handout benchmark v1 all

8/9/2019 Handout Benchmark V1 ALL

1/48

12TH INTERNATIONAL

BENCHMARK WORKSHOP

ON NUMERICAL ANALYSIS

OF DAMS

International

Commission

on

Large

Dams

2nd ‐ 4th October 2013

Graz – Austria

THEME A: FLUID STRUCTURE INTERACTION ARCH DAM ‐ RESERVOIR AT SEISMIC LOADING

THEME B: LONG TERM BEHAVIOR OF ROCKFILL DAMS

THEME C: COMPUTATIONAL CHALLENGES IN CONSEQUENCE ESTIMATION FOR RISK ASSESSMENT

OPEN THEME: CONTRIBUTIONS TO ANALYSES OF DAMS


2/48

12th INTERNATIONAL BENCHMARK WORKSHOP ON NUMERICAL ANALYSIS OF DAMS

General Organizer

Committee on Computational Aspects of Analysis and Design of Dam

Chairman

I. Escuder-Bueno Spain

Vize-Chairman G. Mazza Italy

Technical Advisory Team

RESTELLI, F. ArgentinaZENZ, G. Austria

CURTIS, D. CanadaCHEN, S. ChinaMARULANDA, C. ColombiaVARPASUO, P. FinlandTANCEV, L. Form. Yug. Rep. of MacedoniaFROSSARD, E. France

BEETZ, U. GermanyDAKOULAS, P. Greece

NOORZAD, A. IranMEGHELLA, M. ItalyUCHITA, Y. JapanANDERSEN, R. NorwayPOPOVICI, A. RomaniaGLAGOVSKY, V. RussiaMINARIK, M. Slovakia

HASSANZADEH, M. SwedenGUNN, R. SwitzerlandMATHEU, E. United StatesCARRERE, A. (HONORARY MEMBER) FranceFANELLI, M. (HONORARY MEMBER) Italy

Technical Organizer Graz University of Technology Institute for Hydraulic Engineering and Water Resources Management Stremayrgasse 10/II, 8010 Graz, Austria www.hydro.tugraz.at

Contact Univ. Prof. Dipl. Ing. Dr. techn. Gerald Zenz [email protected] Dipl. Ing. Markus Goldgruber

[email protected]

12TH

INTERNATIONAL

BENCHMARK

WORKSHOP

ON

NUMERICAL

ANALYSIS

OF

DAMS


3/48

Introduction to Theme AGerd‐Jan Schreppers

TNO DIANA

Georgios MaltidisFederal Waterways Engineering and Research Institute ‐ Karlsruhe

Giorgia FaggianiRSE ‐ Italy

Shervin ShahriariGraz University of Technology

Richard MalmKTH Royal Institute of Technology

10:00 ‐ 10:30 Coffee Break

Anton D. TzenkovStucky SA

Marion ChambartStucky SA

Adrian PopoviciTechnical University of Civil Engineering Bucharest

Marco BrusinAF Consult Switzerland

Antonella FrigerioRSE Italy

Abdoul Diallo

EDF – CIH

Discussion ‐ Result Comparison

12:00 ‐ 13:30

Panos Dakoulas ‐ Seismic Analysis of a Concrete Arch DamUniversity of Thessaly

Herman B. Smith ‐ Earthquake Assessment of Slab and Buttress DamsNorplan AS

Øystein Eltervaag ‐ Influence of Surface Roughness on Sliding Stability

Norwegian University

of

Engineering

Science

and

Technology

Camilo MarulandaIngetec

Etienne FrossardTractebel Engineering SA

Ignacio Escuder BuenoUniversidad Politécnica de Valencia



Lunch

12th International Benchmark Workshop on Numerical Analysis of Dams

Final Program

07:30 ‐

08:30

ICOLD Committee

Meeting

08:30 ‐ 10:00

10:30 ‐ 12:00

Wednesday 2nd

October

Registration

13:30 ‐ 15:30

16:00 ‐

17:00

18:00 ‐ 21:00 Social Meeting (Aula TU Graz)


4/48

Brantley ThamesUSACE

Mark JourdanU.S. Army Engineer Research and Development Center

Phil DyeUSACE

Omid SaberiGraz University of Technology


Mark Davison

HR Wallingford

Mustafa AltinakarUniversity of Mississippi

Leonardo MancusiRSE ‐ Italy


12:00 ‐ 13:30

Violeta

Mircevska ‐

Dam‐

fluid

interaction

using

ADAD‐

IZIIS

softwareIZIIS‐Skopje, Macedonia

Ignacio Escuder Bueno ‐ Need for transient thermal models, with daily inputs,

to explain the displacements of arch‐gravity damsUniversidad Politécnica de Valencia

Sujan Malla ‐ EQ safety assessment of arch dams based on non‐linear dynamic

analysisAxpo Power AG

Emmanuel Robbe ‐ Behaviour of an arch dam under the influence of creep,

AAR and opening of the dam/foundation contact EDF ‐ CIH

Bachir Touileb ‐ Long Term Deformations of a CFRD Dam due to Annual

Emptying and Filling CyclesGroupe Artelia

Guido Mazza ‐ The rehabilitation of Beauregard Dam: the contribution of the

numerical modeling RSE Italy

Gerald Zenz ‐ The Koelnbrein Dam (Excursion)TU Graz

15:30 ‐ 16:00

16:00 ‐

17:00 Farewell

13:30 ‐ 15:30

12th International Benchmark Workshop on Numerical Analysis of Dams

Final Program

Thursday 3rd

October

09:00

07:30 ‐

09:00 Registration

Lunch

‐ 10:00

10:30 ‐ 12:00


5/48


6/48


7/48


THEME A:

FLUID STRUCTURE INTERACTION

ARCH DAM – RESERVOIR AT

SEISMIC LOADING


8/48

HFTD Analysis of an arched dam at seismic loading

W. Kikstra1, F. Sirumbal

1 and G. Schreppers

1

1 TNO DIANA BV, Delftechpark 19a, 2628 XJ Delft, NETHERLANDS

E-mail: [email protected]

Abstract

Hybrid Frequency-Time Domain (HFTD) method is applied to analyze the response of a dam-

foundation-reservoir system. With this method the effect of frequency dependent properties

such as compressibility of fluid, reservoir-bottom absorption and far-field reflection can be

considered. HFTD results are compared with transient Newmark-type results and effect of

frequency dependent properties is found to reduce amplitudes and stresses in the dam for the

chosen bottom absorption in the reservoir.

Conclusion

The given foundation-dam-reservoir system was analyzed for the full duration of the

earthquake. With the HFTD method implemented in the standard version of DIANA for the

case of frequency independent properties the same results could be reproduced as with

implicit time stepping with Newmark’s method. With HFTD the effect of frequency

dependent properties such as compressibility of fluid, reservoir bottom absorption and infinite

extend reflection have been analyzed and quantified, resulting in an interesting method to be

applied specially in the dynamic analysis of dam-reservoir interaction models.


9/48

Fluid Structure InteractionArch Dam – Reservoir at Seismic Loading

G. Maltidis1

1 Federal Waterways Engineering and Research Institute, Kußmaulstrasse 17, 76187Karlsruhe, GERMANY

Abstract

The fluid structure interaction is an important issue that must be taken into account for the

analysis and design of hydraulic structures. Since the first attempts to calculate the

hydrodynamic pressures on structures analytically (Westergaard, von Kármán, Mononobe,

Housner, Chwang, Zangar) the engineers and researchers have the last years a very useful

tool, the finite and boundary element method, in order to analyze complicated structures

taking into account different sophisticated phenomena. However, even nowadays, the

common praxis is to use the early developed techniques, because of their simplicity andcapability of implementation in the most finite element codes.

Conclusion

The earthquake analysis of an arch dam-reservoir-foundation system was performed with

different modeling aspects. The results show the almost identical values for the two added

mass approaches, with the one of Zangar to be a little bit more favorable than the one of

Westergaard. The acoustic elements models give generally lower values of the stresses due to

reservoir hydrodynamic loading.

It is remarkable that the maximal or the minimal values can differ up to 5 MPa for the same

mesh model and the stress shape can differ comparing the model with the two meshes.


10/48

Finite Element Modelling of Seismic Fluid-Structure

Interaction for a large Arch Dam

G. Faggiani1 and P. Masarati

1

1 Ricerca sul Sistema Energetico - RSE SpA, via R. Rubattino 54, 20134 Milano, ITALY


Abstract

The linear dynamic fluid-structure interaction at seismic loading for the artificially generated

large arch dam provided for the Theme A was modelled using the approach of acoustic

compressible elements, with both the coarse and fine meshes provided by the formulators of

the 12th ICOLD Benchmark Workshop.

