Académique Documents
Professionnel Documents
Culture Documents
17/12/13
11:34
Page 19
Impact of Near-Fault vs. Far-Field Ground Motions on the Seismic Response of an Arch Dam with Respect to Foundation Type
M A Hariri-Ardebili(1) and V E Saouma(2)
ABSTRACT
In the present paper, the impact of input ground motion characteristics on the seismic response of a typical concrete arch dam has been investigated by relying on the foundation numerical model. For this purpose six ground motions, including near-fault and far-field recordings, with various characteristics in the acceleration response spectrum, have been used to analyze the coupled system. Three different options have been introduced for the dam-foundation interaction: 1) foundation rock is assumed to be rigid; 2) foundation rock is modelled as a massless medium; and 3) a massed foundation is assumed with the inclusion of infinite elements at the far-end boundaries. The reservoir medium is assumed to be compressible, and has been modelled according to the Eulerian approach. It has also been observed that modelling the foundation as a massed element decreases stresses relative to the massless model, while use of a rigid foundation model leads to stress concentration at the dam-foundation interface. In all these foundation types, near-fault ground motions have led to higher responses than far-field motions, especially for the upper parts of the dam body, with less scattering of results in far-field ground motion. Keywords: Dam-foundation interaction, near-fault ground motion, massed foundation, seismic performance evaluation.
1. INTRODUCTION
Dams are complex infrastructures, and their behaviour depends on various parameters and parametric combinations. Chopra[1] proposed a set of factors that significantly influence the 3D numerical analysis of arch dams. These factors consist of: the semi-unbounded size of the reservoir and foundation rock domains; dam-water interaction; wave absorption at the reservoir boundary; water compressibility; dam-foundation-rock interaction; and spatial variations in ground motion at the dam-rock interface.
(1)Ph.D.
Student, Email: mohammad.haririardebili@colorado.edu; (2)Professor, Email: saouma@colorado.edu; Department of Civil, Environmental & Architectural Engineering, University of Colorado at Boulder, Boulder, CO, US.
19
17/12/13
11:34
Page 20
In considering the complex nature of materials and loads, as well as their interaction in a coupled dam-reservoir-foundation system, either all sources of system non-linearities should be considered, or they should all be neglected in favour of analyzing the system based on a linear elastic assumption of materials, which entails use of the appropriate toolset in order to interpret results. In the present paper a typical double curvature arch dam will be studied for the purpose of investigating the influence of the foundation numerical model, and its boundary conditions, on the seismic response of the system when subjected to both near-fault and far-field ground motions. Various types of foundation models, i.e. rigid, massless and massed foundations, will be used in the finite element model of a dam-reservoir-foundation system. Moreover, a set of site-specific ground motions with a range of characteristics (near-fault and far-field) will be introduced to determine the effects of seismic input mechanisms on the safety evaluation of dams. A straightforward methodology will be proposed in an effort to systematize the seismic assessment of concrete dams. Lastly, the results will be compared in terms of stresses, displacement and demand-capacity ratios. In addition the effects of seismic input for arch dams, in terms of ground motion components, rotation and spatial permutations, will all be examined.
