, (3) passenger vehicles, (4)adults, and (5) children.

12. Figure 1 is a modification by the authorof a study performed by Black (ref. 3). Thecurves in Figs 2-5 were derived theoreticallyby the author. Figure 3 is in reasonableagreement with a theoretical analysis performedby Simons, Li and Associates (ref. 4). Thelower curve in Fig. 4 is in reasonableagreement with a theoretical analysis performedby David J. Love and Associates, Inc (ref. 5),and a laboratory flume study performed atColorado State University by Abt and Wittier

142

PAPER 2 1 : TRIESTE

5

4

C 3

Velocity (m/8)1.5 2.0 2.5 3.0

HIGH DANGER ZONE- Almost any size adult is indanger from flood water.

JUDGEMENT ZONE - Danger level is based uponengineering judgement.

LOW DANGER ZONE- Almost any size adult is notseriously threatened byflood water.

5 6 7

Velocity (ft/ s)

Figure 4. - Depth-velocity flood danger level relationship for adults.

Velocity (m/s)

1.0 1.5 2.0— , —

HIGH OANGER ZONE - Almost any size child is in clangerfrom flood water.

JUDGEMENT ZONE - Danger level is based uponengineering judgement.

LOW DANGER ZONE - Almost any size child (excludinginfants) is not seriously threatenedby flood water

Figure 5. - Depth-velocity flood danger level relationship for children.

using monoliths (ref. 6). Very little researchhas been done on this topic; even if this werenot the case, there would always bediscrepancies which could not be avoided due tothe many initial assumptions that would have tobe made, very large number of variables thatwould have to be considered, and personalphilosophy. The relationships presented inFigs 1-5 are very reasonable for estimatinglives-in-jeopardy for downstream hazardclassification needs, and satisfy one of thepurposes of the Guidelines - to bringconsistency and objectivity into downstreamhazard classification.

13. The depth-velocity flood danger levelrelationships are divided into three zones: lowdanger, judgement, and high danger. Anexplanation of these three zones follows:Low-danger zone. - If a possible downstream

hazard is subject to a depth-velocitycombination plotting within this zone, then thenumber of lives-in-jeopardy associated withpossible downstream hazards is assumed to bezero.High-danger zone. - The low-danger and high-

danger zones represent the two extremes ofreasonable certainty regarding the occurrenceof no lives-in-jeopardy associated withpossible downstream hazards is assumed to bezero.Judgement zone. - The low-danger and high-

danger zones represent the two extremes ofreasonable certainty regarding the occurrenceof no lives-in-jeopardy and some lives-in-jeopardy, respectively. Between those twoextremes exists a zone of uncertainty withrespect to assessment of lives-in-jeopardy.

Because every flood situation is unique, it isimpossible to account for all of the variablesthat may result in lives to be in jeopardy.Thus, in this case, it is left up to theanalyst to use engineering judgement fordetermining lives-in-jeopardy. Wheneverpossible, several opinions and a commonagreement among analysts should be reached inmaking this determination.

14. In many downstream hazardclassifications, especially where large damsand catastrophic flooding are involved,reference to Figs 1-5 is superfluous because ofthe obvious flood danger. But, for situationswhere the downstream hazard classification of adam is solely dependent upon an isolated floodsituation where occupants of a dwelling orvehicle may be in danger, or a person having noprotective environment (e.g. house, vehicle)may be in danger, Figs 1-5 should be used. Insuch situations, the analyst will havepredicted a reasonable maximum depth andvelocity, with confidence, at the possibledownstream hazard sites.

ESTIMATING INUNDATED AREA15. Determination of downstream hazard

classification based on the downstream hazardclassification system presented in table 1, andthe flood danger via Figs 1-5, is straight-forward, providing the extent of flooding isknown. The following methods for determiningthe extent of flooding are recommended in theGuidelines: (1) use of an existing inundationstudy, (2) engineering judgement, and (3) adam-break/inundat ion study.Use of existing inundation study

16. Many dams have comprehensive dam-break/ inundat ion studies prepared for thedownstream area. If these studies exist, theyshould be used as the basis for downstreamhazard classification.Engineering judgement

17. In some situations the downstream hazardclassification may be obvious - a large damwith a populated area located in the floodplain immediately downstream from the dam, forexample. In these cases the downstream hazardclassification is based solely on engineeringjudgement using information from a field surveyand/of other current topographic maps.A dam-break/inundation study

18. If a comprehensive dam-break/inundationstudy does not exist, or the downstream hazardclassification is not obvious, then an analysisshould be performed to define the inundatedarea. There are three main phases to a dam-break/ inundat ion study: (1) assume a damfailure scenario, (2) determine the downstreamterminal point of flood routing, and (3)perform the recommended analytical procedure.

19. Assume a dam failure scenario. Theresults of a dam-break.inundation study wouldbe the most accurate if we knew the failurescenario a priori. However, for dam-break/ inundat ion studies, this is uncertain andcan only be assumed.

20. The failure scenario possibilities arenearly infinite. Because of this, and for

143

RISK, HAZARD AND SAFETY

safety of dams conservativeness, the Guidelinessuggest a procedure that seeks the highestdownstream hazard classification that isreasonable. This procedure begins with theevaluation of a sunny-day failure. If thesunny-day failure scenario results in a highdownstream hazard classification, no furtheranalysis is necessary because this is thehighest downstream hazard classificationpossible. However, if the downstream hazardclassification is less than high, thenadditional analysis is necessary to determinethe incremental flooding, that is, if a moresevere downstream hazard classification can beobtained if the dam should fail in combinationwith the "incipient danger flood" prior to thedam failure. The incipient danger flood is thenatural runoff flood that results in thepossible downstream hazard to experienceincipient flooding. For example, a runoffflood that causes flooding at thefoundation/ground interface of a house is theincipient danger flood for that particularhouse. After the dam-break flood is combinedwith the incipient danger flood, for allpossible downstream hazards, the magnitude offlooding is compared to Figs 1-5 and the lives-in-jeopardy and, subsequently, the downstreamhazard classification determined.

21. Determine the downstream terminal pointof flood routing. A dam-break flood routingneeds only to be performed for a distancedownstream from the dam until the downstreamhazard classification can be ascertained, oruntil "adequate flood water disposal" isreached. Adequate flood water disposal is thatpoint below which potential for loss of lifeand significant property damage caused byrouted floodflows appear limited.

22. Recommended analytical procedure. TheGuidelines recommend use of the NationalWeather Service Simplified Dam-Break model(SMPDBK) (ref. 7) for estimating flood depthsand velocities, and offer specific criteria forbreach parameters. SMPDBK is recommendedbecause of its ease of use together withreasonable accuracy. However, it is cautionedthat the analyst be knowledgeable of SMPDBK'slimitations. Other methods should be used,such as the National Weather Service DAMBRKmodel (ref. 8), when more accurate results aredesired and/or the analyst has the background,time, and resources needed to apply suchmethods. Additional information regarding dam-break/ inundat ion studies is provided in theappendix of the Guidelines.

CONCLUDING REMARKS23. While downstream hazard classification

may be obvious in situations pertaining tolarge dams and populations, it often requiresdetailed analysis for other situations such asthose involving small dams and populations,However, detailed analysis does not alwaysresult in a firm downstream hazardclassification. Many unknowns exist withregard to loss-of-life to persons in dwellings,vehicles, or on foot. Due to these unknowns,agency guidance is important to giveconsistency in assignment of downstream hazardclassification. The Guidelines are intended toprovide such assistance.

REFERENCES1. Downstream Hazard ClassificationGuidelines. ACER Technical Memorandum No. 11,U.S. Department of the Interior, Bureau ofReclamation, Assistant Commissioner -Engineering and Research, Denver, Colorado,December 1988.2. "Departmental Manual", Part 753, Dam SafetyProgram", U.S. Department of the Interior,January 1981.3. BLACK R. D. Flood Proofing RuralResistance. Department of AgriculturalEngineering, Cornell University, A "ProjectAgnes" Report, prepared for the U.S. Departmentof Commerce, Economic DevelopmentAdministration, May 1975.4. RUH-MING L. Car Floatation Analysis.Simons, Li and Associates, SLA Project No CO-CB-05, February 7, 1984.5. DAVID J. LOVE and ASSOCIATES, INC.Analysis of a High Hazard Flood Zone, preparedfor the City of Boulder, Colorado, Departmentof Public Works, October 1987.6. ABT S.R. and WITTER R.J. Project NumberFlood Hazard Concept Verification Study.Department of Civil Engineering, Colorado StateUniversity, Fort Collins, Colorado 80523,Prepared for City of Boulder Flood Utility,Department of Public Works, Boulder, Colorado80306.7. WETMORE J.N. and FREAD D.L. The NWSSimplified Dam-Break Flood Forecasting Modelfor Desk-Top and Hand-Held Microcomputers,Hydrologic Research Laboratory, Office ofHydrology, National Weather Service, NationalOceanic and Atmospheric Administration, SilverSpring, Maryland, June 1988.8. FREAD D.L.. The NWS-DAMBRK Model,Hydrologic Research Laboratory, Office ofHydrology, National Weather Service, NOAA,Silver Spring, Maryland, June 1988.

144

22. Mitchell's House reservoirs - investigations, monitoring andremedial works

B. H. ROFE, Rofe, Kennard and Lapworth, Sutton, UK, and D.B. WICKHAM andJ. K. HOPKINS, North West Water, Warrington, UK

SYNOPSIS

Mitchell's House reservoirs are retained by an earth embankment and form a single impoundingreservoir divided by an intermediate embankment.

Survey stations installed on the embankments indicated that differential settlements wereoccurring. Together with damp areas observed on the face of No. 2 embankment, this led to ageotechnical investigation being carried out.

Analysis of the data showed that settlement of the main embankment was continuing; drainage of thebase of the dams was taking place into the foundations; high pore pressures existed in thedownstream shoulders; and the stability of the downstream shoulders of the embankments wasinadequate.

Sand drains have been installed immediately downstream of the core to intercept seepage and drainit into the foundations. A rockfill berm installed on the downstream face of No. 2 embankment hsreduced the rate of settlement. A similar berm is to be placed against No. 1 embankment.

INTRODUCTION

1. Mitchell's House reservoirs are situated 3 kmSE of Accrington, Lancashire. They were builtin the latter half of the 19th Century to theconventional design of the time with earthembankments and puddle clay cores. No. 1reservoir was built first, with embankmentsfacing West and South, and No. 2 reservoir wasformed by extending the Western embankment,leaving the South embankment as a dividingembankment. The total length of the mainembankment is 525m, with a maximum height of20m. The reservoirs share a single overflow.

HISTORY

2. Construction of No. 1 reservoir began in1855, the embankment being built with a puddleclay core connected by a clay blanket atformation level to a cut-off trench along theupstream toe. However, the dam was notwatertight until extensive repairs were carriedout to the core between 1881 and 1891. As wasthe practice at the time, a number of headingswere driven into and under the dam to find thesource of the leak, and eventually a large partof the core was excavated and reconstructed.

3. Work began on No. 2 reservoir some time after1876, but apparently even before No. 1 wassatisfactory. Again there were problems inmaking the reservoir watertight, and these werenot overcome until 1892, when a wing trench inthe Southwest corner was filled with clay.

4. Coal mining took place beneath thereservoirs, but ended in 1905, after which nofurther problems appear to have occurred,although some reservations were expressed aboutthe possibility of long term effects. Work wasdone in 1958 to raise the main embankment andpuddle clay core and enlarge the overflow. Atthe same time some of the exploratory headingsremaining from the 1881 repairs were sealed.

5. The reservoirs appear to have beensatisfactory for some years following this work.However following an Inspection in 1980 theEngineer reported that the embankment of No. 2reservoir was "distinctly soggy" in places. Herecommended that a system of survey stationsshould be installed, and that the old headingsystem should be inspected, with the nextStatutory Inspection to take place in 1983.

GEOLOGY

6. The reservoirs are constructed on asilty/clay glacial till, 10 to 15m thick, whichoverlies horizontally bedded sandstones andmudstones of the upper coal measures. The upper2m of rock is moderately weathered with waterpassing through it - lower down the rock issolid and less permeable.

7. The strata below the till is intersected byNortheast/Southwest faults, one of which liesupstream of the main embankment, with a widthvarying between 12 and 30 metres.

The embankment dam. Thomas Telford, London, 1991 145

RISK, HAZARD AND SAFETY

8. The Lower Mountain coal seam i s l o c a t e dbetween sound sandstone at a depth of 70m.Mining of t h i s seam took p lace in the areabeneath the reservoirs between 1886 and 1905.

MINING

9. The first report of coal mining under thereservoirs was in 1886 when it was stated that aconsiderable quantity had been worked under No.1 reservoir, but very little under No. 2;another report in 1892 drew attention to therisk of settlement and fracture if the pillarsunder the reservoirs were removed. Furtherreports in 1893 and 1896 detailed the risks tothe reservoirs if all the coal were to beremoved, and in 1897 agreement was reached withthe mine owners to leave 50% of the coal aspillars immediately beneath the main embankment,although it appears that the coal beneath thereservoir basins was removed completely.

10. In 1903 mining to the West of No. 2embankment caused subsidence at the groundsurface which damaged sand filters underconstruction. It also resulted in No. 2reservoir having to be "emptied for repairs" butthere is no record of what this involved.Mining was abandoned in 1905.

EXPLORATORY HEADINGS

11. The headings driven below the North end ofNo. 1 embankment in 1881 were on two levels - anupper level varying between 6m and 9m below theoriginal ground surface, and a lower level some8 to 9m deeper. The upper headings were in theglacial till, and lined with masonry, the lowerones were in sandstone, and unlined. The upperheadings, and short drives from them, appear tohave followed the line of the centre of the dam,while remaining on the downstream side of thecore. The lower headings went beyond the centreline, and originally had two spurs extendingwell under the reservoir basin. These spurswere sealed with concrete in 1958, as were partsof the upper system.

12. An inspection of the heading system in 1958found deposits of ochreous sludge in the lowerheadings, and water flowing in them. There wasconcern that this might indicate that materialwas being removed from below the embankment.

SURVEY STATIONS

13. Following the 1980 Inspection, 63 permanenttarget stations were installed on the crests anddownstream shoulders of the main embankment tomonitor surface settlements.

14. By the next Statutory Inspection in 1983four sets of levels had been taken. Althoughthere were some inconsistencies betweenindividual results, the general indication wasthat settlement of 20 - 30mm had taken placeover a period of 30 months on parts of theembankment, affecting both reservoirs.

SITE INVESTIGATION

15. In his report on the 1983 Inspection of thereservoir, the Engineer recommended that a siteinvestigation should be carried out: (a) toestablish whether settlement was taking placewithin or below the embankment; (b) to establishthe main parameters of the material used in theconstruction of the embankment; and (c) toestablish the piezometric gradient through andunder the embankment.

16. It was decided to investigate one sectionthrough No. 1 embankment, and two through No. 2,selected as representing areas of settlement andwet areas on the downstream face. Six boreholeswere drilled on each section, andinstrumentation including both pneumatic andstandpipe piezometers, and verticalextensometers, was installed. A number of trialpits were also excavated to establish the coreposition at the crest of the embankment.

17. A geophysical survey was also carried outusing resistivity and conductivity methods toidentify seepage zones and variations inmaterials. The results confirmed theinterpretation of the ground investigation andpiezometer readings, and indicated that whileseepage from No. 1 reservoir was concentrated inthe bottom layers of fill, that from No. 2appeared to occur at a number of levels.

INTERPRETATION

18. The results from the Site Investigationestablished that the embankments consist of asilty clay fill of a soft firm consistency,poorly compacted, and containing a variablequantity of gravel sized fragments of sandstoneand shale. However, the fill material isgenerally of low permeability of the order of 1x 10~7m/s. The embankments overlie a firm siltyclay glacial till which is 10 - 15m thick withlenses of peat and other organic matter.

19. The bedrock is a horizontally bedded strongsandstone with mudstone bands, moderatelyweathered to approximately 2m below rock head,forming a permeable layer. Below this, theworked out lower mountain mine coal seam wasencountered at 70m depth and had sandrock aboveand below. There was no evidence of subsidenceat this depth in the single borehole drilled tothis level. The rock below the seam was used asa base for a surface benchmark.

20. Evidence of a puddled silty clay core in No.1 Reservoir was firmly established, but this hadnot been brought up to the raised height in the1958 remedial work and stopped at a depth of1.2m below the crest. No recognisable core wasencountered in the investigation into No. 2Embankment, and no cut-off excavation into thefoundation material or rock below. Again, thecentral clay material does not appear to havebeen carried up to the crest in the raisingcarried out in 1958.

146

PAPER 22: ROFE, WICKHAM AND HOPKINS

The general findings from the site investigationare shown diagrammatically in Figures 1 and 2for Nos. 1 and 2 Embankments respectively.

21. Seepage Analysis

Figure 1-IMo.1 EMBANKMENT CROSS-SECTION

Dry I

NATURAL GROUND(TILL)

Figure 2-No. 2 EMBANKMENT CROSS-SECTION

A finite element model of the embankments wasformed with sections divided into five soiltypes and their permeability characteristicsaltered within realistic limits in order toreproduce the piezometer readings at the time ofthe investigation. Flow patterns arising fromthese assumptions were then produced, indicatingthat the weathered rock head horizon acts as adrainage lair and has an appreciable effect onthe piezometric surface, and indicated generallygood drainage in the downstream shoulder.However, there is also indication of someseepage flow at higher levels emerging in thedownstream slope indicated by some surfacepiezometers. The typical flow pattern isillustrated on Figure 3 for No. 1 Reservoir.

22. Stability

A number of features gave cause for concern inregard to the stability of the embankments:-

(a) the downstream slopes of the banks weresteep and near the equilibrium point for surfaceslips;

(b) the downstream face of No. 2 Embankment wasirregular;

(c) no obvious drainage facilities were found inthe toes of the embankments, and damp patcheswere common on the downstream faces;

(d) at high reservoir levels water issues ontothe road crossing the downstream shoulder of theNo. 1 Embankment.

180

160

U0

120

100

60

60

40

20

C'-O

20 £0 60 80 W0 120 140 160 160 200 220 2£0 260 280 300 320 340 360 380

Figure 3 -FLOW PATTERN THROUGH No.1 EMBANKMENT

Figure 4 - EFFECTIVE STRESS PIEZOMETER RESULTS

Based on the stress parameters obtained fromthe consolidated undrained triaxial compressivetest results, shown plotted on Figure No. 4, theeffective stress parameters used in thecalculations were:- 0' = 32° and Cf = 0. Checkcalculations were also carried out in the range0f = 3 0 - 3 4 ° but this was found to have littlepractical effect on the results obtained: Themain reason for this being that factors ofsafety were dominated by the pore pressures.The analysis was carried out for circular slipswith a range of piezometric surfaces under theworst assumptions, factors of safety less than 1were recorded but the piezometric surface levelssupported by actual readings indicated thefactors of the safety of the order of 1 to 1.25.

23. It was concluded from this analysis thatsteps should be taken to improve the drainage inthe downstream shoulder to control the porepressures and at the same time increase thefactor of safety against shallow slips by addingberms to both embankments. The extent of theberms was to some extent dictated by the spaceavailable but it was decided that berms of 4mwidth to safe slopes could be established onboth embankments, leading to increases infactors of safety between 15 and 25%, whichshould relieve the stress on the embankments andensure continuing stability.

147

RISK, HAZARD AND SAFETY

24. Remedial Measures

A number of alternative remedial measures wereconsidered to improve the factor of safetyincluding the following:-

(a) reducing the level of the spillway and hencethe top water level in the reservoir. This wasunacceptable as the reduction of 2m in waterlevel would have reduced the capacity by over25%. Thus, making the storage inadequate forthe source;

(b) The reservoirs could be taken out ofcommission. This would involve substantialworks and lead to the possibility of additionalflooding below the site of the embankments. Inany case it was decided as a matter of principlethat the source was a valuable commodity andshould be maintained;

(c) provide a new core seal to the embankmentsto obviate the possibility of further seepageand settlement. Alternative methods includinggrouting, sheetpiling, pile replacement andslurry changing were considered but the cost ofsuch works was considerably in excess ofalternative solutions;

(d) the installation of additional drainage.This could be achieved by installing trenchdrains, a series of finger drains in the uppersections of the embankments, installinghorizontal drainage holes within the embankment,or installing vertical drainage holes from thecrest. On the grounds of cost and effectivenessthe vertically inclined holes were finallyinstalled because it was felt that this wouldprovide further site investigation data andenable control on the drainage to be exercisedwhilst further longer term stability solutionswere considered;

(e) to improve the stability by installing bermson the downstream toe of the embankment as afree draining weight block to increase thefactor of safety against a slip. This wasadopted as an economical solution but subject toprogramming of other works has yet to becompleted.

25. To summarise; the agreed remedial measureswere to install the inclined vertical drainageimmediately downstream of the crest of theembankment. This was achieved by drilling holesinclined at 7.5° from the vertical at 3m centresand installing a 75mm slotted tube surroundedwith gravel. A section is shown on Figure 5.

STONEWAVE

Berms were to be installed on both embankments:No. 2 Berm was placed during the winter period1988/89, and appears to be performingsatisfactorily. The berm on No. 1 Embankmentwill be installed in 1990/91 as soon as the newtreatment works have been completed allowing fordemolition of the existing works at the toe ofNo. 1 Embankment.

26. The berms were designed with a width of 4m(dictated by the needs of the constructionplant) and slopes of 1:3 (vertrhoriz) along thelength of No. 1 Reservoir, and a slope of 1:2.5along No. 2. These required a fill volume ofapproximately 7,000m3 and consisted of wellgraded limestone compacted in layers to give adensity of 2.1 Mg/m3. A crushed sandstonefilter layer was installed beneath the limestoneto dissipate excess pore pressures generatedduring construction. Additional counterfortdrainage was provided above the berm laid to adrain along the inner edge of the berm and pipedacross to a ditch along the toe. During theconstruction of No. 2 Berm some old stone slabcounterfort drains were discovered in the slopeof No. 2 Embankment, indicating that this hadbeen a long standing problem with thisembankment.

MONITORING

27. Throughout the period following the initialsite investigation in 1983 through to thepresent time the embankments have been monitoredto assess the ongoing situation in regard toboth seepage and settlement. Piezometric levelshave been measured in piezometers installed inthe original and subsequent investigation work,and also recorded in the drainage systeminstalled in 1986. Settlement has been measuredon the extensometers installed within theembankment and on surface monitoring pointsinstalled originally in 1981.

28. The piezometer readings have indicatedgenerally that the drainage system is nowperforming satisfactory, and the readings takenon the vertical inclined drains show that thesehave continued to operate and respond slightlyto increases in reservoir level or heavy surfacewater.

29. The degree of settlement has always provedvery much more difficult to evaluate,particularly in regard to readings on thesurface monitoring points. These have been readon average once a year since installation in1980, but unfortunately by different teams ofsurveyors giving results which have beendifficult to correlate. Readings since 1984have been more consistent as they were carriedout by the same independent firm of surveyors.During the period there has continued to be aslow but steady settlement occurring withdifferential movements peaking with theinstallation of the crest drainage system in1986 but generally continuing at a slow rate.At peak, settlements of about 20mm were recordedbut have generally averaged a few millimetresper year.

Figure 5 -VERTICAL DRAINAGE DETAIL

148

30. These have been confirmed by the moreprecise readings taken within the extensometertubes showing settlement taking place of theorder of 5 - 10mm per year. It is difficult tobe certain of the reason for these, but thereappears to be no evidence of the removal ofmaterial from the bank or foundations byseepage, and it is more likely that this is dueto continuing consolidation and settlement of apoorly compacted embankment subject toadditional loading and improved drainage inrecent years. There is no doubt that theunderstanding of the performance of theembankment given by the comprehensive set ofinstrumentation now installed enables it to becarefully and properly monitored. Any suddenalterations in the pattern of settlement orseepage should show up within the pattern ofreadings to give adequate warning of any

PAPER 22: ROFE, WICKHAM AND HOPKINS

problem. These are plotted on an ongoingprogramme with a time base with groups ofinstruments associated with reservoir waterlevels showing each part of the embankment.

CONCLUSION

31. When the remedial works are completed withthe installation of the berm on the No. 1Embankment we believe that the Mitchell's HouseNos. 1 and 2 Reservoirs will then be in a stableand safe condition to give service for manyyears to come. Nevertheless, because of thecomplex nature of the structures and the factthat they overlie old coal mine workings it willbe necessary to maintain vigilance and carefulmonitoring to ensure they continue to remain ina safe and stable condition.

