Hydrodynamic Loadings of Buildings in Floods

8/2/2019 Hydrodynamic Loadings of Buildings in Floods

1/13

HYDRODYNAMIC LOADINGS OF BUILDINGS IN FLOODSGiovanni Cuomo1, Gholamreza Shams2, Sebastian Jonkman3 andPieter van Gelder4

Assessing the vulnerability of buildings in flood-prone areas is a key issue whenevaluating the risk induced by flood events, particularly because of its proved directinfluence on the loss of life during catastrophes. A comprehensive methodology for riskassessment of buildings subject to flooding is nevertheless still missing. Bearing this inmind, a new set of experiments have been performed at TU Delft with the aim ofspreading more light on dynamics of flood-induced loads and their effects on buildingsand to provide the CDF community with state of the art bench-marks. In this paper, abriefly overview is given of flood induced load on b uildings; the new experim ental workis then presented, together with results from preliminary analysis. Initial results suggestthat use of existing prediction me thods might be unsafe and that impulsive loading m ightbe critical for both the assessment of the vulnerability of existing structures and thedesign of new flood-proof buildings.

INTRODUCTIONRecent catastrophic events like the Indian Ocean tsunami in 2004 (Kawata,2005) and the New Orleans flood due to passage of hurricane Katrina in 2005(FEMA 2006, Figure 1) have focused attention of politicians, economists andengineers worldwide on the need for the assessment of risk due to flooding ofanthropized areas when estimating the potential life and economic losses of acatastrophic scenario.

. A / L M i B ' W f t *

Figure 1. Percentage of building damage in the Lower Ninth Ward, New Orleans(main figure) and fatality rate (top-right).

Senior Engineer, Engineering Hydraulic and Structures Group, HR Wallingford, HowberyPark, Wallingford, OX10 8BA, UK; q.cuomoO.hrwallinqford.co.uk; Post-Doc, HydraulicEngineering Dept., TU Delft' PhD candidate, Hydraulic Engineering Dept, TU Delft, R.Shams(5),tudelft.nl1 PhD, Hydraulic Engineering Dept., TU Delft, S.N.JonkmanO.tudelft.nl1 As s. Professor, Hydraulic Engineering Dept, TU Delft P.H.A.J.M.vanG elder.tudelft.nl

3744
http://rwallinqford.co.uk/mailto:[email protected]:[email protected]://rwallinqford.co.uk/


2/13

COASTAL ENGINEERING 2008 3745

While a huge effort is being devoted in the past to the development ofmethodology for the definition of the hazard due to flooding (including flood-ma pping of riverine and coastal areas), little is known on the dynam ics of flood-induced loads on buildings (Kelman and Spence, 2002) and a comprehensiveme thodology for risk assessment of buildings subject to flooding is still missing.In particular, new tools are needed for the evaluation of the fragility of civilstructures to hydraulic loading.In general the structural vulnerability of buildings in flood flows is modeleddependent on the combination of water depth and flow velocity. Criteria havebeen derived based on data from historical floods, (see e.g. Clausen and Clark,1990), but their empirical validation is relatively limited. The need to account forthe large amount of uncertainties involved with such processes is motivating therecent orientation toward probabilistic approaches to risk assessment over themore established and simple deterministic ones. Bearing this in mind, acollaborative research project is being carried out with the aim of improvingknowledge on hydrodynanic loading of buildings in flood-prone areas.

FLOOD LOADS ON BUILINGSStructural fragility to buildings depends on both the demand (the loading)and the resilience (capacity) of the structure. This work focuses on the flood-induced loading on structures. During extreme events, building laying infloodplains within high flood-induced hazard areas can be subject to a series ofloads including both hy drostatic and hydrodynam ic loads (Figure 2).


3/13

3746 COASTAL ENG INEERING 2008

Hydrostatic loads include:hydrostatic pressure on the vertical element of the structure;buoya ncy on the horizontal element of the structure;Hydrodynam ic loads include:Drag/Velocity-dominated (quasi-static) loads, including:o Flash floods (away from source banks/levee/breaches);o Debris-flows (away from source);o Storm surges (inland)o Tsunami (inland)o Non -breaking wave loadsInertia/acceleration-dom inated (impulsive) loads;

o Dam break, close to sourceo Debris-flow, close to sourceo Flash floods in the vicinity of banks/levee/breaches;o Storm surges (along the coast)o Tsunam i (along the cost and inland at the wa ve front)o Breaking wave loadsThe above classes are described in detail in what that follows. Here, it isworth mentioning that in most cases, different loads types co-exist and act on thesame structure at the same time, or at a different times during a single theflooding event.Hydrostatic loads