2. LITERATURE REVIEW
Many researchers have investigated past efforts to model both the dam-foundation interaction and dam-foundation dynamic responses that incorporate various boundary conditions. Researchers such as Chopra & Chakrabarti [2] , Fenves & Chopra [3] and Leger & Boughoufalah[4], have all studied the importance of the foundation interaction with respect to the seismic behaviour of dams. Nuss et al[5], Chopra[1] and Chopra & Nuss[6] all concluded that if only the foundation rock flexibility is taken into consideration in numerical models, then stresses would be overestimated for all cases, in comparison with the massed foundation model. Bayraktar et al [7] analyzed the effect of base-rock characteristics on the stochastic dynamic response of dam-reservoir-foundation systems subjected to different earthquake input mechanisms. Lemos & Gomes[8] investigated the failure mechanism of the foundation in a concrete dam, and demonstrated that if the numerical analysis included the failure of either the rock joints or the foundation interface, then the inertial behaviour of the rock should be taken into consideration. Moreover, the boundaries at the top and sides of the model must be able to simulate energy radiation into the far-field, while the bottom boundary is assigned as a non-reflecting boundary. These authors also reported that both the experimental (i.e. shaking table test) and numerical displacements of a point in the middle of the dam block show good agreement. Qiumei et al[9] performed seismic analyses on reinforcement concrete gravity dams subjected
20
17/12/13
11:34
Page 21
IMPACT OF NEAR-FAULT VS. FAR-FIELD GROUND MOTIONS ON THE SEISMIC RESPONSE OF AN ARCH DAM WITH RESPECT TO FOUNDATION TYPE to near-field, pulse-like ground motions using a massless foundation. They found that the principal stress and displacement of all the specified points with a pulse-like ground motion were greater than those obtained without using any such motion. Zhang et al[10] studied two different models: a massless foundation model, and a viscousspring boundary input model that incorporates radiation damping. The responses of the 3D canyon without the dam were first analyzed using a massless truncated foundation along with a viscous-spring boundary; in a subsequent step, linear and non-linear analyses of the dam-foundation system were conducted. The authors concluded that stresses, displacements and contraction joint openings are all significantly reduced for both the linear and non-linear analyses when using the viscous-spring boundary model. Bayraktar et al[11] and Akkse et al[12] compared near-fault and far-field ground motion effects on the non-linear response of dam-reservoir massless foundation systems; they found a greater seismic demand on stresses and displacements when the dam is subjected to near-fault ground motion. Saleh & Madabhushi[13] examined the dynamic response of dams on rigid soil foundations, in addition to the resulting hydrodynamic pressure on the dam face, according to the dynamic centrifuge modelling technique. They determined that including a flexible foundation significantly reduces the dynamic response of the dam. Lebon et al[14] focused on the 3D non-linear rock-dam seismic interaction, by considering various soil-structure interaction models based on the finite element approach. Saouma et al[15] investigated the 2D and 3D time-history finite element analysis of rock-structure interactions by considering lateral energy dissipation, and the interaction between the far-field and actual numerical model. This approach is quite straightforward, and makes use of the principle of virtual work for deriving the transferred damping and stiffness matrices. Chen et al[16] assessed the influence of damping on the seismic response analysis of an arch dam. They found that the damping ratio greatly influenced the seismic response of an arch dam when applying a massless foundation model; their work also identified good agreement between the results of this viscous-spring boundary input model, and the output from the massless foundation model with a 10% damping ratio. Hariri-Ardebili & Mirzabozorg[17] targeted the non-linear dynamic analysis of a coupled system, featuring a reservoir-dam-foundation in 3D space, using the smeared crack approach. The foundation was assumed to be massed, while infinite elements at the far-end boundary and viscous boundary conditions were input to model the foundation medium. It was found that the response of a system with massed foundations, including infinite elements, is identical to that when the artificial absorbing boundary on the far-end of the foundation has been modelled using a viscous boundary. In another vein the seismic performance and safety evaluation of hydraulic structures were investigated by, among others, Ghanaat[18] and Yamaguchi et al[19]. Ghanaat proposed a damage estimation methodology for concrete dams that was adopted in guidelines published by USACE[20], which suggests a systematic method based on linear time-history results in terms of both local and global performance indices.
21
17/12/13
11:34
Page 22
Hariri-Ardebili & Mirzabozorg[21] studied the seismic performance of concrete arch dams using real ground motions, as well as Endurance Time Acceleration Functions (ETAFs); they found ETAFs to be capable of identifying various performance levels with acceptable accuracy.
where P is the hydrodynamic pressure, and c0 is the pressure wave velocity in water. For earthquake excitation, the boundary conditions of reservoir water can be summarized as follows: at the fluid-structure surface (i.e. dam-reservoir boundary condition), no flow can cross the interface. This condition is based on the fact that the face of concrete dams is impermeable. In the following equation, superscript s refers to the structure:
(2)
s is the normal acceleration of the dam body on the upstream face, n is the normal where an vector on the dam-reservoir interface towards the outer side of the dam body, and w is the reservoir water density. If no energy is being absorbed on the reservoir bottom, then the same boundary condition represented in Equation 2 can be used for the foundation-reservoir boundary; however, due to sediment and energy absorption at the reservoir bottom, the boundary condition can be rewritten as follows:
(3)
where is the wave reflection coefficient at the reservoir bottom and sides. In high dams surface waves are negligible, and hydrodynamic pressure on the free surface is set to zero. For modelling of the far-end truncated boundary a viscous boundary condition (referred to as the Sommerfeld boundary condition) is employed, in order to completely absorb the outgoing pressure waves:
22
17/12/13
11:34
Page 23
IMPACT OF NEAR-FAULT VS. FAR-FIELD GROUND MOTIONS ON THE SEISMIC RESPONSE OF AN ARCH DAM WITH RESPECT TO FOUNDATION TYPE
(4)
. .