149

Discussion

N. CULLEN (Water Research Centre)As part of its 1988/89 subscription researchprogramme for the then Water Authorities, WRchad published a methodology for the preparationof innundation maps for use by emergency bodiesin the unlikely event of a threatened reservoirfailure. The work had been undertaken byBinnie and Partners under a sub-contract, butremained confidential to the original clients(who include the Scottish Regional Councils).The work had been strongly encouraged andassisted by North West Water and YorkshireWater.

Mr. Cullen expressed the personal hope that,because the DAMBRK program was now availableand because of the increased attention beingpaid in all branches of industry to hazardassessment and contingency planning, thatInspecting Engineers would recommend to theirclients that an inundation map be prepared forembankment dams where a significant number ofpeople - say 100 or more - were at risk.

He had recently attended a conference onEmergency Planning aid felt confident thatmechanisms existed to handle such informationsensitively and confidentially, and toincorporate it within the "All Hazards"contingency plans for civil disasters which arecurrently under preparation at County andDistrict level.

Based on the track record of perform :e of UKdams over the past 60 or so years, theprobability of failure in the future could beencapsulated by the "Rule of 63" - there is a63% chance of a major failure during the next63 years. This risk was very low in comparisonwith most others which UK society faces, butthe threat was not inconceivable.

F.M. Law (institute of Hydrology)In Mr. Trieste's paper the hazardclassification for USBR was based onlives-in-jeopardy being within or outside agrouping of 1-6 people. It appeared to laystress on the number of individuals underthreat. By contrast the British approach wasto recognise the unacceptability of anycommunity of people being under threat from anengineered structure. This was evidenced bypublic reaction to the Flixborough and Aberfandisasters, as well as to the Dolgarrog andLynmouth floods. Building a dam with a plannedthreat to a community, even if 10000 to 1against it happening next year, was not our

The embankment dam. Thomas Telford, London, 1991

standard. Society accepted much larger deathtolls on its road systems so long as they wereperceived as random individual threats; howevera coach crash killing, say, a primary schoolgroup drew a far sharper call for action insafety terms.

Unfortunately the USBR Guidelines recommendusing the NWS Simplified Dambreak model(SMPDBK). This was known to have cleardrawbacks (Ref. 1) compared with DAMBRK properand it is recommended that the latter versionbe used in the user-friendly form developed forthe Department of the Environment by BradfordUniversity and Binnie and Partners.

It was interesting to hear Mr. Knight stressthe 1000 year life of dams in Sri Lanka. Onceone recognised that dam life was of that order(and not just until the capital loans are paidoff) then the high risk of building anyimportant spillway with a 1000 year floodcapacity was clear!

Ref. 1 Water Resources Commission of New SouthWales (1986), Dambreak Seminar Papers(especially that by R. Stack)

J.H. PHILLIPSWhat consideration is given or should we give,to the effects of the P.M.F. emerging from ourspillways. Over a number of years when I wasinvolved with reservoirs, land drainage andtidal retaining embankments in the Lower RiverSevern basin, I was concerned that littleattention was given to the large variation inthe return times catered for and accepted bythe public. From 1 in 2 years for theovertopping of tidal embankments which couldcause loss of life in farms and hamlets, to inexcess of 1 in 10,000 for reservoir floodingfrom a dam, where very soon on the flood routewater would be retained by road and railembankments leading to deep flooding ofhousing, or should a breach occur, similar butmore damaging flooding. At best the culvertsin these embankments will only cater for a 1 in150 event.

In the light of these variations I support theflood routing investigation concept, theresults of which should be passed to the TownPlanning Authorities. This may prevent furtherdevelopment on flood routes and lead to a widerappreciation of the risk and acceptance ofrealistic common standards.

151

RISK, HAZARD A N D SAFETY

E.N. GOSSCHALK (Halcrow and Partners)Research on seismic risk to U.K. dams is beingcarried out for DOE by Halcrow and BuildingResearch Station in collaboration. A draftguidance document for those concerned with thesafety of UK dams was in course of preparationand was due to be submitted to DOE by the endof October. DOEfs comments and approval werenot as yet available. It was intended topropose to DOE that the draft should becirculated to representative authorities forcomment before publication.

It is the case that there is a risk, which isdescribed as extremely small, that earthquakesof greater magnitude than generally perceivedcould occur in the region of the UK. Figuresby Ambraseys and Jackson (Ref. 1) showedepicentres and focal depths of recordedearthquakes which had Richter magnitudes of upto about 6 and focal depths generally of 15 kmor less. These suggested that the occurrenceof events of higher magnitude is conceivableand at such relatively shallow depth (ininternational experience) surface effects couldbe very serious.

Risk Factor

Capacity (1(fm3)

Height (m)

Extreme High

Contribution to risk

> 120

(6)> 45

(6)

Evacuation requirementsin case of danger(No of persons) >1000

(12)Potentiai downstreamdamage High

(12)

120-1

(4)45-30

(4)

1000-100(8)

Moderate

(8)

Moderate Low

(weighting points)

1-0.1

(2)30-15

(2)

100-1(4)

Low

(4)

< 0.1

(0)<15

(0)

None(0)

None

(0)

Table 1

Risk Factors

Total Risk Factor Risk Class(Risk Rating)

( 0- 6)( 7-18)(19-30)(31-36)

i (Low)II (Moderate)III (High)IV (Extreme)

Table 2

Risk Classes

152

Figures published by Long (Ref. 2) and Irving(Ref. 3) were being used to assess theprobabilities of exceedance of peak groundaccelerations. The risk classification forreservoirs proposed in ICOLD Bulletin 72 isbeing used (Tables 1 and 2 attached), based oncapacity of reservoirs, height of darn,evacuation requirements in case of danger andqualitative assessment of potential downstreamdamage. Ted Gosschalk commented tha^reservoirs of capacity one million m or lessor liable to necessitate the evacuation of 100persons or fewer would not necessarily beallocated to other than low or moderate classesof risk.

Dr. Roger Musson of BGS had prepared a mapwhich divided the UK on the basis of experienceinto zones of three levels of seismicity andthis map was shown on the screen. Even in thezones of the lowest level of seismicity, someevents would be possible, for example thedamaging event at Colchester in 1884.Unprecedented events do occur even in regionswhich have been previously seismically quiet.A recent example was the events of magnitudes6.3, 6.4 and 6.7 at Tennant Creek in WesternAustralia within 24 hours in January 1988.Well defined associations between events in theUK and geological features are lacking becausethe geological features are obscured bysuperficial deposits and the relatively smallmagnitudes rarely cause surface ruptures. Thusit is difficult to predict the likely locationof epicentres.

DISCUSSION

It is the intention that the results of theresearch will include guidance on seismicloading, based on peak ground accelerationsgraduated in accordance with the risk posed bythe dam in question, together with guidance onappropriate methods of evaluating safety ofdams.

References

1 Ambraseys, N.N. and Jackson, J.A."Long-term Seismicity in Britain",Conference on Earthquake Engineering inBritain, University of East Anglia,Thomas Telford, April 1985.

2. Long, R.E. "A Ground Motion ProbabilityAnalysis for Britain Based onMacroseismic Earthquake Data", ibid.

3. Irving, J. "Earthquake Hazard inBritain", ibid.

D.E. EVANS, BINNIE & PARTNERS

The work leading up to the construction atWoodhead described by Mr. Chalmers in Paper 20was the study of the flood hazard inLongdendale carried out for North West Water byBinnie & Partners between 1984 and 1986. Ourbrief was to study how the cascade of 5reservoirs (Woodhead, Torside, Rhodeswood,

Table A Flood handling capacity of Longdendale system

Reservoir

Catchment:

Direct (km2)Indirect (km2)Total (km2)Total (%)

Available maximum

Woodhead

33.5

33.549.7

flow capacity

Torside

24.533.558.086.1

past dam

Rhodesvood

4.5158.061.5191.3

prior to work at

Valehouse

3.6262.5166.1398.2

Woodhead (m

Bottoms

1.2366.1367.36100

7s)

Required flow capacity past dam prior to works at Woodhead (m3/s)

(a)(b)

PMF10 000 year

Adopted solution with

319215

throttled

471308

spillway at

480309

Woodhead

481314

481316

Heightening (m)

Peak discharge inPMF (m3/s)

Further spillwaymodifications

5.93(+ wave

surchargeallowance)

1.28 0.25 0.56 0.51

128

None

345

Minor

369

Minor

388

None

392

Minor

153

RISK, HAZARD AND SAFETY

Valehouse and Bottoms) could be brought up tomodern standards. We were to assist North WestWater in development of proposals for thereservoirs up to but not including the designstage. In consultation and co-operation withinterested parties (Longdendale lies within thePeak National Park) we were to prepareengineering alternatives and generalarrangement drawings. The scope was extendedto include close definition of works requiredfor the selected alternative.

The studies concluded that the most economicalway of providing adequate flood dischargecapacity for the 5 reservoirs was to throttleoutflow from Woodhead by a substantial amountso as to exploit to the practical maximumtemporary storage in that reservoir. There wassome flexibility in the choice of schemes whichconcentrated the major civil engineering worksat Woodhead and minimised the modificationsneeded at the other four reservoirs. Themethod selected on economic grounds was alsothe preferred method on environmental groundsas it caused least disruption to the valley asa whole. This conclusion was endorsed by theenvironmental appraisal referred to inparagraph 3 of the paper.

The flood hazard problem in Longdendale isillustrated by Table A which shows the flowcapacity of the whole system prior toconstruction of the new works at Woodhead waswell below that needed to meet Category AGeneral Standard. Moreover the two major damswould be overtopped in the 1 in 10000 yearevent. Generally the limitation on theexisting spillways was lack of channel capacitydownstream of the overflow. The cascade effectof 5 reservoirs in series meant that improvedflood attenuation at the head of the system,where Woodhead commands half of the totalcatchment, was particularly effective as allthe reservoirs would benefit. With theadoption of a throttled spillway at Woodheadthe need for new spillways at the damsdownstream is avoided and modest heightening ofthe embankments at Tor side, Rhodeswood,Valehouse and Bottoms combined with some workto the existing spillways is sufficient tobring the whole system up to Category Astandard. The work at these four reservoirsremains to be done.

When developing the design of the 7mheightening of Woodhead dam and adaptation ofthe existing spillway to give throttleddischarge at high flows, Binnies realised itwould be possible to construct what would be,if effect, a flood embankment on top of thewide crest available. This embankment could beset back from the No. 2 dam core and containarrangements to intercept and divert theleakage passing through and under thisembankment so that the water level just

upstream of the Woodhead 2 core would neverrise above the core top. Seepage through thenew embankment and the existing dam fill belowit would occur only in flood conditions under ahead approximately equal to the flood surchargedepth.

Carefully designed filter and drainage layerswould carry the seepage safely to a generouslysized collector pipe set just above reservoirconservation level. This pipe would drainfloodwater leakage to a discharge point at thetop of the spillway chute. There appeared tobe no virtue in trying to tie into the Woodhead1 core which was known to be ineffective.Perhaps Mr. Chalmers would enlarge on thereasons why in the Babtie design (Fig. 2) ithas been felt necessary to provide a clayblanket tied into the Woodhead 1 core.

M. AIREY (Mdtt MacDonald)Referring to Paper 20 on Woodhead Reservoir

(a) As the reservoir was drawndown for a longperiod, were any desilting measures everconsidered?

(b) Did the prolonged period of drawdown haveany adverse effect upon the quality ofwater at the treatment works downstream?

(c) with the very wide crest of the existingWoodhead embankment the "heightening andthrottling" solution adopted seemed themost obvious. With a normal crest widthwould this approach have been technicallyviable and/or economic?

MR CHALMERS made the following response:

(a) The reservoir was maintained in serviceand the silt itself was not exposed overthe majority of the contract. Onlyduring one period when the reservoir wasempty, accidentally, was the silt fullyexposed during the period of thecontract. So there was no considerationgiven to desilting at that stage.

(b) it is a feed reservoir to the Torsidereservoir which was being used throughoutthe contract period and there was noadverse comment made by our client.

(c) Yes, it was the obvious solution andthere would have been considerabledifficulties if the crest had not been sowide. It would definitely have requiredfar more consideration before embarkingupon raising such an old reservoir if ithad a narrow crest.

154

23. The design and operation of flood storage dams forrecreational uses

J. B. ELLIS. M. HALL, and D. L. HOCKIN, Consultant, Middlesex Polytechnic, Enfield, UK

SYNOPSIS The successful development and enhancement of recreational and amenity activities onflood storage ponds in urban catchments depends upon extending the scope of the hydrological andhydraulic criteria normally applied in their engineering design. The improvement of waterquality and ecology within the pond can be achieved through the application of a variety ofmanagement procedures relating to landscaping, the status of the surrounding land and theconfiguration of inlet and outlet control structures.

INTRODUCTIONl.One of the most common flow control

strategies in urban catchments is to providepurpose-built detention storage and developersare being increasingly obliged to provideattenuation storage for new developments(ref. 1). A recent Local Authority surveycommissioned by Thames Water (ref. 2),identified over 200 individual storage schemesin the Greater London area of which the largemajority were associated with surface watercontrol on new (58%) or existing (18%)developments. Although catchment sizes rangedup to 1000 hectares, 82% were less than 100 haand 50% less than 50 ha. Design dischargesand storage volumes are typically based on areturn period of either 10 or 50 years suchthat live storage volumes varied between 360and 185 000 m3. The majority of these floodstorage facilities were excavated basins withan earth embankment retaining structure.2.Although flood alleviation is the primary

function and justification for theconstruction of such impounded basins, the1988 Land Drainage Improvement WorksRegulations, as well as the 1981 Wildlife andCountryside Act, require that such objectivesmust be achieved in a way which protects and(or) enhances nature conservation and theenvironment. In considering suchenvironmental enhancement it is essential tobe mindful of local community interestsincluding the need to make open water bodiesvisually acceptable whilst at the same timeachieving naturalistic landscapes andhabitats. Full amenity development mayrequire the provision of special facilitieswhich need to be landscaped into the overallreservoir design. Additionally, if thestorage reservoir is to be developed formultifunctional uses which include directcontact recreational activities such ascanoeing or sail boarding, there is a need toensure that the retained water quality is1 clean1 and 'safe1.3.It is now widely recognised that surface

runoff from impervious urban areas can presentwater quality and ecological problems in

receiving water bodies (ref. 3). Thedischarged annual loading per unit effectivehectare of a separately sewered catchment isof a similar order of magnitude as from acombined sewered catchment irrespective of thepollutant considered. Therefore urban floodstorage reservoirs can be subject to polluteddischarges which can potentially restricttheir development for recreational, natureconservation and amenity purposes. However, anumber of design criteria can be utilised toimprove the environmental status of animpoverished water quality and wildlife-pooropen water body in an urban area. Theseinclude considerations of water quality,physical characteristics, access and usage aswell as habitat creation features.

BASIN GEOMETRY AND CONFIGURATIONBasin size4.A number of studies in both Europe and

North America (refs. 4, 5) have shown thatpollution retention in reservoirs is primarilya function of water residence time andturbulence. The former factor is directlyrelated to basin size and field studiessuggest that the ratio of basin surface areato drainage area should be 1 to 2% forresidential and 2 to 3% for commercialdevelopments in order to achieve a totalsolids removal in excess of 70% for annualaverage rainfall conditions. Hydraulicresidence times are another way to expressbasin volume. Long detention times, of theorder of 12 to 36 hours, will result in goodpollutant removals; the longer residence timesare needed for settlement and decay ofbacteria which are of obvious significance fordirect contact recreational activities.5.Optimal removals occur at water depths of

between 1.0 to 1.5 metres which will alsoencourage oxygenation as well as serving birdroosting and feeding purposes. Deeperpermanent pools are needed for amphibians andreptiles whilst fishpools need to have aminimum depth of 2.5 m. A shallow fringing(but discontinuous) platform for emergent

The embankment dam. Thomas Telford, London, 1991 155

ENVIRONMENT AND RESEARCH

vegetation is desirable as it is not onlyecologically valuable but will also enhancenutrient, metal and oil uptake as well asabsorbing wave impacts and concealingunaesthetic changes in water level and litter.

Basin configuration6.The use of indented, irregular shorelines

can provide territorial shelter, seclusion andfeeding grounds for a wider range of wildlifeand waterfowl as well as helping to divide thereservoir surface into discrete zones fordifferent recreational uses.7.The underwater bank profile should have

various gradients of 1:6 or shallower to allowthe development of bands of emergentvegetation of varying widths. Both depth andshape can be modified by excavation, althoughthis may necessitate the provision ofprotection to the reformed banks and bed.Rapid and cheap solutions include placingrubble or rip-rap to create an irregularmarginal profile with spits and bays and tograde gently shelving underwater bankssuitable for planting emergent species. Useof sub-soil filled sandbags on the edges of awater body can also provide a habitat foremergent plants whilst anchored willow logscan provide offshore nesting and roostingsanctuaries or can grow to form fringingwillow in shallower water.8.The inclusion of islands into the

reservoir design will help serve as dividingwalls, increasing flow paths and hydraulic

residence times, encouraging mixing andeliminating the formation of 'dead1 zones. Inaddition they can provide extremely valuablerefuge sites for wildlife.9.The use of a two-cell basin design

achieves two objectives. A forebay (ordiversion structure) will act as asedimentation chamber and oil interceptorminimising 'first-flush' of pollutantdischarges into the main pool and thus improvethe water quality. It will also enhance plugflow and minimise short-circuiting. Thislatter phenomena, which exacerbates thedisturbance of polluted bottom sediments, is acommon occurrence in flood storage reservoirsand the smaller the basin in relation to theinflow volume, the more pronounced is theproblem. Short-circuiting can be controlledby lengthening the flow path (minimum 3/4:1length to width ratios are recommended),submerging the inflow below the permanent poollevel or by the use of inflow baffles todiffuse the inflow.

TURBULENCE10.In order to achieve water quality

improvements in the main basin, one of themost important objectives must be to ensure aninlet design which will minimise theturbulence generated by the inflowing waterwhich is the root cause of much of theobserved poor pollutant performance of storagereservoirs. The inlet function shoulddistribute the influent uniformly over the

. Emergent speciesFloating leafed

species

Footpath

Figure 1: Vegetational Sequence Across Reservioi

156

PAPER 23: ELLIS, HALL AND HOCKIN

cross-sectional area of the settling zone.Inlet baffle walls, submerged weirs orgradually expanding inlet openings to reduceflow velocity will all allow a more controlledrelease of water (at reduced energy) to themain basin where quiescent settling could takeplace. Release into the second basin cell atdepth through a horizontally slotted baffle,for instance, would ensure maximum quiescenceas well as encourage plug flow (ref. 6).11.Use of wide, gently sloping grass swale

channels as inlet conduits can reduce inflowenergy due to bottom and side slope friction.Swales also offer the opportunity forbiofiltration of solids and adsorption ofsoluble and toxic pollutants prior to entryinto the flood basin (ref. 7) .12.The accumulation of sediment shoals near

the inlet zone will also help dampenturbulence in the main basin and also supportan emergent mactophyte reed marsh which willenhance biofiltration and biological uptake ofpollutants. Figure 1 provides an idealisedvegetational sequence for a flood storagereservoir based on a framework of dominantmarginal, floating leafed and submergedmacrophytic species, with emergent speciessuch as Phragmites located adjacent to inlets.The sequence provides maximum hydraulicresistance, effective polysaprobic conditionsas well as offering a range of nesting andinsect habitats. Sediment bars and shoalswill also be particularly attractive to wadingbirds and waterfowl.

13.Reductions in turbulence as well asenhanced retention times can be achievedthrough modifications in outlet design. Dualoutflow structures for quantity and qualitycontrol have long been in use in the UnitedStates (ref. 8). Larger storms can becontrolled through the use of high leveloutlets and overflow spillways, whilst thesmaller, more frequent (and more polluted)storm events can be retained for a longerperiod of time through the use of short, smalldiameter outlets risers. A variety of outletdesigns and their quality characteristics arediscussed in references 4 and 9.

LANDSCAPING14.Planting and landscaping of the adjacent

land should be undertaken to create a naturalenvironment as far as is possible. It ishighly desirable to include such landscapingas part of the initial design with futuremanagement in mind, although considerableimprovements and facilities for passiverecreation and amenity can be introducedretrospectively (refs. 10, 11).15.Well designed,. landscaped vegetation

zones around the storage basin can provideexcellent and integrated urban wildlifehabitats particularly for 'edge1 species ofsongbirds and mammals. A minimum 300 to350 m wide buffer strip is recommended forscreening purposes, to overcome the so-called'island1 effect on species diversity and forgeneral preservation of scenery. Figure 2provides some general guidelines which can beused which include the use of vegetativescreens for car parking and buffer zones to

restrict access as well as gravel beaches,board walks and pond dipping platformsdesigned for disabled access.

16.Shoreline tree and shrub planting needsto be done with care as shading of marginalaquatic vegetation and related accumulation ofexcess organic materials can contribute toboth nutrient and oxygen demand problems.Species that are conducive to passiverecreation include dogwood, yellow stemmedwillow and holly, blackthorn and privet mixes.Additional tree plantings could include alder,white and weeping willow with the surroundingmeadow grass laid down with suitable slowgrowing cultivars (ref. 10).17.The surface of embankments, spill banks

and spillway channels which are subject tohigh velocities can utilise reinforced grassdesigns which can provide protection againstvelocities of 3 m/s for as long as nine hours(ref. 12). Proper management of such dense,well-knit grass swards is essential and thisentails regular cutting, the application offertilisers and weed control.

Lake/riverside Planted bufferwalk and emergency with views through

from housingMown grass

> i t h seating M e a n d e r j n g | o c a |

path; touches/ ^ ^ lakeside in places

Penstock controlledinflow with silttrap or forebay

Cascade entry

• Interpretation boards

• Hide

Reedmarshoutlet

All existing trees preserved. ReinforcedSoil from excavation used to grasscreate mounds/ridges spillway

• • - Mown grassedges to footpath

Figure 2: Idealised Environmental Layout forFlood Storage Basin

MAINTENANCE AND SAFETY18.It is essential that the responsibilities

for future maintenance and safety should bedecided during the planning stage and formalarrangements agreed for regular inspection andmaintenance. These should includeconsiderations of amenity, ecology and waterquality. A formal protocol for inspection,maintenance and servicing should be drawn-up,staff allocated and their duties andresponsibilities confirmed in writing.Administrative and managerial procedures canbe simplified by the routine use ofstandardised reporting forms which should becountersigned by senior staff. Operationalinformation can then be placed on a databaseand accessed to check basin performance andoperating costs. The information can also beused to identify problems that might beovercome in future reservoirs by designmodifications. Information on costs would

157

ENVIRONMENT AND RESEARCH

provide a rational basis for calculatingcommutation payments to be made when anauthority takes over responsibilities from adeveloper.19.Where a lessee develops on-site

facilities for recreational and amenity use,they should be required to accept fullmaintenance and safety responsibility for bothwater area and the adjoining land. Theresponsibilities may therefore include landmanagement, keeping the basin free of silt andweed, macrophyte control as well as providingsafety equipment and controlling public usage.However as recreation, amenity and natureconservation are secondary uses, the leaseshould specifically exclude any powersrelating to the control structure or operationof the flood storage basin.

REFERENCES1. CONFLO 88. Attenuation Storage and FlowControl for Urban Catchments. ThamesWater/Hertsmere District Council, Borehamwood,Herts. 1988.

2. HOWARD HUMPHREYS & PARTNERS. Storage PondSurvey. Thames Water, Reading. 1988.

3. ELLIS, J B (Edit). Urban Discharges andReceiving Water Quality Impacts. PergamonPress, Oxford. 1989.

4. HALL, M J, HOCKIN, D L and ELLIS, J B.The Design of Flood Storage Reservoirs.CIRIA/Butterworths, London. 1990.

5. ROESNER, L A, URBONAS, B and SONNEN, M B(Edits). Design of Urban Runoff QualityControls. American Society Civil Eng., NewYork. 1988.

6. HORNER, R, GUEDRY, J and KORTENHOF, M.Highway Construction Site Erosion andPollution Control Manual. Washington StateDept. Transportation, Olympia, Washington.1989.

7. ELLIS, J B. Bioengineering Design forWater Quality Improvement of Urban Runoff.34-48 in BALMFORTH, D J (Edit): Developmentsin Storm Drainage. Proc. CIRCIA/IWEMSymposium, Sheffield City Polytechnic,Sheffield. 1989.