The effects of incompressible fluid (theoretical hypothesis of the added mass models) and of

the partial absorption of the hydrodynamic pressure waves at the reservoir boundary (bottomand sides) were also investigated.

Simulations were carried out using the RSE in-house FEM code CANT-SD, specifically

designed for dynamic linear and non-linear analyses of dam-reservoir systems.

The coarse and fine meshes showed not dissimilar results, except at dam-foundation interface.

The results of the analyses confirmed that incompressible models could result relatively

conservative and highlighted the benefits of the approach of acoustic elements, mainly the

possibility to take into account the damping effect on the fluid boundary.

Conclusion

Theme A of the 12

th

ICOLD Benchmark Workshop on Numerical Analysis of Dams, dealingwith the linear dynamic fluid-structure interaction at seismic loading, has been approached by

using CANT-SD, a RSE in-house FEM code for dynamic linear and non-linear analyses of

dam-reservoir systems. The dynamic fluid-structure interaction was modelled with the

approach of compressible acoustic elements.

The analyses with the reflecting boundary condition on the bottom and sides of the fluid

domain, performed with both coarse and fine meshes to test the effects of different spatial

discretization, showed that the use of a more refined mesh does not lead to significant

differences in the results, though involving quite a higher computational effort.

Simulations considering incompressible acoustic elements or reservoir boundary absorption

were also performed: the results of these analyses confirmed that the incompressible model

could result relatively conservative and that reservoir boundary absorption could significantlyreduce the earthquake response of the dam. The use of an incompressible model instead of a

compressible one, capable to account for damping effect on the fluid boundary too, speeded

up the transition from a monolithic structural scheme to a different one with independent

cantilevers and joints.


11/48

Fluid Structure InteractionArch Dam-Reservoir at Seismic Loading

S. Shahriari1

1 Institute of Hydraulic Engineering and Water Resources Management, Stremayrgasse 10/2,A-8010 Graz, AUSTRIA


Abstract

In the present paper, a direct time domain procedure is used for dynamic linear analysis of the

coupled arch dam-reservoir-foundation system. The hydrodynamic force on the upstream face

of the dam is modeled by added mass method and compressible fluid elements for

comparison. The concrete and massless foundation rock was assumed to be linear elastic.

Connection between dam and foundation is modeled by coupling all DOF of the

corresponding nodes. Also, viscous damping was applied to the materials by using Rayleigh-

Damping. Numerical results showed that the different modeling techniques of the interaction

lead to different response of the system and stress distribution in the dam’s body. Modeling

the interaction by added mass method, increases the period of the system as well as

overestimating or underestimating the stresses compare to the model with fluid elements.

Conclusion

The following conclusions are drawn based on the numerical experiments conducted herein:

Choosing appropriate damping parameters is a very important step in the dynamic

analysis of the dam-water-foundation system. Underestimating the system’s damping

by choosing higher than 7th angular frequency as the second frequency for calculation

of Rayleigh-Damping parameters can lead to overestimating the stresses in the model.

Modal responses of the dam-water-foundation system are calculated using the finite

element software, ANSYS. The results showed that the eigenfrequencies from the

model with added mass approach were lower than those calculated from the model

which utilized acoustic element to model the reservoir.

Direct time integration procedure is used for dynamic analysis of the system and

deformations and stresses are calculated for static and dynamic load cases. Maximumdeformation due to static loading was 7.4 cm at the crest level followed by the

maximum deformation of 13.6 cm after dynamic analysis. The tensions in arch dams

are not desirable; therefore, the tension stresses were studied. Evaluation of hoop

stresses indicated that there is no significant tension developed in the main section

with maximum value of 0.6 MPa. Furthermore, the maximum tension stress value in

the U/S face of the right and left sections is higher than the main section(S=3.6 MPa).

Vertical stresses evaluation showed that the maximum tension was occurred at the

base of the U/S face of the main section and found to be 6.69 MPa. Also, the

maximum compression stress is developed (S=12.5 MPa) at 180 m from the

foundation level of the U/S face of the main section.


12/48

Concrete arch dam at seismic loading with fluid

structure interaction

R. Malm1,2

, C. Pi Rito2, M. Hassanzadeh

3,4, C. Rydell

1,3 and T. Gasch

1,3

1 KTH Royal Institute of Technology, Concrete Structures, Stockholm, Sweden.2 SWECO Infrastructure, Hydropower and Dams, Stockholm, Sweden

3 Vattenfall Engineering, Stockholm, Sweden.4 Lund University, Building Materials, Lund, Sweden.


Abstract

A concrete arch dam have been analyzed during seismic loading with a model based on

acoustic elements to describe the water and infinite elements as quiet boundaries to prevent

wave reflection. The results have also been compared with a simplified model based onWestergaards added mass approach. The simplified model is only used, in this study, for

comparison with the more advanced model with acoustic elements. Therefore the results from

this simplified model are just used as a rough estimate of the induced stresses and

displacements. Despite this, the simplified Westergaard model gives similar results compared

to the more advanced model with acoustic elements for the water and infinite elements for the

boundaries. The largest difference between the models often occurs at the nodes in the base of

the arch dam, which may be due to poor discretization. Generally, the Westergaard added

mass gives higher maximum principal stresses at the base on the upstream side than the

acoustic model, while often underestimating the min principal stresses at the base on the

downstream side. Both models show high tensile stresses near the base of the arch dam that

would result in cracks.

Discussion

A concrete arch dam have been analyzed during seismic loading with a model based on

acoustic elements to describe the water and infinite elements to describe quiet boundaries to

prevent wave reflection. The results from this model have also been compared to a simplified

model based on Westergaards added mass approach. The simplified model is only used for

comparison and should therefore be considered as a rough estimate of the induced stresses

and displacements. The performed analyses have showed that there are several factors that

have been assumed in this study that influence the results, such as

Rayleigh damping – the choice of damping ratio for the different materials and alsochoice of corresponding frequencies influence the results.

Seismic excitation – in these analyses, only the bottom of the foundation is subjected

to the prescribed seismic excitation. Effects such as seismic wave incoherence have

not been considered.

Infinite boundaries – in the analyses infinite boundaries have defined for both the

foundation but also for the reservoir.

The results from the two models are quite similar, where the minimum principal and hoop

stresses generally are more similar between the two models than the maximum principal and

vertical stresses. The largest difference between the models often occurs at the nodes in the

base of the arch dam, which may be due to poor discretization of the mesh at these points.

Generally, the Westergaard added mass gives higher maximum principal stresses at the base


13/48

on the upstream side than the acoustic model, while often underestimating the min principal

stresses at the base on the downstream side. Both models show high tensile stresses near the

base of the arch dam that would result in cracks. The analyses also show that, despite the

symmetric shape of the arch dam, there is a difference between the results at the left and right

section.


14/48

ICOLD 12-th International Benchmark Workshop

on Numerical Analysis of DamsTheme A: Fluid Structure Interaction Arch Dam Reservoir at

Seismic LoadingA. Tzenkov

1, A. Abati

1 and G. Gatto

1

1 STUCKY SA, Rue du Lac 33, PO Box, 1020 Renens, SWITZERLAND


Abstract

One of the main concerns regarding the numerical dynamic analysis of arch dams is the

proper modelling of the fluid-structure interaction between the dam and the impounded water.

There are several approaches to this, which enables accounting for the hydrodynamic

pressures on the upstream face of the dam with different precision and, respectively, with

different computing effort. This work investigates the impact of the hydrodynamic approach

opted for on the computed stresses and displacements of an example 220-m high double-

curvature arch dam. It is shown that, for this particular benchmark problem, it is important to

consider the compressibility of water.

Discussion and Conclusion

The maximum static compressive stress is the hoop stress at ¾ the dam height on the u/s face

of the main section. It amounts to approximately 7 MPa, which is within the admissible limits

regarding the compressive strength of concrete. The maximum static radial displacement

reaches 8.2 cm at the top of the central cantilever. The maximum compressive stresses during

the ground motion reach approximately 12 MPa in the hoop direction on the u/s face. Seismic

tensile hoop stresses varying between 1 MPa and 2 MPa are computed at the top part of u/s

face of the main section; they disappear if the contraction joint opening/closing is modelled.

The maximum seismic tensile vertical stresses exceed 2 MPA at ¾ the dam height on the

upstream face of the main section. The maximum compressive and tensile stresses occurring

during the earthquake are below the admissible limits. The maximum amplitude of the

dynamic vibrations is about 8 cm with respect to the initial displaced shape of the dam.

