Dr. Martin Wieland Chairman, ICOLD Committee on Seismic Aspects of Dam Design Electrowatt-Ekono AG, Hardturmstrasse 161, CH-8037 Zurich, Switzerland
Abstract Hsinfengkiang buttress dam (1962 earthquake in China, 105 m high), Koyna gravity dam (1967 earthquake in India, 103 m high), Pacoima arch dam (1971 and 1994 earthquakes in California, 116 m high), Rapel arch dam (1985 earthquake in Chile, 110 m high) and Sefid Rud buttress dam (1990 earthquake in Iran, 106 m high) are the highest concrete dams, which have been exposed to very strong ground shaking and have suffered different degrees of damage but none of them has failed. Major repair and strengthening works were carried out for the Hsinfengkiang, Koyna and Sefid Rud dams and all dams are in operation. These dams, as most of the existing dams, were designed against earthquakes, using seismic design criteria and/or methods of seismic analysis, which are considered as obsolete or incorrect today. They have also experienced ground motions that were much more severe than those expected at the time of construction. In 1999, a secondary fault passing almost normal to the Shih-Kang concrete diversion dam experienced substantial vertical movements, destroying two spillway gates near the right abutment during the Chi-Chi earthquake in Taiwan,. Water flowed through the damaged bays, but without catastrophic release of the reservoir. Unlike other actions from the natural and man-made environment, which dams have to resist, earthquakes pose probably the greatest challenge to dam engineers as earthquake ground shaking affects all structures (dam and appurtenant structures) and all components (hydromechanical, electromechanical etc.) at the same time. Thus all these elements have to be able to resist some degree of earthquake actions. The paper discusses the seismic design criteria for concrete dam projects, methods of dynamic analysis for strong ground shaking, and provides information on possible antiseismic design features for large arch dams.
Introduction Since the 1971 San Fernando earthquake in California, major progress has been achieved in the understanding of earthquake action on concrete dams. The progress was mainly due to the development of computer programs for the dynamic analysis of dams. However, it is still not possible to reliably predict the behaviour of dams during very strong ground shaking due to the difficulty in modelling joint opening and the crack formation in the dam body, the nonlinear behaviour of the foundation, the insufficient information on the spatial variation of ground motion in arch dams and other factors. Also, considerable progress has been made in the definition of seismic input, which is one of the main uncertainties in the seismic design and seismic safety evaluation of dams.
It has been recognized that during strong earthquakes such as the maximum credible earthquake (MCE), the maximum design earthquake (MDE) or the safety evaluation earthquake (SEE) ground motions can occur, which exceed those used for the design of large dams in the past. Already a moderate shallow-focus earthquake with a magnitude of say 5.5 to 6 can cause peak ground accelerations (PGA) of 0.5 g. However, the duration of strong ground shaking of such events is quite short and the predominant frequencies of the acceleration time history are also rather high. Therefore, smaller concrete dams may be more vulnerable to such actions than high dams, which have predominant eigenfrequencies that are smaller than those of the ground acceleration. We have to recognize that most of the existing dams have been designed against earthquake actions using the pseudo-static approach with an acceleration of 0.1 g. In regions of high seismicity like Iran the PGA of the SEE may exceed 0.5 g at a number of dam sites. Therefore, some damage may have to be expected in concrete dams designed by a pseudostatic method with a seismic coefficient of 0.1. Because of the large differences between the design acceleration and the PGA, which can actually occur during a strong earthquake, and because of the uncertainties in estimating the ground motion of very strong earthquakes at a dam site, mechanisms are needed that ensure that a dam will not fail if the design acceleration is exceeded substantially. In the case of large dams, ICOLD recommends to use the MCE as the basis for the dam safety checks and dam design. Theoretically no ground motion should occur, which exceeds that of the MCE. However, in view of the difficulties in estimating the ground motion at a dam site, it is still possible that larger ground motions may occur. Some 50 years ago, many structural engineers considered a value of ca. 0.2g as the upper bound of the PGA, but today with more earthquake records available, the upper bound has exceeded 1 g and some important structures have already been checked against such high ground accelerations.