8. WHIPPLE, W and HUNTER J. Settleability ofUrban Runoff Pollution. Journ. Water Poll.Control Fed., 53. 1726-1731, 1981.

9. NIX, S, HEANEY, J and HUBER, W. SuspendedSolids Removal in Detention Basins. Journ.Env. Eng. Amer. Soc. Civil Eng., 114(6).,1331-1343. 1988.

10. COPPIN, N J and RICHARDS, I G.Vegation in Civil Engineering.CIRIA/Butterworths, London. 1990.

Use of

11. BAINES, C. A Guide to Habitat Creation.Greater London Ecology Unit, London. 1984.

12. HEMPHILL, R W and BRAMLEY, M E.Protection of River and Canal Banks.CIRIA/Butterworths, London. 1989.

158

24. The use of close-range photogrammetry for reservoirembankment monitoring

J. K. HOPKINS and D. B. WICKHAM, North West Water, Warrington, UK, and D. M.STIRLING, City University, London, UK

Photogrammetry is an established technique for the presentation of three dimensional imagery.Close-range photogrammetric techniques have been developed to measure structures and landforms.This paper details studies carried out to monitor reservoir embankment deformation and developmethods of presentation of data.

INTRODUCTION1. Photogrammetry utilises photography inconjunction with conventional surveyingtechniques to produce plans, sections ordigital data, and the opportunity to presentthese in a three dimensional model of therelevant imagery. Its major use has been inthe preparation of topographical maps fromaerial photography. More recently techniqueshave been developed to apply close-rangephotography to the measurement of structuresand landforms by Cooper et al (ref.l) andChandler et al (ref. 2).2. The majority of earth embankment damsowned by North West Water (NWW) have surveystations installed on the crest and in somecases on the downstream face. Additionalstations have been installed in areas ofinstability. These stations consist of astainless steel pin set in a concrete blockat least 500mm deep. These stations areregularly monitored for level and in some casesfor alignment, using conventional geodeticinstruments. This is effective in monitoringmovement at discreet points but provides noinformation on the rest of the embankments.3. Two reservoir sites were selected as beingsuitable to study the effectiveness of close-range photogrammetric techniques and toexperiment with various methods for representinggraphically any movements which had occurred.Spade Mill Nos. 1 & 2 are liable to rotationalslips, shallow flow slides and sink holes inblockwork pitched internal slopes. Dean CloughLower is a grass covered earth embankment whichsuffered from instability of the downstream facewhilst under construction in the 1870's. Thedownstream face shows much evidence of slidesalthough no major movement has occurred inrecent years. The Engineering PhotogrammetryUnit of City University was commissioned tocarry out these studies.4. The anticipated precision for measuringmovement was +/- 5mm on the pitching at SpadeMill, and +/- 20mm on the grassed embankment atDean Clough, depending on the vegetation.

PHOTOGRAMMETRYTheory5. Figure 1 illustrates the situation thatexists when a camera photographs an object. Thecamera produces a central perspectiveprojection of the object on the negative wherethe centre of the camera lens, 0, is theperspective centre for the projection. A pointA on the object is imaged at a on the negativeand Object point B is imaged at b. The angle9 subtended at 0 by A and B is recreated insidethe camera by 0' the angle subtended at 0 by aand b. Therefore the camera may be regarded asa form of theodolite which instantaneouslyrecords an infinite number of angles between aninfinite number of points on the objectphotographed. In this way taking a photographis a remarkably efficient way of recordinginformation. The mathematics of photogrammetry,in effect, allow any desired angle to berecreated from measurements between two pointson the photograph. By measuring the positionsof a series of image points on a photograph theresulting series of angles produces what isknown as a bundle of rays.

Fig. 1

The embankment dam. Thomas Telford, London, 1991 159

ENVIRONMENT AND RESEARCH

6. By taking more than one photograph of anobject the three dimensional coordinates ofpoints on the object can be computed by theintersection of two or more bundles of rays.For most purposes two photographs, known as astereopair, are sufficient.7. Unless the camera is set up over a pre-determined point and pointed in a knowndirection, a procedure which is very difficultto carry out precisely, it is normally necessaryfor the position and orientation of the camerato be computed from the measurement of imagesof points whose object point coordinates havebeen determined by some other method. Thesepoints are known as control points. This isknown as a space resection and is similar inprinciple to the standard resection techniquecarried out with a theodolite, but is threedimensional.

Equipment8. For photogrammetry two major items ofequipment are required - a camera for takingthe photographs and an instrument for measuringthem.9. When the first series of field visits wasbeing planned, the most suitable cameraavailable was a Carl Zeiss Jena UNK 10/1318photogrammetric camera. This is a wide anglecamera with a nominal focal length of 100mm anda format size of 130mm by 180mm. The photo-graphs were to be recorded on glass plates formaximum image flatness and stability. By thetime of the second series of visits a CarlZeiss Jena UMK 30/1318 camera was available.This camera was fitted with a 300mm telephotolens, with the same format size as theUMK 10/1318.10. The photographs were measured on anIntergraph Intermap Analytic (IMA) photo-grammetric colour graphics workstation. Thissystem consists of a photogrammetric analyticalplotter with computer controlled measuring stageplates interfaced with an Intergraph colourgraphics workstation running Intergraph'sInteractive Graphics Design System (IGDS)software. In this way, as information ismeasured from the stereopair it is immediatelystored in a "design file", a computer model of3-D data, which can then be manipulated byvarious graphics routines to produce plots, orcan be used as input for other Intergraphsoftware packages such as Digital TerrainModelling (DTM).11. Provided with the IMA were a number ofspecialist software packages. These includedIMAN for carrying out space resection andother photogrammetric tasks, IMAPF forcollecting three-dimensional spatial data in adesign file and IMAPD for specialised datacollection for input into the DTM package.12. For the measurement of the control survey aZeiss Oberkochen Elta 2 electronic tacheometerwith automatic recording of data on a MEM 400memory module was used along with a Carl ZeissJena Ni 007 precise level. On later surveys aWild TC-1 Tachymat was also used.

RESERVOIRSSpade Mill Nos. 1 & 213. Spade Mill reservoirs form two adjacentstorage units situated near Longridge,Lancashire(Fig. 2.) No. 1 reservoir wasoriginally constructed in 1862 by forming anearth embankment across the valley. Between1905 & 1908 the reservoir was enlarged byraising the embankment and carrying outexcavations in the basin. All the internalslopes were graded and pitched with sandstoneblockwork. No. 2 reservoir was constructedbetween 1952 & 1959, primarily by excavation,with the internal slopes being graded andpitched with concrete blockwork. Bothreservoirs have a maximum storage depth ofapproximately 10m.14. 68 survey stations are installed on thecrests of the embankments; 36 on No. 1 and 32on No. 2. A control pillar at P provides thelocal datum.

Areas of Instability15. Six areas of instability have beenmonitored (Fig. 2).16. Area A .is a series of depressions situatedin the top third of this natural slope. Thedepressions are associated with flow betweenthe reservoir and a surface water drain in asandy lens in the glacial till, removing finematerial.17. Areas B, D & F are shallow slides probablytaking place in softened fill or regradednatural ground. Two survey stations aresituated within slips B & D.18. Area C is a deep seated rotational slip theupper part of which is in full with the slipplane following the underlying sandstone rocksurface. The rock is subject to artesian waterpressures. 20 survey stations are situatedwithin the slip.19. Area E is a deep seated rotational slide inglacial till. 6 survey stations are situatedwithin the slip.

Field Visits20. Three field visits have been made to SpadeMill reservoirs. These have provided photo-graphy of each area at least twice to enablecomparisons to be made.21. 15th October 1986 - For this first visitAreas A and B in No. 1 reservoir and Areas Cand D in No. 2 reservoir were recorded.22. At this time only the 100mm focal lengthUMK camera was available. In order to obtainphotography at a sufficiently large scalephotographs of the sites were taken from asmall dinghy. One pair of photographs weretaken of each site, between 50 to 100 metresoffshore, depending on the extent of each site.Additional photography of each site was takenfrom the water's edge, and, except Area C,from the back of a Land Rover above the wavewall.23. To provide control for the photography,rotatable targets, set in gaps between theblockwork, were positioned throughout eachsite. The control survey and photographyproceeded simultaneously.

160

PAPER 24: HOPKINS, STIRLING AND WICKHAM

'AREA C

'AREA BAREA A

SPADE MILL RESERVOIR No.2

'AREA D'AREA E

AREASPADE MILL RESERVOIR No.1

Fig 2. Spade Mill

24. Control targets on Areas, A, C and D, andpart of B, were coordinated by bearing anddistance measured with the Zeiss Elta 2electronic tacheometer from pillar P. Targetsbelow the wave wall in Area B and other checkmeasurements were co-ordinated from temporarystations at Q and R. Distances between asample of targets were checked using a steeltape.25. 29th October 1987 - During this second visitArea E in No. 1 reservoir and Area F in No. 2reservoir were recorded in addition to the fourareas recorded on the first visit.26. The 300mm focal length UMK camera was used.This enabled the photography to be taken fromthe opposite side of the reservoir. The higherviewing angle removed the need to photographeach site from above the wave wall at the topof the slope.27. For the control survey fixed targets, largerthan used on the first visit, were positionedaround each site. Two additional stations,S and T, were positioned and the targets co-ordinated as before.28. 30th November 1989 - On the third visit toSpade Mill only area E in No. 1 reservoir andArea F in No. 2 were photographed. The 300mmfocal length UMK camera was used, and controltargets were positioned and coordinated asbefore.

Reservoirs

Dean Clough Lower29. Dean Clough Lower reservoir is situatednear Great Harwood, Lancashire. (Figure 3).It was constructed in the 1870's and 1880's byforming an earth embankment across Dean Brook.The dam is 350m long, with a maximum height of22m. The downstream face has a slope of 3:1,with a berm just below half height. It had anearly history of instability. The embankmentwas not constructed to the full heightintended and the berm was added. Uneven areasnear the top of the embankment may indicatesome slipped zones, but there has been noevidence of any recent movement. The crestand downstream face have a grass turf grazed bysheep and reed patches are periodically mown.43 survey stations in four rows are installedon the crest and downstream face of the embank-ment. Two control stations, "Stream" and"Track East", provide the local datum.

Field Visits30. Two field visits have been made to DeanClough Lower Reservoir. The central area abovethe berm which has indications of earlierinstability, has been photographed twice toenable comparisons to be made.

161

ENVIRONMENT AND RESEARCH

Fig. 3 Dean Clough Reservoir

31. 4th March 1987 - Only the 100mm focallength UMK camera was available.A strip of photography was taken to providecover along the dam face. Control was markedusing large fixed targets. These werepositioned in three rows; along the top, middleand bottom of the embankment. The targets wereco-ordinated by bearing and distance from sitecontrol station Stream, and checked from TrackEast.32. 17th March 1988 - The second site visitwas similar to the first. A strip of photo-graphy of the dam was taken with the 100mmcamera as before. Additionally, obliquestereopairs were taken with the 300mm camerafrom further back on the valley sides atapproximately 45 degrees to the embankment.

ANALYSISComputation of Control Surveys33. After each site visit, the recorded fieldsurvey observations were downloaded from thedata recording module of the Elta 2 into amicrocomputer. The data was run through areformatting file to produce an input file forthe Three Dimensional Variation of Co-ordinates(TDVC) program.34. TDVC is an interactive computer program forleast squares estimation of spatial co-ordinates. Field survey measurements areinput. The output gives estimated co-ordinateswith their standard errors and 'adjusted'observations with their discrepancies.Observations with large discrepancies can beremoved and the adjustment recomputed.35. For all the Spade Mill surveys the standarderrors of the estimated co-ordinates of thecontrol targets were of the order of 5 to 7mmin X, Y and Z axes. No observations wereremoved from the computation.

Measurement of Photography36. The measurement of photography of SpadeMill was delayed until two sets of photographshad been obtained. This was so thatcorresponding photography of each area at bothepochs could be measured in quick succession.37. Each pair of photographs being analysedwere first orientated by space resection usingthe Simultaneous Orientation section of theIntergraph IMAN nucleus package. Where plandetail, such as header stones, was to bemeasured for monitoring purposes both epochsof photography were measured using the IMAPFMap Feature Coding applications program tocreate an Intergraph design file of thisdetail. Where other methods of presentationwere to be used then only one epoch of eachsite was measured using IAMPF to provide abase plot for superimposing additionalinformation. In these instances both epochsof photography were then measured using theIMAPD DTM Data Collection applicationspackage.

PRESENTATION OF RESULTSSpade Mill Nos. 1 & 238. A number of different methods of present-ation were used for depicting the results ofthe measurement of the Spade Mill photography.39. Plots of Plan Detail - This method wasused for the analysis of Area C in ReservoirNo. 2, and required data from both epochs. Itdemonstrated that Area C had sufferedsignificant slumping of the lining.40. Significant physical features - the wavewall, water's edge, edge of pitching andheader stones were measured on each epoch ofphotography to produce colour coded graphicsdesign files. These were then plotted at ascale of 1:500. One plot could then be laidover the other to check for any changes in theplotted features. One of the main drawbacksof this method of presentation is that at1:500 only changes larger than 150mm in plancould be detected. Also, no change in heightcould easily be depicted.41. However, this method of presentation couldprove useful if a graphics screen display wasused instead of a paper or film plot. In thisway the two design files could be referencedtogether. It would then be possible to zoomin to specific areas, to check for changessmaller than 150mm. Also by viewing thethree-dimensional data from the front or side,changes in level could also be seen.42. Digital Terrain Models (DTMs) - One datacollection package used in the study producedthree-dimensional graphics files which wereused as input into the DTM processing package.During analysis of the photographs the normalprocedure was to collect three-dimensionalcoordinates across the site in a regular grid.A suitable grid spacing, 5 or 10 metres, wasselected and a point measured at a corner ofthe site. The IMAPD software then drove thestage plates of the IMA so that the measuringmark had moved the equivalent of the selectedgrid spacing across the site. The operatorthen placed the measuring mark on the surfaceof the site and recorded the Z coordinate

162

PAPER 24: HOPKINS, STIRLING AND WICKHAM

along with the X and Y coordinates the IMA haddriven to. This process was repeated untilcoverage of the site was obtained. Anysignificant changes in the surface of the sitewhich fell between the regular grid were thendigitised either as break lines or as spotheights.43. The DTM processing package was then usedto produce a number of different outputs:44. Visualisation - One output was avisualisation of the surface of a site in theform of a rectangular grid of lines. Whenviewed as an isometric projection, all areas ofslumping could be easily seen. This method wastried on Reservoir No. 2, Area C. Althoughthis method provides a good indication ofirregularities in the surface it is notsuitable for depicting changes between epochsas one grid overlaid on another produces a veryconfusing image.45. Contour Plots - The DTM can be used tocalculate contours. This was also used todepict changes in Area C. A contour plot wasgenerated for each epoch and plotted at a scaleof 1:500. When one plot was laid over theother some indication of change in shape couldbe perceived. This method suffers from thesame disadvantages as plots of plan detail.46. The DTM was also used to produce contourplots of difference by subtracting thegenerated surface of epoch 2 from the generatedsurface of epoch 1 to produce a surface ofchange, i.e. peaks were where the surface hadrisen and troughs were where the surface haddropped. A contour plot was then generated andplotted at a scale of 1:500. Area C was againused for this study. The advantage of thismethod over the previous methods was that itimmediately showed areas where change in shapeof the surface had occurred.47. A slightly different method was used forthe depiction of Area A in Reservoir No. 1where there are a number of small areas ofsettlement. It was decided to try to depictthese sink holes and any changes relative tothe general plane of slope, using a contourplot based on a rotated plane.48. A DTM was measured for each epoch. Thetwo resulting design files were rotated intothe mean plane of the slope. A contour plotwas then generated for each epoch as well as acontour plot of difference, showing movementsrelative to the surface plane.49. The major problem on this site was thatthe large rectangular stone blocks and therelatively small contour interval (100mm)produced a rather jumbled series of plotswhich were difficult to interpret. A soilslope or small blocks would follow settlementmore closely giving a better representation ofmovement.^0• Vectors - Although the plots of plandetail gave some indication of physicalmovement of particular features, all the otherrepresentations described above could onlyindicate changes in shape of the surfaces ofthe sites. In an attempt to represent themovement of individual points it was decided tomeasure the same points of detail - in thiscase corners of stone blocks - at both epochs.

This was done by studying epoch 1 photographyin the IMA and measuring particular corners ofstones. Epoch 2 photography was then mountedin the IMA and the software drove the measuringmark to the previously measured coordinates.This greatly helped the identification of theselected block corners on the second set ofphotographs. The operator then placed themeasuring cursor on the new position of thecorner and recorded the coordinates. A programthen subtracted epoch 1 coordinates from epoch2 coordinates and generated a graphics designfile of the required vectors.51. One problem to be overcome was how todepict a three-dimensional movement on a two-dimensional plot. It was decided to show theX-Y movement by a line and the Z displacementby a circle at the epoch 2 end of the line.The magnitude of the Z displacement wasrepresented by the radius of the circle and thedirection of the displacement by the colour orrepresentation of the circle.52. This method of representation was used forAreas B, C and D.

Dean Clough Lower53. For the Dean Clough photography it wasdecided to carry out a two-stage measurementprocess. The first stage was to determine theareas on the dam where movement had taken placeand roughly indicate the magnitude of themovement. The second stage, if required, wouldbe to measure, in more detail, the areas wheremovement had occurred. To date only the firststage has been undertaken.54. For the first stage the epoch 1 photo-graphy was placed in the IMA and a coarse DTMcollected for a spacing of 10 metres. The DTMcollected from epoch 1 was then used to back-drive the IMA with the epoch 2 photographymounted. If the measuring cursor appeared tolie on the surface of the dam the operatorinstructed the IMA to drive to the next point.If the cursor appeared to lie off the surfacethe operator moved the mark down onto the newposition of the surface, noted the change inthe Z coordinate reading and annotated thischange on an enlarged photograph of the dam.In this way areas of localised movement werevery rapidly identified.55. Settlements of between 5 and 12mm wereidentified in specific areas which coincidewith those considered susceptible to movement.

CONCLUSIONS56. The study has shown that photogrammetryis a very powerful tool for monitoringlocalised movements on reservoir embankments.It is a very efficient data gatheringtechnique. Up to six separate sites at SpadeMill were photographed and control surveyed ina single day.57. The measurement of the photography canalso be carried out efficiently, particularlywith an analytical plotter. The ability tobackdrive the plotter using the results of aprevious set of measurements enables selectedpoints to be rapidly identified, or, in thecase of Dean Clough, to identify areas wherelocalised movement had occurred.

163

ENVIRONMENT AND RESEARCH

58. Of the photography taken at Spade Mill,the second and third sets, taken with the 300mmUMK, proved more successful than the first set.The higher viewpoint, from the opposite bank asopposed to a boat, made measurement easier. Inaddition, vegetation covering sections of AreaC at apoch 1 obscured large areas.59. The main purpose of the investigations wasto develop systems to enable reservoir slopemovements to be monitored and quantified. Two-dimensional representation of three-dimensionalmovement, at an economical price, was seen as acentral requirement.60. Plans and sections are readily produced.These give a good representation of the plan orsectional profile, but at a reasonable scalesmall changes are difficult to detect.Depiction is in two dimensions.61. Digital Terrain Models are very powerfulfor providing input data for a wide range ofapplications. However, if produced on a finegrid spacing suitable for quantifying localmovement they are relatively expensive toproduce. This is justified where the DTM is tobe utilised for a graphics screen display orfurther modelling or measurement, but is notcost effective for monitoring of movement. TheDTM results can be viewed as an isometricprojection or utilised to produce contourplots. All methods have draw-backs in tryingto produce a clear three-dimensionalrepresentation of movement.62. Vector plots give a visual indication ofrelative movement superimposed on a plan. Itis also possible to scale dimensions to

quantify movement in three dimensions. Thisform of presentation meets the study objectivesof producing a cost effective visualisation ofthree-dimensional movement in two dimensions.63. On Dean Clough a simplified first stageanalysis has been utilised as a low costindication of whether any movement has takenplace. Detailed analysis of local areas canthen be carried out if required.64. In addition to the monitoring ofembankments with known or suspected movementsthe technique could be utilised to produce areference photography/control data set againstwhich suspected movement in the future canreadily be quantified.

ACKNOWLEDGEMENTS65. The authors would like to thank North WestWater for permission to publish this paper.66. The assistance of colleagues in North WestWater and Department of Civil Engineering, CityUniversity is gratefully acknowledged.

REFERENCES1. COOPER M.A.R., LINDSEY N.E. and STIRLINGD.M. Monitoring the three-dimensionalmovement of a large stone structure.International Society of Photogrammetry andRemote Sensing Congress, Rio de Janeiro,September 1984. The International Archives ofPhotogrammetry (V).2. CHANDLER J.H., CLARK J.S., COOPER M.A.R.and STIRLING D.M. Analytical photogrammetryapplied to Nepalese slope morphology.Photogrammetric Record, 12(70), pp. 453-458

164

25. Accommodating rare floods over embankments and steepreinforced channels

G. R. POWLEDGE and T. J. RHONE, US Department of the Interior, Denver, USA, and P.E. CLOPPER, Simons, Li and Assoc. Ltd, Fort Collins, USA

SYNOPSISBased on research and case history dataavailable from the United States (US)and the United Kingdom (UK) engineerscan now feel confident in the design ofoverflow protection for low embankmentdams. The introduction of articulatedconcrete block revetments, originallydesigned for coastal protection in waveenvironments, has produced a viableoption for application to steep-slope,high-velocity flow conditions where, ifleft unprotected, the channel banks andfoundation material would be subject todamaging and potentially catastrophicerosion.

INTRODUCTION1. With the growing concern about

dam safety throughout the world andincreasing awareness of the hydro-logic inadequacy of many older dams,civil engineers and dam designers arenow looking to innovative alternativedesigns in modifying these dams. Thetraditional approach for accommo-dating design flows at embankmentdams is to design spillway and outletstructures with sufficient capacityto avoid overtopping of the embank-ment. Improvements in the collectionof historical flood data has resultedin significant increases in thepredicted probable maximum flood(PMF). Therefore, many older damsare now considered unsafe due toinadequate spillway capacity. Con-ventional modifications includeincreasing the spillway size and/orraising the embankment. However, inmany cases, these have shown to becostly or impractical.

2. Since 1983, research in the USand the UK has identified severalinnovative alternative designs to themore costly conventional modifica-tions. These alternative designsprovide methods for protecting thesteeply sloped erodible embankmentfaces to achieve a high dischargecapacity by allowing the entirecrest, or significant portionsthereof, to be overtopped. Tradi-tionally, cast-in-place reinforced

The embarkment dam. Thomas Telford, London, 1991

concrete would be used to achieve the'desired level of performance andstability; however, more recently,roller-compacted concrete (RCC) hasbeen used on several embankment damsin the US at a significant costsavings. Also, the introduction ofarticulated concrete block revet-ments, originally designed forcoastal protection in wave environ-ments, has produced another viableoption for application to steep-slope, high-velocity flow conditions

where, if left unprotected, the chan-nel banks and foundation materialwould be subject to damaging andpotentially catastrophic erosion.

3. Earlier performance testing ofarticulated concrete block systems atJackhouse Dam in the UK (refs. 1-2)stimulated recent testing to thesesystems under high-velocity, steep-slope flow conditions (refs. 3-4).The research revealed the charac-teristics of hydraulic stability andnature and magnitude of destabilizingprocesses associated with thesesystems under bare (unvegetated)conditions. Both cabled and non-cabled concrete block systems weretested. Earlier research (1986-1987)of the articulated concrete blocksystems required installation be instrict compliance with the manufac-turers specifications. However,after some failures, modifications tothe installation procedures weredeemed necessary and tests performedlater in 1988 showed improvement.This paper describes the hydraulictesting program, results, and conclu-sions derived from approximately fourand one-half years of study. Thepaper also briefly discusses currentin-house research by the Bureau ofReclamation (USBR) to evaluate thecapability of RCC and wedge-shapedblocks to protect the downstream faceof an earth and rockfill dam, measur-ing 48.5 m high, during overtoppingflows. Modifications made to otherembankment dams to allow overtoppingare also discussed.