The hydrodynamic effect modelling approaches investigated in the present study lead to

similar results regarding the structural response of the example arch dam. In general, theWestergaard added masses approach yields higher compressive and tensile stresses, as well as

higher radial displacements. The compressible fluid analysis results in lower stresses with

respect to the incompressible fluid assumption, which is due to the increased damping of the

coupled system. In fact, according to 0, if the natural dam frequencies are significantly lower

than this of the dam reservoir, , the behaviour of the latter is similar to the behaviour of

incompressible fluid. In our case 1.98 Hz; this explains to a certain extent the obtaineddifferences in the results with compressible and incompressible fluid.


15/48

Dynamic Analysis of an Arch dam with Fluid-

Structure InteractionUse of an Open source code AKANTU

M. Chambart1, T. Menouillard1, N. Richart2, J.-F. Molinari2 and R. M. Gunn1

1 STUCKY SA, Rue du Lac 33, CH-1020 Renens, SWITZERLAND2 EPFL ENAC LSMS, CH-1015 Lausanne, SWITZERLAND

E-mail: [email protected], [email protected]

Abstract

In this contribution, the fluid-structure interaction is modeled using the added mass technique

and the incompressible fluid model. Results show that the added mass technique is more

conservative. Different meshes are compared with this later method, demonstrating that the

problem is not mesh dependent. Finally, the computation times obtained with two differentsoftware are compared, showing the efficiency of the new open-source software Akantu.

Conclusion

This benchmark allowed to investigate the field of how, in term of numerical method, to deal

with the hydrodynamic pressure on the upstream face of an arch dam. This has then shown

that the Westergaard theory, which consists in simply adding mass on nodes of the upstream

face of the dam, is clearly not mesh dependent. Second, the method dealing with the mesh of

the incompressible fluid, gives smaller envelops in terms of maximum and minimum of

displacement and stresses of the arch dam. In addition, the development of the Westergaard

theory in an open source code did not show any relevant difficulties and the computation timecan be significantly decreased.


16/48

STUDY ON ARCH DAM – RESERVOIR SEISMIC

INTERACTIONA.Popovici

1, C.Ilinca

1, R. Vârvorea

1

1Technical University of Civil Engineering, 124 Lacul Tei Bd. 020396-Bucharest, ROMANIA


Abstract

The arch dam – reservoir seismic interaction is investigated using ABAQUS 6.11 and DESARC

3.1 software. DESARC computer code offers the advantage of simplicity and computation speed

due to the degrees of freedom based on the stresses (Ritter modified method) being very

recommended for arch dams preliminary structural analysis.The coarse mesh given by formulator was used for investigation in ABAQUS and 12 arches

equally spaced on dam height were used in DESARC. The water effect was considered

according to added mass procedure as well as acoustic elements. All analyses were performed inthe linear elastic field.

The results are presented in compliance with formulator requests: eigenfrequencies and mode

shapes, hoop stresses, vertical stresses, min./max. principal stresses and radial displacements inthree different sections for static and seismic loads. A special attention is paid to compare the

results concerning arch dam – reservoir seismic interaction in different hypotheses applying two

software.

Comments on results of analysesIn compliance with some results illustrated in the above captions and other ones resulted from the

analyses carried out the followings comments may be pointed out:

DESARC3.1 a very simply, friendly and fast computer code in all applications led to results closeto those provided by much more sophisticated ABAQUS6.11 computer code. As a result

DESARC computer code is very recommended for arch dams preliminary structural analysis.

Usually, in engineering practice dead loads are applied on arch dam isolated cantilevers, takinginto account that during dam construction the contraction joints are not grouted. However, even

under these conditions, some of the dead weight of cantilevers is transferred to arches, this quota

depending of the dam shape and valley opening. Coming back to the arch dam provided by

formulator it may remark from stresses and displacements to dead loads and hydrostatic pressures

computed with DESARC in both hypotheses of isolated cantilevers and monolithic structure(fig.8a,b). that the effects of isolated cantilever hypothesis in relation with monolithic structure

are not important. Accordingly, the analyses with ABAQUS, for simplicity, were carried out onlyin case of monolithic structure.

The dam displacements due to dead weight reach about 1 cm to upstream at the crest level. Dead

weight is mainly transferred on cantilever. At the bottom of the central section the verticalstresses reach -7 MPa compression at upstream toe and -1.5 MPa compression at downstream

heel. The hypothesis with isolated cantilever is a conservative one. On arches, the maximum

stress reaches -2 MPa compression at the crest level.Hydrostatic pressure as independent load generates maximum radial displacement of 7.5 cm in

central section, dam crest elevation. The vertical stresses vary between 1 MPa tension and -4


17/48

MPa compression at downstream face and, respectively between 8 MPa tension at the dam

upstream toe and -2 MPa compression at upstream face. The hoop stresses vary between 0.5 MPa

tension, -3.7 MPa compression at downstream face, respectively between 0.5 MPa tension, -7

MPa compression at upstream face. The big vertical tensile stress generated by hydrostatic pressure at the dam upstream toe is reduced to 1.5 MPa by vertical compressive stress in the same

point due to dead weight.

Based on results concerning natural periods values computed in full reservoir hypothesis (Table1) (fundamental period 0.556 s ABAQUS, added mass, 0.594 s DESARC, added mass, 0.645

ABAQUS, acoustic elements)it may conclude the hydrodynamic forces developed by acoustic

elements are higher than hydrodynamic forces computed with Westergaard formula to generateadded masses.

A comparison between seismic responses computed in the hypothesis empty reservoir by spectral

analysis and direct time integration with ABAQUS points out a good correlation betweencorrespondent displacements response their maximum reaching about 6 cm but generally

significant differences between correspondent stresses. In both methods the max/min stressesvary between 4.5 MPa tension and -5.5 MPa compression.The same remarks as mentioned above, result comparing the dam seismic response in the full

reservoir case computed with ABAQUS by spectral analysis and direct time integration methods.

The displacement graphs are almost similar, the maximum displacement reaching 8 cm at thedam crest, central section. The stresses between corresponding points show sometimes significant

differences but in both methods the stresses vary between the same limits: 5 MPa maximum

tensile stress and -6 MPa maximum compressive stress.

Maximum displacement resulted in direct time integration at combination dead weight +

earthquake (empty reservoir) is 6.5 cm to upstream in the central section at the dam crest level.

Vertical stresses on downstream face are generally compression reaching -3 MPa., excepting alocal zone to dam bottom in central section where are tensile stresses reaching about 1 MPa. On

upstream face the vertical stresses are also compression reaching -11MPa at the dam upstreamtoe, central section. On horizontal direction the hoop stresses vary between 4.5 MPa tension and -

5.6 MPa compression to the dam base at downstream face and between 2 MPa tension and -5

MPa compression at upstream face. It may conclude, in this case (empty reservoir) the combinedstatic stresses + dynamic stresses due to 0.1g earthquake are in acceptable limits.

Maximum displacement obtained in direct time integration at combination static loads (dead

weight + hydrostatic pressures) + earthquake (full reservoir) reaches 12 cm to downstream at thecrest level in central section. Max/min vertical stresses on downstream face are permanent

compressions reaching -6.5 MPa maximum value. On upstream face at the dam upstream toe the

tensile stresses reach 5MPa. It is a vulnerable point for cracking. On horizontal direction the hoopstresses vary between 4.8 MPa tension and -5 MPa compression and between 2.8 MPa tension

and -8 MPa compression at upstream face. It may conclude that excepting dam upstream toe

where exists cracking risk, the dam withstand earthquake action without notable incident.As a general conclusion based on data presented above, the seismic response parameters of arch

dams, especially seismic stress state differ significantly function of method of analysis.

Unfortunately, the scarcity of recordings concerning arch dams behavior during strong

earthquakes make difficult to validate a method of analysis using field recordings from.


18/48

Fluid Structure Interaction

Arch Dam – Reservoir at Seismic loadingA solution using FEnas

M. Brusin1, Dr. J. Brommundt1 and H. Stahl1

1 AF-Consult Switzerland Ltd., Täfernstrasse 26, CH-5405 Baden, SWITZERLAND


Abstract

This paper represents a contribution to the ICOLD – 12 th International Benchmark Workshop

on Numerical Analysis of Dams, THEME A - Fluid Structure Interaction Arch Dam –

Reservoir at Seismic loading. The formulated problem, modeling of a 220 m high and 430 m

wide arch dam under static and seismic loads, was addressed by the finite element code

FEnas. FEnas is a Swiss code originally developed at the Imperial College London. Theinteraction between the dam and the water in the reservoir was implemented using the added

mass technique in the general form following Westergaard.