Seismic Design Criteria for Hydro-electrical Dam Projects According to ICOLD Bulletin 72 (1989), large dams have to be able to withstand the effects of the MCE. This is the strongest earthquake that could occur in the region of a dam, and is considered to have a return period of several thousand years (typically 10000 years in regions of low to moderate seismicity). Per definition, the MCE is the largest event that can be expected to affect the dam. This event can be very powerful and can happen close to the dam. The designer must take into account the motions resulting from any earthquake at any distance from the dam site, and possible movement of the foundation if a potentially active fault crosses the dam site. Having an active fault in the foundation is sometimes unavoidable, especially in highly seismically active regions, and should be considered as one of the most severe design challenges requiring special attention. It has to be kept in mind that each dam is a prototype structure and that the experience gained from the seismic behaviour of other dams has limited value, therefore, observations have to be combined with sophisticated analyses, which should reflect reality as close as possible. We also have to realize that earthquake engineering is a relatively young discipline with plenty of unsolved problems. Therefore, every time there is another strong earthquake some new unexpected phenomena are likely to emerge, which have implications on regulations and codes. This is particularly true for dams as very few modern dams have actually been exposed to very strong ground motions. As mentioned earlier, the time of the pseudostatic design with a seismic coefficient of 0.1 has long passed. Of course, this concept was very much liked by designers because the small seismic coefficients being used did not require any special analyses and the seismic requirement could easily be satisfied. As a result, the seismic load case was usually not the governing one. This situation has changed and the earthquake load case has become the governing one for the design for most high risk (dam) projects, especially in regions of moderate to high seismicity.
As a general guideline, the following minimum seismic requirements should be observed: (i) Dam and safety-relevant elements: Design earthquakes (i.e. operating basis earthquake, OBE, and SEE) are determined based on a seismic hazard study and are usually represented by appropriate response spectra (and the PGA or the effective peak acceleration). (ii) All appurtenant structures and non safety-relevant elements of dam: Use of local seismic building codes (including seismic zonation, importance factor, and local soil conditions) if no other regulations are available. However, the seismic design criteria should not be less than those given in building codes. As shown in Fig. 1 the earthquake damage to appurtenant structures can be very severe. (iii) Temporary structures and construction phase: The PGA for temporary structures and the construction phase should be of the order of 50% of the PGA of the design earthquake of the seismic building code. For important cofferdams a separate seismic hazard assessment may be needed. According to Eurocode 8 (2002), the design PGA for temporary structures and the construction phase, PGA c can be taken as PGA c = PGA (t rc /t ro ) k
where PGA is according to the building code, t ro = 475 years (probability of exceedance of 10% in 50 years), for k a value between 0.3 and 0.4 can be used depending on the seismicity of the region and t rc = approx. t c /p
where t c is the duration of the construction phase and p is the acceptable probability of exceedance of the design seismic event during this phase, typically a value of p = 0.05 is selected. Therefore, assuming k = 0.35 and a construction phase of an appurtenant structure of 3 years results in PGA c = 0.48 PGA, and for a cofferdam, which may stay for 10 years, we obtain PGA c
= 0.74 PGA (Note: For major coffer dams, t ro , may be larger than 475 years and thus the PGA shall be taken for that longer return period). This numerical example shows that the seismic action for temporary structures and for construction phases can be quite substantial. In many cases the effects of seismic action during dam construction has been underestimated or ignored.