The hydrostatic horizontal load ( Fh static) derives from the difference inwater level on the upstream and the dow nstream sides of the wall. Per unit lengthit is given by:

^static = PS 12 \hu s-h js) (1)In which g is the acceleration due to gravity, hus and h& are the water depthsrespectively upstream and downstream the wall and p is the density of the fluid,which is a function of the amount of solid particles suspended w ithin the flood.When assessing the vulnerability of horizontal structural elements or theoverall stability of a building, buoyancy should also taken into account as itapplies a potentially unbalanced uplift force and affects the resistance of gravity-based structures against sliding and overturning. Buoyancy (per unit length) canbe easily estimated using:

Fy Static = W * \PSKp ~ PSKown ) ( 2 )In which g is the acceleration due to gravity, pghup and pghdown are thehydrostatic heads acting respectively above and below the horizontal elementconsidered and w is the structural element's width
http://pskp/http://pskp/http://pskp/


4/13

COASTAL ENGINEERING 2008 3747Hydrodynamic loadsHydrodynamic loads derivefrom he combination of the inertia and drag andgenerally depend on both the kinematics of the flow and the geometrical anddynamic characteristics of structure. An example force time-history recordedduring dam break experiment presented in this paper is shown in Figure 3showing a sharp (inertia-dominated) peak (Fmax) followed by a less intense butlonger lasting (quasi-static, drag-dominated) load (Fqs+).

' m a x i f i i i i

*


5/13

3748 COASTAL ENGINEERING 2008

than Eq. 3, Eq. 4 has been shown to underestimate most violent impulsiveloading events (Cuomo et al. 2003).Dam-break and tsunam i-induced loads on buildings.Recent catastrophic tsunam i events have focused the attention of researchersworldwide on the need to assess hydrodynamic loads exerted by tsunami onbuildings. Among these, experiments by Arnarson (2005) represent rareexam ples of tests concentrated on m easuring loads exerted by dam-break flow onisolated buildings. Results from the above experiments have been included indesign standards including the recently published by FEMA (2008) in which thehydrodynam ic force acting on an isolated building reached by a tsunami is givenby:

Fh=0.5pC D-b-h-u2 (5)The autho rs of the FEM A guidelines suggest using a drag coefficient CD = 2when assessing the drag component of the load only but assuming C D = 3 toaccount for the impulsive com ponen t of the loading, which implies:

Fh,imp = ^^ Fh,drag ( 6 )

Short wave-induced loadsWhen assessing hydrodynamic loads for use in design, short wave loads(period Is < T < 10s) on buildings are usually accounted for separately. In thecommon practice, they are usually subdivided into loads exerted by nonbreaking, breaking and broken waves, the transition among the above coditionsbeing mainly a function of the local wave height (H) to water depth (h) ratio. Itshould nevertheless be borne in mind that for walls exposed to wave loading,both the incident wave condition and the water depth in front of the structuremight vary during the flood and it is therefore uneasy to define boundaries foreach loading condition.A series of well established prediction methods for non-breaking wave loadsare available to practitioners. Among these, Goda's (1974) prediction method forwave loads on caisson breakwaters represents a landmark in the evolution ofphysically rational approaches to the assessment of wave loads at walls.Takahashi et al. (1994) extended Goda's prediction method to include the effectof wave breaking. Within the PROVERBS research project, a new designmethods was derived for the evaluation of wave forces on caisson breakwaters.described in details in Oumerac i et al. (2001) .

Since breaking is highly dissipative, force exerted by broken waves areassumed to be significantly less intense then for the breaking wave case. It iscommonly assumed that after breaking, the wave propagates like a bore inshoreand in-land, eventually transferring their residual kinetic energy to the structureat time of im pact (see Ramsden, 1993 and references therein).


