Dams and earthquake 2 -6 -<r~c Dams and earthquake Proceedings of a conference held at the Institution of Civil Engineers, London, on 1-2 October 1980 Thomas Telford Limited, London, 1981 Organized by the Institution of Civil Engineers in association with the British Section of the International Commission on Large Dams and the Society for Earthquake and Civil Engineering Dynamics—the British Section of the International Association for Earthquake Engineering: also sponsored by UNESCO Organizing Committee: Mr R. G. T. Lane (Chairman); Mr D. A. Howells; Professor J. K. T. L. Nash; Mr F. F. Poskitt Cover photograph by courtesy of Sir Alexander Gibb and Partners, Consulting Engineers Published for the Institution of Civil Engineers by Thomas Telford Limited, PO Box 101, 26-34 Old Street, London EC1P 1JH First published 1981 ISBN: 0 7277 0123 1 © The Institution of Civil Engineers, 1981 All rights, including translation, reserved. Except for fair copying, no part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the publisher. Requests should be directed to the Managing Editor, Publications Division, Institution of Civil Engineers, PO Box 101, 26-34 Old Street, London EC1P 1JH The Institution of Civil Engineers as a body does not accept responsibility for the statements made or for the opinions expressed in the following pages Printed in Great Britain by Burlington Press (Cambridge) Ltd, Foxton, Cambridge CONTENTS Opening address. Marcus Fox, MP 1 SESSION 1: SEISMICITY 1. Optimum seismic input in the design of large structures. R E. Long 3 7 2. The definition of seismic risk at dam sites. R Chaplow 9 3. Substitute short duration earthquake accelerograms for non-linear analysis. O. C Zienkiewicz, N. Bicanic and R Fejzo 17 4. Seismic risk studies for large dam projects in northern Iraq. M. B. Tosic 23 / 5. Design earthquake recurrence analysis. E. R Ries, N. R Vaidya and A. P. Michalopoulos 31 6. Aseismic design considerations for a large arch dam. R Dungar 37 Discussion on Papers 1-6 45 SESSION 2: MATERIALS BEHAVIOUR WHEN SUBJECTED TO EARTHQUAKE 7. Materials behaviour under earthquake loading. P. Bertacchi 57 8. Liquefaction potential of Wash sands. T. D. Ruxton 61 9. Dynamic behaviour of rockfill dams. M. Nose and K. Baba 69 10. Friction parameters for the design of rock slopes to withstand earthquake loading. S. R Hencher 79 11. Peculiarities of the seismic-resistant analysis of earth dams with pervious gravelly shells. A. J. L. Bolognesi 89 f 12. Lessons from the performance of earth dams during earthquake. H. Bolton Seed 97 13. Seismic stability analysis of Kokubo dam. K. Mori, K. Ishihara, K. Takeya and K. Kanazashi 105 14. The response of concrete to short term loading. J. W. Dougill 113 15. Failure and damage: earth and rock fill. D. A. Howells 125 16. Thirty years of research work on earthquake resistance of hydraulic structures in China C Shen and H. Chen 129 17. Soil amplified earthquake ground motions. J. N. Protonotarios 137 Discussion on Papers 7-17 145 SESSION 3: STRUCTURE BEHAVIOUR WHEN SUBJECTED TO EARTHQUAKE 18. A simplified method for the earthquake resistant design of earth dams. S. K. Sarma 155 19. Seismic analysis and design consideration for concrete dams. K. J. Dreher 161 20. Computational models for the transient dynamic analysis of concrete dams. O. C. Zienkiewicz, E. Hinton, N. Bicanic and P. Fejzo 171 21. Earth dam analysis for earthquakes: numerical solution and constitutive relations for non-linear (damage) analysis. O. C Zienkiewicz, K. H. Leung, E. Hinton and C T. Chang 22. Seismic studies in relation to the design of Alicura dam and appurtenant works. B. Gilg 23. Natural frequencies and response characteristics of gravity dams. D. Altinisik and R T. Severn 24. Aseismic design of arch dams: particularly the contribution from the reservoir, and multiple-support excitation of the base. D. Altinisik, P. A A Back, S. R Ledbetter, R T. Severn and C A Taylor 25. The use of models in assessing the behaviour of concrete dams. G. Oberti and A Castoldi 26. Theoretical assessment of the behaviour of arch dams for seismic loading. G. L. Hutchinson and T. G. Tsicnias 27. Criteria for the earthquake resistant design of concrete dams. J. Laginha Serafim Discussion on Papers 18-27 SESSION 4: ENVIRONMENT AND RISK INCLUDING INDUCED SEISMICITY 28. Dams, natural and induced earthquakes and the environment E. T. Haws and N. Reilly 29. Hazards of natural and induced seismicity in the vicinity of large dams. P. L. Willmore 30. Computed and observed deformation of two embankment dams under seismic loading. M. P. Romo and D. Resendiz 31. Full-scale forced vibration studies and mathematical model formation of arch concrete dams. T. A Paskalov, J. T. Petrovski and D. V. Jurukovski 32. The potential for induced seismicity—geological approaches. B. O. Skipp and M. Higgins 33. Probable risk estimation due to reservoir induced seismicity. S. K. Guha, J. G. Padale and P. D. Gosavi Discussion on Papers 28-33 Closing address. R G. T. Lane and H. Bolton Seed Opening address MARCUS FOX, MP (Parliamentary Under Secretary of State, Department of the Environment) The consequences of failure of a large dam are and the damage extended to at least 150km from many times greater than those of any other type the epicentre. of structure. Civil engineers have recognized this themselves, and I am most impressed by We think of severe earthquakes occurring the world-wide co-operation and exchange of primarily in known seismic zones, but I am told information that takes place through the that these are virtually impossible to predict International Commission on Large Dams. and earthquakes have occured outside those zones in areas considered to be geologically This Conference, which brings together both stable. seismologists and civil engineers, has been convened to consider just one aspect of dam Another phenomenon is the fact that earthquakes safety - their ability to resist earthquake. have been observed following the impounding of Nevertheless, the subject matter is very wide water in large reservoirs, indicating that large starting with the study of seismicity, moving reservoirs can induce their own earthquakes. on to the behaviour