23
17/12/13
11:34
Page 24
(6) where Cni and Csi are the damping factors in the normal and tangential directions respectively, and Ni is the element shape function.
(a)
(b)
(c)
Figure 1. a) Standard massless foundation model; b) massed foundation with viscous boundaries; c) massed foundation with infinite elements
24
17/12/13
11:34
Page 25
IMPACT OF NEAR-FAULT VS. FAR-FIELD GROUND MOTIONS ON THE SEISMIC RESPONSE OF AN ARCH DAM WITH RESPECT TO FOUNDATION TYPE This solution is exact if the P and S waves both impinge the artificial boundary at right angles. These are only approximate solutions for inclined body waves, whose reflected energy is only a small proportion of the total energy. In many cases, the further the artificial boundary is chosen from a source that radiates waves, the more the angle of incidence with respect to the artificial boundary will approach 90 and, hence, the better the viscous dampers will perform[15]. The other way to prevent wave reflection at the artificial boundary is by introducing infinite elements at the foundation exterior (Figure 1c). The use of infinite elements, stiffness and damping pertinent to the semi-infinite medium via the artificial boundary of the structure, has been incorporated into these analyses. The basic idea behind infinite elements is to allow the use of elements with special shape functions for the geometry at the far-end truncated boundary. Two sets of shape functions will therefore be applied: the standard shape function, Ni; and a growth shape function, Mi, which grows without a bound as the coordinate of the ith node approaches infinity. The Ni functions are applied to field variables[17], while Mi is applied to the geometry. These Mi shape functions, and their derivatives, are presented in Table 1 for a 20-node solid element with a face in infinity (Figure 1c). The effect of a semi-infinite medium via the far-end boundary of the foundation is taken into account once the resultant stiffness matrices, and their related proportional damping matrices, have been assembled into both the global stiffness matrix and global damping matrix of the system. The stiffness matrix of infinite elements is calculated as follows:
(7)
where , , and constitute the local coordinate system of the infinite elements, [J] is the Jacobian matrix, and [B] is the matrix transforming the nodal displacement of the considered element into the Gaussian point strains within the element, given as:
(8)
(9)
25
17/12/13
11:35
Page 26
Table 1. Growth shape functions, and their derivatives, for one face in infinity
(10)
where [M], [C] and [K] are the mass, damping and stiffness matrices of the structure (in the case of a massless foundation, the stiffness matrix includes both dam and foundation properties, and the mass matrix includes only the dam body, while the massed foundation model uses both dam and foundation characteristics in [M] and [K] matrices). [G], [C] and [K] are matrices representing the mass, damping and stiffness equivalent matrices of the reservoir, respectively. Matrix [Q] is the coupling matrix; {f1} is the vector that includes both the body and hydrostatic force; {P} and {U} are the hydrodynamic pressure .. and displacement vectors, respectively; and {Ug} is the ground acceleration vector. A detailed definition of all these matrices and vectors is available in published literature by Mirzabozorg et al[26].
26
17/12/13
11:35
Page 27
IMPACT OF NEAR-FAULT VS. FAR-FIELD GROUND MOTIONS ON THE SEISMIC RESPONSE OF AN ARCH DAM WITH RESPECT TO FOUNDATION TYPE
27
17/12/13
11:35
Page 28
(a)
(b)
Figure 2. a) General view of Karaj Dam; b) finite element model of the dam-reservoir-massless foundation system
28
17/12/13
11:35
Page 29
IMPACT OF NEAR-FAULT VS. FAR-FIELD GROUND MOTIONS ON THE SEISMIC RESPONSE OF AN ARCH DAM WITH RESPECT TO FOUNDATION TYPE
Figure 3. Seismotectonic map of the Central Alborz region, including the Karaj Dam[30]
29
17/12/13
11:35
Page 30
(a)
(b)
Figure 4. Comparison of the infinite element vs. viscous boundary approaches for: a) a one component intensifying step-like acceleration function; b) a three component real ground motion
(11)
30
17/12/13
11:35
Page 31
IMPACT OF NEAR-FAULT VS. FAR-FIELD GROUND MOTIONS ON THE SEISMIC RESPONSE OF AN ARCH DAM WITH RESPECT TO FOUNDATION TYPE in which M and K are the mass and stiffness matrices, subscripts s and f refer to the structure and fluid, respectively, and Sfs is the portion of the fluid domain boundary where fluid medium motion is directly coupled to the motion of a structure. On this boundary the fluid and structure have the same displacement normal to the boundary, while the tangential motions are uncoupled. In this equation, both the stiffness and mass matrices are asymmetric. Solving this last equation using Lanczos formulation, or any other appropriate method able to solve the asymmetric coupled equation, yields the undamped coupled frequencies of the dam-reservoir-foundation system.