6/13


Debr is - induced loadsDebris-induced loads represent a kind of their own, both for their high levelof unpredictability and their extremely high dangerousness. Among the rarestudies carried out in this field, we will mention the recent and accurate workcarried out by Haehnel and Daly (2002).Load combina t ionAs already mentioned, in most cases a structure is subject to a combinationof the loadings described above. En example calculation showing differentdegree of load combination acting on a vertical wall is given in Figure 4, formleft to right and from top to bottom :

hydrostatic pressure;hydrostatic pressure + drag;hydrostatic pressure + wav es;hydrostatic pressure + drag + waves;

Figure 4. Example calculation showing the effect of the hydrostatic pressure (top-left) alone, and in com bination w ith a steady current (top-right), waves (bottom-left),and both (bottom-riaht).Note that during a single flood event, the load felt by the wall might varysignificantly within the sketched areas in Figure 4. For example, as the waterreaches the base of the wall and start rising, the wall will experience increasinghydrostatic pressure and drag force due to the current. If the structure is set in a

coastal area, wa ves might as well reach the wall which will therefore experienceloading from broken, breaking or pulsating wav es as a function of the local wavehigh to water depth ratio, which can vary due to tide and storm surge variationduring a single flood.In other cases tsunamis, flash floods and dam break flows might drag alongheavy objects as debris, with catastrophic effects on any structure laying alongthe path.


7/13

3750 COASTAL ENGINEERING 2008NEW PHYSICAL MODEL TESTSA new set of dam-break physical model tests were carried out in one of thewave-flumes of the Laboratory for Fluid Mechanics of the Delft University ofTechnology to study the dynamics of flood-induced loads and their effects onbuildings and to provide the CD F comm unity with state of the art benchm arks.Experimental set-up and measurement instrumentsExperiments were carried out in a 42m long, 0.8m wide, 1.0m high waveflume. A water column was stored in a 2.88m long, 0.67m wide and 1.26m highreservoir. To minimize the effect of the gate opening on the water tonguepropagating downstream, the gate was operated by rotating a stiff metal platearound its top by pulling its bottom by means of a free-falling counterweight.The building was represented by a cubic structure (characteristic lineardimension of 18cm) and made out of thick aluminum to minimise its weight andmaximise its stiffness (dynamics of model structure is discussed further in thefollowing). The following sensors were housed in the model structure:1 accelerator;

I axial force transducer;II pressure sensors.In order to be able to effectively capture the sharpest im pulsive events, datawere logged continuously at high frequency (50KHz), recording was initiated bymonitoring the movem ent of the gate.

Figure 5. Sketch of model structure housing force transducer and acceleration (left)and pressure transducers (right).Experimental programThe experimental program covered a number of layouts and geometricalconfigurations, including (Figure 6):

Single building layout (left): aimed to investigating the effect ofbuilding o rientation and distance from the source;


8/13


City layout (top-right): aimed to investigate complex flow pattern andlocal amplification in urban environmentDebris layout (bottom-right): aimed to investigate the dynamics of thewater-debris-structure interaction.

Figure 6. Model layouts used in the experiments: "single building layout" (left),layout" (top-right) and "debris layout" (bottom-right, dots represent the debris). 'city

Tested param eters included:Building orientation with respect to the main flow,Water level in the reservoir;Distance of the building from the gate;Density and volume of debris;Relative position within the "urban" environmentand breach opening.Each test was repeated three times for quality checking and to enhance thesignificance of the tests.

Dyna mics of the model s t ruc tureWhen investigating impulsive loads, dynamics of model structure andmeasurement instruments plays a fundamental role . It's k nown from the literaturethat to avoid distortion of measured load, it's essential that the experimentalsetup is not significantly dynamically responding within the range of frequencywithin which the loading belongs to. Although in principio easilyunderstandable, a dynamically neutral measuring system is impossible to setup,and the task becom es particularly difficult when the loads to be m easured have avery short duration O(10"2s). Special attention has been therefore paid during thedesign and construction of the experimental setup to obtain the shortest possibleresonance period. A numerical model of the experimental setup was used duringthe design to ensure that the dynamic response of the model structure wouldn'thave significantly affected the measurements.

Results from the num erical mo delling indicated that a very thick wall shouldhave been used in order to reduce oscillation periods below the range offrequency being targeted. T his need w as nevertheless in opposition with the willof saving space inside the building to house an many sensors as possible in themodel structure.


9/13

3 7 5 2 COASTAL ENGINEERING 2008

The optimum wall thickness was 15mm, the shapes of the correspondingfirst 3 modes of vibration of the model structure (instrumented in the pressuresensors) are shown in Fig. 7, corresponding period of vibration were Tn


10/13

COASTAL ENGINEERING 2008 3753INITIAL RESULTSIn the following, initial results from preliminary analysis performed on time

histories recorded during the first set of the "single building layout" tests arepresented. Data recorded during tests carried out with the structure located closeto the gate (Fig. 6) only are presented in terms of effect of tested parameters andbuilding orientation with respect to the direction of the main flow.3 0 0