165

ENVIRONMENT AND RESEARCH

TESTING PROGRAM4. In early 1986, a large-scale

flume and recirculating water supplysystem were constructed by Simons, Liand Associates, Inc. (SLA) to examinethe performance of embankment protec-tion systems under steep-slope, high-velocity flows. The flume, was3.35 m high, 1.2 m wide, and 27.4 mlong. An erodible embankment 1.8 mhigh with a crest surface of 6.1 mand downstream slopes ranging from2H:1V to 4H:1V was placed on theflume, Fig. 1. Various protectionsystems were installed on the embank-ment surface and subjected to over-topping flows of up to 2.8 m3/s.This flow rate yielded 1.2 m of over-topping head, with maximum velocitiesof approximately 5.2 to 6.7 m/s,depending on system roughness, meas-ured near the downstream toe. Thistesting program was unique fromearlier tests in the US and UK inthat the embankment tested was ahighly erodible silty sand (SM) whichtested the effectiveness of theprotection systems rather than theerosion resistance of the embankmentsoils.

5. Initial performance studies ofvarious protection treatments weresponsored by the U.S. Federal HighwayAdministration (FHWA) and USBR. Ingeneral, the high velocities andlarge tractive forces developed onthe downstream slope of the embank-ment caused deformation and/orfailure of meshes, mats, and wire-enclosed riprap. This type offailure was characteristic of shear-stress dominated deformation at over-topping heads typically in the 0.3 to0.6 m range, with measured shearstresses ranging from approximately190 to 720 Pa. Treatments which suc-cessfully resisted the hydraulicstresses at full discharge includedsoil cement, which for purposes ofevaluation is considered the same asRCC, placed in 10-cm-thick steps, andseveral articulated concrete blockrevetment systems.

Point gauge andvelocity probe

Inlet diffuser(Straightens andsmooths incomingflow)

,Flow metereterl

Jpodbox

.-27.4 x 3.4 x i.2m Flume

/ Video camera-Outlet gates

\ (Control tailwater1 depths)—.

embankment-Embankment test section

-914mm Pipe

Fig. I. Profile of -testing facility

6. For the final phase of the SLAstudies, the FHWA and USBR werejoined by the Soil Conservation Ser-vice (SCS) and the Tennessee ValleyAuthority (TVA) to extend thehydraulic testing program to focusdirectly on the performance of thearticulated block systems. Fivesystems were investigated: threecabled systems (Armorflex, Petraflex,and Dycel) and two noncabled systems[concrete construction blocks andwedge-shaped blocks of Soviet design(ref. 5)]. Fig. 2 provides informa-tion relating to the geometric con-figuration, weight, and dimensions ofeach of the five systems.

7. The Dycel system, with thelargest, area-to-thickness ratio,failed during 0.3-m overtopping headby allowing excessive underflow toaccumulate beneath the system. Theother four systems performed success-fully during overtopping heads ashigh as 1.2 m.

HYDRODYNAMIC FORCES AND REVETMENTBLOCK STABILITY

8. Hydraulic Forces - An individualblock surrounded by a matrix of iden-tical blocks is subjected to theforces of lift and drag under theaction of flowing water. The liftforce acts in a direction normal tothe plane of the bed, and is typi-cally comprised of the buoyant force

« 4 4 c m ». 39 cm.

DD1 ! 1 1' I I Ii I I I

1it i i

! 1i i

J [^3 cm Grooves

SYSTEM

A. Armorflex Class 30B. Dycel 100C PetraflexD. Construction BlockE. Wedge Block

UNIT WEIGHTWITHOUT FILL(Kg/m 2 )

176205205195234

Fig. 2. Sketches of five types of concrete blocks tested.(NOT TO SCALE)

166

PAPER 25: POWLEDGE, RHONE AND CLOPPER

and differential pressure across theblock due to local accelerations.Lift forces can be substantiallyincreased due to excessive seepagepressures beneath the block, and byflow separation which causes a nega-tive pressure to occur on the uppersurface of the block. The lattercommonly occurs at sharp transitionsfrom a mild bed slope to a steeperone i.e., at the transition pointbetween dam crest and the downstreamslope (ref. 6).

9. The USBR recently conducted astudy to measure the pressure profilealong a horizontal surface and over asharp transition to a steep slope.This study revealed a large reductionin surface pressure in the vicinityof the change in slope. This pres-sure change is present over a veryshort distance in the direction offlow. Apparently the curvilinearflow over the intersection returns tothe original flow profile in thisshort distance. Fig. 3 shows typicalpiezometric pressure profiles forthree different changes in slope atan overtopping head equal to 0.66 m.

Fig. 3

10. The drag force acts in thedirection of flow, and is comprisedof frictional drag and form drag.Form drag, in particular, can lead tothe creation of forces large enoughto initiate block movement (rotation)where the block in question presentsa frontal profile which is subject todirect impact by the flow (ref. 7).This is possible in instances ofirregular subgrade preparation orpoor installation where an individualblock protrudes vertically above itsadjacent neighbors. Cabled blocksystems have the ability to maintain

& 2.00o

i= 1.50oQ.

CDL

CL

O

CD

O

1.00

S 0 50OL 3.

Overtopping Crest Study0T Head = 0.656 Meters

Scale ( i : i )

2=202=302=40

50 4.00 4.50 5.00 5.50Horizontal distance along embankment (m)

Fig. 3. Pressure profile at a change in slope.

the amount of projecting frontal areaat a practical minimum when cableruns are oriented in the direction offlow, with the maximum height of pro-jection limited to the differencebetween the diameters of the cableand the cable tunnel (Fig. 4 a ) .Wedge-shaped block systems negatethis effect entirely by providing athin upstream cross section and athicker downstream one. All upstreamedges are therefore effectively"shielded" from direct impact(Fig. 4 b ) . Drainage slots are alsoprovided at the downstream edge,thereby relieving seepage and upliftpressures on the underside of therevetment and enhancing the intimatecontact between the blocks and thesubgrade (ref. 8 ) .

Definition of Failure11. Loss of "intimate contact"

between a block, or group of blocks,and the subgrade which they are toprotect has been identified as theprimary indicator of incipientfailure (refs. 2 - 3 ) . Given thenature of revetment mattressinstallation in typical steep-slopeapplications, failure due to slippingor sliding of the revetment matrixalong the plane of the bed is remoteand has never been observed undercontrolled test conditions. Thisincludes steeply sloped embankmentswhere mechanical or vegetative shear

Stagnation pressure

Protuding block-

Geotextile-Fig. 4a. Typical profile of cabled revetmeni-

system cables in direction of flow.

Separation zone

Drainage layer

Geotextile-

Fig. 4b. Typical profile of wedge block system.

167

ENVIRONMENT AND RESEARCH

restraint was not provided. Appar-ently the frictional resistancedeveloped between the blocks, geotex-tile and/or granular filter, andsubgrade soil is usually sufficientto prevent sliding occurrence. Theloss of contact, therefore, is theresult of overturning forces leveredabout the downstream edge, or aboutthe downstream corner point when theblock is located on the sideslope ofan already steeply sloped channel.However, physical dislodgement oreven measurable movement does notneed to occur in order for the un-desirable seepage flow to initiateand progress within the subblockenvironment, causing erosion of theembankment.

12. Therefore, the definition of"failure" of an articulated blockrevetment system is when overturningmoments are exactly balanced byresisting moments. The dominance ofoverturning moments denote thecondition where the ingress of flowbeneath the system is imminent, andloss of contact is initiated. Thisdefinition of failure appears reason-ably conservative in that the addi-tional shear and uplift restraintprovided by vegetative or mechanicalanchorage systems is not dependedupon by the designer. Likewise, anyrestraining force which can beattributed to cables should not beconsidered, because the mobilizationof tension forces in cables can onlycome into play once finite rotationhas occurred, by which time thesystem has already been defined ashaving "failed."

Stability Analysis13. Both lift and drag on a block

produce overturning moments propor-tional to their magnitude and to thelength of the moment arms throughwhich they act. The resistance tooverturning is provided by the sub-merged weight of the block actingthrough the center of gravity and itsmoment arm. Hydraulic stability isthus dependent on the hydraulic con-ditions of flow and the size, weight,and geometric characteristics of theblock. The analytical method fordetermining revetment stability byway of the "factor of safety" methodwas developed originally by Simonsand Senturk (ref. 9) in their deriva-tion of a methodology for evaluatingthe stability of rock riprap in open-channel flow. In their method, thecritical shear stress at whichparticle motion is initiated wasdetermined by the Shields relation-ship. In the case of articulated

block revetment systems, the criticalshear stress is determined throughcontrolled hydraulic testing andmeasurement•

14. The factor-of-safety procedurecan be extended to blocks of differ-ent dimensions and weights, providedthey are geometrically similar to thesystem for which the critical shearwas previously determined throughlaboratory testing. The blocks mustbe of the same "family" in terms ofmethod of interlock, profile config-uration, and characteristics of boun-dary roughness and interaction withthe flow field. Given this basicsimilarity, the weight and dimensionsof the block in question can be com-pared with those of a tested block todetermine the critical shear stressand a force balance approach.

MODIFICATIONS TO ALLOW OVERTOPPING ATEXISTING DAM SITES

15. During the 1980s a number ofcost-effective modifications havebeen made to existing embankment damsand grasslined waterways to preventoverflow erosion. Modifications suchas the use of gabions, grouted riprapand even well-maintained grass lininghave proven to be effective under lowovertopping flows (ref. 6). The useof simple concrete constructionblocks has shown to be effective inpreventing erosion of SCS chutespillways.

16. The use of RCC has beendemonstrated to be a cost-effectivealternative on several embankment

dams in the US by providing erosionprotection during overtopping headsin excess of 0.6 m deep. SpringCreek Dam, located in Colorado, has a15.2-m-high embankment and wasmodified in 1986 to allow overtoppingby the stairstep placement of RCC onthe downstream face.

17. The use of articulated con-crete blocks to prevent overtoppingerosion was first used in the UK(ref. 1), and the wedge-shaped blockswere developed and used in the USSR(ref. 5). These applications allowedfor the articulated concrete mats tobe placed on the downstream face ofan embankment or steep waterway.With the use of soil anchors andgrass vegetation these applicationshave been effective in preventingerosion.

18. Articulated concrete blockmats are also currently being used tomodify three embankment dams in theBlue Ridge Parkway located in the

168

PAPER 25: POWLEDGE, RHONE AND CLOPPER

Eastern United States (ref. 10).These dams range in height from 8.4to 12.0 m at their maximum section.The modifications are designed toprevent breaching and erosion fromovertopping flows 1.2 m deep.

RECENT DEVELOPMENTS AND FUTUREDIRECTIONS

19. With the sufficient researchand case history data currentlyavailable engineers can now feelconfident in preparing designs foroverflow protection for low embank-ment dams. However, USBR now facesthe challenge of protecting muchlarger dams during overtopping flows.A. R. Bowman Dam is a 48.5-m-highearth and rockfill embankment locatedin Central Oregon which is projectedto be overtopped by flow depths of upto 6.3 m during the PMF event.

20. A research program has beeninitiated by USBR, and has thefollowing objectives regarding damovertopping: 1) design criteriautilizing RCC technology for new damsand overlays for rehabilitatingexisting embankment dams that complywith Safety of Dams criteria, 2) todetermine optimal step geometry as afunction of hydraulic forces andenergy dissipation, 3) to developstep geometry for embankment damsthat uses hydraulic forces to enhancesubsystem pressure relief duringoperation (ref. 11).

21. In addition to laboratorystudies, field studies will be made

to test and compare at near-prototypeconditions the performance of RCC,Russian wedge-shaped blocks, and rip-rap embankment protection systems.These data will evaluate flow aera-tion, dynamic pressures and embank-ment drainage as important variableswhich influence protection methodstability under large overtoppingflow conditions.

22. Two adjustable slope testfacilities have been built to developstep geometry. One is to investigateembankment overlays on slopes of 2:1,3:1, and 4:1. The other will be usedfor concrete dam slopes of 0.6:1 and0.8:1. These flumes are each 0.45 mwide by 0.75 m high and have a verti-cal fall of about 4.5m. The maximumunit discharge for each flume is1.67 (m3/s)/m. The sidewalls of theflumes are formed in clear plastic.A rail mounted instrument cart isprovided along each flume for laservelocimetry, air content measurement,other instrumentation, andphotography.

23. These indoor facilities willbe used to optimize the spillway stepgeometry for relief of uplift pres-sures under RCC overlays on embank-ment dams by venting through theoverlay to low pressure zones of thesteps and for increasing of energydissipation to minimize required toeprotection. Pressures, velocity pro-files, flow depths and hydraulic jumpcharacteristics will be measured.

24. Early results of this researchhave confirmed a previous hypothesisthat the velocity attained along thedownstream slope is directly propor-tional to the depth of overflowacross the embankment crest ratherthan the length of chute or slope.Once uniform flow is reached thevelocity remains constant.

25. A new prototype outdoor testfacility will utilize a 2:1 slope,have an approximate 15 m verticaldrop, be 1.5 m wide by 1.5 m high,and have a unit discharge oi:4.65 (m3/s)/m. An existing prototypechute or a university test facilitywill be used for a series of nearprototype tests to evaluate theeffects of aeration, dynamicpressures, embankment drainage, andnatural freeze-thaw phenomena onprotective system stability.

SUMMARY AND CONCLUSIONS26. Based on prototype experience

and/or large-scale testing of manyerosion-protection systems, thoseshowing sufficient stability to per-form reliably under steeply slopedhigh-velocity flow conditions appearto be limited to (1) traditionalcast-in-place reinforced concrete,(2) roller-compacted concrete (or itslower-strength alternative, soilcement), and (3) selected articulatedconcrete block revetment systems.

27. The articulated concrete blocksystems may be the most cost effec-tive alternative in many typical lowhead project settings, subject toavailability and proximity of mate-rials and equipment.

28. The proper selection of blocktype, weight, and dimensions arecritical to design performance.These dimensions can be determined bya factor-of-safety method ofanalysis, provided that initialdeterminations of critical shearstress are performed under controlledconditions as described in thispaper. Also installation proceduresshould conform to site specific con-ditions. The block manufacturers1

specifications have not always beenappropriate.

169

ENVIRONMENT AND RESEARCH

29. The RCC and Russian wedge-shaped blocks appear to have applica-tion for protecting high embankmentdams from overtopping flows.

30. Recent research verifies thatterminal velocities are directly pro-portional to the depth of overflowrather than the length of the slope.

REFERENCES

1. HEWLETT, H.W.M., L.A. BOORMAN,and M. E. BRAMLEY, 1985. Reinforce-ment of steep-grassed waterways.Construction Industry Research andInformation Association (CIRIA)Technical Note 120, 6 Storey's Gate,London, England SW1P 3AU.2. HEWLETT, H.W.M., L. A. BOORMAN,and M. E. BRAMLEY, 1987. Guide to thedesign of reinforced grass waterways.CIRIA Report No. 116., 6 Storey'sGate, London, England SW1P 3AU.3. CLOPPER, P. E. and Y. H. CHEN,1988. Minimizing embankment damageduring overtopping flow. FinalReport, Simons, Li & Associates, Inc.Fort Collins, Colorado, prepared forthe Federal Highway Administration,Washington, D.C.

4. CLOPPER, P. E., 1989. Hydraulicstability of articulated concreteblock revetment systems during over-topping flow. Final Report, Simons,Li & Associates, Inc., Fort Collins,Colorado. Prepared for the FederalHighway Administration, U.S. Bureauof Reclamation, Soil ConservationService, and Tennessee ValleyAuthority.

5. PRAVDIVETS, Y. P. and M. E.BRAMLEY, Stepped protection blocksfor dam spillways. Water Power andDam Construction, Vol. 41, No. 7.,July, 1989, 49-56.6. POWLEDGE, G. R., D. C. RALSTON,PAUL MILLER, Y. H. CHEN, P. E.CLOPPER, and D. M. TEMPLE, 1989.Mechanics of overflow erosion onembankments. II: Hydraulic and DesignConsiderations. ASCE Journal ofHydraulics Engineering, Vol. 115,No. 8, August 1989, 1056-1075.7. BAKER, R., Pre-Cast concreteblocks for dam spillways. (R&DSection) Water Power and DamConstruction, Vol. 41. No. 7, July1989, 60-66.8. BRAMLEY, MERVYN, RICHARD MAY, andRAY BAKER, Performance of wedge-shaped blocks in high velocity flow,CIRIA Research Project 407, Stage 1Report, July 1989.

9. SIMONS, D. B. and F. Y. SENTURK,1977. Sediment transport technol-ogy. Water Resources Publications,Fort Collins, Colorado.

10. WOOTEN, R.L., G. R. POWLEDGE andS. L. WHITESIDE, 1990. Cellular Con-crete Mat Overtopping ProtectionSystems on Three Blue Ridge ParkwayDams. Proceedings of 1990 NationalConference of Hydraulic Engineering,ASCE, San Diego, California.11. FRIZELL, KATHLEEN H., PERRY J.HENSLEY, DAVID L. HINCHLIFF, andCHUCK HENNING., Overtopping protec-tion for embankment dams. Proceed-ings of National Cpnference ofHydraulic Engineering, ASCE,San Diego, California, 1990.

170

26. Deformation of Ramsden dam during reservoir drawdown andrefilling

P. TEDD, J. A. CHARLES, BRE, Watford, UK, J. R. CLAYDON, Yorkshire Water, Bradford, UK

Ramsden dam is an earthfill dam some 25m high with a puddle clay core. It is estimated thatthe crest has settled more than lm since construction in 1883. Measurements have shown thatlong term settlement is largely due to drawdown of the reservoir. The net crest settlement dueto a cycle of drawdown and impounding was affected critically by the magnitude of thedrawdown. Settlements occurred throughout the full depth of the core.

INTRODUCTION1. Ramsden dam was constructed between 1879

and 1883 for Batley Corporation Waterworks andthe engineer was G H Hill. The reservoir is oneof four in the Holme valley in West Yorkshiresupplying water to Dewsbury. The maximum heightof the dam is about 25m and the length is justover 120m. The upstream slope is 1 in 3 and thedownstream slope is approximately 1 in 2. Thelower part of Ramsden dam is submerged byBrownhill reservoir. The original constructiondrawings indicate that selected fill was placedeither side of the core. The puddle core isfounded on a concrete filled cut-off trenchthat has a maximum depth of 22m in the centreof the dam. The clay core is 3m wide at thecrest and increases in width with depth withboth faces having batters of 12 in 1.

2. Ramsden dam shows obvious signs thatconsiderable settlement has occurred. It isestimated from records that there has been morethan lm of settlement since construction in1883 (ref.l). Since 1977 precise surveying ofthe level and horizontal alignment of stationson the crest close to the wave wall has shownthat the continuing and varying rate ofsettlement and downstream movement was relatedto reservoir fluctuations (ref. 2). The maximumsettlement has occurred at the central sectionof the dam where it is deepest. The averagerate of settlement between 1977 and 1985 wasapproximately 8mm per year and the average rateof horizontal downstream movement was 3mm peryear.3. It was originally intended to construct a

new wave wall on Ramsden dam in 1987, but thiswas deferred following consideration of thedetailed design with the panel engineer. Dueallowance could have been made for the presentrate of settlement, but uncertainty over thecause of the settlement raised the possibilitythat adding a substantial wave wall couldaffect local stability. It was decided to delayconstruction pending an investigation in whichthe effects of reservoir fluctuation onembankment deformations were monitored. Hoyle(ref. 3) suggested in 1975 that those engagedin research would find this subject worthy offurther investigation.

The embankment dam. Thomas Telford, London, 1991

INSTRUMENTATION4. Figures 1 and 2 show the location of some

of the instruments installed at Ramsden dam. In1987, surface surveying stations were installedon the downstream slope, the crest and theupper part of the upstream slope. Settlementand horizontal upstream/downstream movementswere measured relative to two survey pillars onthe valley sides. Also in 1987, magnetsettlement gauges and inclinometers wereinstalled in the core to measure sub-surfacemovements. Precise levelling was continued onlevelling points installed in 1977 along thecrest close to the wave wall.

120m -Survey stations along crest

6 5 4 3 2

\ > ,',-j

Key: Inclinometer, I Settlement gauge, 1 Pressure cell-piezometer

Survey stations along crest

6 5 4 3 2

Fig. 1. Longitudinal section through coreand settlement along the crest

171

ENVIRONMENT AND RESEARCH

5. During earlier investigations (ref. 1) toassess the susceptibility of the core tohydraulic fracture and to examine theeffectiveness of the downstream fill to act asa filter, push-in earth pressures cells andpneumatic piezometers were installed in thecore, and standpipe piezometers were installedin the core and the upstream and downstreamfill. The shear strength parameters of thedownstream fill were also measured.

OBSERVATIONS DURING THE 1988 AND 1989 DRAWDOWNS6. Detailed observations of movements are

presented for the reservoir being emptiedcompletely during 1988 when repair work wascarried out to the draw-off works and for thecontrolled drawdown to 6m below TWL in 1989.Six stages are identified in describing theresults (see Table 1). They correspond toperiods of time and various states of Ramsdenreservoir. The dates for the beginning and endof the stages correspond to when observationswere taken. During Stage 2, the "empty11 state,the reservoir was partially refilled, howeverStage 2 can still be regarded as a period whenany movements were due largely to timedependent effects. Because of the silt and theupstream slope of the valley bottom, the emptystate corresponds to a level approximately 18mbelow the crest (16.6m below TWL).

Table 1. Stages during investigation

Stage

Stage

StageStageStageStage

Stage

1,

2,3,4,5,

6,

Dates

24/6/88 -

6/10/88 -24/11/88 -5/4/89 -28/7/89 -

12/10/89 -

6/10/88

24/11/885/4/8928/7/8912/10/89

1/2/90

State of Ramsdenreservoir

Drawn down fromfull to empty"Empty11

ImpoundingFullDrawn down to6.0m below TWLImpounding

Surface movements7. Except for observations on the upstream

slope, all movements presented are relative tothe beginning of May 1988 when both reservoirswere full. Little movement was measured fromwhen the instruments were installed in 1987 toMay 1988 with the reservoir remaining full.Surveying stations could not be installed onthe upstream slope until the reservoir had beendrawn down slightly. Figure 1 shows thelongitudinal section and the crest movements atthe end of various stages and Fig. 2 showssurface movements on the crest, and upstreamand downstream slopes at the deepest crosssection at the end of the various stages. Forclarity the movements for stages 4,5 and 6 areonly shown for the crest survey station. Figure3 shows the level of Ramsden reservoir andvertical and horizontal movement time plots fora selection of the survey stations.

8. Figure 2 shows that settlement andupstream horizontal movement due to drawdown(Stage 1) were largely confined to the crestand the upstream slope. The largest settlementsoccurred in the middle of the crest, above thecentre of the clay core. Figure lb shows thatthe magnitude of crest settlement was relatedto the height of the embankment. Some movementof the downstream survey stations close to the'crest occurred (see Fig 1), but the magnitudeof the movement decreased with distance fromthe crest. The settlement component of thecrest movement was very much larger than thehorizontal upstream movement. On the downstreamslope, the magnitude of the settlementcomponent reduced so rapidly with distance fromthe crest, that the horizontal component ofmovement was greater at stations B,C and D. Atstations E, F and G, the small settlements upto 6mm may be due to the drawdown of Brownhillreservoir. The magnitude of the movements onthe upstream slope will have been greater thanthat shown in Fig. 2 since measurements couldnot be started until the reservoir was drawndown by 2m.

Ramsden

TWL

10mm I Movement scale Stages: 1. Empty2. Empty3,4. Full5. 6m Below TWL6. Full

Brownhill

Fig. 2. Summary of surface movements at the end of the stages

172

1988

M J J A S O N D

1989

PAPER 26: TEDD, CHARLES AND CLAYDON

1990

J F M A M J J A S O N D

A U2

Fig. 3. Development of surface movements and changes in reservoir level with time

9. Figure 3 shows that settlement andupstream movement occurred as soon as thereservoir drawdown began. The rate ofsettlement of the crest stations was notsignificantly larger than those on the upstreamslope particularly in the early stages ofdrawdown. Settlements on the crest and upstreamslope continued during Stage 2 when thereservoir was empty and even when partialrefilling occurred between the 3rd October and7th November 1988 (see Fig. 3). Partialrefilling caused the rate of settlement todecrease. The rate of decrease was larger thefurther away from the core. Emptying thereservoir again in November caused a rapidincrease in settlement, again with themagnitude being greatest at the crest. Furthersettlement occurred when the reservoir wascompletely empty. There was little movement ofthe survey stations on the downstream slopeduring Stage 2.

10. Refilling the reservoir from empty tofull (Stage 3) caused only 6mm of heave at thecrest, about 10% of the total settlement.Settlement at the crest was still continuing(see Fig 3) when refilling commenced andcontinued until the reservoir level wasapproximately llm below the crest level.Measurable heave did not occur until thereservoir was 5m below crest level. In contrastheave of the survey stations on the upstreamslope was measured shortly after reservoirimpounding began. A much larger proportion ofthe settlement on the upstream and downstreamslopes recovered on reservoir refilling (seeFig. 2). Again measurements could not be takenon the upstream stations when the reservoir wasfull. Refilling of Brownhill reservoir duringDecember 1988 probably caused the heave of thestation G.