The simulations performed included a comparative analysis of two meshes with different

spatial discretization. The computed results show that the impact of the discretization is rather

small. Moreover, the results of the analyses have been compared to typical deformations

observed at Swiss dams and similar studies on existing dams. All results here are in good

agreement with preceding studies of this kind.

Discussion and Summary

Stress-strain analysis of arch dam during seismic activity including reservoir-structureinteraction using added masses technique according to generalized Westergaard approach is

presented. Interaction of an arch dam with the impounded water leads to an increase in the

dam vibration periods. The fact that water moves with the dam increases the total mass that is

in motion. This added mass increases the natural periods of the dam, which in turn affects on

the effective earthquake inertia forces. The Westergaard method usually gives the largest

added-mass values, which is evident by its increasing the vibration periods the most.

However, this does not automatically give the largest stresses, because response of the dam

also depends on the characteristics of the earthquake ground motion. Although Westergaard

approach is widespread in practice, for different variants of quasi-static and dynamic analysis

of seismic effects, the above-mentioned facts lead to the conclusion that the results must be

taken with caution.After comparison of two different sizes of spatial discretization of meshes we can conclude,

that the coarser mesh also gives satisfactory results which can be used in engineering practice.

In fact, the differences are small. Whereas the performed simulations with linear material

behavior are not very costly in terms of computation time and required computer power for

both meshes, the coarser mesh could be very useful as soon as more complex processes, e.g.

non-linear material behavior, are to be considered.

Maximum earthquake stresses are located in the central upper portion of the dam as well

along the dam-foundation contact zone. Maximum radial deformation occurs in the crest, and

decreasing towards lateral sides, as well as towards fixed end.

The vertical stresses observed at the upstream surface show some artefacts for the load cases

with the earthquake as they are supposed to be zero at the top of the dam like it is observed

for self-weight and self-weight with hydrostatic pressure. These non-zero values are


19/48

considered to be artefacts. The same artefact in a much smaller magnitude is seen at the top of

the downstream surface.

Stress time histories curves shows several significant tensile stress cycles that can lead to

open the contraction joints, especially in crown zone, while on the time histories of radial

deformation in addition to timing of deformation could be noted as dominant periods of the

first few natural frequencies.The results that are shown have been compared to typical deformations observed at Swiss

dams. The overall behavior of the dam here is in good agreement with what AF has seen in

reality. The comparison of the results of the earthquake simulations with previous studies of

this type performed at AF was positive.

We can conclude that the added mass technique in combination with the direct dynamic

analysis relatively quickly and easy produce results useful in engineering practice.


20/48

The Seismic behaviour of an Arch Dam-Reservoir-Foundation System

A. Frigerio1 and G. Mazzà

2

1 Ricerca sul Sistema Energetico – RSE S.p.A., via R. Rubattino 54, Milan, ITALY


Abstract

The paper summarizes the main features and the related results of the linear static, modal and

seismic analyses carried out on the concrete arch dam proposed as Theme A of the 12 th

ICOLD International Benchmark Workshop.

The analyses have been performed by means of the COMSOL Multiphysics software, making

reference to the coarse finite elements mesh provided by the Formulators, but some slightly

changes.In the modal and dynamic analysis, the water-dam-foundation interaction has been taken into

account. The impounded water has been modeled by means of acoustic finite elements,

assuming appropriate boundary conditions to simulate the wave reflecting conditions between

the fluid and the foundation, as well as the fluid and the upstream face of the dam, and the

non-reflecting condition at the end of the reservoir channel.

Results have been provided according to the requests of the Formulators in terms of radial

displacements, hoop and vertical stresses on three vertical sections of the dam.

Conclusion

The paper presents the dynamic response of the dam-reservoir-foundation system according tothe basic requests of the Formulators. Only the coarse mesh has been considered as some

preliminary analyses have demonstrated that, compared to a high computational effort, results

showed only slight differences.

A second aspect to be emphasized refers to the boundary conditions assumed for the fluid

domain at the rock interface: if a fully reflecting condition is assumed, the stresses computed

on the dam are considerably higher than those attained in the case discussed in this paper

where a partial absorption conditions has been assumed. This means that experimental

investigations should be undertaken in order to define more realistic boundary conditions.

A suggestion for the Formulators, in the synthesis phase, refers to the opportunity to make

some comparisons between the main outcomes of Theme A and those of the previous

Benchmark Workshops, cited above.


21/48


THEME B:

LONG TERM BEHAVIOR OF ROCKFILL

DAMS


22/48


23/48

Assessment of ICOLD Benchmark Case Study using

Flood2D-GPU and HEC-FIA

Brantley A. Thames 1 and Alfred J. Kalyanapu

2

1 US Army Corps of Engineers, 801 Broadway, Nashville, TN 37203, USA2 Department of Civil & Environmental Engineering, Tennessee Technological University,

1020 Stadium Drive, Prescott Hall 334 Box 5015, Cookeville, TN 38505, USA


Abstract

This paper presents a benchmark case study of a dam failure using Flood2D-Graphics

Processing Unit (GPU), a two-dimensional hydraulic modeling software, coupled with the

Hydrologic Engineering Center’s Flood Impact Analysis (HEC-FIA) software, a flood

consequence model, to evaluate flood risk in a hypothetical test bed environment. Thedefined objective of this study is to present the findings prescribed by the ICOLD Theme “C”

Formulation Team, which include inundation mapping, population at risk, loss of life, and

direct economic impacts. The increased computational capability afforded by Flood2D-GPU

allows for a much deeper analysis of the hypothetical scenario. Many studies have been

conducted to evaluate the uncertainty in the estimation of breach parameters (breach width

and time of failure); however, how breach parameter estimation uncertainty affects the

uncertainty in consequence estimation has been rarely evaluated. A secondary objective of

this study is to determine the affects of breach parameter estimation uncertainty on

consequence estimation uncertainty through a comparative analysis of four breach parameter

estimation (BPE) methods and a sensitivity analysis of the main breach parameters such as

breach depth, breach width, time of failure, and breach side slope. Through these analyses, breach depth, typically assumed using engineering judgment, is found to be the most sensitive

breach parameter. Similar breach formation times are computed for three of the BPE

methods, Froehlich (1995a), Froehlich (2008), and Von Thun and Gillette, evaluated while

varying breach widths are computed; however, the consequence results are similar despite the

varied breach width computations highlighting the importance of breach formation time.

Conclusions

This paper presents a solution to the problem statement using Flood2D-GPU coupled with

HEC-FIA offering possible hydraulic and consequence outcomes under the hypothetical

scenario provided. The comparative analysis of various BPE methods provides recognition ofthe similarities and differences between the methods. The sensitivity analysis of the breach

parameters demonstrates the importance of specific parameters by illustrating the effects of

these parameters on not only the hydraulic outputs but also the consequence results. These

analyses combined begin the conversation into utilizing the most appropriate BPE method and

focusing in on which breach parameters are the most critical to approximate the most accurate

hydraulic and consequence results in a study of this nature.

The results of the comparative analysis suggests that similar consequence results can be

expected when using the Froehlich 1995a, Froehlich 2008, and VTG breach parameter

estimation methods but, without additional case studies using this same approach, relative

replication of these results is not substantiated. However, all of these approaches are based onstraightforward regression relationships and should produce similar results. The MLM BPE


24/48

method is similar in many aspects to the other three BPE methods with the exception of the

inclusion of a volume of material eroded term. Additional case studies could provide insight

into the differences of results experienced during this study.

In general, the results of the sensitivity analysis follow the expected trends. As breach depth,

width, and side slopes increase and the time of breach formation decreases, the breachdischarge hydrograph shape becomes more compressed resulting in higher peak discharges,

which ultimately results in more significant flood impacts and consequences. Several

conclusions can be drawn from these analyses.

Final breach invert is very difficult to estimate and is an assumed value not calculated by any

of the BPE methods. Unfortunately, the results of this analysis show that the breach

hydrograph and consequence results are extremely sensitive to the breach invert assumption.

When conducting a risk-based assessment of a dam failure, special care should be taken to

ensure that the best estimation of breach invert is assumed based on the sensitivity of breach

hydrograph and consequence outcomes to the estimation of this parameter. Ultimately,

additional research could provide a better approach to estimating this parameter drasticallyreducing the uncertainty in these types of analyses.