In general, building codes exclude dams, powerplants, hydraulic structures, underground structures and other appurtenant structures and equipment. It is expected that separate codes or regulations cover these special infrastructure projects. However, only few countries have such regulations. Therefore, either the ICOLD Bulletins (ICOLD Bulletin 123, 2002) are used as a reference or the (local) seismic building codes (Eurocode 8, 2002). The seismic building codes are very useful reference documents for checking the design criteria for a dam. As the return period of the SEE of a large dam is usually much longer than the return period of the design earthquake for buildings, which in many parts of the world is taken as 475 years (in the USA and Canada the seismic forces are determined based on a reference return period of 2500 years in the new building codes and the resulting ground accelerations are multiplied by a factor of 2/3; this procedure results in higher seismic forces in areas of low to moderate seismicity as compared to the concept of a constant return period of 475 years for low and high seismicity areas), the PGA values of the SEE should be larger than that of the design earthquake for building structures with a 475 year return period multiplied with the importance factor for high risk projects. If this basic check is not satisfied, then a building located at the dam site would have to be designed for stronger ground motions than the dam. In such controversial situations further investigations and justifications are needed. In the past, the pseudostatic acceleration of 0.1 g was used for many dams. This acceleration can be assumed as effective ground acceleration, which is ca. 2/3 of the PGA. Therefore, the SEE for a dam with large damage potential should not have a PGA of less than say 0.15 g as this would be less than
what was used in the past. However, this is mainly a concern for dam projects in regions of low to moderate seismicity and not for Iran. For the seismic design of equipment, the acceleration response calculated at the location of the equipment (so-called floor response spectra) shall be used. It should also be noted that the peak acceleration at the equipment location is normally larger than the PGA. For example, the radial acceleration on the crest of an arch dam is about 5 times larger than the PGA of the SEE (nonlinear dynamic response of the dam) and up to 8 times larger in the case of the low intensity OBE (linear- elastic behaviour of dam).
Effects of Strong Ground Shaking on Concrete Dams The most important cases of earthquake damage and performance of concrete dams under severe ground shaking are: Pacoima arch dam in USA: During the 1971 San Fernando earthquake, the 113 m high dam was subjected to a base acceleration of about 0.7 g; and a peak acceleration of 1.25g was recorded on rock at the left abutment, slightly above the dam crest. The dam did not develop any structural cracks. The left abutment was strengthened through the installation of post- tensioned anchors to stabilize two large rock wedges that had moved during the earthquake. The peak acceleration near the top of the left abutment during the January 17, 1994 Northridge earthquake were 1.76 g horizontal and 1.6 g vertical. The joint between the left abutment and the left end of the arch, opened up to 5 cm due to movement of the rock wedges on the upper abutment. Minor horizontal cracking of concrete at the left end of the dam, and several minor horizontal and vertical block offsets at the joints occurred. The reservoir was at about mid- dam height during both earthquakes. Koyna dam in India, a 105 m high straight gravity dam, and Hsinfengkiang dam in China, a 103 m high buttress dam, were shaken as the result of nearby earthquakes of Magnitudes 6.5 (1967) and 6.1 (1962), respectively. Both earthquakes were suspected of being caused by reservoir-triggered seismicity. Both dams developed substantial longitudinal cracking near the top. Damage was attributed to design or construction details that would be avoided in modern structures. The two dams were strengthened and are still in service. Sefid Rud buttress dam in Iran: The 106 m high dam suffered damage during the Manjil Earthquake of June 21, 1990. The dam was located within the epicentral area and the peak ground acceleration was estimated at 0.7 g. The dam experienced various forms of damage, including severe cracking along horizontal lift joints in the upper part of the buttresses. The dam suffered significant damage, but had overall satisfactory performance considering that the Manjil Earthquake was probably not far from what can be expected as a maximum event at that site. Lower Crystal Springs dam in USA: The 38.5 m high curved concrete gravity dam built of interlocking concrete blocks, withstood the 1906 San Francisco earthquake without a single crack. The primary fault rupture was located less than 180 m from the dam. During the 1989 Loma Prieta earthquake, the dam again experienced moderate shaking but was once more unaffected. Further case studies are given in ICOLD Bulletin 120 (2001) and Wieland (2002, 2003). Up to now, the performance of large concrete dams has been quite satisfactory. However, it is premature to conclude that they are intrinsically safe or safer than embankment dams, as the number of embankment dams is significantly higher than that of concrete dams. Moreover, concrete dams may have been built to higher seismic design standards than those used for some of the earlier embankment dams. In recent years, roller compacted concrete (RCC) dams have become viable alternatives to traditional concrete dams. Again, there exist hardly any observations about the behaviour of RCC dams during very strong ground shaking. As the horizontal lift joints in RCC dams have lower strength properties than the parent mass concrete, cracks are likely to develop along the lift joints similar to the cracks that have developed in the upper art of the Sefid Rud dam during the 1990 Manjil earthquake.
