ALCRUDO & MULET 1

The Tous Dam break case study F. ALCRUDO, J. MULET

Area de Mecánica de Fluidos, CPS-Universidad de Zaragoza, 50.018-Zaragoza, SPAIN

alcrudo@posta.unizar.es

SUMMARY This document presents a description of the IMPACT Project Case Study adopted for Theme Area 2, Breach Formation, and Theme Area 3, Flood Propagation. The case study is based upon the failure of Tous Dam and the flooding of Sumacárcel, a small town about 5Km downstream of Tous Dam that broke due to overtopping on October 20, 1982. The event is pertinent to the IMPACT Project because it regards the failure of a major flood control structure, and considers risk to population and damage to downstream located properties. The case study described in this document refers to Flood Propagation issues encompassing both flood wave travel along a natural river valley and its impact to urban areas. The information provided allows for mathematical modelling of the event and an estimation of the uncertainties present in the data as well as those introduced by the methodology and models used. 1 INTRODUCTION During 20 and 21 October 1982 a particular meteorological condition consisting of a cold, high-altitude depression surrounded by warm air with high moisture content led to extremely heavy rainfall in the hinterland of the central Mediterranean coast of Spain. Average intensities as high as 500mm in 24 hours were recorded in wide areas. As a result the Júcar River basin, directly affected by the rains, suffered heavy flooding all along. Particularly dramatic was the flooding of the downstream part of the basin that is densely populated, with towns such as Alcira or Algemesí and the city of Valencia. The last flood control structure of the Júcar River basin, Tous Dam, located only several kilometres upstream of the two aforementioned towns ultimately failed on October 20, at about 19:00 hours with devastating effects downstream. The Júcar River basin covers some 21600 km2 of hinterland in the central part of the Mediterranean coast of Spain. Of these, 17820 km2 lead to Tous dam (see Figures 1 to 3). Meteorological conditions in the area, of mediterranean character, are prone to extreme rainfall events associated to high insolation and proximity to warm sea waters. These characteristics are accentuated in the lower part of the basin, closer to the Mediterranean sea. Further, two mountain ranges (Sistema Ibérico running NW-SE and Sistema Bético running SW-NE) come together in the area producing a converging channelling effect.

ALCRUDO & MULET 2

The basin is well known for its remarkable floods that are historically documented in the area since the XIV century. Their nature is markedly random. For instance no catastrophic floods were registered in the 1923-1982 period, but from 1982 and in only 5 years two such phenomena took place (October 1982 and November 1987).

Figure 1: Location of the Júcar River basin in the Spanish peninsula (from Ferrer et al. 1999).

Figure 2: The Júcar River basin (from Ferrer et al. 1999).

ALCRUDO & MULET 3

Figure 3: Location of Tous Dam and area under its influence (from Ferrer et al. 1999).

Figure 4: Total rainfall distribution in mm in the Júcar River basin (from Estrela 1999).

ALCRUDO & MULET 4

Figure 5: Atmospheric situation at the time of the heavy rains in October 1982.

Figure 6: The remains of Tous Dam after the failure looking upstream (from Estrela 1999).

ALCRUDO & MULET 5

Existing records state that on September 27, 1517, a flood swept away the whole town of Alcocer. Other extreme floods described in the records took place in 1753, 1766, 1779, 1785, 1791, 1801, and 1805. The last one knocked down 61 houses in the town of Alcira and more than two hundred had to be shored up. High water marks along the streets reached between 2m and 3m. Specially well documented are the two most important floods since 1600: The San Carlos flood on 4-5 November 1864 with an estimated peak discharge of 13000 m3/s and the October 20, 1982 flood that ultimately led to the collapse of Tous Dam. The former was referred to by local witnesses during a visit to Sumacárcel in quest for documentation of the 1982 flood. The old Tous Dam (i.e. the one that failed, since a new, larger dam is presently in operation in the same place and with the same name) is the result of a series of projects that, in some cases never left the drawing boards, in others entered construction stage but were later on halted and so on. The solution finally adopted consisted of a rock fill dam with an impervious core, leaning on both banks against two concrete blocks pertaining to an older, concrete dam design that was never concluded. The main data concerning the old Tous Dam and corresponding reservoir are listed below: Crest Elevation 98.5 m Water Elevation at Normal Operation 84 m Reservoir Capacity: At Normal Operation 52 Hm3

At Crest Elevation 122 Hm3

Spillway elevation 87.5 m (Gates closed) Spillway Elevation 77 m (Gates lift open) When rain started to fall in the basin during the late night of October 19 and early morning of October 20 streams merged together running downstream towards Tous Reservoir. The strongest intensity took place during the morning and around noon on October 20 (from 8:00 to 14:00) with other peaks later in the afternoon. Total rainfall over the basin reached almost 600 Hm3 (million m3), largely exceeding the capacity of old Tous reservoir (120 Hm3) that bore a peak inflow hydrograph of 9900 m3/s. A plot showing total rainfall distribution over the basin during the event can be seen in Figure 4. In Figure 5 a satellite view of the Spanish Peninsula at the time of the heavy rains is shown.