When comparing the breach parameter estimations of SS, ElB, W b, and tf , using the three BPE

methods that produced similar consequence results, with the results of the sensitivity

simulations, the breach formation time proves to be a critical parameter in the development of

breach discharge hydrographs. Even with a W b varying between 49 and 146 m, the resulting

breach hydrograph shape and peaks are similar due to the similarity in formation time, which

ranges from 0.6 to 0.9 hours.


25/48

Risk Assessment Analysis of a Hypothetical Dam

Breach Using Adaptive Finite Element MethodsAdaptive Hydraulic Model, ADH

McVan, Darla1, Ellis, Jeffrey2, Savant, Gaurav3, PhD, P.E., M.ASCE, and Jourdan,

Mark1, D.Eng, P.E.

1Research Hydraulic Engineer, Engineer Research and Development Center, U.S. Army

Corps of Engineers, Vicksburg, MS, 39180.


2Engineering Technician, Bowhead Science and Technology, LLC, and Onsite Contractor,

Engineer Research and Development Center, U.S. Army Corps of Engineers, Vicksburg, MS

39180.


3Research Water Resources Engineer, Dynamic Solutions LLC and Onsite Contractor,

Engineer Research and Development Center, U.S. Army Corps of Engineers, Vicksburg, MS

39180.


Abstract

A virtual computational test bed was developed by a committee sponsoring a session at the

International Commission on Large Dam (ICOLD) 12th International Benchmark Workshop.

The purpose of this virtual test bed was to serve as a platform for comparison of dam breach

models and methodologies. This paper provides the results of one of the model comparisons.

The ADH model, developed by the U.S. Army Engineer Research and Development Center

(ERDC) was used to simulate a catastrophic dam failure and gradual dam failure. The ADH

Model is described, as well as the domain, setup, and boundary conditions for the specific

simulations.

Results are provided as gridded and tabular data, as well as hydrographs. The gridded results

include peak flood depths, flood wave arrival times, peak unit flow rates and total population

at risk. Tabular results include flooded area and population as risk (including age

demographics). Hydrographs are presented for both the dam failure discharge and at five

down stream cross sections. Loss of life and economic impacts were not calculated.

Conclusion

The results from the catastrophic breach event showed a total flooded area, excluding the

lake, of 41.1 km2 which affects over 28,800 people. The total number of 12-year and under

affected by the maximum inundated area is over 4.600 and the total number of 65-year and

over is over 4,200. The flooded area with depths greater than 8 meters is 3.3 km2 and affects

about 1,700 individuals, over 400 are 12-year and under and over 130 are 65-year and over.

The greatest number of the population at risk occurs within the first hour of the dam breach

resulting in over 26,000 being affected. Three hours after the initial dam breach, the majority

of the flood wave has reached the lake and no longer affects the city of Hydropolis.


26/48

The gradual breach scenario assumed breaching was caused by overtopping resulting in 80-

percent of the dam being removed. The time to reach the maximum breach and the maximum

width were determined using equations developed by Froehlich (1995a). The results showed

a total flooded area excluding the lake of over 20 km2 which affects over 9,200 people. The

number of 12-year and under affected by the maximum flooded area is 1,600 and the number

of 65-year and over is 1,699. The flooded area with depths greater than 8 meters is 1.99 km2

affects less than 300 individuals with over 50 being 12-year and under and less than 30 being

65-year and over. The majority of the population at risk occurs within the first 90 minutes of

the dam breach; however the total number is skewed because the inundated area for each time

interval overlaps.

These results represent the impacts of a dam breach, whether that breach is catastrophic or

gradual. Though loss of life was not calculated using the described model, the population at

risk was significant. Additionally, economic impacts were not considered, as this is not part of

the applied model. Even without these values available for comparison, there are many data

elements that will be useful in comparing the ADH model to other models presented at this

workshop. The use of the virtual test bed should prove to be a valuable tool, both now and inthe future, for comparison of such models.


27/48

Dam Failure and Consequence Assessment with

Standardized USACE MMC ProceduresInternational Commission on Large Dams 12

th International

Benchmark Workshop on Numerical Analysis of Dams (Theme C:Computational Challenges in Consequence Estimation for Risk

Assessment)

D. Williams1 and K. Buchanan

2

1 U.S. Army Corps of Engineers, Tulsa, Oklahoma, USA

2 U.S. Army Corps of Engineers, Huntington, West Virginia, USA

Email: [email protected]

AbstractThe U.S. Army Corps of Engineers’ (USACE) Modeling, Mapping and Consequences(MMC) Production Center has submitted an entry for the International Commission on LargeDams (ICOLD) 12th International Benchmark Workshop on Numerical Analysis of Dams(Theme C: Computational Challenges in Consequence Estimation for Risk Assessment). As

part of this exercise, standard MMC processes were used to develop hydraulic modeling of adam break inundation as well as the associated consequences. Modeling techniques includedone- and two- dimensional numerical models, use of geographic information systems, andeconomic and life loss estimation. Breach parameters were developed from standardregression equations. Based on the rules and regulations that USACE has established for theestimation of risk, economic consequences and life loss modeling results from this breach

analysis corresponded to a high hazard dam. This analysis demonstrates the standardizedMMC process for consistently evaluating all USACE dams while providing the flexibility toanalyze projects that range from small and simple to large and extremely complex.

Conclusion

This exercise demonstrates the practices and procedures that the MMC uses to analyze thelarge inventory of dams that are owned and operated by USACE. Most of these dams arelarger and more complex than the hypothetical dam that was analyzed in this project. Someinclude complex reaches that extend for hundreds of miles downstream with travel times thatlast for days or weeks. Elements of each of these dams and their downstream reaches are

unique, and it is important to emphasize that the sophistication of MMC models, whileincorporating a standardized process, varies by project.

Since the reach in this exercise was both steep and short, flood wave travel times to thedownstream boundary of the model were very quick. As a result of these conditions,comparisons of the inundation areas resulting from the Froehlich (1995), Froehlich (2008),and Von Thun and Gillette equations were very similar. Modeling with dam breach

parameters from all three equations resulted in a flood wave with a steep peak and fast traveltime through the confined mountain valley below the dam followed by lateral attenuation ofthe flood wave in the broad plain below the base of the mountain range.

Breach parameters from Froehlich (1995) were carried forward in the modeling used forconsequences analysis. Estimation of economic damages and life loss was challenging


28/48

because FEMA HAZUS data, a standard input for MMC HEC-FIA models, was not availablefor the benchmark exercise. Therefore, average values were used, which may or may not berepresentative of Hydropolis. Total damages of $657M and a life loss of nearly 2,000 personswere estimated for this dam breach.

Based on the rules and regulations that USACE has established for the estimation of risk,these results correspond to a high hazard dam. The procedures that were used in developingthe hydraulic modeling and consequences estimation for the benchmark exercise wereconsistent with USACE MMC processes. These guidelines, while providing consistency inthe analysis of all USACE owned and operated dams, also provide enough flexibility toanalyze projects that range from small and simple to large and extremely complex.


29/48


30/48

The above mentioned results used in the successive 2-D depth-averaged calculations with

TELEMAC-2D. The following conclusions can be made based on the 2-D simulation

results:

The height of water and the maximum height of water which are important parameters

in risk management fields are calculated to define inundation regions. These results

also can be used to define warning zones according to the flooded areas.

The wave arrival time is analysed for 15 minutes intervals. This time model shows us

how long it takes for the flood to reach each part of the downstream area which is

useful for reducing loss of human lives in risk management studies.


31/48

A Benchmark study on dam breach and consequence

estimation using EMBREA and Life Safety Model

M. Davison1, M. Hassan

1, O. Gimeno

1, M. van Damme

1and C. Goff

1

1 HR Wallingford, Howbery Park, Wallingford, Oxfordshire, OX10 8BA, UNITEDKINGDOM

E-mail: m.davison@hr wallingford.com

Abstract

This paper presents the modelled consequences of a hypothetical dam breach as laid out in

Theme C of the 12th international Benchmark Workshop on Numerical Analysis of Dams. TheEMBREA model was used to model the failure of the dam due to headcut erosion and toderive a breach hydrograph which was then used in InfoWorks ICM to model the 2D flood

spreading. The results of the flood model have been used to calculate how severe the floodwould be in terms of the total population at risk, loss of life using the Life Safety Model andeconomic impact. The paper shows that the consequences of the hypothetical flood will besevere in terms of casualties and economic damage.

Conclusions

The paper presents the consequences of the breach of the hypothetical dam as described in

Theme C of the 12th International Benchmark Workshop on Numerical Analysis of Dams.