(a)
(b)
Figure 5. a) Comparison of the undamped coupled dam modes; b) sample of key dam mode shapes
31
17/12/13
11:35
Page 32
Figure 5 shows the vibration periods for the first 30 modes of the dam-reservoir system, considering foundation effects. In addition, this figure depicts some of the selected dam mode shapes on a rigid foundation. As observed in this figure, the massed foundation model leads to higher periods than the massless foundation, and both models produce higher values for the period than the rigid foundation model. It can be noted that based on this figure, the influence of foundation on the period covering the first ten modes (T > 0.15sec) is more significant than the subsequent modes.
32
17/12/13
11:35
Page 33
IMPACT OF NEAR-FAULT VS. FAR-FIELD GROUND MOTIONS ON THE SEISMIC RESPONSE OF AN ARCH DAM WITH RESPECT TO FOUNDATION TYPE Acceleration and velocity time-histories of all scaled ground motions are depicted in Figure 6; in addition, Figure 7 shows the horizontal acceleration response spectrum of the Karaj Dam site (as a target spectrum) with a maximum credible level (MCL) obtained by conducting a hazard analysis of the dam site. The ground motions have been scaled in order to provide acceptable consistency to the target spectrum over a range of structural periods, Tmin to Tmax, as defined. The dams small-amplitude fundamental period of vibration is denoted here by T1. Tmax is set at 2T1, and Tmin should typically be set at 0.2T1. If a substantial response and damage can occur due to responses in modes with periods shorter than Tmin, then Tmin should be selected sufficiently small so as to capture this important behavioural pattern[32,33]. It is also worth pointing out that the selected period range considers all effective modes contributing to the vibrational behaviour of the dam (i.e. the contributed effective mass equals at least 90% of the total system mass). For high concrete arch dams, setting the lower bound to 0.1sec is generally appropriate for the purpose of spectrum matching[21,27]. Based on Figure 4, the value of T1 differs for the various models. To achieve iso-intensity ground motions in all cases, the value of 0.45sec was selected as the fundamental period; consequently, the values of Tmin and Tmax were determined as 0.09sec and 0.9sec, respectively. As seen in Figure 6, an acceptable level of consistency is reached between ground motions and target spectrum over both the selected period range and fundamental period. The scaling factor is defined in a way that satisfies the following equation within the selected period range:
(12)
where SaEQGM and SaTARGET are the acceleration response spectrum of selected ground motions is the linear scaling factor for each ground motion. and (targeted) site spectrum, and It is noted that other methods may be used to fit the ground motion response spectrum on the target spectrum to any desired accuracy. However, these methods alter the intrinsic nature of ground motion and additional corrections are usually required on the acceleration time-histories. Based on Table 2, the originally selected near-fault ground motions are stronger than far-field motions, yet comparing the PGV-to-PGA ratio for all ground motions reveals that this ratio tends to be higher for far-field motions. All ground motions were proposed herein for selection within the same magnitude range, i.e. M = 7.3 0.4, and surface magnitude range, i.e. Ms = 7.4 0.4. Near-fault ground motions were also selected at a distance of less than 12km from the centre of the rupture, while far-field motions were chosen to be at a distance greater than 50km. From Figure 5, the original duration of near-fault motions equals approximately 30-40sec, while their value reaches 60-130sec in far-field motions. As previously mentioned, in order to reduce the computational effort for the present case the significant ground motion duration was calculated based on 90% of their total energy. This significant duration equals approximately 9-14sec in near-fault and 30-31sec in far-field motions.