of materials and structures when subjected to earthquake, and ending with The engineer is faced with many unknown factors, a study of the environment and risks, including and the object must be to achieve greater induced seismicity. understanding of seismic activity, to improve designs to allow for these movements and to A study of an atlas indicating the localities acquire greater knowledge of the characteristics of the most severe earthquakes reveals that the of the materials used. All these are essent­ UK is relatively stable, although not completely ials for the improvement of the earthquake stable because occasional tremors do occur and resistance of dams. some occasional slight damage to dams is report­ ed. There was an earthquake near Carlisle in Engineers must always bear in mind the dev­ December last year, and there have been recent astation that a community can suffer on the tremors in the Stoke-on-Trent area. Scotland, failure of a large dam. Their design decisions too, has had tremors around the villages of must always be of the highest order. I trust Comrie and Menstrie. But the majority of prob­ that this Conference helps them to achieve this lems arise in other parts of the world. If high standard and it gives me great pleasure, British consulting engineers are to compete therefore, that the Institution of Civil with engineers from other countries in obtain­ Engineers has organized the Conference in ing commissions to design and supervise the association with the British Section of the construction of large dams in those parts of International Commission on Large Dams and the the world which are physically less stable, Society for Earthquake and Civil Engineering they must ensure that they are familiar with Dynamics, the British Section of the Inter­ the problems encountered and also practised in national Association for Earthquake Engineering all the modern techniques available to combat and UNESCO. the effects of earthquake. I am most impressed that some 140 delegates We all know how devastating earthquake can be. are here from over 20 countries. I welcome The disaster in Tangshan in north-east China, you on behalf of Her Majesty's Government and which occurred on 27 July 1976, is reported our people. to have killed no less than 650 000 people Dams and earthquake, TTL, London, 1981 1 1 Optimum seismic input in the design of large structures R. E. LONG, BSc, PhD, FRAS, MBCS, University of Durham From a consideration of the nature and characteristics of seismic waves, a model for testing design for earthquake failure is proposed. The model, referred to as an integrated model, considers an earthquake as a fault slip and places this within the geology. Direct calculation of energy entering the dam is made by modelling source, geology and dam as one complete unit. The problems of implementation of this model are discussed and its potential value estimated. INTRODUCTION both P and S. This illustration shows in a 1. This paper aims to place the design of simple way the waves generated by a moving dams (and particularly earth dams) in their source. To this simple model can be added true seismological context. The seismologist plausible variants which can account for has a growing understanding of motions in the the wide range of near field motions found in near field of a large earthquake. The dam practice. designer, however, generally approximates these motions to give a simple model on which These waves propagate through the geology to test a design. It is not apparent that from the fault source to the dam. That the such simple models necessarily test a design geology modifies the motions is well known. with all the types of motion that could cause What is less well known is that the trans­ failure. Thus this paper seeks a closer mission function varies with the angle 6f fusion of seismology and design methods. incidence of the input motion onto the geology. This problem was studied in detail by Haskell 2. The aim is not to present a complete new (2). In engineering terms the peaks of design method but to investigate the possib­ amplification of a layer occur at frequencies ility of a method where seismological data can which are angle of incidence dependent. form an integral part of design. Normally seismic motions are assessed on one model by 5- The effects of a layered geological the seismologists, and then transferred to foundation can be expressed either as a another model for engineering design. In the frequency dependent transmission function or as ideal situation these two models should be a series of separate wavelets emerging at the combined. Thus we investigate an integrated surface for a particular input. This latter design method where earthquake source, found­ approach reveals the basic mechanism. A series ation geology and dam are considered as one of multiply reflected and refracted arrivals overall single unit. are generated by each of the input wavelets. These reflections and refractions will be The Seismological Problem accompanied by changes of wave type, P to S and 3. The basic reason for proposing such an S to P. Thus a simple P input gives rise to approach to design is that it allows full a mixture of P and S energy at the surface of account to be taken of the nature and charac­ a layered geology, and similarly for an S input. teristics of seismic motions. The need for this becomes evident by considering such 6. The input waveform therefore becomes motions in general terms. Motions from an extended in time with each input wavelet trans­ earthquake are generated by relative displace­ formed into a series of (interfering) P and S ments of the sides of a fault. The waves wavelets which are input to the dam itself. propagated from such a moving source can be These several P and S arrivals interfere to considered as a set of individual wavelets. give a complex particle motion which is rarely The multiplicity of wavelets received at rectilinear, but more ellipsoidal. Thus the points in the near field is well demonstrated dam is subjected to motion which contains by considering simple models. Fig. 1 shows virtually all possible motion in a generally accelerations generated at the surface