11. Refilling also caused the crest to movehorizontally downstream by an amount greaterthan it had previously moved upstream,resulting in a net downstream movement.However, on the downstream slope, at surveystations A,B,C and D, the upstream movementcaused by drawdown was generally equal to thedownstream movement on refilling with no netdownstream movement. As with the heave onrefilling, the majority of the horizontaldownstream movement occurred during the laststages of refilling.

12. Very little movement occurred duringStage 4 when the reservoir remained full for 4months, but as soon as the reservoir waslowered (Stage 5), settlement and upstreammovement of the crest occurred. Refilling thereservoir caused the same pattern of movementsas with the earlier refilling, but the amountof crest settlement recovered as a percentageof the maximum settlement was much larger thanwhen the reservoir was completely emptied, seeTable 2.

Table 2. Crest settlement and recovery

Maximum settlementon drawdown

Settlement recoveredon refilling

Percentage recovered

Completedrawdown

58mm

6mm

10%

Partialdrawdown

16mm

8mm

50%

Note: The maximum settlement includes Stages1,2 and part of 3 for the complete drawdown

173

ENVIRONMENT AND RESEARCH

Sub-surface movements13. All sub-surface movements presented in

Figs 4 and 5 are absolute movements having beencorrected for movement at the top of theinstruments. Figure 4 shows the horizontal andvertical movements with depth at the end ofvarious stages. All movements are relative toMay 1988, except for the settlements presentedin Fig. 4c. where they are relative to the endof stage 4. Figure 5 shows the verticalmovements with time at the surface, and 9.1mand 18.5m below the crest together with thelevel of Ramsden reservoir.

14. Complete drawdown of the reservoir(Stage 1) resulted in vertical displacementthroughout the depth of the core. Linearextrapolation of the settlement below 22.5mgives zero settlement at the top of the cut-offat depth of 25m. The average strain over thefull depth of the core was approximately 0.16%although it was significantly smaller above adepth of about 9m and no strain appears to haveoccurred in the upper 2m of the core.

Horizontal movement, mm Settlement, mm

u/s 20 0 20 d/s 0 20 40 60

15

- 20

(a)

= 10

I

i15

20

25

• OD/ uI II• OD

J

IIOD

3-*//*-2I. o'n'

/ //• O D

m• OD

. ///

(b)

- Top of cut-off

Settlement t mm

0 20

10

T5

201

t

(c)

— 25

Fig. 4. Sub-surface movement of the core at the end of the stages(a) horizontal movement, (b) settlement, stages 1,2 and 3 and (c) settlement, stages 5 and 6

1988 1989

M J J A S O N D J F M A M J J A S O N D J F

I

20

40

60

80

1990

Stages

10

15

- 20

( ) Depth of settlement, metres below crest

Fig. 5. Development of sub-surface vertical movements in the core with time

174

PAPER 26: TEDD, CHARLES AND CLAYDON

15. The inclinometer results (see Fig. 4a)showed upstream horizontal displacements toincrease with depth to about 5m as a result ofreservoir drawdown. From 5m to 10m there was noincrease in upstream movement. The smalldiscontinuity in the measurements below 10mcorresponds with the discontinuity in verticalstrain between llm and 13m, (Fig 4b).

16. The effect of partial refilling ofRamsden reservoir during Stage 2 wasaccompanied by a decrease in the rate ofsettlement at all depths. A very small heavemay have occurred below 18m depth, (see Fig 5).Emptying the reservoir again in November causeda rapid increase in settlement at all depths atboth sections. Further settlement occurred atall depths when the reservoir remained emptyduring November.

17. On refilling (Stage 3) heave began tooccur at depths equivalent to the reservoirlevel while the crest was still settling. WhenRamsden reservoir was full again, a netsettlement had occurred at all depths. Thediscontinuity in vertical strain between llmand 13m was more evident. Above and below thissmall discontinuity the strain was reasonablylinear. The strain above this depth was lessthan the strain below, see Fig 4b. Refillingcaused little change in the horizontal movementprofile except for the apparent totaldownstream movement of the whole core.

18. Very little movement occurred duringStage 4 when the reservoir remained full.Figure 4c shows the profile of settlement withdepth caused by drawing the reservoir down to6m below TWL,(end of Stage 4 to end of Stage 5)and then refilling (end of Stage 6). Thesettlement measurements in Fig. 4c are relativeto the end of Stage 4 when the reservoir wasfull.

19. Piezometric levels in the upstream anddownstream shoulders close to the core followedthe water levels in Ramsden and Brownhillreservoirs respectively with little time lag.Permeability measurements in the downstreampiezometers gave values of 10"7m/s. Themeasurements showed that water pressure fromRamsden and Brownhill reservoirs act againstthe core of Ramsden dam. Above the level ofBrownhill reservoir the piezometric levels inthe downstream fill were very small.

20. Piezometric and earth pressuremeasurements in the core showed a steadydecrease on reservoir drawdown and increase onrefilling. The ratio of change in horizontalearth pressure and piezometric pressure tochange in reservoir head was approximately 0.3.This is similar to that measured in other dams(ref. 4).

DISCUSSION21. Long term continuing deformation

(settlement and horizontal movement) can be animportant indicator of performance of a dam.Apart from localised surface disturbancescaused by, for example, traffic and animalactivity, there are four main mechanisms thatcould cause long term deformation.

a. Internal, erosionb. Slope instabilityc. Secondary consolidation of puddle clay and

creep of shoulder filld. Stress changes due to fluctuations in

reservoir level

22. The first two are due to some malfunctionof the embankment and will at some stagerequire some remedial work to be carried out tothe dam. Earlier investigations, (ref. 1 and 4)indicated that internal erosion of the core asresult of hydraulic fracture was unlikely to beoccurring at Ramsden dam and even if it wereoccurring, the fill immediately downstream ofthe core should act as a filter and preventfurther erosion. It would also seem unlikelyfrom the movement observations, the shearstrength parameters and the pore pressures thatslope instability of the downstream fill was acause of the movements at Ramsden.

23. The second two mechanisms are due tonormal behaviour of the dam, but it isimportant to be able to quantify the magnitudeof these effects. To determine whethersettlements measured many years after thecompletion of a dam can be attributed tosecondary consolidation of puddle clay core andcreep of the shoulder fill or erosion or slopestability, Charles (ref. 5) has proposed thefollowing settlement index (Si).

Si = s1000 H Iog(t2/ti)

where s is the crest settlement measured in mmbetween times ti and t2 since the completion ofthe embankment at a section of the dam H metreshigh. Where values of Si are greater than 0.02it is suggested that some mechanism other thancreep may be causing the settlement and thesituation should be seriously examined.Settlement measurements at Ramsden dam (ref. 5)give a Si of 0.077 between 1977 and 1985, and0.12 between 1983 and 1985. During 1988 whenthe reservoir was emptied, the Si increased to0.46.

24. At Ramsden dam the majority of thedeformations measured during 1988 and 1989 wereclearly due to reservoir fluctuations. Similarbehaviour has been observed at a number ofother embankment dams with puddle clay cores(ref. 6). A complete drawdown of Ramsdenreservoir such as occurred in 1988 has been arare event and therefore the relatively largenet crest settlement of 50mm that was measuredis unusual. Partial drawdown of the reservoirto 6m below TWL has been more common andtherefore a net crest settlement ofapproximately 8mm per 6m drawdown might beexpected. However, prior to the presentinvestigation, one particular cycle of drawdownto 6m below TWL and refilling in less than 2months in 1986 caused less than lmm of crestsettlement. The rate of drawdown on thisoccasion was much less than in 1989, butrefilling was much quicker. It would thereforeappear that the rate as well as the magnitudeof the drawdown influences the net settlement.

175

ENVIRONMENT AND RESEARCH

Principal mechanisms causing deformations dueto reservoir fluctuation

25. Upstream fill: The piezometric levels inthe upstream fill follow the reservoir levelwith little time delay. Therefore reservoirdrawdown causes a rapid increase in thevertical effective stress approximatelyequivalent to the decrease in reservoir head.The effective stress may be nearly doubled. Thesettlement of 30mm is equivalent to a modulusof approximately 100,000 kPa. This would bevery large for initial loading of such a fill,but would be reasonable for a reload situation.Resubmergence of the upstream fill causes adecrease in effective stress and more elasticrecovery of surface settlement than on thecrest.

26. Puddle core: Two mechanisms could accountfor the movement of the core. Firstly, loweringthe reservoir level reduces the lateralpressure on the upstream side of the core andleads to horizontal upstream movement.Initially this movement would be due toundrained deformation of the core. On refillingthe reservoir the horizontal pressure of thereservoir water against the upstream side ofthe core causes a horizontal downstreammovement of the core and probably some heave ofthe core. Horizontal downstream movement of thecore also allows some downstream movement ofthe upstream fill close to the core.

27. Secondly removal of the water from theupstream side of the core also changes thedrainage conditions and allows consolidationand time dependent settlement of the core. Mostof the settlement that occurred during Stage 2,when the reservoir was "empty11, will have beendue to the dissipation of pore water pressures.Refilling will have caused swelling of thecore.

SUMMARY AND CONCLUSIONS28. Reservoir drawdown caused significant

surface settlement and smaller upstreamhorizontal movement of the crest and theupstream slope. Impounding reversed themovements, resulting in a net settlement anddownstream horizontal movement at the crestwhen the reservoir was full. The magnitude ofthe net crest settlement was affectedcritically by the magnitude of the drawdown interms of height and time. Only 10% of crestsettlement due to complete drawdown wasrecovered on refilling, but 50% was recoveredfollowing refilling after drawdown to only 6mbelow TWL. There does not appear to be anyappreciable net surface movement of thedownstream fill following a cycle of drawdownand impounding. Drawdown caused settlement andchanges in pore water pressure throughout thedepth of the core.

29. Settlements occur in the upstream fillbecause reservoir drawdown increases theeffective stress. Settlement of the core occursbecause reservoir drawdown causes a reductionin lateral support on the upstream side of thecore and hence undrained deformation. Furthertime dependent deformation of the core occursas excess pore water pressures dissipate.

30. The magnitude of net crest settlementfollowing a cycle of drawdown and refillingdepends on the magnitude and rate of thedrawdown, and on the time the reservoir is leftin a drawndown state. Research is continuing toinvestigate the effects of reservoirfluctuations on embankment deformations atother dams.

ACKNOWLEDGEMENTS31. The work described in this paper forms

part of the research programme of the BuildingResearch Establishment and is published bypermission of the Chief Executive. The mainclient for the work is the Water Directorateof the Department of the Environment. Theinitiative of Mr J D Humphreys in instigatingthe work at Ramsden dam is appreciated. TheGeneral Manager of Yorkshire Water WesternDivision, Mr J.R. Layfield, permitted the useof the dam for research purposes and encouragedpublication. Mr I R Holton has made a valuablecontribution to the field work and the analysisof the results. Mr D Burford designed andinstalled the surveying system.

REFERENCES1. Tedd P. Claydon J.R. and Charles J.A.Detection and investigation of problems atGorpley and Ramsden Dams. Proc. of ReservoirsRenovation 88 Conf., Manchester, 1988, paper5.1,ppl-15. BNCOLD: London.2. Claydon J.R. Discussion to reference(5),Discussion Proceedings of BNCOLD-IWESConference Reservoirs. 1986, pp69-71.3. Hoyle N. Discussion. BNCOLD, Proc. Symp. ofInspection, operation and improvement ofexisting dams. Newcastle. 1975, D5/10.4. Charles J.A. and Watts, K.S. Themeasurement and significance of horizontalearth pressures in the puddle clay cores of oldearth dams. Proc. Instn. Civ. Engrs. 1987, Partl,vol. 82,February, ppl23-152.5. Charles J.A. The significance of problemsand remedial works at British earth dams.Technical Papers of BNCOLD-IWES Conf Reservoirs1986, ppl23-141.6. Robertshaw A.C. and Dyke T.N. The routinemonitoring of embankment behaviour. BNCOLD1990.

Crown copyright

176

27. The routine monitoring of embankment dam behaviour

A. C. ROBERTSHAW and T. N. DYKE, Yorkshire Water, Bradford, UK

The paper outlines the various requirements for the structural monitoring ofembankment dams within the Western Division of Yorkshire Water which isresponsible for the operation of over 70 such dams. It details the ways inwhich the work is routinely carried out with a particular emphasis on themonitoring of deformations. The methods adopted for the storage of theserecords is described in some detail and the paper concludes with a presentationof some of the results that have been observed.

INTRODUCTIONYorkshire Water - Western Division

1. The Western Division of Yorkshire Water isresponsible for the supply of water andtreatment of waste water for a population ofjust over one million in the MetropolitanDistricts of Bradford, Calderdale and Kirkleesand in part of North Yorkshire County Council.In order to supply water to this populationthere are 80 reservoirs within the Divisionwhich have a capacity greater than 25tcm and aretherefore covered by the provisions of theReservoirs Act 1975.

2. Over 70 of the reservoirs within theDivision are formed either wholly or in part byearth embankments with the remainder consistingof either concrete dams or covered concreteservice reservoirs. The average age of the damsis approximately 100 years with the oldest beingcompleted in 1827 and the newest in 1985. Thecapacities of the reservoirs vary from just over25tcm to approximately 22,000tcm in the case ofGrimwith reservoir and the highest embankment isScammonden Dam which, with a height of over 60m,is the second highest earth dam in the UnitedKingdom.

Reservoir Safety Management3. The reservoirs in Western Division are

operated by water supply staff in the Bradford,Calder and Skipton Operational Areas. ADivisionally-based Reservoir Safety section hasbeen in place since 1983 and currently consistsof a Reservoir Safety Manager, a ProiectEngineer, a Surveyor and a Technician. Thesection is responsible for the followingactivities:-

(a) Ensuring compliance with the provisionsof the Reservoirs Act 1975.

(b) Providing advice to the Operational Areason Reservoir Safety matters.

(c) Coordinating and progressing theimplementation of repair and improvementschemes at reservoirs.

(d) Monitoring the behaviour and structuralperformance of the dams within theDivision.

(e) Maintaining the statutory and otherassociated records for all the reservoirswithin the Division.

4. The remainder of the paper describes thevarious methods that have been adopted withinthe Division to monitor the structuralperformance of the dams and to keep records oftheir behaviour. The survey methods that aredescribed should not be regarded as the onlymethods that are available but they have beenfound to provide acceptable results over a largenumber of years.

EMBANKMENT MONITORING REQUIREMENTS5. The extent of embankment monitoring in

Western Division has primarily been governed byrecommendations made by Inspecting Engineers intheir statutory inspection reports under theReservoirs Act 1975. The introduction ofSupervising Engineers in the mid-1980fs alsoresulted in an increase in monitoring work withrecommendations for new monitoring systems andrequests for the reading of established systemsat increased frequencies.

6. The monitoring of embankment behaviour orperformance can essentially be considered tofall.under three main headings:-

(a) Deformation monitoring(b) Pore^water pressure monitoring(c) Leakage/drainage flow monitoring

7. In the case of deformation monitoring,prior to 1970 recommendations from InspectingEngineers do not appear to have been specificenough as to necessitate the setting up ofpermanent monitoring systems and the only workthat was undertaken was to level along the crestof a dam to establish its freeboard. This wasobviously not a suitable arrangement for the

The embankment dam. Thomas Telford, London, 1991 177

RISK, HAZARD AND SAFETY

on-going monitoring of behaviour as no permanentstations were involved. Similarly, the earlymethods for monitoring groundwater levelsinvolved the use of "observation wells" whichwere slow to react and unsuitable for detailedmonitoring. Flow monitoring is generally theonly measurement procedure which provideslong-term records against which currentbehaviour can be assessed.

8. The mid 1970fs witnessed a new emphasis onembankment monitoring with more detailedrecommendations from Inspecting Engineersrequiring the setting up of permanent levellingand/or alignment stations on dams. This newapproach also coincided with a dramatic growthin site investigation work and since then therehas been a continuing increase in the number ofpiezometers installed throughout the Division.

9. In addition to the requests fromInspecting and Supervising Engineers, WesternDivision has also adopted a policy wherebyadditional measurements are taken at reservoirswhich are undergoing substantial drawdown andrefilling operations. These additionalmeasurements primarily relate to the monitoringof deformations but the frequency of otherreadings may also be increased. It is also apolicy of the Division to annually monitor thevertical movement of all earth embankment damswhether or not this is the subject of arecommendation from an Inspecting Engineeralthough permanent stations have not yet beenestablished at all dams.

10. The frequency of monitoring variesaccording to the requirements of a particulardam, however, broadly speaking the most commonfrequencies can be summarised as annually formovement monitoring, monthly for the reading ofpiezometers and weekly for drainage/leakageflows. There are obviously a considerable numberof exceptions to these frequencies with, forexample, movement monitoring reducing to monthlyin some cases and leakage monitoring increasingto three monthly. In some cases it isrecommended that monitoring be carried out withthe reservoir at its highest and lowest waterlevels during the year although in practice thisis often difficult to achieve.

EMBANKMENT MONITORING SYSTEMS11. Table 1 sets out the full extent of

monitoring work which is undertaken in WesternDivision at all its 80 reservoirs andillustrates the considerable workload that isinvolved. The monitoring of piezometric levelsand leakage/drainage flows generally use thetraditional measuring systems that are inwidespread use. Since such basic systems asV-notch weirs, the timed filling of measuringcontainers and the details of standpipe,hydraulic and pneumatic piezometers will befamiliar and commonplace to most readers theremainder of the section will concentrate ondeformation monitoring and will describe thesystems adopted in Western Division to establishpatterns of vertical and horizontaldisplacements.

MovementVerticalHorizontal

Piezometers

Flows

Table 1.

Deformation

Earth

669

18

39

Dam Performancein Western

Monitoring

No of DamsConcrete

11

1

3

Monitoring SystemsDivision

12. As can be seen from Table 1, the majorityof embankment dams in Western Division havesystems for the monitoring of verticaldeformations, with most of the levellingstations being positioned along dam crests,usually in or above the core material. However,individual circumstances sometimes dictate thesiting of additional levelling stations andthere are several sites within Western Divisionwhere such monitoring stations can be found onthe upstream and downstream sides of the crests,on wave walls and on the upstream and downstreamshoulders.

13. In the case of horizontal movementmonitoring, the most common system used is againto install stations along the crests of damswith the upstream-downstream component of themovements being treated as the most important tobe monitored. Should the need arise to monitorother parts of an embankment such as theupstream and downstream shoulders, systems canbe readily designed and adapted to cater forsuch needs.

Surveying Techniques - Levelling14. Vertical deformations are invariably

monitored by precise levelling techniques usinga Wild NA2 Automatic Level with parallel platemicrometer attachment which enables readings tobe taken to the nearest 0.lmm. Sightings aretaken to a 4 x 1 metre sectional aluminium-alloystaff with centimetre graduations. The use ofan invar staff has been considered but has notbeen adopted as it is thought to be impracticaland unnecessary, mainly due to the adverseweather conditions which are prevalent at manyof the sites within Western Division but alsobecause acceptable results can be obtained usingthe more traditional staff.

15. Crest-levelling stations are spaced at nogreater than 30 metre intervals and at least onebench mark is established on stable ground awayfrom the dam. A point on the overflow crest isalso used as an additional datum whereverpossible*

16. A closed loop system of levelling hasbeen adopted with each station being levelled onthe outward and return legs, thus providing acheck on the readings. The level is set upadjacent to every other monitoring station witha short intermediate sight being taken to thisstation. Backsight and foresight distances arekept the same to eliminate collimation errorbetween the change points. No adjustment ismade to the readings to distribute any closing

178

PAPER 27: ROBERTSHAWAND DYKE

error and the calculated levels fromobservations on the outward leg are used if theclosing error is within acceptable limits andits pattern of accumulation is satisfactory.

17. In addition to traditional levellingtechniques, trigonometrical heighting is alsoundertaken at some sites incorporating precisetheodolite observations and distancemeasurements. This method usually forms part ofa larger system whereby three-dimensionalcoordinates for stations on a dam areestablished from remote instrument stations.This is discussed in more detail in thefollowing section.

Surveying Techniques - Alignment18. Whilst the monitoring of vertical

deformations on old embankment dams has becomestandard practice since the late 1970fs, thesame prominence has generally not been given tothe checking of horizontal deformations. In thecase of new embankment dams, such monitoringsystems are established as a matter of course inorder to monitor their initial behaviour.However, the setting up of similar systems onolder dams has only been undertaken as a resultof areas of instability becoming evident in theform of increased settlement, cracking, slumpingetc.

19. Observations are taken to alignmentmonitoring stations using a Wild T2000Electronic Theodolite with Wild DI4Electromagnetic Distance Measuring attachment.Raw field data is transferred directly to aHusky Hunter data-logger which is equipped withOptimal survey software to present the resultsin coordinate form as and when required.

20* Method 1: The basic method used for themonitoring of horizontal movement within WesternDivision is to construct a control pillar onstable ground in line with the long axis of thedam crest. The theodolite and EDM are set up onthis pillar and observations are first taken toreference points around the site and then to themonitoring stations on the dam. The smallangular changes to the monitoring stations withtime represent upstream or downstream movementwhich, along with a horizontal distancemeasurement, can be converted to an actualmovement in millimetres. The limitations ofthis method of monitoring include:

(a) It is difficult to incorporate on longdams, curved dams and dams with more thanone embankment.

(b) Whilst it provides excellent results formonitoring stations on the crest andwave-wall, this system becomes lessaccurate for stations on the upstream anddownstream shoulders.

(c) It only detects horizontal movement inthe upstream-downstream direction.

21. Method 2: An alternative method that isused is the network approach whereby a number ofsurvey pillars are set up on stable groundaround the site and coordinated to provide localcontrol. Bearings, vertical angles and slopedistances are then taken from an appropriatepillar to the monitoring stations on the dam

from which X, Y and Z coordinates for eachstation can be calculated. Positioning of theinstrument stations is of vital importance andwherever possible steep sights are avoided asslight errors in the measurement of the verticalangle or dislevelment of the instrument willresult in exaggerated inaccuracies in thecalculated coordinates of the monitoringstations. This method is perhaps a moreflexible system of monitoring and is favoured toMethod 1 where dams have monitoring points onboth crest and upsteam and downstream shoulders.Movement is also detected in all directions andtherefore a more comprehensive analysis can bemade of the deformation characteristics of thedam. However, the accuracy suffers slightlywith this system, especially in the verticalplane, due mainly to the more indirect way ofarriving at such a value as compared to spiritlevelling. This method also tends to be muchmore time-consuming which is especiallyimportant with such a large number of dams to bemonitored.

Monitoring Stations22. The first crest levelling stations in

Western Division were set up in the late 1970fsand consisted of a length of reinforcing bar setin approximately 0.3 cu m of concrete with thetop of the bar positioned at or just aboveground level. Such stations have provedunsuitable for long-term monitoring due to theirsusceptibility to damage and disturbance by

Extension bar andsighting target

Stop tap box and supportn mortar surround

Threaded top to fitextension bar

_Ughtly compactedill material

150mm 0 flexibledrainage pipe

25mm 0 stainless

4 No. mild steel vanes

p 300mm *\

Fig 1. Embankment Crest: Monitoring Station

179

RISK, HAZARD AND SAFETY

vandalism, heavy plant, grass cutting equipment,frost heave and corrosion. A programme ofreplacing old stations with a new installationdesign has therefore been implemented (see fig1). The location of the monitoring stationbelow ground level and the introduction of aflexible sleeve provides adequate protectionagainst damage.

23. A facility for monitoring horizontalalignment has also been incorporated into thedesign whereby an extension bar with a surveytarget can be attached to the undergroundlevelling pin. This brings the installationabove ground level and therefore able to beobserved from an instrument station off the dam.This new design of monitoring station is beingrestricted to use on embankment crests only asthey are relatively easy to locate and lines ofsight are not likely to be obscured by excessivevegetation cover. When monitoring downstreamshoulders the use of conventional pillar-typemonuments is preferred which can be set atvarying heights above ground level thus allowingunobscured lines of sight and ease of location.

Surveying Errors associated with monitoring work24. Apart from mis-readings by the Surveyor,

the other errors associated with the preciselevelling of monitoring stations are generallyeither weather related or instrument related.