Earthquake Design Aspects of Concrete Dams There are several design details that are regarded as contributing to a favourable seismic performance of arch dams, i.e.: Design of a dam shape with symmetrical and anti-symmetrical mode shapes that are excited by along valley and cross-canyon components of ground shaking. Maintenance of continuous compressive loading along the foundation, by shaping of the foundation, by thickening of the arches towards the abutments (filets) or by a plinth structure to support the dam and transfer load to the foundation. Limiting the crest length to height ratio, to assure that the dam carries a substantial portion of the applied seismic forces by arch action, and that nonuniform ground motions excite higher modes and lead to undesired stress concentrations. Providing contraction joints with adequate interlocking. Improving the dynamic resistance and consolidation of the foundation rock by appropriate excavation, grouting etc. Provision of well-prepared lift surfaces to maximize bond and tensile strength. Increasing the crest width to reduce high dynamic tensile stresses in crest region. Minimizing unnecessary mass in the upper portion of the dam that does not contribute effectively to the stiffness of the crest. Maintenance of low concrete placing temperatures to minimize initial, heat-induced tensile stresses and shrinkage cracking. Development and maintenance of a good drainage system. The structural features, which improve the seismic performance of gravity and buttress dams are essentially the same as that for arch dams. Earthquake observations have shown that a break in slope on the downstream faces of gravity and buttress dams should be avoided to eliminate local stress concentrations and cracking under moderate earthquakes. The webs of buttresses should be sufficiently massive to prevent damage from cross-canyon earthquake excitations.
Dynamic Response Analysis of Concrete Dams The main factor, which governs the dynamic response of a dam subjected to small-amplitude ground motions, can be summarized under the term damping. Structural damping ratios obtained from forced and ambient vibration tests are surprisingly low, i.e. damping ratios of the lowest modes of vibrations are of the order of 1 to 2% of critical. In these field measurements the effect of radiation damping in the foundation and the reservoir are already included. Linear-elastic analyses of dam-foundation- reservoir systems would suggest damping ratios of about 10% for the lowest modes of vibration and even higher values for the higher modes. Under earthquake excitations the dynamic stresses in an arch dam might be a factor of 2 to 3 smaller than those obtained from an analysis with 5% damping where the reservoir is assumed to be incompressible and the dynamic interaction effects with the foundation are represented by the foundation flexibility only (massless foundation). Therefore, the dam engineer may be willing to invest more time in a sophisticated dynamic interaction analysis in order to reduce the seismic response of an arch dam. Unfortunately, there is a lack of observational evidence, which would justify the use of damping ratios in seismic analyses of concrete dams. Anyway, linear-elastic seismic analyses are only suitable for the OBE checks or smaller dams in regions of low to moderate seismicity where the dynamic tensile stresses in a dam are relatively small. Under strong ground shaking caused by the MCE or SEE tensile stresses are likely to occur that exceed the dynamic tensile strength of mass concrete. In addition, there are contraction joints and lift joints, which have tensile strength properties that are inferior to those of the parent mass concrete. Therefore, in the highly stressed upper portion of an arch dam the contraction joints will start to open first and cracks will develop along the lift joints (this coincides with the seismic behaviour of Sefid Rud dam). Along the up- and downstream contacts between the dam and the foundation rock local stress concentrations will occur, which will lead to the formation of cracks in the concrete and the foundation rock. Little information exists on this type of cracks. But such cracks are also likely to
develop at the upstream heel of the dam under the hydrostatic water load. Thus, the dynamic deformations of the dam will mainly occur at these contraction and lift joints and along a few cracks in the mass concrete. The remaining parts of the dam will behave more or like as rigid bodies and will exhibit relatively small dynamic tensile stresses. Joint opening and crack formation will also lead to higher compressive stresses in the dam. However, these dynamic compressive stresses are unlikely to cause any damage and the engineer can focus on tensile stresses only. For the analysis of joint opening other numerical models are needed than for the linear-elastic dynamic analysis of a dam. Two concepts are used for the nonlinear analysis of a cracked dam: smeared crack approach: In this method the finite element model used for the linear dynamic analysis of the dam-foundation-reservoir system can be used by replacing the elastic concrete model by a generalised constitutive model of cracked mass concrete. This method was proposed by researchers but has hardly