The dam breach analysis undertaken using EMBREA highlights the importance of choosing

the correct erosion method and the impact of choosing the correct erodibility of the soil.Based on the data provided, it was estimated that the embankment fails according to headcuterosion leading to a steep hydrograph with a high peak discharge, the consequences of whichare more severe than when an embankment fails due to surface erosion or piping.

The effects of the different roughness scenarios considered on the flood model was negligiblecompared to the impact of the different hydrographs used as input, emphasizing again theneed for a good dam breach model. The flood model results provide useful information forwarning and evacuation planning. The model indicates that the first flow from the dam breachreaches the populated area 1.5h to 2h after the breach initiates, but the peak flow travels from

the dam to the city in less than 15min, reaching the further edge in less than 90min.

The total direct economic impact is estimated at more than US$1.24b. The loss of lifecalculations performed with the static Life Safety Model show a loss of life of 4,966, whichrepresents nearly the 20% of the total people at risk.


32/48

Uncertainty in Two-Dimensional Dam-Break Flood

Modeling and Consequence Analysis

M.S. Altinakar1, M.Z. McGrath

1, V.P. Ramalingam

1, D. Shen

1, Y. Seda-Sanabria

2 and

E.E. Matheu3 1 National Center for Computational Hydroscience and Engineering, The University of

Mississippi, Brevard Hall Room 327, P.O. Box 1848, University, MS 38677-1848, USA2 Critical Infrastructure Protection and Resilience Program, Office of Homeland Security, U.S.

Army Corps of Engineers, Headquarters, Washington, DC 20314, USA3 Critical Lifelines Branch, Sector Outreach and Programs Division, Office of Infrastructure

Protection, National Protection and Programs Directorate, Washington, DC 20528, USA


Abstract

This paper presents the preliminary results of an extensive study to evaluate and quantify

uncertainty in two-dimensional numerical dam-break flood modeling and consequence

analysis based on a benchmark test case. The benchmark problem consists of a hypothetical

61m-high embankment dam in a mountainous region with lightly populated urban areas

located downstream. Although the benchmark problem was intended mainly for estimating

the uncertainty in the evaluation of the consequences of failure of a dam near populated areas

with complex demographics, infrastructure and economic activity, the present study also

investigated the uncertainty in two-dimensional numerical modeling based on three control

variables. DSS-WISE™ numerical model was used to calculate dam-break flood and potential

loss-of-life for a total of 120 cases, which represented 40 random pairs of breach width and

breach formation time, each computed with three different sets of Manning’s coefficientsdefined based on land use/cover type. Computed results include raster maps of maximum

flood depth, maximum flood discharge per unit width and flood arrival time as well as

hydrographs at 7 cross sections. Analysis of population at risk (PAR) impacted by the flood

and loss-of-life were also computed for all 120 simulations. The paper presents some

preliminary results based on the statistical analysis of results.

Conclusion

The uncertainty in 2D numerical modeling of dam-break flood and the resulting loss of life

were studied using 120 simulations representing 40 random pairs of breach width and breach

formation time calculated with three different sets of Manning’s coefficients. The resultsshow that large variations in the extent of the inundated area, water depths, and loss-of-life

occur based on the particular combination of control parameters. The paper presents the

average value and standard deviation of selected computed results as well as their upper and

lower bounds. Peak discharge at the breach cross section and the cross sections downstream

seem to be highly correlated with the breach formation time rather than the breach width.

Only part of the results could be presented. Additional findings will be presented in

subsequent publications.


33/48

Risk assessment for hypothetical dam breakA method for the rapid and consistent evaluation

L.Mancusi1, L.Giosa

2, A.Cantisani

2, A.Sole

2 and R.Albano

2

1 Sustainable Development and Energy Resources Department, Research on Energy Systems(RSE) Spa, via R. Rubattino 54, 20134 Milano, Italy

2 School of Engineering, University of Basilicata, viale dell’Ateneo Lucano 10, 85100

Potenza, Italy


Abstract

This paper details the technical contribution to the theme of flood risk analysis as

consequence of a dam failure. According to the numerical problem proposed for the

workshop, the analysis consists of the evaluation of the dam break and its consequences. The

simulation includes two scenarios of dam breach: the scenario 1 that represents the case of an

easy erodible dam and the scenario 2 the case of an erosion resistant dam. For each scenario, a

dam failure discharge hydrograph was calculated and the subsequent flood wave and

consequences have been evaluated.

The methodology adopted involves, for a first part, the use of standard models for the

hydraulic modelling of the dam breach and flood wave propagation. For the second part, a set

of GIS scripts was written, tested and developed using the python scripting language to obtain

a rapid appraisal of consequences for the population and to assess the direct economic

damages for residential, commercial, and industrial buildings.

Since the latter elaboration depends greatly on the type and the level of detail of available

data, in this study we have been used data as generic as possible and GIS scripts that allow,for a great variety of cases, a rapid initial assessment.

Conclusion

In this paper we present the results of the analysis of a possible dam failure. The development

of a dam break is a complex process involving numerous uncertainties: the methodology

adopted in this work is a medium-scale approach type and can be used for the rapid and

consistent evaluation of consequences for the population and to assess the direct economic

damages for residential, commercial, and industrial buildings. Rapidity is allowed by using

aggregate data: maps of land-use, population census and parcel zone. Consistency is required

to ensure comparability between evaluations. For that reason the method can be used to prioritize corrective actions to achieve the greatest and quickest possible risk reduction or for

identification of the most effective and better-justified measures of risk mitigation.

The comparison carried out for the two scenarios is an example of use of the methodology to

estimate the sensitivity of results with respect to an uncertain parameter which is the breach

formation time.


34/48


OPEN THEME:

FREE CONTRIBUTIONS


35/48

Seismic Analysis of a Concrete Arch DamConsidering Concrete Heat Generation Damage Effects

P. Dakoulas1

1 University of Thessaly, Volos, GREECE

E-Mail: [email protected]

Abstract

This study evaluates the seismic performance and safety of Tavropos arch dam (Greece)

considering the dynamic canyon-dam-water interaction. Tavropos dam has a height of 83 m

and crest length of 220 m. The numerical model of the arch consists of 17 concrete cantilevers

separated by vertical joints, whereas the foundation-abutment model represents a large region

of the canyon. The interface behavior at the vertical joints and the horizontal base joint is

modeled in three different ways: as bonded surfaces, as surfaces in contact with frictional

strength, and as a combination of bonded and contact surfaces. To account for possible prior

damage, the study simulated the thermo-mechanical phenomena that took place during dam

construction, related to heat generation due to concrete hydration. The seismic analysis is

performed against a MCE having a magnitude of 7.5 and peak ground acceleration of 0.47g.

The seismic evaluation is based on an extensive parametric investigation of the effects of

various key factors, including concrete strength, joint interface behavior, water level,

hydrodynamic pressures, ambient temperature, canyon rock flexibility, spillway geometry,

earthquake excitation characteristics, etc. It is concluded that the expected overall stability

and performance of Tavropos concrete arch dam is satisfactory.

ConclusionsEvaluation of the results of all parametric studies leads to the general conclusion that, for the

maximum earthquake intensity anticipated, the dam is safe against any instability and its

overall performance is satisfactory. More specifically, the main conclusions are the following:

1. The concrete-hydration heat generated during construction did not cause any significant

cracking within the main body of Tavropos dam.

2.

The maximum seismic relative displacement of the dam for the case of grout-bonded

joints is small (about 3 to 4.5 cm). There are no permanent relative displacements at the

end of shaking.

3. For the extreme scenario in which the vertical joints and the joint at the base are already

ruptured before the earthquake, the relative displacement of the dam at the end of shakingis limited to values less than 5 cm. Thus, even for this extreme scenario, the stability of

the 17 cantilever dam parts is fully assured.4.

Based on the results of all analyses, the maximum compressive stresses developing duringseismic shaking range between -10 MPa to -16 MPa and do not cause any plastic strains.

5. The magnitude of tensile stresses within the concrete parts depends on the behavior of the

vertical joints. The results of the parametric analysis show that the tensile plastic strains

(or cracking) developing within the dam body are very small. Therefore, the performance

of the dam with respect to the development of tensile cracking during shaking is

satisfactory.

6. The summer environmental temperature condition is the most critical one, as it results to

higher tensile plastic strains within the dam body.


36/48

7.

Among the four reservoir water elevations investigated, the most critical one is the

maximum water level (792 m), which results to relatively higher tensile plastic strains

within the dam body.8. The stiffer rock, having a wave velocity

sV =2800 m/s, yielded consistently larger values

of tensile plastic strains compared to those obtained fors

V =2000 m/s.