25. Windy and sunny conditions both result ina marked drop in the accuracy of readings. Themajority of reservoirs in Western Division are^ited above 250m AOD and in very exposedlocations so surveying in windy conditionsbecomes virtually unavoidable if monitoringrequirements are to be met. Also irregularrefraction on hot sunny days makes reading ofthe staff extremely difficult with a resultingreduction in accuracy. This is especiallyapparent in conditions of intermittent sun wherefluctuations in the same staff reading of up to0.3mm occur when lighting conditions changeduring sighting.

26. The main instrumental error is associatedwith the line of collimation of the level beingout of adjustment and therefore not projecting atruly horizontal line of sight. Errorsresulting from this factor can be kept to aminimum by equalising the backsight andforesight distances between change points andusing the same approximate instrument positionsat each visit. In order to achieve the maximumcontinuity between readings, adjustment of thelevel is avoided unless the errors becomeunacceptable. In practice, criticalexaminations of the levelling errors have neverindicated a requirement for adjustment.

27. The errors mentioned previously have beenfound to have a cumulative effect on theaccuracy of the levelling run resulting in aclose relationship between the length of surveyand the closing errors arising. Thisrelationship is shown in Table 2, theinformation being derived from precise levellingwork undertaken on dams in Western Division overseveral years.

Crest Length (m) Expected Closing Error (mm)

100200300400500600

Table 2. ClosingWork

-t- 1.0+ 1.5+ 2.0+ 2.5+ 3.0+ 3.5

Errors during Monitoringin Western Division

28. Table 2 assumes a closed loop oflevelling along the dam crest and back. Theexpected closing error indicates the error whichcan be expected to accumulate using theautomatic level with parallel plate micrometerand aluminium staff in calm, overcastconditions. It must be stressed that thesevalues for expected closing errors can only beused as a rough guideline and that other factorsmust be taken into account when assessing thedegree of reliability of a particular set ofreadings.

29. In alignment monitoring the weatherconditions are again a major factor influencingaccuracy, with sunny conditions having a markeddetrimental effect on angular readings, anddamp, misty conditions causing great problemswhen attempting precise distance measurementsusing EDM. Obviously in these extreme weatherconditions it would be inadvisable to attempt tocarry out such precise surveying work, however,if one was to be entirely governed by theever-changing nature of British weather themonitoring work would become virtuallyimpossible. Therefore, as ideal weatherconditions cannot be guaranteed, a compromisehas to be reached whereby as many sightings aspossible are taken to each monitoring stationfrom which mean values are extracted.

30. As with levelling, instrumental error canagain be a major problem when monitoringhorizontal alignment and regular checks areundertaken to check whether instruments are incorrect adjustment or are in need of repair.

RECORD KEEPINGGeneral Introduction

31. There are generally three types ofrecords which must be maintained for the propermanagement of large reservoirs and these can besummarised as follows:-

(a) Statutory.(b) Monitoring.(c) Operational.

32. For the owner of a large number ofreservoirs there is a requirement to keep largeamounts of information which is of the samebasic format and, as such, is amenable to theuse of computer-based storage systems for theefficient handling of the data. With this basicpremise in mind a decision was taken in 1983 toimplement the use of such systems and a startwas made on the first of the monitoring systemswhich was for the storage of settlement records.The original proposal was for the storage of

180

PAPER 27: ROBERTSHAW AND DYKE

monitoring and operational data only but thephased introduction of the Reservoirs Act 197Sfrom April 1985 onwards provided an opportunityto extend the scope of the systems to provide acomprehensive Reservoir Safety Records Systemcovering the three items listed above,

33. The original decision to computerise thestorage of reservoir-based data coincided withthe introduction of a "Distributed DataProcessing" approach to the provision ofcomputer services across the whole of YorkshireWater, This was a proactive approach wherebystaff were actively encouraged to develop andimplement computer-based solutions to theirproblems and was backed up by a heavyexpenditure on the introduction of new hardwareand support facilities. The initial investmentconsisted of the purchase of five DEC VAX 750mini-computers to supplement the existing ICLmain-frame together with a considerableexpansion of the hardware and software supportarrangements including the introduction of alarge training programme. The VAX computerswere located at 5 sites within Yorkshire Waterand interconnected by a communications networkwith terminals connected to each VAX by a WideArea Network. The basic configuration has beenconsiderably uprated since 1983 with theintroduction of additional and more powerfulVAXfs which can be clustered together to work asa more efficient unit.

34. The individual components of theReservoir Safety Records Systems were completedby 1986 and have remained essentially unalteredsince then apart from minor modifications toimprove their efficiency and the introduction ofarrangements whereby other persons outside theReservoir Safety section can have access to thedata.

Reservoir Safety Records System35. Access to the VAX network is by a series

of hierarchical Users which are each allocatedan amount of disc-storage space. Each User isprotected by a password and has access to thefollowing software facilities :-

(a) Database Management (DATATRIEVE)(b) Spreadsheet (DECALC)(c) Graphics (DECGRAPH)(d) Office Management (Word Processing,

Electronic Mail).

A major advantage of the multi-user approach isthat it is possible, by the establishment ofappropriate path names within the software, forUsers to have shared access to data records.The access capabilities of the various Users canbe controlled by the granting of appropriateprivileges to give Read, Write or Modify accessto each individual User depending on theirrequirements•

36. The Reservoir Safety Records are held inDATATRIEVE databases which are located in twointerlinked Users named RESSURV and RESSAFE andthe various links with other Users are as shownin Figure 7.

HEAD OFFICE(Read)

RESSAFE

SUPSTAT RESSURV(Read, Write)

- RESSKIP(Read, Write)

RESSUP(Read)

Fig 2. User Access to Reservoir Safety Records

This figure illustrates the use of privileges torestrict the type of access granted to eachUser. The SUPSTAT User is controlled by theDivisional Water Supply Statistics section andis the means by which reservoir levels aretransferred into Part 1 of the prescribed formof record at weekly intervals after having firstbeen used to calculate reservoir stocks. TheRESSUP User is available to all DivisionalSupervising Engineers for read-only access toall the statutory and monitoring records. TheRESSKIP User is controlled by the Skipton Areaoffice which is remote from the DivisionalHeadquarters and is used for the input of rawmonitoring data and the output of calculatedresults (drainage flows, piezometric levels,reservoir levels, etc). The HEAD OFFICE User islocated on a different VAX but network linkshavebeen established to enable Regional staff tohave access to data from all Divisions relatingto the appointment of Panel Engineers and theirprogress with Inspections.

37. All components of the Reservoir SafetyRecords System are accessed by a series ofmenus. The three main sub-divisions of thesystem are as follows:-

(a) Head Office Statistics.(b) Prescribed forms of record.(c) Dam Performance Monitoring System.

The Head Office statistics have been brieflydescribed above and the next two sections willdescribe the other components of the system.

Prescribed forms of record.38. The relevant information for each Part of

the prescribed form is held in database form andcan be accessed from menus for storage,deletion, modification and reporting purposes.A reporting procedure has been established foreach Part so that the printout resembles, asclosely as possible, the form that is set out inthe Regulations. Hard-copy printouts are keptin individual files for all 80 reservoirs andare continually updated as changes occur withall superseded information being retained sothat the reason for the change can be deduced ata later date. Copies of the appropriatemonitoring information for each reservoir isalso contained in the file and is periodicallyupdated with information from the DamPerformance Monitoring System.

181

RISK, HAZARD AND SAFETY

39. There are a number of uses to which theinformation contained in the prescribed form canbe put. As mentioned previously, weekly recordsof water level are automatically transferredinto Part 1 of the system from an associatedsystem and these levels are used to assist inthe interpretation of the structural monitoringdata. Part 10 of the prescribed form containsdetails of all recommendations made byInspecting Engineers in the interests of safety.The relevant database has been expanded toinclude all other recommendations which may havebeen made by either Inspecting or SupervisingEngineers and also includes details of any workarising from valve tests. Reports can beproduced of only those items to which Part 10refers or alternatively they can be produced of

all recommendations, complete with target andcompletion dates, which can be used byOperations for maintenance programming purposes.In addition to the two specific applicationsreferred to above, the information in therecords can also be used for the production ofad-hoc reports and even for distribution lists.

Dam Performance Monitoring System (DPMS)40. Records are not incorporated into DPMS

without some consideration being given to thecapabilities of existing systems and, forexample, the manual system for recording flowsin the Calder Area has been retained as it isconsidered to give a satisfactory presentationof the results. Despite this, however, the vastmajority of the monitoring information from damswithin Western Division is now recorded in DPMS.

41. The information contained within DPMS isheld in a number of databases covering thefollowing subject areas:-

(a) Movement.(b) Piezometric Level.(c) Drainage/Leakage Flows.

Each database is accessed by a menu for storage,reporting and plotting purposes althoughdeletion and modification are only available bydirectly entering the database managementsystem. All the databases include records ofreservoir level for the appropriate date and thepiezometer and flow records can also includerainfall if required.

42. The movement system has been set up sothat surveying information in both thehorizontal and vertical planes can be stored.The information is held in absolute form so thatrelative values can be calculated from any givenstarting date, eg. levels are stored in metresabove Ordnance Datum and routines have beenwritten to calculate settlement values from agiven date for plotting purposes.

43. The piezometric information is entered inthe form of either depth readings to the watersurface in the case of standpipe piezometers orpre-calculated pressures. Tables of top ofstandpipe levels or appropriate datum levels areheld within the computer from which piezometriclevels can be calculated for either standpipe,hydraulic or pneumatic piezometers. Tip levelsare also held in tabular form so that the headabove the tip can be calculated for reporting

purposes.44. The flow system allows for the entry of

readings in either time taken to fill a givencontainer, head measured over a weir or directentry of previously calculated flows.

Head-discharge tables for various weir types areheld within the computer so that flows can beautomatically calculated and whichever method ofentry is used, the flows are always presented inlitres per minute for reporting and plottingpurposes.

45. With such a large number of reservoirs itcan be appreciated that a correspondingly largeamount of data has been, and is constantlybeing, entered into the various systems. In thecase of the settlement system alone there arenow over 10,000 records held in the computer andthis figure is considerably exceeded by thenumber of reservoir levels contained in Part 1of the prescribed form. The capabilities of theDATATRIEVE system are such, however, that evenwith such large quantities of data the retrievaltimes are still perfectly adequate at thepresent time although arrangements are beingconsidered for the archiving of data on magnetictape so that it can be stored off-line shouldthis ever be required.

ANALYSIS OF MONITORING DATAProduction of Results

46. Results are initially analysed by theproduction of reports and a number of standardreports can be selected from the appropriatemenus of DPMS. The reports are designed suchthat all the data for the chosen monitoringpoint, ie. survey station, piezometer or drain,between selected dates, is printed in date orderenabling a rapid appraisal of the records withrespect to time, reservoir level and/or rainfallto be carried out. If further investigation isfelt to be justified then a number of standardgraphs can be generated or the data can betransferred to the DECALC spreadsheet packagefor further manipulation prior to plotting.

47. Although any graphics terminal can beused to produce the plots, a more versatileworkstation has been established within theReservoir Safety Management Section and therelevant details are shown in Figure 3.

DEC VAX

DEC LA 210Line Printer

IDEC VT 340

Colour GraphicsTerminal

HP7550AColourPlotter

3* Diagrammatic Representation ofDPMS Workstation

182

PAPER 27: ROBERTSHAW AND DYKE

S. «-20-

I -30.

J -40• -50-

-P -60

; -70.

(0 go

UIDDOP - SETTLEMENT

STATION 06 14-Dec-i988 to 30-Apr-1990

• — • Res Level (w)0 — 0 Settlement (MM)

^ ^ -- ^ B ^

r f. 4--''""

100 200 300 400 500 0

Time in Days

&0

Fig 4. Crest settlement at Widdop Dam

Selection routines have been established withinDPMS to choose the data to be plotted and theactual plotting is carried out by either theline printer using the DECGRAPH package or bythe plotter using a more complex package calledSIMPLEPLOT. The main difference between the twopackages is that the DECGRAPH system has beenset up to plot only one point plus reservoirlevel whereas the SIMPLEPLOT system can plot upto nineteen points plus reservoir level to ahigher standard of presentation in up to eightcolours with a selection of line-types. Figure4 shows a graphical representation of thesettlement of a survey station on the embankmentcrest at Widdop reservoir and was produced froma menu in a matter of seconds by the DECGRAPHsystem.

Analysis of movement data48. Figure 4 has been specifically included

as it illustrates a number of points which havebeen observed at Widdop and at other dams of asimilar construction. The dam at Widdopconsists of an earth embankment over 20m inheight with a relatively wide crest and apuddle-clay core which extends into a shallowpuddle clay filled trench. The graph refers toStation 06 which is located at the upstream sideof the crest in the middle of the dam and showsthe settlement of the point as the reservoir wasemptied and then refilled between the beginningof 1989 and the beginning of 1990. It can beseen that the drawdown caused a sharp increasein the rate of settlement which was of the orderof 10 times the previous long-term rate andamounted to approximately 40mm from thereservoir full to the reservoir empty condition.Whilst the reservoir was empty the settlementcontinued but at a reduced rate such that by thetime the reservoir was ready to be refilled afurther 20mm of settlement had occurred.Initial refilling of the reservoir did notresult in any discernible change in thesettlement rate but a rise of approximately 8mmdid occur as the reservoir approached the fullcondition. The whole full-empty-full cycleresulted in a net settlement of just over 50mra.

49. Analysis of the associated horizontaldata shows that the crest of the dam asrepresented by the wave-wall moved upstream byapproximately 40mm during drawdown and thenmoved downstream by approximately 20mm onrefilling. Observations to the wave-wall in

this case may have produced a false impressionfor the movement of the dam as a whole asexperience on other dams has indicated that, ingeneral, dams are left with a net downstreammovement as a result of the reservoir beingdrawn down and refilled.

50. Movement observations of the large numberof dams within Western Division over the last 13years have led to the formation of the followinggeneral conclusions:-

(a) The movement of a dam is directly relatedto the height of fill. Any departure fromthis relationship is a clear indicationof either survey errors or thedevelopment of a problem.

(b) Drawing a reservoir level down is likelyto produce a rapid increase in the rateof settlement accompanied by an upstreammovement. Completely emptying a reservoiris likely to produce disproportionatelylarge movements.

(c) Maintaining a reservoir in an emptycondition eventually causes thesettlement to return to the long termreservoir full settlement rate.

(d) Refilling a reservoir does not initiallyalter the rate of settlement buteventually heave does occur although thisdoes not usually fully compensate for theinitial settlement. Generally the damalso moves back downstream more than itmoved upstream during emptying of thereservoir.

(e) It has generally been found thatrelatively short dams in steep-sidedvalleys move more than longer dams of asimilar height.

51. The basic patterns of movement describedabove enable a fairly accurate prediction of thethree-dimensional movement of a dam caused byreservoir drawdown to be made so that anydepartures from the pattern can be criticallyexamined. Tedd, Claydon and Charles (ref. 1)examine this matter further in their analysis ofRamsden dam which is also located in WesternDivision.

REFERENCES

1. TEDD P, CLAYDON J.R. and CHARLES J.A.Deformation of Ramsden Dam during reservoirdrawdown and refilling. BNCOLD 1990.

183

28. Embankment dam behaviour: the contribution of geo-chemistry

A. MACDONALD and J. M. REID, Babtie Shaw and Morton, Glasgow, UK

Geochemical degradation can have a significant effect on the geotechnical properties ofembankment fill materials and on the quality of drainage waters. The most susceptible materialsinclude mudstones, weathered igneous and metamorphic rocks, and rock with a high proportion ofsecondary minerals. Geochemical information should be collected at the site investigation stage,including from trial embankments. The design should allow for any geochemical degradation likelyto occur and contractors should be made aware of the problems. Chemical monitoring should becarried out during and after construction to check the design assumptions.

INTRODUCTION1. When material is excavated and placed

as fill in an embankment, its properties willbe liable to change from those in its naturalstate. This is due in part to the mechanicalbreakdown of the material during theearthworks operations, but also to chemicalreactions resulting from the change ofenvironment. The greater the contrastbetween the conditions under which thematerial formed and those pertaining in theembankment, the greater is the potentialinstability of the minerals of which thematerial is composed. Thus bedrock is morepotentially susceptible to geochemicaldegradation than soil, in particular igneousand metamorphic rocks which formed underconditions of high temperature and pressurevery different from those pertaining at theearthfs surface.

2. In practice, however, thesusceptibility of a material to geochemicaldegradation is dependent on a number ofkinetic factors rather than on its mineralcomposition alone; fresh igneous andmetamorphic rocks form some of the strongestand most durable aggregates available. Theinitial weathering grade of the material isof vital importance to both its physical andchemical stability, as are the grain size,presence of secondary minerals and degree ofleaching to which the material is subjectedboth during placement and in the longer termdam environment.3. The subject of geochemical degradation

or weathering of rock as a factor in thedesign of engineering structures such asembankment dams has been attracting attentionin recent years. The Department of theEnvironment recently commissioned a researchcontract on chemical deterioration of fillmaterial in earth dams in the United Kingdom(ref.l). A number of papers have beenpublished over the years by Taylor, Spearsand co-workers on the breakdown ofCarboniferous Coal Measures mudrocks in

colliery spoil tips and tailings dams, eg(refs 2,3) The subject of rock weathering inengineering time has been comprehensivelydiscussed by Fookes et al (ref.4), whoproposed Rock Durability Indices to measurethe susceptibility of materials in variousengineering situations.

4. The processes which can causegeochemical degradation of fill materials arebasically those involved in the naturalweathering of rocks and soils, namelysolution and precipitation, oxidation andreduction, hydration, hydrolysis, dispersionand cation exchange. These processes aredescribed in more detail in (ref.l) and instandard geochemical and geological texts, eg(refs 5,6). The effect of excavation andplacement in an aerated, active leachingenvironment such as an embankment can be togreatly accelerate these processes comparedto the natural state.

5. Because of the large volumes of fillinvolved in dam embankments, even a smallpercentage of chemically active material canhave a major effect on the quality ofdrainage water or the generation of hazardousgases. If the proportion of chemicallyactive material is larger, there may besignificant effects on the geotechnicalproperties of the fill, such as density,shear strength, compressibility, grading andpermeability. These effects need to beinvestigated and quantified at the designstage if the embankment is to functionsatisfactorily throughout its design life.6. Materials susceptible to geochemical

degradation can vary widely. Fookes et al(ref.4) found that in-service deteriorationoccurred mainly in basic igneous materialsand that the cause of degradation was eitherthe presence of secondary minerals or activein-situ weathering. The study carried outfor the Department of the Environment (ref.l)found that a number of rock types could besusceptible to deterioration in dams in theUnited Kingdom : these are; (1) calcareous

The embankment dam. Thomas Telford, London, 1991 185

ENVIRONMENT AND RESEARCH

materials in contact with acidic water; (2)clays or raudstones with a high proportion ofexpandable clay minerals; (3) clays ormudstones with a high exchangeable sodiumpercentage; (4) clays and mudstonescontaining pyrite; and (5) highly weatheredrock, particularly of igneous and metamorphicorigin. Apparently similar materials canbehave very differently in differentsituations, depending on slight variations inmineralogy and weathering state. This isillustrated later by a comparison of thebehaviour of Carboniferous mudstones atRoadford and Carsington dams.

7. Geochemical degradation is likely to bemost intense during the initial stage of damconstruction, when the material is newlybroken up and exposed to an oxidising andleaching environment. Once the embankment iscompleted and the reservoir impounded therate of reaction will decrease. If thematerial is below the water table the supplyof oxygen will be greatly reduced and therate of removal of weathering products willbe limited by the rate of seepage through theembankment. This will generally be veryslow. Further reactions may occur betweenthe seepage water and other materials in theembankment, however, and it may take manyyears for all the reaction products to beleached out of the embankment and into thedrainage system. Above the water table,reactions may be limited by the reduced rateof percolation of surface water once thesurface of the dam is sealed by topsoil anddrainage installed.

8. In a loosely tipped 50 year oldcolliery spoil tip, Spears et al (ref.7) showedthat the maximum depth of chemical weatheringwas 3.8lm below the crest, while the surfacezone of intense weathering only extended to adepth of lm. Excavations in the downstreamside of Burnhope dam, County Durham (ref.8)showed that marks formed by the bucket teethof the excavator were still visible in themudstone fill after 25 years, indicatingminimal deterioration of the material. AtCarsington dam, the highly weathered mudstonefill was found to have undergone furtherchange to a depth of 0.3m between 1984 and1987. The fill was left fully exposed to theatmosphere during this time, and the surfacelayer had degraded to a soft dark grey clay.This material formed an impermeable layerprotecting the underlying fill from furtherphysical and chemical deterioration.9. From the foregoing, it can be seen that

the geochemical behaviour of fill materialsshould be investigated and taken into accountin embankment dam behaviour. In thefollowing sections guidance is given as to

the actions required at various stages in thedesign and construction process, illustratedby examples from the authors' experience andother examples. It must be emphasised thatevery site is unique and that strategies forevaluating and dealing with geochemicaldegradation have to be developed to suit theparticular circumstances of each site.Awareness of the potential problems is thekey to success.

DESK STUDY10. At the desk study stage information

should be gathered on the mineralogy, degreeof weathering and geochemical behaviour ofany potential fill materials and of the damfoundation. In the U.K., published data maybe available in memoirs and reports of theBritish Geological Survey, who should beapproached for advice. Data may be availablefrom articles in learned journals and siteinvestigation reports. Inquiries should bemade regarding any problems encountered withsimilar materials in existing embankments orother works.

11. The chemistry of the waters to bestored in the reservoir should be determinedand compared with the mineralogy of thepotential fill materials. Natural waters inthe U.K. are generally not aggressivechemically, but acidic waters resulting frompeat in upland areas or pollution in lowlandareas can cause dissolution of carbonates infill material (ref.l). Reaction betweenstored effluent and fill material is asignificant factor in the design of tailingsdams (ref.9).

12. The aim at the desk study stage is toidentify potentially susceptible materialsand prepare a programme of sampling andtesting to be carried out at the siteinvestigation stage.

SITE INVESTIGATION13. The site investigation is the stage

at which a comprehensive databank of thechemistry and mineralogy of all potentialfill and foundation materials, groundwater,surface water and gases should be obtained.This should enable prediction of the likelygeochemical reactions to be made, which canthen be verified by monitoring fill, drainagewater and gas composition during constructionand operation. It is much harder to assessproblems which arise during construction iflittle or no pre-construction chemical ormineralogical data is available.

14. The nature and scope of the testingwill depend on the materials involved. Anessential requirement is high qualityengineering geological logging of all

186

PAPER 28: MACDONALD AND REID

exposures, trial pits, cores and soilsamples. Correct assessment of weatheringgrade and identification of secondaryminerals are particularly important.Geochemical problems are very often due tosecondary minerals such as calcite, pyrite,gypsum, jarosite, zeolites and clay minerals.Methods of rotary coring should be specifiedto ensure that these minerals are notdestroyed during drilling. The use of tripletube core barrels and foam or mud flush maybe required.

15. The sample description should be backedup by an appropriate programme of chemicaland petrographic analyses. For igneous rockssuch as basalt, this may involve making thinsections and examining them in a polarisingmicroscope as well as carrying out X-ray andchemical tests. Fookes et al (ref.4) give areview of proposed petrographic indices forassessment of unsound materials. For claysand mudstones X-ray diffraction tests todetermine the clay mineralogy are essential.Standard chemical tests such as pH, sulphateand carbonate should always be carried out insufficient numbers to determine the chemicalcharacteristics of the materials. Sulphideor total sulphur tests should also be carriedout. Details of test procedures are given inBS1377 (ref.10), BRE Digest CP2/79 (ref.ll),

Head (ref.12), BS1O47 (ref.13) and Taylor andSpears (ref 14).

16. The chemical composition of naturalmaterials is often highly variable, becauseof the irregular distribution of secondaryminerals. Sufficient tests should thereforebe carried out to assess this naturalvariability. One method of achieving this isto carry out intensive chemical testing in asmall number of trial pits or boreholes,combined with more general testing at a muchlower frequency. This enables a geochemicalprofile of the material to be built up andzones of active and inert material to beidentifed. This approach was adopted for thesite investigation for the redesign ofCarsington dam in 1987. The proposed fillmaterials are highly weathered mudstones ofCarboniferous age and were thought to containhigh concentrations of sulphates andsulphides. Testing was carried for pH, totaland water soluble sulphate, sulphide,carbonate, and organic matter. A profilefrom one of the trial pits is shown on Fig.1. Sulphide and carbonate were found to bevery low throughout the weathered strata, butthe sulphate content was high and the pHacidic. The sulphate was in the form ofjarosite, visible as a yellow-green powder onbedding planes and joint surfaces. Only the

Organic Content (%) pH Total S 0 3 (Vo) Carbonate C 0 2 (%)

G.L .0-25.