been applied in practise. The concrete models need a large number of material parameters (up to 30 different parameters) most of them are unknown and would have to be determined by parameter identification methods. In addition, the smeared crack approach ignores the formation of cracks along existing joints and, thus, does not grasp the main features of the dynamic response of the dam. discrete crack approach: In this method a number of contraction and lift joints and also the dam-foundation contact, where the dynamic tensile strength is significantly less than in the parent mass concrete, are modelled by interface elements. This method has been used for several dams (Wieland, 2003). The main advantages of discrete crack models are their simplicity (only information on strength properties of joints is needed) and their ability to model the observed behaviour of dams during strong ground shaking. Further developments are still needed. In nonlinear dam models radiation damping may play a less prominent role as in the case of a linear- elastic analysis where this issue is still controversial. Once cracks along joints are fully developed, the viscous damping in the cracked region of a dam may be replaced by hysteretic damping in the joints. As these damping mechanisms are still not well understood or are too complex to be considered in practical analyses, it is recommended to perform a sensitivity analysis in which the effect of damping on the dynamic response of a dam is investigated. The dynamic tensile strength of mass concrete, f t , is another key parameter in the seismic safety assessment of concrete dams. It depends on the following factors: uniaxial compressive strength of mass concrete, f c , different correlations between f t and f c
have been proposed in the literature; age effect on concrete strength (it may be assumed that the OBE or SEE occurs when the dam has reached about one-third of the design life); strain-rate effect (under earthquake loading, the tensile strength increases); and size effect (the tensile strength depends on the fracture toughness of mass concrete and the thickness of the dam). If the size effect is considered, the dynamic tensile strength of mass concrete in relatively thick arch- gravity dams drops to below 3 to 4 MPa. An additional factor, which is hardly considered in arch dam analyses, is the spatial variation of the earthquake ground motion. This feature can be simulated approximately by adding mass to the foundation model close to the dam abutment. The criteria for the assessment of the safety of the dam foundation during strong ground shaking need further improvements. Today, the dam analyst delivers the seismic abutment forces to the geotechnical engineer, who performs stability analyses.
Seismic Strengthening of Arch Dams When methods for the seismic risk reduction of concrete dams are evaluated, the engineers usually arrive at the following options:
Lowering of the reservoir level in an existing dam: This option is feasible for existing dams where the seismic risk of a dam has to be reduced immediately, but this is only a temporary solution; Modifications in the shape of a dam (change of the dam type, thickening of the crest, addition of filets at the abutments, shape modifications to achieve symmetry/antisymmetry in the lowest mode shapes, etc.): This is a very costly solution for existing dams, however, shape optimization is standard practice for new dams; Post-tensioning of monoliths with vertical anchors: This method is feasible for the heightening, strengthening and repair of gravity-type dams; one has to note that very high prestressing forces would have to be applied, to prevent high tensile stresses during strong ground shaking. In arch dams the high seismic tensile stresses are primarily in arch direction near the crest of the dam. Shear keys: Shear keys prevent relative movements of adjacent concrete blocks and prevent them from falling into the reservoir as any relative movement leads automatically to a an increase in arch stresses. Special solutions for stabilizing concrete blocks are needed for arch dams with large crest spillways where the crest arch is discontinuous. Design of a seismic belt: This option is feasible for new arch, arch-gravity and gravity dams in zones of high seismicity, but it may be too costly for existing dams. Seismic belts prevent detached concrete blocks in the crest region of an arch dam from falling into the reservoir similar to shear keys. In addition, any joint contraction opening will lead to energy dissipation. The following arch dams have a seismic belt: Rapel (Chile), Sir (Turkey), Katse (Lesotho), and Inguri (Georgia). Dampers: Damping elements have not yet been used for massive dams. The most effective means of energy absorption would be by using concrete with good damping behaviour and high fracture toughness, by contraction joint reinforcement (seismic belt) in the upper portion of the dam, by shear keys with special energy absorption capabilities or by designing joints where controlled sliding can take place (friction). For efficient energy absorption zones with high strains or large displacements (dam crest) have to be targeted, but these are generally small and, therefore, energy dissipation is not yet a feasible option.