37/48

Earthquake Assessment of Slab and Buttress Dams

H. B. Smith1,2

and L. Lia2

1 Norplan AS, Oslo, Norway2

Department of hydraulic and environmental engineering, NTNU, Trondheim, NorwayE-mail: [email protected]

Abstract

This paper outlines work on the examination of the dynamic properties and seismic safety of

flat slab and buttress dams. The work has been carried out with a linear elastic numerical

model established in the Finite Element Model program Abaqus.

One single dam section in a typical Norwegian flat slab dam with heights of 12 and 25 meters

has been modeled in regards to varying reservoir water level and lateral bracing. Abaqus has

been used for the frequency analysis and the dynamic time-history analysis.

The natural vibration modes of the dam represent movements in separate directions. Through

a seismic event the greatest response will be represented by the buttress deflection in direction

along the dam axis (axial direction). The resistance towards earthquakes will depend

significantly on individual stability of elements rather than the global stability. In particular,

the tensile stresses occurring in the buttress when deflecting in the axial direction are found to

be a potential failure mechanism.

Providing lateral bracing by struts positively influences the response of the slab and buttress

dam in seismic events. When lateral bracing is provided, the ability of the dam to transfer the

inertial forces to the abutment is important.

Conclusions

The results demonstrate how the dam vibration modes represent movements in separate

directions. An increased dam section height results in an increased response because of the

greater oscillating mass. Meanwhile, increased water level in the reservoir increases the

period for oscillation in the upstream-downstream direction.

Through a seismic event, the greatest response will be represented by the buttress deflection

in the axial direction. The resistance towards earthquakes will be mainly dependent on the

individual stability of elements rather than the global stability. In particular, the tensilestresses occurring in the buttress, as a result of deflection in the axial direction, is found to be

a potential failure mechanism.

Providing lateral bracing by struts positively influences the response of the slab and buttress

dam in seismic events. For the Norwegian slab and buttress dams in particular (Ambursen

dams), the contribution of the existing insulation walls is considered to be small but positive.

When lateral bracing is provided, the ability to transfer the inertial forces to the abutment is

important.

In addition to the assessment of simplified analytical methods, it is recommended that further

numerical studies should focus on the flat slab and buttress dams nonlinear response, on the

importance of global dam geometry, and verification of the numerical input parameters by

physical vibration tests of an existing slab and buttress dam.


38/48

Influence of Surface Roughness on Sliding StabilityTests and numerical modeling

Ø. Eltervaag1, G. Sas2, L. Lia3

1Department of Structural Engineering, Faculty of Engineering Science and Technology,Trondheim, Norway

2 Northern Research Institute, Narvik, Norway3Department of Hydraulic and Environmental Engineering, Faculty of Engineering Science

and Technology, Trondheim, Norway


Abstract

Lightweight concrete dams slide when the shear capacity of one or more sliding planes in the

dam’s structure or foundation is exceeded. Several laboratory shear tests were carried out on

concrete rock samples. The two materials were mated trough teeth-sawed interfaces withdifferent inclination profile. This paper presents the results of the numerical modeling of those

tests.

The results of the shear tests were compared to the predictions of the model used in the

Norwegian guidelines. It has been found that the model used in the guidelines do not predict

the shear capacity accurately. Through finite element analyses a better representation of the

tests has been achieved, especially regarding the influence of roughness.

Conclusion

Through finite element modeling, more sophisticated analyses regarding stability towardssliding might be obtained. This allows more sustainable design of structures subjected to

sliding instability. However, it must be noted that due to the limited amount of samples

analyzed and the scale effects encumbered with shear capacity, further work is needed to

utilize the potential of finite element modeling of stability towards sliding for full scale dams.

A full parametric study is needed to improve the method.


39/48

Solution of dam-fluid interaction using

ADAD-IZIIS software

V. Mircevska1, M. Garevski

1, I. Bulajic

2, Simon Schnabl

3

1 Institute of Earthquake Engineering and Engineering Seismology,

univ. “Ss. Cyril and Methodius”, Box 101, 1000 Skopje, R.Macedonia2 Mining institute, Batajnicki put 2, 11000 Belgrade, Serbia

3Faculty of Civil and Geodetic Engineering, University of Ljubljana,

1000 Ljubljana, Slovenia


Abstract

Fluid dam interaction has a remarkable impact on the dynamic response of dams and could

play an important role in assessment of their dynamic stability. This is particularlyemphasized when dams are subjected to strong seismic excitations. The phenomenon has been

for the first time physically explained and mathematically solved by Westargaard. The very

first dynamic analysis based on application of “Added Mass Concept”; underestimated the

random earthquake nature in assessment of the hydrodynamic effects. In the recent years,

various BEM-FEM and FEM-FEM techniques have been developed to account for many

significant parameters that influence the accuracy of calculated hydrodynamic effects. This

paper presents a BEM-FEM orientated solution based on the use of the matrix of

hydrodynamic influence as a very effective tool for analyses of extensive domains of fluid-

dam-foundation rock systems for two major reasons: the computation time is far more

effective than that in direct or iterative coupling methods and stability of the solution. The

presented analyses are based on the use of genuine software originally written for static and

dynamic analysis and design of arch dams.

Conclusion

ADAD-IZIIS software is based on BEM-FEM oriented solution of the coupled structure and

incompressible and inviscid fluid domain. The software offers process of generation of the

mathematical model that runs parallel and interactively with the process of design of the arch

dam body. Automatically are generated the finite element mesh of the dam and part of the

foundation mass for accounting the phenomena of dam-foundation interaction, and boundary

element mesh that presents the boundaries of the fluid domain for accounting the fluid

structure interaction. The solution of the coupled system is accomplished by use of matrix of

hydrodynamic influence utilizing the concept of virtual work of “unit accelerations”.

Comparison of the calculated hydrodynamic effects using both BEM-FEM solution and added

mass method is made. The added mass method provides acceptable results only in the range

of Westergaard restricted hypothesis. Since it neglects the dam flexibility and water

compressibility and does not require any discretization of the reservoir domain wherever these

features have an impact on the magnitude of hydrodynamic effects there will be discrepancy

of the obtained resultants.


40/48

Need for transient thermal models, with daily inputs,

to explain the displacements of arch gravity dams

I.Escuder1, D. Galán

2and A. Serrano

3

1 Instituto de Ingeniería del Agua y Medio Ambiente. Universitat Politècnica de València

(Spain) http://www.ipresas.upv.es2 División de Seguridad de Presas. Canal de Isabel II Gestión (Spain).

http://www.gestioncanal.es3 iPresas, SPIN-OFF UPV. web: http://www.ipresas.com


Abstract

La Aceña Dam, with 66 meters of maximum height, belongs to the typology of concrete

gravity dams, and is operated as part of the water supply infrastructures of a Spanish major

city. Many instrumentation records were available but among them, the ones provided by four

direct pendulums outstood by its quality and consistency. The range of the values of the

movements registered by those pendulums, of almost 4 cm and of totally elastic nature

(showing no irreversible movements), set some interpretation challenges. The apparent

incapacity of the models to reproduce the observed behaviour was used as a starting point for

the diagnosis of the main sources of uncertainty: the nature of the foundations and the state of

the joints, among others. These aspects have constituted the developmental axis of a series of

works that have led to a number of effective and efficient actions on the dam during the last

years. However, an updated transient thermal model with daily inputs on external and internal

measured temperatures, coupled with a mechanical model has shown an spectacular

improvement in the explanation of the displacements recorded by the instrumentation.

Summary of the results and conclusions

The main conclusions that can be extracted from the current analysis of the stress-deformation

behaviour of the dam of La Aceña through the simulation methodology described along the

former sections, established in order to make the most of the whole set of records of thermal

monitoring and movements, are:

The elaborated models reproduce in a reasonably satisfactory way the thermal

behavior of the structure as well as the movements observed in the dam.

The fact of having carried out a transient thermal simulation for each day, capturing

the impact of the direct sun exposure on the downstream side of the dam and

estimating with accuracy the water temperature have been key elements to the

goodness of the obtained numerical model.

Figure 4 shows a summary of the average errors incurred by the different models. As it can be

observed, the best model is the so called “reference one” (3D, with foundation and the

hypothesis of improved temperature).


41/48

Figure 4: Average errors in mm for each model (shown for each pendulum, from P1 in darker

red to P4, in a lighter grade.)