33
17/12/13
11:35
Page 34
34
17/12/13
11:35
Page 35
IMPACT OF NEAR-FAULT VS. FAR-FIELD GROUND MOTIONS ON THE SEISMIC RESPONSE OF AN ARCH DAM WITH RESPECT TO FOUNDATION TYPE
(a)
(b)
Figure 6. a) Scaled acceleration time-histories; b) scaled velocity time-histories for near-fault and far-field ground motions
Figure 7. Scaled acceleration response spectra of selected records based on the target spectrum
35
17/12/13
11:35
Page 36
Figure 8 represents the general flowchart used in the present paper to conduct the seismic safety assessment of a concrete dam, as based on USACE methodology[20]. This procedure constitutes a standard method available for use with any other concept in dam engineering for the purpose of assessing the seismic behaviour of a dam. In this paper such a method has been combined with the concept of near-fault/far-field ground motions, as well as with the foundation type effect. It is clearly apparent that this methodology starts with a seismic hazard analysis of the dam site to identify the sites seismic characteristics along with the condition of the various faults, their activities, maximum credible level, and a set of suitable near-fault and far-field ground motions for seismic analyses. Earthquake ground motions should be selected based on source characteristics, source-to-site transmission path properties, and site conditions. Current practice usually dictates using at least three ground motions and then taking their maximum results, or using up to seven ground motions and taking their average responses. The finite element model of the dam-reservoir-foundation system is then developed, based on previous instructions offered for the various foundation types. Three component ground motions (i.e. two horizontal and one vertical) are required for 3D seismic analysis of arch dams. The structure must be capable of resisting maximum earthquake ground motions occurring in any direction. In time-history analysis it may be necessary to identify the stronger horizontal component in order to obtain the highest-magnitude system response. The orthogonal components of earthquake ground motion are commonly applied along the principal axes of the structure[20], while the maximum response may occur in any direction other than the principal axes. This effect can be determined from component rotation, and may be significant in some cases, especially when the second ground motion component is not as strong as the first, and when the arch dam is not symmetric along the dam axis. This effect is shown schematically in Figure 9, which also displays the maximum first principal stress of the Karaj Dam under an N1 ground motion for the massless foundation case, with the rotation of horizontal components. In this case, applying the orthogonal components along the principal dam axes (x and y) leads to the generation of a maximum first principal stress in the dam body equal to 3.96MPa, whereas applying ground motion at an angle of = 45 with respect to the principal axes (x45 and y45) leads to a higher maximum first principal stress of approximately 4.59MPa. A = 315 angle of application with respect to the principal axes (x315 and y315), yields a lower maximum first principal stress (i.e. about 3.01MPa). For 3D time-history analysis of arch dams a complete permutation of all three components, with positive and negative signs, may be required to obtain the most critical directions causing the greatest structural response[20]. Figure 10 shows the effects of considering all eight permutations of an N1 ground motion on the massless foundation for both stress and
36
17/12/13
11:35
Page 37
IMPACT OF NEAR-FAULT VS. FAR-FIELD GROUND MOTIONS ON THE SEISMIC RESPONSE OF AN ARCH DAM WITH RESPECT TO FOUNDATION TYPE displacement responses. As can be observed, despite some differences between results, these responses are negligible for the current case due to a completely symmetric dam. In this figure, the displacement response differs for all load combinations, while the stress response no longer seems to be sensitive to the direction of the second horizontal ground motion component. This finding suggests that the load combinations are reduced to four different cases under this condition. Let us note that in the present paper, the criterion for selecting a critical direction (for both component rotations and permutations) consists of the higher maximum first principal stress. The last step entails scaling ground motions, based on the site response spectrum for the predefined period range. This scaling step may be either linear, by means of multiplying the ground motion response spectrum by a scalar factor that reasonably matches the target spectrum, or non-linear by changing the original shape of the ground motion response spectrum to better match the target spectrum, and then generating some kind of artificial ground motion that uses the original motion. These scaled ground motions should then be used for linear analysis of the coupled system, with all required results being extracted to perform a seismic safety assessment of the dam. The most common required responses are: displacement and tensile stress time-histories, cumulative inelastic duration (CID) for the most critical node within the dam body, and a 20% limitation of the overstressed area (i.e. the area where tensile stresses exceed the tensile strength of concrete) on both the upstream and downstream faces. The last two major criteria are shown in Figure 8. In these two plots the curve calculated from our case study is compared with the threshold: the case is deemed to be safe if the curve lies below the threshold[21].