ALCRUDO & MULET 6

As a result, water level in the reservoir rose steadily, soon exceeding the Normal Operation elevation (84 m). Early in the morning of October 20 the water started to spill over the gates of the spillway that remained closed (closed gate elevation 87.5 m) but water level continued to increase. The crest elevation (98.5 m) was reached at about 17:30 hours that day and about 15 minutes later spilling over the crest started, beginning the erosion process. It seems that water elevation in the reservoir never exceeded 99.5 m. Dam failure was a progressive process as it happens with loose material dams. However it can be said that collapse took place at about 19:15 hours, when a significant part of the dam fell, including the leftmost gate and rig, and was swept away by the stream. A picture of the remains of Tous Dam can be seen in Figure 6. The outflow hydrograph is discussed in Section 4, Initial and boundary conditions, of this report. The effects of the flood downstream of Tous Dam were catastrophic: 300 km2 of inhabited land, including many towns and villages were severely flooded, affecting some 200,000 people of which 100,000 had to be evacuated, totalling 8 casualties. The first affected town is Sumacárcel (population 2000), about 5 km downstream Tous Dam, lying at the foot of a hill on the right bank of river Júcar (see Figure 7). The terrain is moderately mountainous and most of the buildings lie on a slope what protected them from the flood. The ancient part of the village, however, is located closer to the river course and was completely flooded, with high water marks reaching between 6m and 7m. A sight of the town before and after the flood is shown in Figures 8 and 9 respectively. About 7 km downstream Sumacárcel, where the valley begins to expand leading to the lower course of Júcar river, lies the town of Antella (population 1500). It is located in much flatter ground than Sumacárcel and was hit by the flood wave that knocked down most of the houses to such an extent that a new town has been built at a higher elevation. The case of Antella could have also been used as a case study for urban flooding save for the fact that almost no building remained standing and those who did were so badly damaged that were finally demolished. Therefore data collection proved almost impossible.

ALCRUDO & MULET 7

M

Figure 7: Aerial viewcorner) to Sum

OLD TOUS DA

(scale 1:25000) of the Júcar River reach

acárcel (upper right corner) about one w

SUMACÁRCEL

from Tous Dam (lower left eek after the flood.

ALCRUDO & MULET 8

M

Fi

F

TOUS DA

gure 8: The town of Sumacárcel before the flood looking upstream.

igure 9: The town of Sumacárcel after the flood looking upstream.

ALCRUDO & MULET 9

2 DATA COLLECTION This section describes the process and sources of data to prepare the IMPACT project Flood Propagation Case Study described in this document. The range of material collected spans from topographic to hydraulic data, including timing of the event to the extent possible. 2.1 Topographic data Topographic data regard a sufficiently resolved numeric description of the valley bathymetry to allow modelling. Two sets of data have been found. One dates back to a few weeks after the disaster (1982 data) and the other was produced in 1998. Very unfortunately no data prior to the flood have been found, what would have allowed the study of sediment transport. Both sets are now in digital format, more precisely in the form of a Digital Terrain Model (DTM). Also a series of aerial cartographic pictures covering the Júcar river reach and Tous reservoir are available in digital format. There are two sets of photographs, one to a scale of 1:10000 and the other to a scale of 1:25000 taken a few days after the disaster. These were found in the Diputación Provincial de Valencia (Local Government) in the form of plates and were digitised with the highest possible resolution. The 1982 DTM has been generated by CESI (Italy) from a paper copy map at scale 1:5000 owned by CEDEX (Center for Studies and Experimentation of the Ministry for Public Works, Spain) with terrain elevation curves drawn every 1m. The remains of the dam and evidence of the sediment movement process that took place can be observed in the map. However sediment movement is very difficult to quantify. The horizontal spatial resolution is 5m although finer grids could be generated if needed. However, this possibility seems unlikely. The 1998 DTM was generated at CEDEX (Ministry of Public Works, Spain) from cartographic data of 1998 with the new Tous Dam (considerably taller and larger than the old one) already built and in operation by that time. The horizontal spatial resolution is 5m. Furthermore listings of the buildings present in the area (the town of Sumacárcel plus dispersed buildings in the valley) with their shape and coordinates of the shape vertices plus rooftop elevation were provided by CEDEX. Additional documentation was drawn from the Catastro Provincial de Valencia (Land registry of Valencia). They were corrected to the time of the flood with the aid of local maps and discussions with citizens of Sumacárcel. 2.2 Hydraulic data The 1982 flood has been studied extensively by CEDEX (Ministry for Public Works, Spain) and by the Confederación Hidrográfica del Júcar (Júcar River Basin Authority). In most studies both bodies worked together because they report to the same Ministry. A physical model of Tous Dam and downstream reach, as well as several mathematical models of the flood (mostly of the 1-D and flat pond type) were also developed.