It can be stated that thanks to the recently developed numerical models a new tool is at handto interpret the strain-deformation behaviour of the dam of La Aceña. This tool can

complement the auscultations that are currently being done on the dam and that have provided

an excellent level of information about both the thermal and movement behaviour. Also, the

numerical tool has permitted the satisfactory and consistent reproduction of the stress-

deformation behaviour auscultated until the end of 2011 and can act in the future as a way of

contrasting the impact of new actions planned on the dam as well as a form of detecting

potential behaviour change trends.

Figure 5 shows the comparison between the reference model (with the hypothesis of improved

temperature) and the model with the original hypothesis of temperature.

Reference

model

Impact of

no

cosidering

foundation

Impact of

no

cosidering

3D

Impact of

no

updating

thermal

parameters


42/48

Figure 5: Extent of the improvement of the thermal model (reference model in red, non-

updated thermal model in blue)


43/48

Earthquake safety assessment of arch dams based on

nonlinear dynamic analyses

Sujan Malla1

1Axpo Power AG, Parkstrasse 23, CH-5401 Baden, SWITZERLAND

[email protected]

Extended Abstract

A program of systematic assessment of the earthquake safety of all significant dams in

Switzerland is scheduled to be completed by the end of the current year 2013. A document

containing state-of-the-practice guidelines for this investigation program was issued by the

Swiss Federal Office of Energy in 2003. The primary goal of this program is to ensure the

safety of the downstream population against loss of life and property damage (Panduri et al.,

2012). For this purpose, it must be shown that the safety evaluation earthquake (SEE) would

not cause a dam failure resulting in an uncontrolled release of reservoir water. Depending on

the dam height and the reservoir volume, a dam is categorized as class 1, 2 or 3 and has to be

checked for the SEE ground motion with a return period of 10,000, 5,000 or 1,000 years,

respectively.

At the time of the design of most of the existing dams, the earthquake action was usually

considered as a pseudo-static loading and typically a horizontal seismic coefficient of 0.1 g

was assumed. This is clearly inadequate since the dynamic earthquake motion would be

considerably amplified in a dam resulting in much higher accelerations. In practically all

major arch dams, the crest region would experience horizontal accelerations exceeding 1.0 g

in the event of the SEE, even when the ground motion at the rock level has a horizontal peak

ground acceleration (PGA) of the order of 0.20 g only.This paper presents two examples of earthquake safety assessment of arch dams in

Switzerland using nonlinear dynamic analyses. All the dynamic calculations were performed

with the help of the general-purpose finite element (FE) software ADINA (ADINA R & D,

2008). For the dynamic analyses, the three components of earthquake excitation were

simulated by artificially-generated spectrum-compatible acceleration time histories. The

hydrodynamic pressure was modeled as added masses acting perpendicular to the upstream

face of the dam.

The linear-elastically computed tensile stresses during the earthquake shaking would usually

exceed the dynamic tensile strengths of the contraction and lift joints, which are substantially

weaker in tension than the monolithic mass concrete. Thus, the vertical contraction joints

would open and cracks would develop at the horizontal lift joints, possibly leading toformation of detached concrete blocks in the crest region, which is subjected to high

earthquake accelerations. Such detached blocks could thus undergo sliding and rocking

motions during a strong earthquake.

The seismic stability of potentially detached concrete blocks of various heights in the crest

region was analyzed using two-dimensional (2D) models. In view of the geometric constraints

in an arch dam, such a detached concrete block cannot move beyond the downstream face of

the dam. This was simulated in the numerical model by means of gap elements. The cracked

lift joint was represented in the model as contact surfaces. The base of the detached block was

subjected to the amplified acceleration at the crack level, as obtained from the 3D linear-

elastic analysis of the arch dam (Malla and Wieland, 2006).


44/48

Earthquake safety assessment of Roggiasca arch dam

The 68 m high Roggiasca arch dam was completed and first impounded in 1965. With a crest

thickness of 2.5 m and a maximum thickness at the base of 7.5 m only, the dam has a

relatively high Lombardi slenderness coefficient of 21. At the location of this dam, the

10,000-year SEE has a horizontal PGA of 0.17 g.

The earthquake response of the dam was analyzed using a 3D FE model, which comprised the

dam, the foundation rock, the sediment deposit on the upstream side and the soil fill on the

downstream side. The largest accelerations and stresses due to the earthquake shaking occur

at the middle and also the two quarter points of the dam crest. The analysis results showed

that the upstream sediment deposit and the downstream soil fill would not play a significant

role in the dynamic behavior of the dam.

The linear-elastic analysis showed that the largest compressive stress due to the combination

of static and dynamic loads is about -10 MPa, which is not a problem for the dam concrete

with a static compressive strength of 46 MPa. During the earthquake shaking, high horizontal

tensile stresses of up to about 4 MPa oriented in the arch direction occur in the crest region, inspite of the compressive stresses under the static loads (self-weight and hydrostatic pressure).

If the thermal stresses in winter and the effect of the ongoing chemical expansion of the dam

concrete are also considered, the tensile stresses would be even higher.

As the Roggiasca arch dam is rather thin, any detached block in the crest region would be

quite slender. Such a detached block has a tendency to undergo only rocking with virtually no

sliding. However, the nonlinear dynamic analysis of the possibly detached blocks showed that

the rocking motion would not be very large and would cause dynamic crack openings of only

a few millimeters at the base. The detached blocks remain dynamically stable even when

subjected to peak horizontal accelerations about twice as large as the pseudo-static

overturning acceleration. This behavior can be explained by the fact that acceleration spikes

would have a relatively high frequency of about 5 Hz corresponding to the dominant natural

frequency of the dam. Hence, such a spike would exceed the pseudo-static overturning

acceleration only for a very short time of less than one-tenth of a second, which is too short to

result in any significant block rotation. A review of literature on dynamic overturning of rigid

blocks subjected to ground motion also confirmed that acceleration peaks would have to be

many times larger than the pseudo-static overturning acceleration for a block with dimensions

of the order of a few meters to be toppled by a dynamic base excitation at such a frequency

(Shi et al., 1996; Makris und Roussos, 2000).

Earthquake safety assessment of Gigerwald arch dam

The 147 m high Gigerwald dam was completed and first impounded in 1976. The results of

the linear dynamic analysis of a 3D FE model of this double-curvature arch dam showed that

the crest region would be subjected to high amplified horizontal accelerations of up to about

2.0 g during the 10,000-year SEE, which has a horizontal PGA of 0.19 g. Moreover, high

horizontal tensile stresses of up to about 6 MPa would occur in the central crest region. In

reality, such horizontal tensile stresses cannot develop due to the presence of the vertical

contraction joints, whose tensile strength is very small.

The next step was to perform the dynamic analysis of a 3D FE model in which all 23 vertical

contraction joints were simulated as frictional contact surfaces and the dam concrete was

assumed to be uncracked. This analysis showed that the earthquake shaking would cause

relative sliding displacements of up to about 10 mm between the adjacent blocks and thecontraction joints would open by up to about 4 mm. The maximum compressive stress in the


45/48

dam obtained from this nonlinear analysis approaches nearly -19 MPa (excluding the corner

singularity at the dam-rock interface), which is not a problem for the dam concrete. In

comparison, the maximum compressive stress in the case of the linear analysis is

about -16 MPa. In spite of the joint displacements, the dynamic displacements and

accelerations of the dam computed in the nonlinear analysis do not deviate significantly from

those in the corresponding linear analysis. The main difference lies in the absence of anysignificant horizontal tensile stresses in the arch direction in the central crest region in the

case of the nonlinear analysis owing to the presence of the contraction joints. During the brief

openings of the contraction joints, the upper portion of a dam block acts temporarily as a

cantilever, due to which relatively high transitory vertical tensile stresses exceeding 6 MPa

appear on the downstream face of the dam. Hence, horizontal cracks are likely to form,

especially at the lift joints, possibly resulting in the detachment of the uppermost portion of a

central block from the rest of the dam body.

The safety of the possibly detached concrete blocks in the central crest region subjected to the SEE

shaking was assessed using simplified 2D models. This analysis showed that an 8 m high

detached block could slide during the SEE by up to about 50 cm towards the reservoir and therocking motion of this detached block would result in crack opening displacements of up to

about 7 cm. However, the detached block would remain stable during and after the earthquake

and the earthquake damage would not lead to an uncontrolled release of water.

For a less conservative dynamic stability assessment, a nonlinear analysis was also performed

using a 3D model in which the uppermost 8 m of the central block was assumed to be

detached from the rest of the dam along the contraction joints at the sides and an assumed

horizontal crack along a lift joint. This 3D analysis showed th

handout benchmark v1 all

Documents