37
17/12/13
11:36
Page 38
Figure 8. General flowchart for the seismic safety assessment of concrete dams considering various foundations and ground motion types
38
17/12/13
11:36
Page 39
IMPACT OF NEAR-FAULT VS. FAR-FIELD GROUND MOTIONS ON THE SEISMIC RESPONSE OF AN ARCH DAM WITH RESPECT TO FOUNDATION TYPE
(a)
(b)
Figure 9. a) Rotational effect of horizontal ground motion components; b) variation in maximum first principal stress due to N1 with respect to component rotation
39
17/12/13
11:36
Page 40
(a)
(b)
Figure 10. Effects of considering permutations of all three earthquake components on: a) maximum first principal stress; b) displacement along the crest
40
17/12/13
11:36
Page 41
IMPACT OF NEAR-FAULT VS. FAR-FIELD GROUND MOTIONS ON THE SEISMIC RESPONSE OF AN ARCH DAM WITH RESPECT TO FOUNDATION TYPE
41
17/12/13
11:37
Page 42
(a)
(b)
(c)
(d)
Figure 11. Non-concurrent envelope of displacement in the stream direction at the central cantilever for: a) massless foundation; b) massed foundation; and c) rigid foundation d) shows the average results
Figures 12 to 14 present the non-concurrent envelope of maximum first principal stress (MFPS), and minimum third principal stress (MTPS), on the upstream dam face for three types of foundations. For the massless configuration, the higher MFPS values typically occur on the lower parts of the upstream face near the foundation; moreover, these tensile stresses are primarily characterized by cantilever stresses due to hydrodynamic pressure of the reservoir being transferred to the bottom of the dam. The upper part of the dam, near the crest, also experiences tensile stress (but less than the bottom part), whose primary characteristic is the arch stresses due to the arch action of the monolithic body. The central part of the dam generally exhibits low tensile stresses and, in some cases, these parts are completely equivalent. The central and upper parts of the upstream face have a higher compressive stress (in terms of MTPS); also, the central parts of the downstream face, in the vicinity of the foundation (not shown in these figures), experience high compressive stresses.
42
17/12/13
11:37
Page 43
IMPACT OF NEAR-FAULT VS. FAR-FIELD GROUND MOTIONS ON THE SEISMIC RESPONSE OF AN ARCH DAM WITH RESPECT TO FOUNDATION TYPE
Figure 12. Non-concurrent envelope of maximum first principal stress and minimum third principal stress on the upstream face for the massless foundation (Pa)
43
17/12/13
11:37
Page 44
Figure 13. Non-concurrent envelope of the maximum first principal stress and minimum third principal stress on the upstream face for the massed foundation (Pa)
44
17/12/13
11:38
Page 45
IMPACT OF NEAR-FAULT VS. FAR-FIELD GROUND MOTIONS ON THE SEISMIC RESPONSE OF AN ARCH DAM WITH RESPECT TO FOUNDATION TYPE
Figure 14. Non-concurrent envelope of the maximum first principal stress and minimum third principal stress on the upstream face for the rigid foundation (Pa)
45
17/12/13
11:38
Page 46
Table 3 summarizes the MFPS and MTPS for all cases, along with their comparisons. As can be seen in all cases, the rigid foundation model leads to higher MFPS and MTPS values than the massless and massed models. In adopting the massed foundation model as the basis, it can be concluded that using a massless foundation model leads to a difference in the range of 17-20% under near-fault ground motions, and 21-28% under far-field motions. On the other hand, using a rigid foundation model increases the percentage of MFPS differences to about 57-86% for near-fault motions, and 33-53% for far-field motions. The percentage increase in compressive stress (in terms of MTPS), due to the use of a massless foundation, amounts to about 13-20% and 16-24% for near- and far-field ground motions, respectively, while these results equal 16-50% and 30-35% for the rigid foundation model. In all cases, the assumption of a massless foundation is closer to the benchmark model (i.e. massed foundation) than the rigid foundation. The percentage error in the difference between massless and massed foundation models under near-fault ground motions is less than that under far-field motions for both MFPS and MTPS; in contrast, these results are completely reversed for the percentage of difference between rigid and massed foundation models (except for N3 ground motion, which produces a lower percentage of difference than for far-field motions). Table 3 also lists the values of the maximum demand-capacity ratio, DCRmax, for the various models. Based on USACE guidelines[20] the dam response to MCE is assumed to be linear elastic, with little or no possibility of damage if the computed DCR values are less than or equal to 1. The amount of contraction joint opening at DCR 1 is expected, however, to be small with negligible or no effect on the overall stiffness of the dam. The dam is considered to exhibit a non-linear response, in the form of contraction joint opening/closing, and lift line cracking provided the estimated DCR > 1. The level of non-linear response or joint opening and cracking is deemed acceptable if DCR < 2, and the overstressed region is limited to 20% of the dam surface area. For cases where DCR 2, the dam body is expected to behave in a strong non-linear manner in terms of mass concrete damage and/or permanent relative drifts between blocks. Based on this table DCR > 1 in all cases, yet remains limited to 2 for the massless and massed foundation cases, while approaching or even exceeding 2 for some ground motions in the rigid foundation model. Using a massed foundation leads to lower DCRmax values, in comparison with the massless foundation model. Comparing near-fault with far-field ground motion effects on numerical models reveals that far-field motions tend to yield a higher DCRmax than near-fault motions on both the massless and massed foundations. No general rule is available for comparing ground motion type on the rigid foundation model, though near-fault motions show a slightly higher value. Figure 15 presents the percentage of the overstressed area of the dam under various conditions, with respect to the pre-defined threshold curve proposed by USACE[20]. Even