ALCRUDO & MULET 10

The CEDEX issued several reports on the subject of which those cited in the references under CEDEX (1984), CEDEX (1989) and CEDEX (1989b) could be accessed by the authors of this report. The first one provides a thorough description of the event, including recorded measurements and statements of eye witnesses and citizens mostly in populated areas and hot spots (dams, power stations) but also all around the basin. Hydrologic data are given that were used to determine inflow hydrograph to Tous Reservoir. Hydraulic parameters concerning Júcar River and its control sections were used to determine Tous Reservoir outflow hydrograph. A sequence of events explaining Tous Dam failure can also be found there. Five years later the second report (CEDEX 1989) issued is a revision of the first one refining some figures in view of work carried out at the CEDEX laboratory in a physical model of Tous Dam and corrected hydrologic calculations. In this task assistance was sought from external institutions (Laboratorio Nacional de Engenheria Civil, Portugal and the U.S. Soil Conservation Service) to assess the CEDEX figures and conclusions. The two series of cartographic pictures mentioned above, provide excellent resolution and show quite clearly the envelope of water elevation along the banks of the Júcar River reach. This was recognized by A. Zuccalá at CESI who devoted some work to prepare an Autocad file displaying the shore line envelope during the flood along the river banks. The pictures themselves, upon careful analysis, provide much information including sediment transport (deposition), but are difficult to work with because of their size (between 100 and 250 Mbyte each). As regards hydraulic data concerning the flooding of the town of Sumacárcel, and besides the information included in the reports by CEDEX, a field study was carried out. A visit to the town was made by the authors of this document during 26-29 March 2003. Several eye witnesses of the flood were interviewed, including a member of the local police with whom all the streets flooded were covered and data regarding high water marks and timing of the flood were taken. These data were translated onto paper maps. Regarding the issue of timing it must be borne in mind that the town was evacuated, hence a very limited set of data could be recorded.

ALCRUDO & MULET 11

3 VALLEY BATHYMETRY 3.1 General valley bathymetry There are two Digital Terrain Models (DTM) available covering the Júcar River reach from Tous Dam to about 1 Km downstream the town of Sumacárcel. The first or older one, named DTM_1982.dat, dates back to one week after the disaster. The second or later one, named DTM_1998.dat was made in 1998 and covers the whole Júcar River lower basin although only the reach of interest to IMPACT project is described and will be used here. Both bathymetries were prepared from topographic surveys obtained from aerial photogrammetry with a 5m spacing between nodes. The two files are described below in greater detail. a) DTM_1982.dat The DTM was generated by CESI (Italy) from paper maps at scale 1:5000. The paper maps were firstly digitized with high enough resolution so that terrain isolines could be easily identified. The isolines were then digitised into an Autocad file and from these the DTM model was generated. The DTM is made up of a grid with 5m spacing. The file itself is a structured, ordered ASCII collection of points. The file header lists the number of rows (J=1040) and columns (I=793) with the I-index running first. Each point is defined by its (X,Y,Z) coordinates that are all three printed in a single row. The remains of old Tous Dam are present in the upstream part of the DTM. Also, the maximum elevation of the terrain has been cut off to a fixed value in certain areas outside the river valley. These are easily noticed in a plot view as plateaus. This issue has no influence as regards flood simulation because water never reached those areas. b) DTM_1998.dat The 1998 DTM was generated by CEDEX (Spain) from digital photogrammetry data. The file is a non-structured collection of 258674 points covering about 5 Km of the valley, from the new Tous Dam to about 1 km downstream of Sumacárcel with 5m nodal spacing. The upstream part of the DTM features the new Tous Dam, finished in 1995, that can easily be seen in a contour or 3-D plot of the terrain. The 1998 DTM covers a considerably longer reach of the river than the 1982 one. Hence it should be cut short downstream the town of Sumacárcel at the discretion of the modeller (about 1 km should be more than enough). A general view of the 1982 and 1998 bathymetries can be seen in Figures 10 and 11 respectively below.

ALCRUDO & MULET 12

M

Figure 10: Th

M

Figure 11: Th

OLD TOUS DA

SUMACÁRCEL

e 1982 Bathymetry.

e 1998 ba

SUMACÁRCEL

NEW TOUS DA

thymetry.

ALCRUDO & MULET 13

Differences between both DTMs Some remarkable differences can be found between the two DTMs available. They regard mainly the river bed and to a lesser extent the upstream part of the valley where construction works for the new Tous Dam have been performed. The main reason for the differences in river bed is the dredging that took place during conditioning of the reach after construction of the new Tous Dam. Also considerable amount of sediment transport took place during the 1982 flooding that in some places exceeded 2m of sediment deposition. The 1982 DTM shows this sediment in place. As an example, the river bed at the section by the town of Sumacárcel after the flood was 1.25m above the original bed. Minor differences in bottom elevation might also have arisen because of the different cartographic sources and processing techniques used. 3.2 Building and town topographic data A complete listing of building position and data are available in file Buildings.dat. This is an ASCII file defining the building shape by listing its vertices. Hence the entry for a given building is the building number followed by the number of vertices of its boundary and the roof elevation (assumed flat), followed by a list of the (X,Y) coordinates of each one of the vertices of its boundary. The following example reads: Building number 186 is a polygon with 4 vertices and its rooftop elevation is 67.5m. The (X,Y) coordinates of the given 4 vertices follow.

186 4 67.5 2100.75 3819.50 2105.56 3820.50 2100.50 3816.00 2097.00 3819.50

This file allows for inclusion of the town into any of the two DTMs described above (it must be said that the raw DTMs contain no data of the buildings or town) by appropriate computational algorithms. However, in order to ease the task for the different modellers two add-on DTMs covering only the town of Sumacárcel have been generated. They are named Town_1982.dat and Town_1998.dat. Town_1982.dat and Town_1998.dat, contain a structured, ordered ASCII collection of points (in fact they form a mesh or computational grid). The file header lists the number of rows and columns with the I-index running first. Each point is defined by its (X,Y,Z) coordinates that are all three printed in a single row. It must be borne in mind that Town_1982.dat must be used in conjunction with DTM_1982.dat and the same holds true for Town_1998.dat with respect to DTM_1998.dat. In Figure 12, shown below, it can be seen an example where the buildings that make up the town of Sumacárcel have been introduced as a bottom elevation in the DTM.