2 50

2 0 0

>I S 0

100

5 0

1 T I 1 T ~ > ^l i l t \s^I I I I ^C ^ ^I T I I - ~~f \j^ ~1 , 1 1 S / Ai j_ i y_ y_ Ti i i ~/\'f ~ ~i i i / / ii l YJ^-- - - - =.i i^cll ^>~* '

4 00

| 30020 0

1 00

1 I 1 I ^11 1 I I / \ ^ - I I I 1/ yT\i i i A / i1 I 1 / rf \l ' i / A 'i i i / / 1 i1 i ~/~/~ I ~^"^l i / y ' ^ * ^ ^! l_ y M _^KT I

6 0H [ a n ]6 0

Figure 9. Effect of water level in the reservoir and angle of inclination of the buildingwith respect to the main flow direction (0 black circles, 30 triangles, 60 circles) onquasi-static (left) and impulsive (right) loads. Best fit lines obey Equation 7.Quasi-static a nd impu lsive loadsQuasi-static loads (Fqs+) and impulsive loads ( F ^ ) on the front face of thebuilding are plotted respectively on the left hand side of Figure 9 as a function ofthe water level in the reservoir (H), showing an increasing non-linear trend withlinearly increasing H over the whole range of parameter tested. Best fit curves inFigure 9 (over the range of parameter tested) obey the expression:

Fh=a- exp (H-b^2V c J

(7 )

with a, b , c being best-fit empirical parameters summarised in Table 1, togetherwith R-square of the corresponding fit.Table 1. Empirical parameters for horizontal forces in Eq 7, as a function of struc tureorientation with respect to the main flowLoadFas+FQS+FQS+F1 i m pF-1 im DF-1 im D

Orienta t ion0

30600

3060

a254222126535441233

b127131169115119138

c6870101535470

R 20.990.990.990.990.990.99


11/13


The ratio of the impulsive to the quasi-static loads on the front face of thebuilding is plotted in Figure 10 for increasing values of H. Best fit curves inFigure 10 obey the expression:FhMp={a-H'+b-H 2+c-H + d)-Fh ,qs+ (8 )

where a = -4.8 10'6, b = 9.45 10"4 , c = -4.2 10"2 and d = 1.79 are best-fitempirical parameters, Equation 8 is valid in the range 20cm < H < 120cm andhas an overall R2 = 0.795.For increasing H, the impulsiveness of the loading Fimp/Fqs+ initiallyincreases up to reach a maximum of 2.5+ at H = 100cm, but then decreases. Thisis probably due to the fact that for water levels exceeding lm, the flux has notdeveloped (accelerated) enough at location of the building to become critical forthe stability of the building.

2.5

2

%. 1.5a

l0.5

!it

i

"" ~ 'Ii

T ~1

Iii

I I I !

I i X -" " " A - ~" " >

| , ' , l lJ - - - - 1 t 1 i

i t l Il i i I

l l I l

60H[ori]Figure 10. Impulsiveness of the loading Fimp/Fqs+ as a function of the water level atrest in the reservoir. Best fit solid line obey Equation 8, 95% confidence bounds arealso shown (dashed) together with prediction by Equation 6 (dashed-dotted)Effect of building orientation

Effect of building orientation with respect with the main flow direction onquasi-static and impulsive component of the loading is shown respectively on theleft and right hand side of Fig. 9, as expected, the loading decreases withincreasing angle of inclination of the exposed face to the main flow drection.Com parison with previous findingsWhen compared with results from previous researchers, impulsive loadingmeasured during the new set of experiments appear to be significantly higherthan their corresponding quasi-static loads. In particular, Figure 10 showsimpulsiveness ratio up to more than 2.5, confirming limitations inrecommendations derived from previous studies (Eq. 6, also shown in Figure asan horizontal line for comparison).Discrepancies between observation during the present tests and those carriedout by Arnarson (2005) and Nouri et al. (2007) are probably due to thedifference in the dynamics of the experimental setup used in each case. Indeed,


12/13

COASTAL ENGINEERING 2008 3755looking at time-histories shown by previous authors, it appears likely that theexperimental setup used in previous experimental work could have damped outmost intense impacts as they were acting over a range of frequency higher thanthose corresponding to the natural frequency of vibration of the model structure.In the present experiments, since the model structure had very fast response, anyrecorded load was acting over a range of frequencies significantly lower thanthose correspo nding to the natural frequencies of the mode l structure, resulting inthe load being felt as "quasi-static" by the experimental setup-up and hence in aminimum distortion of recorded signal.FURTHER INSIGHTS ON LOAD TIME-HYSTORYFurther insight on the relative importance of impulsive loading on buildingscan be derived by looking at pressure time-histories recorded during testing.Impulsive pressures up to 5.5 times their corresponding quasi-static com ponentsare plotted on the left hand side of Figure 11. M ost energetic co mp onentfrequencies identified by wavelet transform of an example pressure time-historyrecorded during testing (right hand side of Figure 11) fall within the range of thenatural periods of vibration of multi-story buildings and their non-structuralelements, suggesting that impulsive loading might become critic when thevulnerability of building in flood-prone areas has to be assessed.