46
17/12/13
11:38
Page 47
IMPACT OF NEAR-FAULT VS. FAR-FIELD GROUND MOTIONS ON THE SEISMIC RESPONSE OF AN ARCH DAM WITH RESPECT TO FOUNDATION TYPE though the demand curves for all models lie below the threshold curve, the rigid foundation generally produces a greater overstressed area than either the massless or massed foundation at various DCR values. Moreover, Figure 14d compares these curves for near- and far-ground motion recordings. It is clearly apparent that the difference between these two types remains small for massless and massed foundation models; nonetheless, near-fault motions show a bigger jump than far-field motions for the rigid foundation model.
Table 3. Comparison of extreme stresses within the dam body for different cases
47
17/12/13
11:38
Page 48
(a)
(b)
(c)
(d)
Figure 15. Percentage of overstressed area on the dam face at various DCR values: a) massless foundation; b) massed foundation; c) rigid foundation d) shows the average results
11. CONCLUSION
This paper has provided a discussion on foundation numerical models, i.e. massless, massed and rigid models, relative to the seismic safety analysis of a dam-reservoir-foundation system. Two sets of near- and far-field ground motions, consistent with the site characteristics of the Karaj Dam in Iran, were chosen for the purpose of dynamic analysis. The results were compared in terms of displacements, principal stresses and demand-capacity ratio for the dam. The rotational effects of horizontal components were closely investigated for this specific case. It was found that applying the orthogonal components at an angle of = 45, with respect to the principal axes, leads to an increase in the maximum first principal stress of approximately 16% within the dam body. It was also determined that the current case is rather insensitive to the component permutation rule.
48
17/12/13
11:38
Page 49
IMPACT OF NEAR-FAULT VS. FAR-FIELD GROUND MOTIONS ON THE SEISMIC RESPONSE OF AN ARCH DAM WITH RESPECT TO FOUNDATION TYPE Based on the analyses conducted herein, the massless foundation yields higher displacement values along the central cantilever of the dam than the massed foundation, with both these values exceeding the rigid foundation displacement. In all foundation types, near-fault ground motions lead to a greater response than those generated by far-field motions, especially over the upper half of the cantilever. The scattering of results in near-fault ground motion is more pronounced than that in far-field motion, yet remain within an acceptable range. For the massless foundation, higher MFPS values typically occur at lower parts of the upstream face, near the foundation. The central part of the dam generally exhibits very low tensile stresses and, in some cases, these parts are completely equivalent. The central and upper parts of the upstream face, and the central parts of the downstream face in the foundation vicinity, revealed high compressive stresses. For all ground motions, the rigid foundation model leads to higher MFPS and MTPS than either the massless or massed models. The percentage differences between massless and massed foundation models for stress response under near-fault ground motions is less than the corresponding percentage differences under far-field motions, while these results are completely reversed for the percentage differences between rigid and massed foundation models, with the exception of N3 ground motion (which has a lower percentage of difference than far-field motions). Based on current research, DCRmax > 1 in all cases; however, it is limited to 2 for the massless and massed foundation cases, while it approaches or even exceeds 2 for some ground motions in the rigid foundation model. In contrast, the overstressed area demand curves lie below the threshold curve in all ground motion cases. The rigid foundation tends to produce a greater overstressed area than either the massless or massed foundation model at various DCR values. It can be concluded that use of the massed foundation model with infinite elements at far-end boundaries is more appropriate for simulating the seismic behaviour of arch dams than the massless model. In addition, introducing a rigid foundation increases stresses within the dam body, and would not be a suitable model for assessing the safety response of arch dams under earthquake conditions. Lastly, near-fault ground motions may generate different responses than conventional far-field ground motions; hence their effects should be investigated separately should an active fault be identified near a dam site, or should a new fault be detected after the construction of a dam.