ALCRUDO & MULET 14

Figure 12: Buildings introduced in a DTM as bottom elevation.

ALCRUDO & MULET 15

4 INITIAL AND BOUNDARY CONDITIONS 4.1 Upstream boundary condition The upstream boundary condition for use in Flood Propagation modelling is the outflow hydrograph from Tous Dam. This hydrograph is synthetic since no discharge records exist. However it has been obtained from a consistent blend of observations, measurements in a reduced scale physical model of Tous Dam and downstream reach, and hydrologic and hydraulic calculations. It is the most likely curve representing the phenomenon and after several revisions of the study they can be considered the best estimate of what actually happened. The peak of the outflow hydrograph is obtained by means of a calibrated section of the Jucar River situated just downstream the dam. The rest of the hydrograph is obtained using the historical water levels in the reservoir together with the calibrated discharge rating curve of the dam, collating the results with measurements from a physical model of the event.

Tous outflow

0

2000

4000

6000

8000

10000

12000

14000

16000

19/10/198212:00

20/10/19820:00

20/10/198212:00

21/10/19820:00

21/10/198212:00

22/10/19820:00

22/10/198212:00

Time

m3/s

Figure 13: Tous Dam outflow hydrograph. The Tous outflow hydrograph numeric data are stored as a table in a Microsoft Excel book named Outflow.xls. It should be imposed as an inflow boundary condition at the upstream section of the river reach considered. Since the upstream section of any of the two DTM files includes either the breached old Tous Dam (1982) or the new Tous Dam (1998) several options are open to the modeller. Either impose this hydrograph flowing over the dam

ALCRUDO & MULET 16

(or dam remains), across the breach or just start the simulations just downstream the dam location. It should be noticed that the outflow hydrograph of Figure 13 represents a situation where nothing happens until about 7:00h October 20 (i.e. 25200 seconds starting at 0:00h October 20). Hence considerable computing time can be saved if modelling starts at that time. 4.2 Downstream boundary condition The river reach considered extends about 1 km downstream the location of Sumacárcel. There are no data regarding the flow state at that section and hence the modeller should consider the most appropriate boundary condition to represent this situation. 4.3 Initial condition The initial depth of water in the river reach is unknown prior to the rain events. The normal flow of Júcar River is roughly 50m3/s which is totally negligible in comparison with the scale of Tous outflow hydrograph (see Figure 13). Since Tous Dam spillway gates remained closed until the dam failure, the flow in the river was only due to the river base discharge plus the surface runoff coming from the partial basin covering the reach. The surface runoff until spilling started can be inferred from the fact that an estimated 100mm rain had fallen until 8:00 hours of 20 October. An additional 100mm fell during the rest of October 20 and October 21. Given the huge volume of water represented by Tous outflow hydrograph, calculations can be started with a dry bed as a first approximation. In order to be consistent among different modellers it is advised that the time origin be set at 0:00h October 20.

ALCRUDO & MULET 17

5 ROUGHNESS COEFFICIENT Given the diversity of soils, vegetation coverage and crop fields present in the river reach, it is clear that the friction levels vary substantially all over the area. As a very simple approximation a base Manning coefficient can be used but main features of the flow can be lost. Table 1, below, shows Manning coefficients used during calibration of mathematical models (one dimensional) of the Tous event by CEDEX. Crop field or vegetation distribution can be found from the aerial pictures of the reach.

SOIL TYPE FRICTION COEFFICIENT (MANNING) S.I. units

RIVER BED 0.025-0.045

COVERED RIVER BED 0.014-0.017

ORANGE TREE ORCHARD 0.05-0.1

RICEFIELD 0.02-0.025

VEGETABLE GARDEN 0.025-0.04

Table 1: Estimated Manning coefficients for different land use.

Especially important for the flooding of Sumacárcel are the orange tree orchards that surround the town. The flow impedance due to these trees dammed up water and greatly contributed to the urban flooding. As a suggestion a base Manning of 0.03 can be used for the whole river reach and then take into account the orange tree orchards as zones of increased Manning coefficient. From the table above, and given the size and density of the trees the maximum value of the span should be selected, thus yielding a Manning of 0.1. Two remarkable areas covered by orange trees are described as four sided polygons whose vertices are listed in the Table 2 below:

Vertex points ZONE 1 ZONE 2

X Y X Y

P1 2025 4040 2430 3250

P2 2500 3500 2730 2810

P3 2675 3800 3030 3120

P4 2130 4200 2770 3725

Table 2: Vertices of the polygons covered by orange trees.

ALCRUDO & MULET 18

6 PROBES Field observations and or measurements have been collected at 21 locations in or close to Sumacárcel. These are the only means for model assessment together with the envelope of the shore line along the river reach. The latter has been obtained by the CESI team from the traces observable in the aerial pictures taken some days after the flood and have been enclosed in an Autocad file. The available data on the 21 locations are limited to high water marks and, in some places, a rough evolution of water depth with time. It is important to note that since the ground in the town area was paved, the flood did not erode it. Put in other words, there is no concern as for sediment movement and bathymetry uncertainties in that area. The Table 3 below, lists the location (X,Y) of the selected points in the coordinates of the bathymetry files.