n u n iJtonASC

-

Figure 11. Example pressure time-history of a particularly impulsive event recordedduring present experiments (left). Amplitude contour plot (bottom right) of thewave let transform of a different exa mple pressure time-history (top right).INITIAL CONCLUSIONS AND FURTHER WORKAssessing the vulnerability of buildings in flood-prone areas is a key issuewhen evaluating the risk induced by flood events, especially for his strongcorrelation with loss of life during most catastrophic events.Bearing this in mind, a new set of experiments have been performed at TUDelft with the aim of spreading more light on dynamics of flood-induced loadsand their effects on buildings and to provide the CDF community with state ofthe art bench-marks. Initial results suggest that impulsive loading might besignificantly higher than those predicted by available prediction methods and


13/13


should be regarded as potentially critical when assessing the vutaerability ofexisting structure and designing flood-proof bu ildings.ACKNOWLEDGMENTSThis work is part of a research project funded by BC Hydro under contractGOCE-CT-2004-505420 and FLOODsite under contract C71B44. Prof. W .Uijttewaal and Prof. H. Vrijling (TU Delft) are thanked for the many interestingdiscussion and suggestions. The authors are thankful to Aimilia Pistrika, S. deVree and A. den Toom for their precious help during testing.REFERENCESArnason, H. (2005) "Interactions between an incident bore and a free-standingcoastal structure" PhD The sis, University of Washing ton, 172 pagesClausen, L. and Clark, P.B. (1990) "The development of criteria for predictingdambreak flood damages using modeling of historical dam failures" Int.

Conf. on River Flood Hydraulics, pp. 369-380Cuomo G., Allsop N.W.H. and McConnell K. (2003) "Dynamic Wave Loads onCoastal Structures: Analysis of Impulsive and Pulsating Wave Loads" Proc.of Coastal Structures 2003 Conference, Portland, Oregon, USA.FEM A 549 (2006) "Hurricane Katrina in the Gulf Coast. Building PerformanceObservations, Recommendations, and Technical Guidan ce"FEMA P646 (2008) "Guidelines for Design of Structures for VerticalEvacuation from Tsunamis"Goda, Y . (1974) "New w ave pressure formulae for com posite breakwater" Proc.of 14th ICCE, Copenhagen, Denmark, ppl7 02-1 720 . ASC E New Y ork,Haehnel R. H. and Daly S. F. (2002) "M aximum Im pact Force of Woody D ebrison Floodplain Structures", US Army Corps of Engineers, Cold RegionsResearch and E ngineering Laboratory, Report ERDC/CRRE L TR-02-2Kelman I. and Spence R. (2002) "An overview of flood actions on buildings"Engineering Geology 73 , 297-309Kawata Y. et al. (2005) "Comprehensive analysis of the damage and its impacton coastal zones by the Indian Ocean tsunami disaster"Nouri Y., Nistor Y., Palermo D. and Saatcioglu M. (2007) "Structural analysisfor tsunami-induced force and debris impact" Proc. of Coastal Structures2007 Conference, Venice, Italy (in press).Oumeraci, H., Kortenhaus, A., Allsop, N.W.H., De Groot, M.B., Crouch, R.S.,Vrijling, J.K. & Voortman, H.G. (2001) "Probabilistic Design Tools forVertical Breakwaters.",Balkema, Rotterdam (392 pp).

Ramsden, J.D. (1993) "Tsunamis: Forces on a vertical wall caused by longwaves, bores and surges on a dry bed" PhD thesis, California Institute ofTechnology, Pasadena, California, Report No. KH-R-54Takahashi, S., Tanimoto, K. and Shimosako, K. (1994) "A Proposal ofImpulsive Pressure Coefficient for the Design of Composite Breakwaters"Proc. of Hydro-Port '94, Port and Harbour Research Institute, pp 489-504

Hydrodynamic Loadings of Buildings in Floods

Documents

Transcript of Hydrodynamic Loadings of Buildings in Floods