Seismic Belt in Arch Dam At the crest of large arch, arch-gravity and gravity dams the earthquake ground acceleration is often amplified by a factor of about 5 in gravity dams and up to 8 in arch dams. In addition, relatively large tensile stresses occur in the upper central region of arch and arch-gravity dams during an earthquake. These stresses can easily exceed the dynamic tensile strength of mass concrete, which is in the range of 3 to 4 MPa. The tensile stresses lead to cracks in the upper portion of the dam. The cracks may develop preferably along the grouted vertical construction joints and the horizontal lift joints. Various experimental studies have confirmed that earthquake cracks are to be expected in the upper central portion of large concrete dams. Thus the concrete blocks separated by cracks may overturn or slide into the reservoir as integrity of the dam gets lost in this zone. The stabilising effect of the arch thrust in this region disappears when the vertical joints open. This leads to stress redistributions and further damage in a dam. The integrity of detached concrete blocks can be greatly improved by tying them together at the crest. This will prevent them from overturning. The sliding failure mode of detached concrete blocks is not really feasible as long as horizontal cracks will develop along the lift elevations. Even during very strong earthquakes like the 1990 Manjil earthquake, the maximum sliding movement along horizontal cracks was less than ca. 3 cm. Overturning of slender blocks in thin arch dams is therefore a more feasible failure scenario than sliding. Overturning can be easily be prevented by a relatively small stabilising force acting at the top of the slender block. This force can be produced by a seismic belt, which consists of reinforcing steel bars or cables placed horizontally along the dam axis at the crest. The length of the belt could be limited to about 80% of the crest length as the central upper portion of the dam is the one, which exhibits the highest seismic stresses. As the vertical contraction joints may
open near the crest, the steel bars must be ductile and able to plastify over a larger length across the joints. Special detailing of the reinforcement across the joints will be needed to ensure proper energy dissipation and durability. The reinforcing steel should have a high ratio of tensile strength to yield strength and a large ultimate strain. The main design criterion for a seismic belt is to provide an anchoring force to stabilise a detached concrete block. This force has to balance the overturning moment of the detached concrete block under consideration. The main problem in this analysis is to determine the maximum size of a detached concrete block, which has to be stabilised. A seismic belt should be considered as a feasible option in the earthquake-resistant design of the crest of thin arch dams, as the central crest region forms the weakest part in most arch dams, in particular when the water level in the reservoir is low.
Feasibility of Joint Reinforcement During very strong ground shaking, joint movements (joint openings, shear displacements) can take place in high arch dams. The joint opening is more or less confined to the central upper portion of arch dams. Contraction joint opening may lead to damage of water stops located close to the dam surface. Studies have been carried out in China to prevent or limit contraction joint opening displacements in very high arch dams by means of steel reinforcement of the joints. As the contraction joint at the upstream face will be closed after an earthquake due to the hydrostatic pressure in the reservoir and in view of the fact that an earthquake, which can damage conventional water stops occurs very seldom and since the water seepage through the damaged joint will be small, this damage may be accepted. However, if it is not possible to lower the reservoir after an earthquake to replace or repair the water stops it might also be considered to install pipes in the joints, which could be used for re-grouting of leaking joints. As indicated above the largest joint openings will take place in the upper portion of the dam where the hydrostatic pressure is still relatively small, therefore, water stop damage may be accepted during a rare earthquake.
Hydromechanical Equipment Because of the large amplification of the acceleration response in the crest region of most concrete dams, it is necessary to design any equipment located in this region for the corresponding floor response spectra. In view of the large crest amplification it may become uneconomical to install equipment with eigenfrequencies in the rage of the maximum spectral amplifications. High or low tuning of the equipment would be the right solution. Moreover, the hydrodynamic actions on gates and valves must also be determined based on the proper floor response spectra. The correct seismic analysis and design of the equipment are very important, as these elements have usually not been studied as thoroughly as the dam. An incorrect seismic coefficient of 0.1 is still used by hydromechanical and electromechanical engineers independent of the seismic hazard and the location of the equipment!