GAUGE X Y EvolutionAvailable Comments

1 2410 3290 Yes River bed 2 2400 3335 No Cinema 3 2355 2315 No Church street 4 2345 3380 No Condes de Orgaz

Street 5 2335 3175 No Júcar Street 6 2335 3420 No Proyecto C Street 7 2330 3365 Yes Old city hall 8 2315 3450 Yes Clocks site 9 2310 3590 No Era Square 10 2303 3255 No Júcar Street 11 2285 3425 No Stairs Street 12 2285 3500 No Condes de Orgaz

Street 13 2280 3280 No Valencia Street 14 2266 3550 No Pintor Sorolla Street 15 2265 3400 No Valencia Street 16 2259 3530 No Pintor Sorolla Street 17 2250 3440 No Pallecer Street 18 2230 3525 No Severo Ochoa Street 19 2205 3445 No Virgen Street 20 2195 3440 No Virgen Street 21 2190 3485 No West Avenue

Table 3: Probe locations.

ALCRUDO & MULET 19

And Figure 14 below shows the location of the probes in the town.

Figure 14: Probe locations in Sumacárcel.

The modellers are encouraged to record water level history at those points during the event in order to assess the accuracy and uncertainty of their models. The following series of pictures help giving an impression of the places where the probes are located and the maximum water elevation attained during the flood.

ALCRUDO & MULET 20

Figure 15: A view of Sumacárcel main square from its main access, across the bridge.

MAXIMUM WATER LEVEL

Figure 16: Probe 3, the man behind the car marks 2m elevation above the ground.

ALCRUDO & MULET 21

MAXIMUM WATER LEVEL

Figure 17: Probe 6.

MAXIMUM WATER LEVEL IN THE SAN CARLOS FLOOD (1864)

MAXIMUM WATER LEVEL IN THE 1982 FLOOD

Figure 18: Probe 7.

ALCRUDO & MULET 22

Figure 19: Probe 8.

.

MAXIMUM WATER LEVEL

Figure 20: Probe 10, the man marks 2m elevation above the ground.

ALCRUDO & MULET 23

MAXIMUM WATER LEVEL

Figure 21: Probe 16.

ALCRUDO & MULET 24

7 FLOOD PROPAGATION MODELLING ADVICES In this section some advices are given concerning modelling of the Tous event. They have been drawn from the authors’ experience during preliminary modelling of the flood in view of preparation of the Tous Flood Propagation Case Study for the IMPACT project (i.e. this document).

1. During collation of field data, evidences have been found that the Tous outflow hydrograph timing could be erroneous. In particular it seems that the whole curve should be advanced in time (i.e. shifted to the left) by about one half to one hour. The reasons for this are several:

- The outflow hydrograph is discretised by the hour, and the peak reading is at 20:00h

(October 20). It seems logical that the peak outflow matches the timing of definite dam collapse that has been reported around 19:15h.

- The maximum water depths in the town were registered at 19:40h by a (pendulum)

clock that stalled upon arrival of water to the first floor room of a house. It looks reasonable that peak flooding levels occur after peak outflow discharge and not the opposite.

- Outflow discharge can be computed with the recorded water levels at Tous reservoir

(prior to the collapse) using the Tous dam outflow rating curve. The figures obtained only match the Tous outflow hydrograph (Figure 8) if this is advanced in time by one half to one hour.

It seems that the Tous outflow hydrograph was computed attending to fit the overall

water volume and outflow trends (shape) but disregarding accurate timing since it is discretised only every hour.

2. The influence of the orange tree orchards surrounding Sumacárcel is of paramount

importance in damming up water due to the increased flow resistance. This has a strong impact on the flood characteristics, in particular, in steering the flood into the town. After witnesses, the flood was directed towards the river bank opposite to the one where the town lies due to the topography (see figures above and aerial pictures etc…). However that bank was covered with very tall orange trees (5 to 6 m high) that dammed up water leading to a lateral flooding of Sumacárcel.

3. The flood duration is very high (about 2 days). Therefore model computations are very

long and good judgement has to be exercised in order to find a compromise between resolution (particularly in the town area) and computational effort. It should be born in mind the particular stress that IMPACT project puts on urban flooding and that a reasonable description of the flow inside the city would be desired.

4. In order to assess uncertainty, models should be run with the two available bathymetries.

Even though they show considerable differences, specially, regarding the river bed due

ALCRUDO & MULET 25

mainly to the dredging performed during river conditioning, it could cast light into the influences associated with topographic uncertainties. It should be recalled that even the 1982 DTM does not represent the pre-flood bathymetry due to the strong sediment movement that took place during the event.

8 LIST OF FILES AVAILABLE FOR FLOOD PROPAGATION MODELLING

Outflow.xls → Numerical values of Tous Dam outflow hydrograph.

DTM_1982.dat → This file is the 1982 DTM bathymetry (just after the event) prepared by the CESI team, with the format explained in the text, section 3, Valley Bathymetry.

DTM_1998.dat → This file is the 1998 DTM as explained in the text.