by a random sequence. slip propagating over a finite area of a fault (1). The particular source parameters are 7. This refers specifically to the motions, here chosen to show the structure of the motion but the direction of propagation is not random The several impulses of acceleration corres­ and is predominantly directed away from the pond to the start and stop of the slip, each source. The source is not static but extended of which generate an acceleration pulse in over the slip zone of the faulting starting at Dams and earthquake, TTL, London, 1981 3 SEISMICITY one point on the fault and propagating outward clusions. Firstly the type of energy incident from that point. Thus the effective angle of is not restricted to any particular wave type. incidence of this energy through the geology Secondly the response to a particular wave and into the dam varies with time during the type cannot be considered separately from 'any earthquake. For a large, close event the other wave type. The word type here not only change in angle can be considerable. The effect refers to P or S but also the several separate of the layered geology changes with angle and arrivals of P and S. Thirdly, since the therefore effectively with time. several wave types have differing velocities and differing angles of approach into the 8. Thus we may summarise the incident motion structure, the composite motion cannot be as a time variant multiplicity of interfering considered as a single input time history over wavelets generated by a moving source and any one surface that might be drawn in the travelling into the dam via a geology which structure. This implies that application of acts as a time dependent filter. This pattern a single time history, say to the base of a of input motion is of totally different model, is not realistic, nor is it a realistic character to motions applied in a quasi-static approximation. environment where the travelling nature of the wave and the effective time dependence of the The model for design geological filter is ignored. The result is 13. These observations suggest that if input to widen the total range of input motions. An motions are to be totally realistic the test integrated model with the source modelled in model should not only include the dam and the geology automatically includes a solution foundations, but also the fault with earth­ of the attenuation problem and the problem of quake placed in the foundation geology. assessing motion from specific events. Checks for failure would then be made by allowing an appropriate slip on the fault and The inclusion of a dam observing resultant stress conditions. Such a 9. Into this environment we may now place a design method is clearly capable of providing dam. A structure whose dimensions are much a close approximation to the real problem. smaller than a wavelength would tend to move as a unit with acceleration and deceleration 14. The implication of this method however forces providing the major hazard. For such a presents a number of problems. Firstly the case the application of a specific motion to elastic properties of the geology must be the foundations or to the whole structure known sufficiently accurately with accurate would be appropriate. location of faults at depth. Secondly the source parameter (area of slip, distribution 10. Such an approach is not applicable to a of the amplitude of slip motion over the large dam whose dimensions are of the same slipped region, etc.) need to be known or order as the wavelength of the seismic energy assessed. Thirdly the mathematical technique input. For such large structures the waves required to calculate the motion generated by are propagated into the dam so as to form a a fault slipping in a layered geology have yet pattern of motion over the dam. This pattern to be developed. Fourthly the integrated is the result of the interference of waves method would normally be implemented using from the multiplicity of ray paths from source finite element or different techniques, when through the intervening geology to the struct­ the size of the model in comparison with the ure. seismic wavelength would require a large number of nodes with appropriate computer problems of 11. The surface motions resulting from a handling large arrays. plane wave incident on an earth dam can be cal­ culated by considering the earth dam as a 15. Consider these problems in order: simple hill in a homogeneous geology. Fig. 2, The geology is normally investigated by a based on Bouchon (3), shows how surface motions borehole. These are of limited depth and vary with position on the surface and angle of therefore give data of only the upper few incidence. These studies explaining the metres normally restricted to the site itself. amplification of motion by topography also To- gain accurate knowledge of deeper layers show in essence the problem of inputting seis­ would require deeper boreholes than normal and mic energy into an earth dam. The complexity then -bhese would not show up lateral variation of motions arise from a scattering of energy in sufficient detail. Such investigations by the surface discontinuity formed by the would be required over an area much larger dam. The implication for dam design is that than the site. The solution in an earthquake motion and the implied stress distribution are environment is to measure the transmission assymetric and clearly very different from function from fault to dam directly. Small results obtained from applying non-travelling earthquakes occurring on the fault can be waves to the structures. In the real environ­ recorded at site. These records can be ment several such input waves, each with a decomposed to remove the source function different angle of approach and therefore when the transmission function in the time different stress patterns within the dam, domain remains. This analysis can be done for would interfere to form a continuously changing a variety of events distributed over the fault pattern of stress with time. to give the changing transmission function with position. A generalised function for a 12. This approach leads to a number of con­ station in Iran is shown in fig. 3- This 4