Earthquake Triggered Landslides, Rockfalls and Avalanches The possibility of earthquake-triggered landslides, debris flows, rockfalls and avalanches etc. must be considered when assessing the general hazards of mass movements into reservoirs. The freeboard should account for large waves caused by landslides into the reservoir. Landslides (and turbidity currents in the reservoir) also could block flow into appurtenant structures and safety devices such as the bottom outlet. Mass movements are very common features during very strong ground earthquakes in mountainous regions. During the 1999 Chi-Chi earthquake in Taiwan and also during the 1990 Manjil earthquake in Iran (Fig. 1) over 10,000 mass movements were observed. The magnitude 7.6 earthquake of January 21, 2003 (Colima earthquake) has triggered thousands of landslides.
Conclusions Concrete dams of all types can be designed and constructed to satisfactorily resist earthquake loadings in areas of high seismicity. Experience shows that modern dams that have been shaken by strong earthquakes have experienced good performance. Most of the dams constructed before the 1980ies have been designed against earthquakes by methods, which are considered as obsolete today. Therefore, a comprehensive seismic evaluation program has to be undertaken, to bring the safety of the existing dams up to modern seismic safety standards. Dams, which are seismically deficient and pose a great seismic risk, would either have to be strengthened or their reservoirs have to be lowered to achieve an acceptable earthquake safety. Besides structural safety, the earthquake risk of a large dam can be reduced by (i) seismic safety monitoring with e.g. strong motion instruments, and (ii) emergency planning for worst-case scenarios, which includes a water alarm system, evacuation plans and a team of dam and earthquake experts, who can assess the safety of a dam after a strong earthquake. Earthquake-resistant design is carried out on the basis of OBE. The optimum dam exhibits symmetrical mode shapes and has a relatively strong crest with filets along the abutments to minimize the maximum dynamic tensile stresses in the dam and the foundation. For the stress analysis, finite- element models of the dam-reservoir-foundation system are used. Seismic safety evaluation is carried out on the basis of MCE, MDE or SEE. During these events, cracks will develop and joints will open. Most of the deformations in a concrete dam are due to deformations at joints and cracks. For the safety check, the dynamic stability of detached concrete blocks has to be verified. Relatively simple rigid body models can be used to analyze the rocking and sliding motion of detached concrete blocks. All concrete dams with grouted contraction joints (including buttress and gravity dams) and shear keys exhibit a three-dimensional behaviour during strong earthquakes. This leads to large amplifications of the acceleration response in the central upper portion of a dam which must be taken into account. Earthquake observations show that well designed dams do not fail after the development of cracks along joints and that sliding movements along cracks and joints are in the order of a few centimetres. The stability of detached concrete blocks can be improved by a seismic belt located at the crest of an arch dam.
References 1) Eurocode 8 (2002): Design of structures for earthquake resistance, Part 2: Bridges, European Standard, prEN 1998-2:200X, Draft 2, May 2002 2) ICOLD Bulletin 72 (1989): Selecting Seismic Parameters for Large Dams, Guidelines, Committee on Seismic Aspects of Dam Design, ICOLD, Paris 3) ICOLD Bulletin 120 (2001): Design features of dams to effectively resist seismic ground motion, Committee on Seismic Aspects of Dam Design, ICOLD, Paris 4) ICOLD Bulletin 123 (2002): Earthquake design and evaluation of structures appurtenant to dams, Committee on Seismic Aspects of Dam Design, ICOLD, Paris 5) Wieland M. (2002): Lessons learnt from the earthquake behaviour of large dams and their implications on seismic design criteria, Proc. Third International Conference on Dam Engineering, Editors: M. Wieland and J.S.Y. Tan, CI Premier Pte Ltd., Singapore, March 20-22, 2002 6) Wieland M. (2003): Seismic aspects of dams, General Report Question 83: Seismic aspects of dams; International Commission on Large Dams; Proc. 21 st Congress of ICOLD, Montreal, Canada, June 16-20, 2003
Fig. 1: Earthquake damage caused to various facilities at the dam site of the Sefid Rud buttress dam, 1990 Manjil earthquake, Iran top: overturned ceramic insulators in substation (left) and damage in control room of powerplant (right) centre: large boulders blocking access road to dam crest (left) and overturned transmission tower caused by rockslide (right) bottom: severely damaged structures located on the dam premises