Buildings.dat → This file is a listing of the buildings, represented as polygons. The vertices of the polygons and the rooftop elevation are listed as explained in the text (section 3.2, Buildings description)

Town_1982.dat → This is a fine mesh covering the town with the buildings

placed on top of the terrain represented by the 1982 DTM (for this, use has been made of Buildings.dat and DTM_1982.dat files)

Town_1998.dat → This is a fine mesh covering the town with the buildings

placed on top of the terrain represented by the 1998 DTM (for this, use has been made of Buildings.dat and DTM_1998.dat files)

Tous_water_levels.dwg → Autocad file showing the envelope of the shoreline during the event on top of a series of aerial pictures covering the river reach. The picture files are listed below.

1131_tagliata_ridotta.tif 1132_tagliata_ridotta.tif 1134_tagliata_ridotta.tif 1136_tagliata_ridotta.tif Important Notice: The five files listed above MUST be in the same directory in order that the Autocad program can find and display them appropriately. Furthermore their names MUST NOT be changed either.

ALCRUDO & MULET 26

9 TOUS DAM Tous Dam is located in the lower part of the Jucar River Basin and only about 15 km far from de Mediterrenean Sea. Itt constitutes a fundamental piece for the defence against the Jucar River floods. It also serves as the last exploitation step for regulation and irrigation. The first works to build a dam in Tous started in October 1958, with a project of a concrete dam about 80m high above the river bed. During construction, geological conditions of the foundation terrain forced to paralyse the works in December 1964, after location of two faults that delimit the river bed. The works were resumed in April 1974, resorting to a rock-fill dam with clay core between the remaining concrete blocks previously built finished at both sides of the river bed. The old Tous Dam was a rock-fill dam, composed largely of fragmented rock with an impervious core. The core is separated from the rock shells by a series of transition zones built of properly graded material. The rock-fill zones are compacted in layers by heavy rubber-tired or steel-wheel vibratory rollers. Rock-fill dams are generally divided in three parts with very diverse characteristics. The impervious zone, whether inclined or central, which should have sufficient thickness to control through seepage, permit efficient placement with normal hauling and compacting equipment, and minimize effect of differential settlement and possible cracking. The drainage layers, which are a critical part of the embankment design, because it is essential that the individual particles in the foundation and the embankment are held in place and do not move as a result of seepage forces. Finally, the rock-fill zone, which prevents the erosion of the clay core and provide stability to the dam.

Figure 22: Old Tous Dam cross section Figure 21 shows a cross section of the old Tous Dam. The three parts in which it is divided are very clear: The clay core, the drainage layers around the core and finally the rock-fill body that surrounds the other two zones. Figure 4 shows the two phases of the modified project. At

ALCRUDO & MULET 27

the time of the disaster, only the first phase was finished. The crest level of the first phase was set at 98.5m with a Normal Reservoir Operation level at 84m. The notable elevations of the project’s first phase are highlighted to make easier the location of the berms. The berms in the downstream dam’s face are situated at two different elevations, 84.00 and 90.50 m. A top view of the old Tous Dam can be seen in the Figure 22. The surrounding terrain, the concrete blocks in both sides of the dam, the spillway, the berms and some important elevations are easily appreciated in this view.

Figure 23: Old Tous Dam top view Figure 24 shows a detailed view of a standard cross section of the old Tous Dam where the slopes pf each side of the core are indicated. Whereas the slope in the upstream side of the core is uniform, it changes in the downstream side at the elevation of 64 m. In Figure 25 it can be observed that the level of the spillway is 77 m with the sluicegates opened. If the sluice gates are closed, as it happened to be the case during the disaster, the crest level in the spillway is set to 87.5 m, because the sluice gates are 10.5 m high.

ALCRUDO & MULET 28

Figure 24: Detailed view of the old Tous Dam core

Figure 25: Detailed view of the spillway section.

ALCRUDO & MULET 29

10 MATERIALS USED IN THE CONSTRUCTION OF TOUS DAM As it was mentioned before, the old Tous Dam was a rock-fill dam. It was made up by different materials distributed in several areas. The embankment should be zoned to provide an adequate impervious zone, transition zones between the core and the shells, seepage control and stability. Materials used in the different zones of the rock-fill dam are defined by means of their grading and other physical properties. The following sections describe the main characteristics for the rock-fill, the clay and the filters respectively. ROCK-FILL From information obtained directly from staff involved in the works of the old Tous Dam, it can be assumed as a good approximation of the real grading of the rock-fill a uniform distribution between a particle size of 3-5 cm and a particle size of 40-50 cm. Figure 26 shows a plot of the two estimated boundaries of the rock-fill grading. It is supposed that the real grading of the rock-fill is between the two red lines.

Rock-fill grading

0102030405060708090

100

0 5 10 15 20 25 30 35 40 45 50

Particle size (cm)

% in

wei

ght

Figure 26: Rock-fill estimated grading during construction of old Tous dam The estimated values for the specific weight and the friction angle are:

γ (103 Kgr/m3) = 2.65

Φ (º) = 30-40

ALCRUDO & MULET 30

The value for the friction angle is set as the standard value for gravels with similar characteristics. The project requirements for the rock-fill material of the new Tous Dam are given below. The new Tous dam is also a rock fill dam built on top of the remains of the old one. In fact part of the old clay core could be spared for the new dam. The available information regarding the new dam is evidently much more comprehensive than the old one. However, the materials characteristics of the new Tous dam may give some guidance on the value of some materials parameters not available for the old dam. As can be seen the estimated specific weight for the old Tous Dam is a bit larger than the values for the new Tous Dam. The estimated friction angle, on the other hand, is a bit lower than the value for the new Tous Dam. The characteristics for the rock-fill in the new Tous Dam are detailed below:

γdry (103 Kgr/m3) = 2.1-2.2

γwet (103 Kgr/m3) = 2.4

Φ (º) = 45

DLA (%) = 39-44

Elasticity modulus (103 Kgr/m2) = 20000-30000

Poisson modulus = 0.3 where DLA is the Los Angeles wear coefficient. CLAY CORE The physical characteristics for the clay used in the old Tous dam core are detailed below: Modulus number K = 400 Modulus exponent n= 0.8 Poisson modulus = 0.8 Cohesion C (103 Kgr/m2) = 20 Friction angle Φ (º) = 10 Rupture ratio Rf = 0.8 Thrust coefficient at rest Ko = 1 The grading of the clay for the old core can be assumed to be similar as in the new Tous Dam because it was obtained from the same quarry. Figure 27 shows the plot of the project grading limit requirements for the new Tous Dam. The clay grading used during construction must lie between the red lines.

ALCRUDO & MULET 31

Project clay grading requirements

50

60

70

80

90

100

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160

Particle size (mm)

% in

wei

ght

Figure 27: Grading limit requirements The graphic shown in the Figure 27 has been obtained from Table 1 below:

Sieve (ASTM) Sieve size (mm) % in weight

6’’ 152.4 100

3/4’’ 19.05 85-100

Nº 4 4.75 70-100

Nº 40 0.42 65-95

Nº 200 0.075 50-85

Table 1. Grading limits for the old Tous Dam project

The grading characteristics for the clay actually obtained from the quarry are shown in Table 2 below:

GRADING (ASTM STANDARD SIEVE) %

in weight 1’’ (25.4 mm) Nº 4 (4.75 mm) Nº 200 (0.075 mm) < 5 µ

AVERAGE 90.0 84.2 66.3 35.1

STANDARD

DEVIATION

7.4 9.3 11.4 9.8

MAXIMUM 100.0 100.0 97.0 59.0

MINIMUM 51.7 33.9 15.0 3.0

Table 2.Grading clay characteristics

ALCRUDO & MULET 32

The data of the Table 2 can be seen in Figure 28, where the red line represents the maximum limit, the yellow line the minimum limit and the green line the average grading.

Clay grading

0102030405060708090

100

0 2 4 6 8 10 12 14 16 18 20 22 24 26

Particle size (mm)

% in

wei

ght

Figure 28:. Clay grading

As it can be seen from Figure 28, the average clay grading is between the project grading limits for the new Tous Dam. Besides the grading, the clay plasticity must be defined. The liquid limit, the plastic limit and the plasticity index are detailed in Table 3:

PLASTICITY

LIQUID LIMIT PLASTIC LIMIT PLASTICITY INDEX

AVERAGE 29.5 13.5 16.1

STANDARD

DEVIATION

4.0 1.3 3.4

MAXIMUM 50.7 19.6 31.3

MINIMUM 18.8 11.1 4.5

Table 3: Clay plasticity

The available data of the natural humidity of the clay used in the core of the old Tous Dam are detailed in Table 4 below:

NATURAL HUMIDITY (%)

AVERAGE 9.7

STANDARD DEVIATION 1.7

MAXIMUM 13.4

MINIMUM 4.7

Table 4. Clay humiditty

ALCRUDO & MULET 33

And finally, the lower limit for the dry specific weight of the clay is set as: γdry (103 Kgr/m3) = 1.7 FILTERS It can be assumed that the filter materials employed in the old Tous Dam are very similar to the filter materials in the new Tous Dam, which is not far from reality. Actually, this hypothesis is not too relevant for the dam breach modelling because the filter volume is not very large in comparison with the core or the rock-fill. The filter materials represent the basic tool to prevent from internal core erosion. The filters must be arranged in grading transition between the core and the rock-fill zone. Hence the particle size will increase from the core to the outer part of the filter zone. The materials employed in the filters of the new Tous Dam are divided in two types depending on the grading. The available data of the grading for both materials employed in the filters of the new Tous Dam are described in the Table 5.

SIEVE (ASTM

STANDARD)

FILTER

(% in weight)

DRAIN

(% in weight)

3’’ 100

3/4’’ 60-100

Nº 4 100 10-40

Nº 10 70-100 0-17

Nº 20 30-65 0-10

Nº 40 15-35 -

Nº 100 0-15 -

Nº 200 0-5 0-5

Table 5. Filter and drain grading

The filters are situated in the inner part of the filter zone near the clay core whereas the drains are situated in the outer part of the filter zone, in transition to the rock-fill zone. The physical properties for the filters are detailed below: γdry (103 Kgr/m3) = 1.67 γwet (103 Kgr/m3) = 2.1 Φ (º) = 35

ALCRUDO & MULET 34

and for the drains: γdry (103 Kgr/m3) = 1.83 γwet (103 Kgr/m3) = 2.15 Φ (º) = 35 Finally, the elasticity modulus and the Poisson modulus for both materials can be estimated as: Elasticity modulus (103 Kgr/m2) = 8000 Poisson modulus = 0.3 11 HYDRAULIC CHARACTERISTICS OF TOUS DAM AND RESERVOIR This section describes some of the hydraulic characteristics of Tous Dam and Tous Reservoir that may be needed for breach modelling of the failure. Figure 29 below shows the reservoir volume as a function of water level:

0

20

40

60

80

100

120

140

160

180

50 60 70 80 90 100 110

water level (m)

volu

me

(hm

3)

Figure 29: Water volume – Elevation curve for Tous Reservoir Figure 30 represents estimated values of the discharge over Tous dam with sluice gates closed as they were in fact during the failure. It is to be remarked that as far as the water level does not overtop the crest (98.5m), water flows only through the spillway, but not over the gates. Once water level rises above the dam crest, water spills over the entire crest length, plus across the spill way, between its upper bridge and the gates, hence acting now as a submerged

ALCRUDO & MULET 35

structure. These data were obtained from a reduced scale model of the dam. Finally Figure 31 shows recorded water level in Tous reservoir until the dam failed by overtopping.

reservoir rating curve with closed sluicegates

8687888990919293949596979899

100101

0 500 1000 1500 2000 2500 3000 3500 4000 4500

outflow (m3/s)

wat

er le

vel (

m)

Figure 30: Reservoir rating curve with closed sluice gates

WATER LEVEL IN THE TOUS DAM (m)

87.0088.0089.0090.0091.0092.0093.0094.0095.0096.0097.0098.0099.00

100.00101.00

8:00 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00

TIME

WA

TER

LEV

EL (m

)

Figure 31: Water level recorded in Tous Dam during the event

ALCRUDO & MULET 36

If a very refined simulation of the 2-D lake flow is considered, it may be interesting to know the hydrograph of the Escalona River into Tous reservoir. The Escalona River flows into the reservoir just upstream the Tous Dam. The junction of the Escalona River can be seen in the Figure 32.

Tous Dam

Tous Reservoir

Escalona River junction

Figure 32: Situation of the Escalona River mouth.

Figure 33 shows total inflow hydrograph into Tous Reservoir that was estimated using hydrologic methods. It comprises both the inflow coming from the upstream section of Tous reservoir (or tail), plus the contribution due to the Escalona River. In order to accurately model the flow in the lake it would be advisable to subtract the Escalona outflow hydrograph from the Tous inflow hydrograph, and then use both as inflow hydrographs coming from the appropriate points: The Escalona outflow hydrograph should be imposed at the junction of the Escalona river with the Tous reservoir, and the modified (subtracted) Tous inflow hydrograph should be imposed at the tail or upstream section of the Tous reservoir. The corresponding outflow hydrograph from the Escalona River into Tous Reservoir can be seen in the Figure 34.

ALCRUDO & MULET 37

0

2000

4000

6000

8000

10000

12000

20/1

0/19

82 0

:00

20/1

0/19

82 6

:00

20/1

0/19

82 1

2:00

20/1

0/19

82 1

8:00

21/1

0/19

82 0

:00

21/1

0/19

82 6

:00

21/1

0/19

82 1

2:00

21/1

0/19

82 1

8:00

22/1

0/19

82 0

:00

22/1

0/19

82 6

:00

Time

m3/

s

Figure 33: Estimated inflow hydrograph into Tous Reservoir during the flood

ESCALONA RIVER OUTFLOW HYDROGRAPH (m3/s)

0

500

1000

1500

2000

2500

3000

3500

0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00

TIME

OU

TFLO

W (m

3/s)

Figure 34: Escalona River outflow hydrograph

ALCRUDO & MULET 38

12 ACKNOWLEDGEMENTS The authors are grateful to CEDEX and Confederación Hidrográfica del Júcar and in particular to T. Estrela, A. Jiménez and C. Mateos for their help in retrieving and disclosing some material as well as for their explanations and discussions concerning the Tous Dam event. The release of the 1998 DTM owned by CEDEX is very much appreciated. Many thanks are also due to A. García Gumilla from the Local Police of Sumacárcel for his guide during the very informative visit to the town. The authors would like to show their gratitude to J. L. Perales and V. Vilella from Diputación Provincial de Valencia for their help and availability in finding and copying the cartographic plates of the flooded area as well as for their interpretation and explanations on the subject. 13 REFERENCES CEDEX (1984), Estudio hidrológico de la crecida ocurrida en los días 20 y 21 de Octubre de 1982 en la Cuenca del Júcar, Madrid, Spain, 1984. CEDEX (1989), Revisión del hidrológico de la crecida ocurrida en los días 20 y 21 de Octubre de 1982 en la Cuenca del Júcar, Madrid, Spain, October 1989. CEDEX (1989b), Revisión del estudio en modelo matemático de las inundaciones de Octubre de 1982 en la zona del Júcar, Madrid, Spain, November 1989. Ferrer, J., Estrela, T., Jiménez, A. (1999), Plan Global frente a las inundaciones en la Ribera del Júcar: Líneas de actuación. Confederación Hidrográfica del Júcar, 1999. Estrela, T. (1999), Hydraulic Modelling of the Tous Dam Break, Fourth CADAM (Concerted Action in Dam Break Modelling) Workshop, Zaragoza, Spain, November 1999. Estrela, T. y L. Quintas (1996), El modelo de flujo bidimensional GISPLANA. Revista de Ingeniería Civil. Vol 104. Páginas 13-21. Spain 1996. Utrillas, J.L., (2000), La presa de Tous. Ed. MOPTyMA. Spain