1-20

V1 . _

5-0

Topsoil

Head

CompletelyWeatheredMudstone

0 10 2 0 3 0 4-0 2 3 4 5 6 0 0-5 1-0 1-5 20 2-5 30 0 0-2 0U 0 6 0-8 10t I l I i i i i i i I i i i i i i i i i i i i

Siltstone

HighlyWeatheredMudstone

w ^w0W 0

0 W - Water Sample

Fig.l: Geochemical Profile of Trial Pit, Carsington

187

ENVIRONMENT AND RESEARCH

surface layers of head and residual mudstonewere chemically inert, the jarosite havingbeen further oxidised to limonite andgoethite.

17. Physical and mechanical tests can alsobe used to assess the susceptibility of fillmaterials to degradation. Fookes et al(ref.4) proposed rock durability indicesbased on point load strength index, magnesiumsulphate soundness test, water absorption,specific gravity and modified aggregateimpact value. These indices are of greatvalue for assessing the durability of rockfill in engineering structures. However theydo not give any information on chemicalproperties of the fill and should not be usedas a substitute for chemical analyses.

18. Most site investigations for damsinvolve the construction of trialembankments. These provide excellent modelsfor the final structure. The chemistry ofthe fill materials and, particularly, theseepage waters from the embankment should beclosely monitored. As trial embankments aregenerally left unsealed and free standing,they present an environment which is veryfavourable to accelerated weatheringreactions. If geochemical degradation isgoing to be a problem, a trial embankmentshould give some indication of it. AtRoadford dam, trial embankments wereconstructed from local fresh to moderatelyweathered mudstone and sandstone ofCarboniferous age. Seepages from theembankments were found to be acidic with highconcentrations of sulphate, iron andmanganese. Chemical and mineralogicalanalyses of the embankment fill indicatedthat some oxidation of pyrite and formationof jarosite had taken place. Similarseepages were observed from the trialembankment at Carsington.

19. Simulated weathering tests can becarried out on fill materials in thelaboratory in an attempt to obtain morequantitative data on anticipated rates ofweathering and their effect on thegeotechnical properties of the material.These tend to be expensive and are ofteninconclusive. One of the main difficultieswith such studies is the difference inphysical and chemical mechanisms operating ina small, closed system in the laboratorycompared to those operating in a damembankment. In-depth studies are bestcarried out as research projects over aperiod of at least a year and requireappropriate resourcing. Shorter, semi-quantitative tests may give useful guidanceon the likely behaviour of the materials.Some examples are given in paragraphs

20 and 21.20. At Colliford dam in Cornwall, the

embankment was constructed with sand wastefrom china clay workings. This materialcontained about 12% of kaolinised feldspar,and there was concern that this might degradeto clay during the lifetime of the dam.Accelerated weathering tests were carried outby passing hot water continuously throughcolumns of the fill and monitoring theleachate. The results indicated that somedegradation was taking place, but thereaction tended to be self-limiting; as morefeldspar was converted into clay thepermeability decreased and so did the rate ofthe reaction (ref.15).

21. At Roadford dam, Devon, simulatedweathering tests were carried out on themudstone fill by leaching with peaty waterand acid and measuring the resulting changesin chemistry and mineralogy. The resultswere used as a guide to the likely long termproperties of the fill assuming fulldegradation of the material. At Carsingtondam, where the mudstone fill was known to beacidic, tests were carried out by SheffieldUniversity Departments of Geology and CivilEngineering in collaboration with Babtie Shaw& Morton to assess the effect of acidleaching on the shear strength of thematerial. A slight decrease in shearstrength was noted, but this was dependent onthe density of the sample. There was adecrease in the grain size of the materialduring the tests. Acid leaching tests werealso carried out on site at Roadford Dam todetermine the effect on the filter materialof acid leaching from the embankment fill.Leaching with sulphuric acid over a period of48 hours was found to produce a very slightdecrease in the permeability of the filterand a weight loss of 1%.

DESIGN22. Having obtained the required

information, due allowance for the effects ofgeochemical degradation must be made in thedesign. For some effects, such as sulphateand acid attack on concrete and otherconstruction materials, guidance is availablefrom standard publications (refs 16-18). Forothers the criteria are less clear cut andsome degree of interpretation is necessary.

23. A number of reactions may affect thedensity and permeability of fill material.For example, solution of calcite or otherminerals may decrease density and increasepermeability. Conversely, precipitation ofgypsum following pyrite oxidation andreaction of the acid with calcite results ina volume increase of over 100% and can

188

PAPER 28: MACDONALD AND REID

generate significant heave pressures (ref.3).Several examples of this phenomenon have beenreported in the U.K. recently on the floorslabs of buildings (refs 19,20). Nixon(ref.21) gave guidelines for potentiallytroublesome shales viz; (1) total sulphurcontent in excess of the acid solublesulphate content, and (2) acid solublecalcium content not less than 0.5%. Theextent to which these reactions take placedepends critically on the availability of airand the rate of seepage through the fill.

24. In general the effects of geochemicaldegradation are to cause acceleratedmechanical breakdown of the fill, leading toa decrease in grain size, permeability andshear strength. This can be accomplished byacid attack on cementing materials or clayminerals, or simply by uptake of water bycertain clay minerals; the so-called"swelling clays" such as montmorillonite,vermiculite and mixed layer illite -montmorillonite intergrades. The mudstonesat Carsington, for example, break downrapidly an exposure to moisture. X-raydiffraction tests revealed that mixed layerillite - montmorillonite accounts for aboutW of the clay fraction. At Roadford theclay minerals consisted dominantly of illite,kaolinite and quartz, and the mudstones aremuch more stable when exposed to moisture.

25. The effects of weathering processes onthe shear strength of Coal Measures mudstoneshas been studied extensively by Taylor (refs2,3). He gives lower bound peak shearstrength parameters of X1 = 22° and cf = 0for loose tipped colliery spoil heapmaterial, and suggests that this is similarto the fully weathered peak shear strength ofthe weakest Coal Measures mudstones andclays. However, he highlights the fact thatintense chemical weathering of the surface ofa spoil tip over a 50 year period had onlyreduced j6r by about 1.5°. Skempton and Coats(22) stated that at Carsington damdeterioration of the mudstone fill used inthe dam shoulders had caused a decrease in 0f

from about 28° to 25°. This was thought tobe largely due to breakdown of the fill byabsorption of water by swelling clayminerals.

26. If laboratory simulated weatheringtests have been carried out, it may bepossible to determine the likely drop inshear strength directly from them. Inaddition some allowance for deteriorationbased on comparison of the shear strength offresh and weathered material from thepotential borrow area can be made. This dualapproach has been adopted at Colliford

(ref.15), Roadford and Carsington,the appropriate parameters for the mainshoulder fill materials being indicated onTable 1.

CONSTRUCTION28. Within the tender documents for a

project it is important that potentialproblems caused by chemical degradation ofthe fill material are clearly set out forcontractors in order that their tender priceadequately reflects the difficulties ofdealing with such material. It is equallyimportant that these are explained conciselyto the Client beforehand. For example, forthe Carsington reconstruction contract theContractor priced the items for pollutioncontrol at around £900,000, some 5% of thetender total, even though extensive pollutioncontrol lagoons were already in place. Thereare three main aspects which should beconsidered during the construction stage,namely safety, pollution control and fillworkability.

29. Safety - Leaching of minerals fromthe fill can result in gaseous emissions.These may be toxic or explosive in theirown right, or simply lead to an oxygendeficiency in confined spaces on the site.Such confined spaces include trialtrenches excavated in the fill for testingpurposes as well as manholes in the drainagesystem. Monitoring for gases should becarried out in all boreholes and confinedspaces; many of the more common geochemicalreactions generate carbon dioxide, andmethane and hydrogen sulphide may also beproduced within the fill or migrate upwardsthrough the foundation in certain strata, inparticular mudstones. Good design, forexample by limiting manhole depths, shouldhelp to mitigate these problems. However thecontractor must be made aware of the dangersin order that he can take the necessarysafety precautions.

Table 1. Examples of allowances made forchanges in shear strength due todegradation.

Dam

Colliford(ref. 15)

Roadford

Carsington

Shoulder fill type

China clay sand waste

Moderately weatheredto fresh sandstonesdno mudstonos

(i) Low normal stress(ii) High normal stress

Highly weatheredmudstone

Moderately weatheredmudstone

Design shear strength parameters

Newly won fill

C' : KN/—2

0

00

10

15

<t>'

39°

40°32°

22°

25°

Degraded fill

c': KN/m2

0

031

10

15

0'

35°

35°' 29-5°

22°

24°

189

ENVIRONMENT AND RESEARCH

30. Pollution Control - Pollution ofdownstream watercourses by high suspendedsolid loads has long been recognised as aproblem during embankment construction and ithas been good practice for many years toinstall settlement lagoons to deal with this.Geochemical changes in the fill and on borrowarea surfaces can result in water which isacidic in nature and has high metalconcentrations from underdrainage systems andfrom rainfall run-off. This must be dealt withpossibly by chemical dosing and settlement,if the run-off is of unacceptable quality todischarge to the river. At Carsington,caustic soda and lime dosing facilities werefound to be required together with largesettlement lagoons. Prior consultationwith the authority responsible for monitoringdischarges should enable parameters to beincluded in the contract documents for waterquality immediately downstream of the sitecompared to that upstream. The contractor isthen responsible for installing temporaryworks to ensure these parameters are met.Alternatively the Employer can design andoperate pollution control facilities, hencerelieving the contractor of the risk ofsystem failure provided he diverts all siterun-off through the facilities.

31. Fill Workability - Geochemical reactionscan result in rapid changes in materialproperties with the effects of air andmoisture on the fill. This is furtheraccelerated by vehicle movement, and ittherefore has to be made clear in thecontract documents that for the moresusceptible fill materials, operations mustcease whenever rain threatens.

32. During construction, procedures shouldbe established to monitor the physical andchemical properties of the material in theborrow areas, the fill placed in theembankment, the quality of run-off anddrainage water, and the presence of gases inrelief wells, manholes and other confinedspaces. This will enable a check to be madeagainst design assumptions. Such informationwill also be required to enable decisions tobe taken on the need for treatment in thepost-construction stage. A database ofresults can be established to enable trendsto be reviewed and long-term monitoringprocedures and frequency adjusted to suit.

IN-SERVICE CONDITION33. Once the dam has been completed, and

responsibility for operating the reservoirhas been taken over by the Client, continuedmonitoring will be required to determine ifgeochemical reactions are continuing, or havestarted to occur, within the fill.

Monitoring of drainage flows is good practiceon any dam, both during initial filling andsubsequent operation, and is regarded as oneof the best means of highlighting problemswith the structure at an early stage.Similarly, monitoring of drainage waterquality is the best means of establishing therate of geochemical change within anembankment. It should be carried out atfrequent intervals from commencement ofimpounding until the mineral composition ofthe drainage water becomes stable. Waterquality analysis should be continuedthereafter, but on a less frequest basis, tomonitor any long-term changes.

34. If the pattern of usage of thereservoir changes at any time, for example bythe transfer of water from a differentcatchment to supplement yield, the frequencyof water quality analysis may have to beincreased to monitor the effect of this newsource. Similarly the testing frequencyshould be increased if there are majorfluctuations in water level, such as couldoccur in a drought situation.

35. The drainage waters emanating from theembankment may be of too poor a quality fordischarge to the downstream watercourse.Some form of treatment may therefore have tobe installed. The cost implications of thiscan be considerable and so an attemptshould be made to assess at an early stagethe period for which treatment will berequired in order to provide the mosteconomic solution for what is likely to be atemporary condition. The chemical testingcarried out on the fill during constructioncoupled with the results of water qualitytesting can be used to obtain an approximateestimate of rate of leaching of minerals fromthe fill.

36. It is not possible to be precise aboutthe form of treatment required to deal withgeochemical pollution as this will depend onthe nature of the fill and the quality of thewatercourse into which the discharges are tobe made. The economics of each situationwill have to be assessed. However typicalexamples that have been considered for somesites are:(a) trickle discharge over wooded hillsidesdownstream of the embankment(b) pumping back into the reservoir andrelying on dilution(c) full treatment in a fpackage1 skid-mounted plant(d) utilisation of downstream pollutioncontrol facilities installed for embankmentconstruction purposes.

37. Operations staff must be made aware ofthe potential dangers from gases, and safe

190

PAPER 28: MACDONALD AND REID

entry procedures for manholes, tunnels etc.should be clearly set out. Wherever possiblegases should be allowed to vent naturally.However manholes may require to be purgedusing small mobile plant, and a permanentforced ventilation system is likely to berequired in all tunnels and galleries.

CONCLUSIONS38. Considerations of the geochemistry of

the fill material should be an inherent partof embankment dam design. The financialimplications of geochemical degradation canbe considerable. Embankment side slopes willrequire to be flatter to allow for lowerlong-term shear strengths and measures willhave to be taken both during and afterconstruction to control pollution ofwatercourses. The design of concretestructures and specification of importedmaterials should take cognisance of thepotential for aggressive groundwaters.

39. Of prime importance in predictinglong-term geochemical behaviour is thecollection of data from chemical testing ofthe fill and drainage waters at all stagesfrom initial site investigation through toconstruction. Testing and analysis shouldcontinue throughout the operation of thestructure until a stable environment isachieved.

REFERENCES1. BABTIE, SHAW & MORTON: Chemicaldeterioration of fill material in earth damsin the United Kingdom. Department of theEnvironment Research Contract NoPECD/7/7/193, 1986.2. TAYLOR, R.K.Composition andengineering properties of British collierydiscards, p244. British Coal, Mining Dept.,London,1984.3. TAYLOR R.K. Coal Measures mudrocks:composition, classification and weatheringprocesses. Quarterly Journal of EngineeringGeology, 1988, Vol.21,85-99.4. FOOKES P.G., GOURLEY C.S. and OPIKERSEC. Rock weathering in engineering time.Quarterly Journal of Engineering Geology,1988, Vol.21,33-57.5. KRAUSKOPF K.B.Introduction togeochemistry, p721. McGraw-Hill, NewYork, 1967.6. FITZPATRICK E.A. Pedology, asystematic approach to soil science, p306.Oliver and Boyd, 1971.7. SPEARS, D.A., TAYLOR, R.K. and TILL, R.A mineralogical investigation of a spoilheap at Yorkshire Main Colliery. QuarterlyJournal of Engineering Geology, 1971, Vol.3,239 - 252.

8. KENNARD M.F., KNILL, J.L. and VAUGHAN,P.R. The geotechnical properties andbehaviour of Carboniferous shale at theBalderhead dam. Quarterly Journal ofEngineering Geology, 1967, Vol 1, 3 - 24.9. PENMAN A.P.M. Tailings dams andlagoons. Proc. Symp. Building on Marginaland Derelict Land, p25-45, Thomas Telford,London, 1986.10. BRITISH STANDARDS INSTITUTION. Methodsof test for soils for civil engineeringpurposes. BS 1377 : 1981.11. BOWLEY M.J. Analysis of sulphate-bearing soils. Building ResearchEstablishment, CP2/79, 1979.12. HEAD K.H. Manual of soil laboratorytesting Vol. 1, p339. Pentech Press, Plymouth,1980.13. BRITISH STANDARDS INSTITUTION.Specification for air-cooled blastfurnanceslag aggregate for use in construction.BS1047 : 1983.14. TAYLOR, R.K. and SPEARS, D.A.Laboratory investigation of mudrocks.Quarterly Journal of EngineeringGeology,1981, Vol.14, 291 - 310.15. JOHNSTON T.A. and EVANS J.D.Colliford dam sand waste embankment andasphaltic concrete membrane. Proceedings ofthe Institution of Civil Engineers, Part 1,1985, Vol.78, 689-709.16. BUILDING RESEARCH ESTABLISHMENT.Concrete in sulphate bearing soils androcks. BRE Digest 250, 1981.17. HARRISON W.H.Durability of concretein acidic soils and waters. Concrete,1987,February, 1 8 - 2 4 .18. BRITISH STANDARDS INSTITUTION.Structural use of concrete, Part 1, Code ofpractice for design and construction. BS8110:Part 1: 1985.19. HAWKINS A.B. and PINCHES G.M. Causeand significance of heave at LlandoughHospital, Cardiff - a case history of groundfloor heave due to gypsum growth. QuarterlyJournal of Engineering Geology, 1987, Vol.20,41-57.20. WILSON E.J. Technical note: Pyriticshale heave in the lower Lias at Barry,Glamorgan. Quarterly Journal of EngineeringGeology, 1987'. Vol.20, 251 - 255.21. NIXON P.J. Floor heave in buildingsdue to the use of pyritic shales as fillmaterial. Chemistry and Industry 4, 1978, 160- 164.22. SKEMPTON A.W. and COATS D.J. CarslngtonDam Failure. Proc. Symp. Failures inEarthworks, p203-220. Thomas Telford, London,1985.

191

29. Reservoirs - a legacy of opportunity

K. E. COXON, J. E. EMBERTON, and R. M. PAGE1T, Gibb Environmental Sciences, Abing-don, UK

Whilst reservoirs are a vital management tool for the water industry, their constructionand management impinge upon the local built and natural environment. A wide range ofimpacts have been identified in this paper both relating to the reservoir's primary useand any secondary uses which may result. Both direct and indirect impacts are outlinedand a discussion of mitigation measures is included*

Introduction

The impoundment of water, forming reservoirsbehind embankment dams, is necessary in bothdeveloped and developing na t ions .Reservoirs are an important water managementtool , supplying water for domestic andindus t r ia l use and r e g u l a t i n g wateravailability. In the UK the equivalent of350 l i tres of water is added to the publicsupply per head per day, of which 270 litresper head per day is actually delivered tocustomers. Of t h i s , half the water isdelivered to households (135 litres per headper day) and half to industry. Reservoirsare especially important where groundwater isof inferior quality or is unavailable.

While reservoirs are clearly necessary, theenvironmental implicat ions of theirconstruction and management, their effect onthe local community and economy, and thepotential for secondary usage and developmentcannot be ignored.

A wide range of impacts can be identified.These impacts can be classified as short orlong term, primary (direct), secondary andtertiary (indirect). They can be due to theconstruction operation itself, changes insi te afteruse (including secondary use andsecondary development) or more indirectimpacts such as change in groundwater leveland hydrological gradients.

Impacts

The impacts of dam and reservoir constructioncan basically be divided into two clearlydistinguished classes, those of shoit-termconstruction phase impacts, and those oflonger term, post-construction phase,impacts. These are discussed in thefollowing sect ions. However, beforediscussing these it is important to considerwhat we mean by 'Environment1. Do we onlymean aspects of the natural surroundings suchas ecology and geology? Do we include

The embankment dam. Thomas Telford, London, 1991

aspects of the site such as rights of way,land use and human perception of the site?Or, do we include financial aspects such asland value, effect on the local economy andsecondary development? These are ofteninvestigated separately as Environmental"habitat" assessment, social impact andeconomic impact. The inclusion of social andeconomic aspects often dramatically affectsthe cost benef i t equat ion of manydevelopments, including reservoir and damprojects. However, the Chambers Englishdictionary defines environment as "externalconditions influencing development or growthof people, animals or plants". Indeed the ECDirective specifically states that impacts onthe human environment should be included (1).Thus environmental assessments shouldinvestigate all aspects of impact. Oneshould remember that impacts need not alwaysbe adverse. Benefits of dam and reservoirconstruction should be included in any EA.

Short Tera Impacts

These usually tend to occur during theconstruction phase and re la te to theconstruction operation itself together withassociated social and economic changes.Impacts due to aspects of the constructionoperation itself include landscape and landuse changes, noise , v i b r a t i o n andtransportat ion impacts, and temporarydisruption of ecology. Landscape changesinclude landscape alteration and consequentvisual impact. The visual impacts ofembankment dam construction are not as greatas those of concrete dam construction.However, they are often the most obvious tothe local populace during the constructionphase. Where relocation of residents in thelands to be flooded or local opposition tothe scheme has occurred i t is the visualchanges in topography that induce most ofthe social problems experienced.

193

ENVIRONMENT AND RESEARCH

It is often necessary to restrict access tooperat ing areas of the s i t e duringconstruction in the interests of publicsafety* This a l ters land use and mayinterrupt public rights of way* Accessroutes also often sever public rights of waysuch as footpaths or b r ida l ways*Alternative routes must be provided and wellsignposted* It is advantageous if thesealternative routes are those which will beused post construction ie they do not crossland to be flooded* This avoids having tochange routes more than once* This type ofdisruption is caused not only by theconstruction site itself but also 'transitcamps1, stores and vehicle parking andmaintenance areas* These operations alsocreate changes within the land use itself andmay lead to the severance of agriculturallands belonging to a single farm*

The construction phase involves muchmachinery movement, releasing energy into theenvironment in the form of noise andvibration* Transportation of men andmaterials to the site spread these effectsaway from the construction site itself* Poorsite access and insufficient road capacityoften cause problems which directly affectthe local population which may already be'sensit ised1 by the development* Suchproblems must be mitigated wherever possible*

Disruption of the local ecology often occursduring construction, either due to release ofcomponents into the natural environment, suchas particulates into local water courses, orseverance of habitats. Disruption of waterflow in local streams, or release ofcomponents such as particulates or oils,intolocal water courses can cause dramaticeffects on the 'downstream1 ecology, eitherby restricting oxygen exchange at the watersurface, cutting down photosyntheticefficiency by particulate settlement onaquatic plants leaves, direct phytotoxiceffects, or changing the flow regime and thusthe stream bed substratum*

The influx of men and money into a localcommunity often causes short term social andeconomic effects* There is frequently arequirement to house many workers in thelocal community* Temporary manual work isoften available to the local unemployed, andthere is always the need for goods andservices for the construction operation andmen employed on site* This can causeproblems, especially in small communities*

The last group of temporary impacts relate tothe reservoir f i l l ing operation itself*Progressive land take and disruption in thedownstream water flow regime are the mostimportant effects on the natural environment.The sudden removal of construction workersfrom the local community itself, can cause asmany problems as those created by the intialinflux of workers at the beginning of theconstruction phase*

Long Tern Effects

Long term effects relate to the after-use ofthe site rather than aspects of constructionitself, and remain long after the reservoiris constructed and filled* The most obviousimpact is land take. Reservoirs remove largetracts of land from their previous use. Thefilling operation also causes changes in thegroundwater regime of lands bordering the newreservoir which influences the land useoptions on adjacent lands. Many reservoirsare also used for leisure activities such assailing and fishing* The leisure uses oftenencourage secondary development of adjacentlands for club-houses, restaurants, holidayfaci l i t ies , marinas and boat yards with allof their associated infrastructure*

The ecological impacts may be diverse*Changes In water flow and nutrient status ofthe water (by precipitation of dissolvedions, settlement of par t icula tes andutilisation of nutrients by the lake's floraand fauna) can substant ia l ly a l t e r thedownstream ecology of the watercourse.Construction may also disrupt fish migrationroutes and introduce species not previouslyassociated with the area* Changes in thegroundwater regime adjacent to the site maycause vegetation and consequent faunalchanges within the lakes l i t tora l zone*Lastly the infrastructure associated with thereservoir and any secondary development maycause severance of habitats or disruption toterrestrial flora and fauna*

The increase in use of a particular area dueto the reservoir/dam construction may inducethe local authority or statutory conservationagency to place s tatutory ecologicaldesignations (eg Site of Special ScientificInterest status in the UK) on that area forprotection of the local ecology* Theincreased use will also lead to increases innoise and transportat ion effects andpermanent changes in the local economy toservice any secondary development that takesplace.

Reservoir management techniques to preventeutrophication or mitigate si l t accumulationwill also affect the environment by releaseof particulates and/or potentially toxicchemicals into the environment* These willalso lead to increases in traffic and noiseimpacts* Toxic chemicals may also bereleased by any secondary use of the site egsailing (eg TBT paints now banned on smallcraft) or secondary development (solid andliquid wastes)*

Mitigation

The mitigation strategy should consider,careful ly , each a spec t of r e s e r v o i rconstruct ion and operat ion balancingenvironmental susceptibility with the controlmeasures that would need to be implemented.

194

PAPER 29: COXON, EMBERTON AND PAGETT

The mitigation process is facilitated by acomprehensive review of existing data and ofthe nature and character is t ics of thereservoir. Where adverse impacts areapparent, mitigation measures for minimisingand/or offsetting these impacts, andopportunities for enhancing na tura lenvironmental values can be explored.

I t i s e s s e n t i a l to understand theutilisation, alteration and impairment ofnatural resources affected by the reservoirso that mitigation measures can be proposedand incorporated during the reservoir designprocess.

Changes in microclimate in the vicinity ofthe reservoir should be reviewed in terms ofhumidity levels which may affect insectpopulations. An analysis should be made ofthe hydrologic regime of the stream or riversystems to be modified by the reservoir.During this review, it is essential to assessthe overall impact on the monthly and annualmass water balance so that potential changesin the hydrological balance can beidentified. Likely modifications to averageand seasonal water quality should be detailedfor both the reservoir and the downstreamflow(s).

An assessment should be made of the effectson groundwater quantity and quality in thevicinity of the reservoir and downstream,alterations to water table, wells/aquifers,and infiltration rates in the watershed.

Sediment influx from watershed runoff anddownstream erosion, tectonic/seismicactivity, mineral resources, physical andchemical weathering, landslide and subsidencecharacteristics should be considered,

A reservoir development may have effects onsoi l erosion in the watershed, slopes t a b i l i t y , bea r ing c a p a c i t y andsettlement/heave and soil structure.

The assessment should include details of theimpact on fauna and flora in the watershedarea above the reservoir and in thedownstream zone(s), including those caused byassociated developments such as access roads.Inundated fauna and flora may need to berescued, relocated or re-established in newareas free from encroachment.

Anticipated physical, chemical and biologicalchanges within the reservoir may affect thewater column (nutrient trapping and thermalstratification) and in the benthos (animalcommunities living in or on the reservoirbottom sediment). Existing fisheries maysuffer as a result of changes to the quantityand quality of downstream flow directly orindirectly during migratory per iods .Mi t iga t ion measures may need to beimplemented to prevent overfishing orproblems of storage, processing and marketingof fish. An altered fishery may haveimplications for local nutrition and diet.

Any potential for the growth of weeds in thereservoir would need to be considered asshould the uses of the drawdown zone foragricultural and other purposes.

The mitigation plan should consider andidentify the impacts resulting from floodcontrol, including potential reduction inflood damage and reclamation of lands foragr icul tural use. The effects of thereservoir on navigation such as those relatedto low flow, proposed transhipment facilitiesand landing facilities may require mitigationand have wider implications for land usepatterns and land capability. For instance,downstream aquaculture could be improved bylow flow augmentation.

The social/economic conditions of the localpopulation in the region affected by thereservoir should be examined in order toidentify the need for mitigation in terms ofwelfare, new roads and new industr ies .Mi t iga t ion plans for managing theresettlement of populations (wholly orpartly), provisions for rehabi l i ta t ingfamilies in their new living/workingconditions and al ternat ive choices forresettlement (rural and urban) need to beassessed. If resettlement necessitateschanged agricultural methods (eg growingupland crops, use of i r r iga t ion) , themi t iga t ion programme should containprovisions for assisting the resettlers.

The mitigation study should assess the degreeto which the reservoir may affect thelikelihood of contracting water-borne diseaseand should propose corrective measures. Anassessment should also be made of plans foradequate community sani ta t ion, both toenhance quality of life and to minimisepollution of the reservoir. The study shouldassess the possible recreation and aestheticvalues of the reservoir and the plans foreffective development of i ts recreationalpotential, such as the planting or replantingof borrow areas and other disturbed areas.Measures for any inundated archaeological,historical and cultural sites should beassessed and the appropriateness of salvagingor preservation schemes be explored.

Secondary Uses

The 1980fs saw the "Water for All" decade andp o s i t i v e e f f o r t s to improve watera v a i l a b i l i t y for use in Less DevelopedCountries where the main primary uses ofembankment dam reservoirs are for hydro-e l e c t r i c power and for i r r i g a t i o n .Approximately one and half billion hectaresare used for agriculture worldwide. About13% of this area receives i r r igat ion water,providing 40% of the crop production.

A secondary use of i r r igat ion dams is thecontrol of the seasonal floods, although suchcontrol is not always beneficial. Some

195

ENVIRONMENT AND RESEARCH

populations rely on seasonal flooding for theestablishment of flood plain fisheries andhave to learn new techniques for lake-fishingof the reservoir to replace their establishedriverine fishing methods. The type of fishcaught often changes also. Annual floodsalso have the advantage of the deposition ofnutrient-rich sediments as the flood recedesfrom the flood plain, improving theagricultural quality of the land. Also theimpoundment of water within a reservoir cangreatly reduce the nutrient s tatus andsuspended sediment load of river waterdownstream of the dam. Major changes in theflow regime downstream can have seriouseffects downstream, as occurred with theAswan Dam in the 1960fs, Reduction in freshwater flow led to the loss of 4000 km of theNile delta, a vital food producing area andallowed the ingress of saline water from theestuary, destroying a highly valuable sardineindustry.

However, irrigation dams can support valuablesecondary uses. The creation of Lake Voltaresulted in the establishment of veryimportant Tilapia and freshwater sardinefisheries, valuable year round commercialresources.

In the UK the 1973 Water Act enforced themulti-purpose use of reservoirs, provided thesecondary uses were commensurate with theprimary purpose for the construction of thedam.

In the UK such multi-purpose uses areactually recreational but there are potentialconflicts with the primary management aims ofthe reservoir. The main primary use is theprovision of a potable water supply.

Fisheries

A popular secondary use of reservoirs isfishing, but the establishment of fisherieswithin reservoirs can cause managementproblems, A major component, and expense,in the treatment of potable water supplies isthe removal of algae through fi l tration.Dense algal growths rapidly clog f i l ters .Some algal species produce toxins which maytaint the water even after standard treatmentprocesses. Corrective treatment of such'tastes1 and odours is expensive and notalways possible. In extreme cases toxinsfrom algal blooms can ki l l livestock, thatdrink untreated water. This occurred in theUK in the summer of 1976 and 1989, Fish eatzooplankton, the organisms which normallygraze on algae, A good zooplanktonpopulation can significantly reduce watertreatment costs by 'cropping1 much of thealgae. The zooplankton are much larger thanalgae and are removed much more easilywithout clogging f i l ters . Introduction offish into a reservoir can therefore increasewater treatment costs.

If, through public pressure, or other reasonsthere is a management policy to maintain afishery within a reservoir, there are twoal ternat ives : ei ther a coarse or gamefishery. Coarse fish species are more suitedto high nutrient lowland waters but can causemanagement problems by shoaling and damagingpumping equipment. The use of ground baitcan cause clouding of water, increasing theorganic loading and treatment costs. Theestablishment of a game fishery, usually a'put-and-take1 trout fishery causes fewerproblems with respect to r e se rvo i rmanagement. But the lobby in favour ofcoarse fishing is very strong within the UK,

Coarse fishing attracts a very differentconsumer to that for game-fishing. It is amuch more social activity with competitivematches, whereas game-fishing tends to be amore so l i ta ry occupation, A differentmanagement strategy is required for the'consumer' as well as the game fish stockitself, A lowland reservoir is not a naturalhabitat for game fish such as trout whichwill need frequent restocking. The high costis generally reflected in licence fees, againemphasising the different type of consumerat t racted and the management approachrequired.

Sailing

Reservoir sai l ing has become extremelypopular within the UK, Generally onlysailing boats are encouraged as potable watersupply reservoirs are often too vulnerable topollution from oil and fuel spillages frompower boats. Reservoir sa i l ing can beseasonal, restricted to the winter monthswith boats used elsewhere in the summer andrequiring thorough cleaning before they areallowed to be returned to the reservoir,

A large reservoir may support simultaneoususes of both fishing and sailing if spatialseparation is possible. Alternativelytemporal separation is used with summerfishing and winter sailing.

Nature Reserves

A general reduction in the extent of waterbodies within the UK over the last 50 yearsor so has encouraged the use of reservoirsas wild life sanctuaries, especially forwaterfowl. The establishment of newreservoirs which can provide good waterfowlhabitats is often actively encouraged by thenation's "birders" who have an extremelystrong lobby. The potential for themultipurpose use of reservoirs can be pushedvery succesfully at planning enquiries whensuch secondary uses are proposed. Naturereserves tend to be res t r ic ted to theshallows of the reservoir , which areunsuitable for other secondary uses and thusgive the undisturbed conditions necessary forwildl i fe . Conflicts with the primarymanagement aim of the reservoir occur due

196

PAPER 29: COXON, EMBERTON AND PAGEn

to the seasonal drawdown of the reservoir - aproblem which has been very successfullysolved at Bough Beech in Kent where smallretaining bunds maintain shallow pans ofwater throughout the year. The provision ofislands, for breeding pairs of fowl, alsoprovides some protection against eggcollectors.

Sightseers

Day trippers or sightseers have been found toremain very close to their carsv Habitaterosion and disturbance can be prevented bychannelling such v i s i to rs to fhoneypotf

sites where lookout posts are provided withinformation boards, picnic sites, toilets andparking facilities,

A major source of friction in the multiplerecreational use approach is between ramblersand all other users. Dedicated footpaths,well signposted, with suitable informationboards etc will keep ramblers away from otherusers.

One of the largest secondary multi-usereservoirs in the south east UK is Datchetwhich successfully integrates year roundsailing, a put-and-take trout fishery andnight-clubs and restaurant facilities.

With the privatisation of a water industry itis anticipated that profitable secondaryuses of reservoirs will continue. It is verymuch hoped that the less profitable secondaryuses such as waterfowl reserves will notbecome less attractive propositions.

Concluding Remarks

Reservoirs are important in water managementterms. They provide essential supplies ofwater for domestic and industrial usage yetthe environmental implications of theirconstruction and operation cannot be ignored.Awareness of the potential short and longterra impacts of reservoirs can assist inincorporating mitigation strategies so thatthe benefits of reservoir development are noteroded by adverse effects . Subsequent,secondary use of the reservoir can enhancethe value of the development in addition tothat of the primary purpose of potable watersupply. Existing reservoirs provide a legacyof opportunity whilst those at the planningor construction stage can have theirreal isable benefit programmed into thereservoir development programme.

197

i n i i i m i n i Severn Barrage Proposed Alignment

Fig. 1. The Severn Estuary region

The embankment dam. Thomas Telford, London, 1991

20 30 Km

199

ENVIRONMENT AND RESEARCH

H.J. MOORHEADThe Authors of paper 29 are to be congratulatedon producing a paper which gives an excellentoverview of the main environmental impacts andeffects of the construction of a dam and theimpoundment of water. The paper is a goodstarting point for consideration of theenvironmental effects of dams. Saying startingpoint, does not in any way intend a criticismof the paper. The subject is a complex one andinvolves many disciplines and there isdifficulty in communicating across theinter-disciplinary boundaries. The paperindicates what has to be covered. Thefollowing diagrams indicate how a particularenvironmental study was organised using therecently completed Severn Tidal Power Studiesas an example.

Figure 1 shows the extent of the area whichwould be affected by a tidal barrage on theCardiff-Weston alignment, and Figure 2 showsthe effect of the barrage operation on the tidelevels where it will be seen that in thewith-barrage situation the level will go from

normal full tide level to half tide level withconsequential environmental effects. Figure 3shows the organisation of the studies and itwill be seen that a separate co-ordinator wasappointed for Tidal, Environmental and RegionalAspects. Figure 4 shows the links between theEnvironmental Studies and other projectaspects. The central role of hydrodynamics,sediments, water quality and salinity can beseen leading to two main streams ofenvironmental work:

- the Natural which comes down through thefood chain and also includes geology andgeomorphology, and

- the Anthropogenic which runs down thehuman related aspects as shown.

Finally all are drawn together in the AppraisalAssessment to produce the EnvironmentalStatement. In this particular case the resultsare being made publically available so thatthere can be a full debate on the EnvironmentalEffects.

PUMPING

S

GENERATINGAFTER PUMPING

GENERATING- NO PUMPING

FILLING

BASIN WATER LEVEL- WITH PUMPING

BASIN WATER LEVEL- NO PUMPING

TIOE LEVELSEAWARD OFBARRAGE

TIME

Fig. 2. Estuary and water level basins

200

PR

IVA

TE

IND

US

TRY

SE

VE

RN

TI

DA

L PO

WER

GR

OU

P

LS

ir R

ober

t M

cAlp

ine

& S

ons

Ltd

Bal

four

Bea

tty

Ltd

G.E

.C.

Tu

rbin

e G

ener

ator

s L

tdN

ort

he

rn E

ngin

eerin

g In

dust

ries

pic

Ta

ylo

r W

oodr

ow C

on

stru

ctio

n L

tdW

impe

y M

ajor

Pro

ject

s L

td

CO

-OR

DIN

AT

OR

Tid

alE

nvi

ron

me

nta

lR

egio

nal

GO

VE

RN

ME

NT

DE

PA

RTM

EN

T O

F E

NE

RG

YaE

CT

RIC

ITY

A

UTH

OR

ITY

CE

NTR

AL

E

LEC

TRIC

ITY

G

EN

ER

ATI

NG

B

OA

RD

STE

ER

ING

CO

MM

ITTE

E

MA

NA

GE

ME

NT

BO

AR

DS

EV

ER

N T

IDA

L P

OW

ER G

RO

UP

PR

OJE

CT

MA

NA

GE

ME

NT

BA

LFO

UR

B

EA

TTY

CO

-OR

DIN

AT

OR

Ele

ctri

cal

an

dM

echa

nica

lE

ngin

eeri

ng

CO

-OR

DIN

ATO

R

Sit

eIn

vest

igat

ion

san

dC

ivil

Eng

inee

ring

CO

-OR

DIN

ATO

R

Eco

nom

ics

and

Fina

ncin

g

CO

-OR

DIN

AT

OR

Leg

alan

dP

rom

oti

on

al

Fig

. 3. P

roje

ct o

rgan

izat

ion

HY

DR

OD

YN

AM

ICS

(Wav

es.T

ldo

s.C

urr

ento

)

SE

DIM

EN

TS.W

ATE

R

QU

ALI

TY.S

ALI

NIT

Y

70 m CO 55 TO O

EN

VIR

ON

ME

NT

NA

TUR

AL

Alg

aeO

ther

P

lan

tsIn

vert

ebra

toa

Fis

hB

ird

sM

amm

als

Geo

log

yG

eom

orp

ho

log

y

AN

THR

OP

OG

EN

IC

Lan

d

Dra

inag

eS

oa

Oef

on

cas

Ag

rlcu

ltu

roW

otor

P

oll

uti

on

/T

reat

men

tA

rch

aeo

log

yH

erit

age/

Co

nse

rvat

ion

Vis

ual

A

men

ity

Not

soC

on

stru

ctio

n

Sit

esT

ran

spo

rtM

iner

al

Ext

ract

ion

RE

GIO

NA

LD

EV

ELO

PM

EN

T

AP

PR

AIS

AL

-

AS

SE

SS

ME

NT

EN

VIR

ON

ME

NTA

L S

TATE

ME

NT

EC

ON

OM

ICS

AN

DFI

NA

NC

E

LAW

AN

DO

RG

AN

ISA

TIO

N

DE

CIS

ION

O

N C

ON

STR

UC

TIO

N

Fig.

4. L

inks

bet

ween

envi

ronm

enta

l stu

dies

and

othe

r pro

ject

s asp

ects

DISCUSSION

M. BRAMLEY (R & D Co-ordinator, NationalRivers Authority)

With regard to Paper 25, it is worth adding somefurther words about work carried out in the USon the subject of dam overtopping and theprotection of embankments and auxiliaryspillways from erosion. It needs to bementioned that George Powledge, the author,chaired the recent ASCE Task Committee onMechanics of Overflow Erosion on Embankments.

The Group's two reports are commended. Bothwere recently published in ASCE Journal ofHydraulic Engineering (their second report isquoted as Reference 6 to Paper 25, their firstreport on "Research Activities" in the same ASCEvolume). Reference 26 in particular provides auseful tabulation of American overtopping eventsat dam and levee embankments. Both reports alsodraw on UK information exchanged with UKengineers during the DTI-sponsored OverseasScientific and Technical Expert Mission (OSTEM)mission with which Michael Kennard and I wereinvolved in mid 1988.

Further information from the US on protectionand provision for safe overtopping of dams andflood banks is given in the CIRIA Report on thismission (1).

The paper refers to the Russian design ofwedge-shaped protection blocks (Figure 4b)whereby the stepped upper surface providesenhanced stability to the protection. The sixstepped block service spillways which have beenbuilt in Russia over the past 10 years have allperformed satisfactorily. However it is worthnoting that in cases the stepped surface hasbeen achieved using standard precast concreteroad slabs (generally of length in direction offlow about lm and thickness about 150mm). Theseare placed in a "stair-step" fashion on ahand-trimmed drainage layer. Thissimplification has enabled economies to be madein the manufacture of the otherwise complexblock shape.

With regard to Paper 29, the authors arecongratulated on a useful review of theenvironmental issues associated with reservoirsand reservoir construction. It may bebeneficial for dam engineers, who may feel thatenvironmentalists regard all effects ofreservoir development as negative, to reflect onthe very positive interaction which has now beendeveloped between engineers and environmentalinterest groups developed in flood defence andland drainage works. Here the conservationofficer or environmental interest groups haveprogressively been involved earlier and earlierin the planning of the works. This is notsimply to address environmental assessments, butalso to identify the environmental parameterswhich - alongside engineering and economicparameters - must form the basis for the design.

It is important to emphasise that positiveimpacts occur, particularly through thesubstantial amenity value which water space nowprovides. At Milton Keynes, property and landvalues in the vicinity of the flood storagereservoirs are considerably higher than atdistance.

Reservoir water quality is becoming an

increasingly important issue, particularly withthe recent emergency of problems ofeuthrophication and algal blooms. Reservoirengineers are likely to become increasinglyinvolved in mitigation strategies, not only fornew reservoir development but also for existingreservoirs, which enable reservoir water qualityto be managed and controlled. Such works might,for example, involve improved mixing, aeration,draw-off or dosing facilities.

Finally, it is important to put the reservoirwithin the context of the catchment as a wholeand recognise that problems of conflictinginterests in water use are prevalent throughout.In England and Wales, the NRA is now developingthe Catchment Management Plans as the basis ofits multi-functional water managementresponsibilities. These involve water quality,water resources, fisheries, conservation, flooddefence, navigation and recreation interests.The Catchment Management Plan, and the relateddesignation of use- related environmentalquality objectives for controlled waters withinthe catchment, should in the medium term providea clearer framework within which environmentalassessment of reservoirs is carried out inEngland and Wales.

Reference

1. Protection and provision for safeovertopping of dams and flood banks;Report to Department of Trade andIndustry on OSTEM mission to USA, ProjectReport 2, November 1987. CIRIA, London.

G. STEPHENSCN (Research Manager, CIRIA)CIRIA has recently started work on two projectsthat will be of interest to dam's engineers.They are:-

The Performance of Wedge Shaped Blocks inHigh Velocity Flows (Stage 2)

The Performance of Block Work Protectionfor Dam Faces

The first project on the performance of wedgeshaped blocks for spillways, has been mentionedin paper 45 of the proceedings and the firststage of the work was more fully reported inthe July 1989 edition of Water Power and Dams.(Pravdivets and Bramley)

The original concept behind the stepped blockwas developed in Russia by Dr. P.I. Giordienkoof the Moscow Institute of Civil Engineeringwhen looking into methods of protectingerodible surfaces subject to high velocityflows. It was further developed by ProfessorYuri Pravdivets of the same Institute into apracticable workmanlike solution. Eightsuccessful stepped block spillways have beenconstructed in the USSR in the past decade.

Outside of the USSR relatively little interesthad been shown in the concept prior to theCIRIA study, most probably because of a lack offirst-hand understanding and experience intheir use. The CIRIA project came aboutbecause of interest by USA and UK engineers inlow-cost methods of upgrading existing

203

ENVIRONMENT AND RESEARCH

spillways using new design concepts. Stage oneof the study concentrated on:

. Demonstrating the potential benefits andmode of functioning of the wedge-shapedblocks.

. Setting out a design methodology.

. Providing the information necessary toundertake stage two.

The work undertaken included:

. A review of published data.

. Assembling of existing practicalknowledge.

. Model testing.

. Outline specification for the designguide.

Work already undertaken by Salford Universityand Hydraulics Research Ltd. under asub-contract to CIRIA has shown that the blocksshow considerable advantages over moreconventional methods of chute construction. Inparticular:

. The upstream edge of the block isshielded from potentially dirsuptive flowstagnation pressure, which can give riseto extreme lift and drag forces on aprotection system.

. The flow pattern produces a low-pressureseparation zone downstream of each step.This zone is connected by drainage ventsto the underlayer and controls thebuild-up of seepage flow.

. The block shape is inherently stable. Ifany block moves perpendicularly off theslope (either by lateral displacement orby rotation about one end), the slopingupper surface experiences a stabilisingdownthrust.

. The stepped upper surface has a highroughness which helps to dissipate theenergy of the flow and reduce flowvelocity, hence reducing the amount ofenergy to be dissipated at the tailwateror toe.

Initial work on the design guide (to becompleted as part of stage 2) has shown that itis feasible to provide a simple frameworkwithin which a competent engineer can developsite-specific solutions. In stage one of theproject an outline specification for the guidewas set out.

In stage 2 of the project, due for completionlate 1991, further laboratory tests at SalfordUniversity are planned. These will look moreclosely at:

1. Pressure distribution, with the datacollected on a data logger and analysedstatistically.

2. The effect of longitudinal joints on thestability of a panel of blocks, by layingthe 17mm average thickness blocks withincreasing gaps between them untilfailure occurs.

3. Block lifting with a 50mm block restingon a micro-switch that detects when theblock first lifts. These tests will beconducted with and without interblockrestraint on the test block.

4. The force parallel to the slope on a 50mmblock resting on rollers with and withoutinterblock restraint.

5. The stability of the blocks in ahydraulic jump and the design of a toearea using all three block sizes. Blocksof a different weight will also beinvestigated by attaching metal strips tothe base of the existing model blocks.

The design guide will incorporate theconclusions from the laboratory and deskstudies together with any further informationobtained from Professor Pravdivets. Work onthe design guide will involve the expansion ofthe draft included in the Stage 1 report toaddress the gaps identified in the designapproach.

The work is being funded by the Department ofthe Environment (Water Industry Directorate),United States Corps of Engineers, Severn TrentWater pic and Salford University.

The second project which has just started is ajoint project between CIRIA and HydraulicsResearch Ltd. The objective of the study is toproduce a guide for the use of engineersresponsible for the design, maintenance andrehabilitation of blockwork protection againstwave attack. Specific phases of the work are:

. Conduct a general survey of blockworkprotection to UK dams including presentand past practices.

. Examine in detail selected samples,including instances of blockworkmovement.

. Derive local wind, and hence waveconditions, at selected sites for pastextreme events.

. Calculate conditions for blockworkmovement, movement/stability for selectedsites using results of previous researchand other members for conditionsidentified above.

. Describe and consider options forimproving stability of existing blockworkprotection.

. Produce a report to give guidance on bestpractice in design, maintenance and/orrehabilitation of blockwork protectionfor dams.

Publication of the final document is due spring1992 and funding is being provided bv the

204

Department of the Environment (Water IndustryDirectorate).

For further information on either projectplease contact Garry Stephenson at CIRIA, 6Storey's Gate, Westminster, London, SW1P 3AU.

N. OOYLE (Colquhoun, formerly N.W.W.)In relation to Robertshaw and Dyke's paper,elbows in embankments, and more especiallycorners of hillside reservoir embankments, wereareas prone to disturbance. Supervisingengineers should pay particular attention tosuch locations. These areas were particularlydifficult to cover by dimensional monitoring,due to the difficulty of establishing adequateinstrument stations and sight lines.

Had the authors encountered such difficultiesand had their monitoring revealed movement atcorners?

MR. ROBERTSHBW in reply agreed with Mr. Hoyle'sremarks about both the difficulty of surveyingand the tendency to movement at corners of

DISCUSSION

embankments. His experience had neverthelessbeen that both vertical and horizontal movementwere largely related to the height of theembankment.

G.P. SIMS (EPD Consultants)Would the authors please confirm that it ispossible to use photographs taken by 35mm orsimilar cameras for photogrammetric purposes?Can advantage be taken of pre-existingphotographs? How would such photographs beused?

In response MR. HOPKINS confirmed that 35mmcameras could be used, but with a much lowerprecision because nothing is known about thecamera. Historical photography can be used,using new photography and identifying featuresaround the site which didn't move and whichappeared on the previous photography. Thesecould be used to work backwards and work outsome of the properties of the earlier, unknowncamera. The results would not be as good, butmay be all that is available.

205