REFERENCES
[1] Chopra, A K, Earthquake Analysis of Arch Dams: Factors to be Considered, Journal of Structural Engineering, Vol 138, Issue 2, pp205-214 (2012). [2] Chopra, A K & Chakrabarti, P, Earthquake Analysis of Concrete Gravity Dams Including Dam-Water-Foundation Rock Interaction, Earthquake Engineering & Structural Dynamics, Vol 9, Issue 4, pp363-383 (1981).
49
17/12/13
11:38
Page 50
[3] Fenves, G & Chopra, A K, Earthquake Analysis of Concrete Gravity Dams Including Reservoir Bottom Absorption and Dam-Water-Foundation Rock Interaction, Earthquake Engineering & Structural Dynamics, Vol 12, Issue 5, pp663-680 (1984). [4] Leger, P & Boughoufalah, M, Earthquake Input Mechanisms for Time Domain Analysis of Dam-Foundation Systems, Engineering Structures, Vol 11, Issue 1, pp37-46 (1989). [5] Nuss, L K, Munoz, R L, Jackmauh, F J & Chopra, A K, Influence of Dam-Foundation Interaction in Seismic Safety Evaluation of Two Arch Dams, Proceedings, 12th World Conference on Earthquake Engineering, Auckland, New Zealand (2000). [6] Chopra, A K & Nuss, L K, Seismic Safety Evaluation and Upgrading of Arch Dams, Proceedings, 23rd Congress on Large Dams, Brasilia, Brazil (2009). [7] Bayraktar, A, Hancer, E & Akkse, M, Influence of Base-Rock Characteristics on the Stochastic Dynamic Response of Dam-Reservoir-Foundation Systems, Engineering Structures, Vol 27, Issue 10, pp1498-1508 (2005). [8] Lemos, J V & Gomes, J P, Modeling Seismic Failure Scenarios of Concrete Dam Foundations, Applications of Computational Mechanics in Geotechnical Engineering, (Sousa, Fernandes, Vargas Jr & Azevedo (eds)), Taylor & Francis Group, London, UK, pp341-349 (2007). [9] Qiumei, H, Yaqi, L, Aiwen, L & Xiaojun, L, Seismic Analysis of Gravity Dam Subjected to Near-Field Pulse-Like Ground Motions, Proceedings , 14th World Conference on Earthquake Engineering, Beijing, China (2008). [10] Zhang, C, Pan, J & Wang, J, Influence of Seismic Input Mechanisms and Radiation Damping on Arch Dam Response, Soil Dynamics & Earthquake Engineering, Vol 29, Issue 9, pp1282-1293 (2009). [11] Bayraktar, A, Sevim, B, Altunisik, A C, Turker, T, EmreKartal, M, Akkse, M & Bilici, Y, Comparison of Near and Far Fault Ground Motion Effects on the Seismic Performance Evaluation of Dam-Reservoir-Foundation Systems, Dam Engineering, Vol 19, Issue 4, pp201-239 (2009). [12] Akkse, M, Non-Linear Seismic Response of Concrete Gravity Dams to Near-Fault Ground Motions Including Dam-Water-Sediment-Foundation Interaction, Applied Mathematics & Modeling, Vol 34, Issue 11, pp3685-3700 (2010).
50
17/12/13
11:38
Page 51
IMPACT OF NEAR-FAULT VS. FAR-FIELD GROUND MOTIONS ON THE SEISMIC RESPONSE OF AN ARCH DAM WITH RESPECT TO FOUNDATION TYPE [13] Saleh, S & Madabhushi, S P G, Response of Concrete Dams on Rigid and Soil Foundations Under Earthquake Loading, Journal of Earthquake & Tsunami, Vol 4, Issue 3, pp251-268 (2010). [14] Lebon, G, Saouma, V & Uchita, Y, 3D Rock-Dam Seismic Interaction, Dam Engineering, Vol 21, Issue 2, pp101-130 (2010). [15] Saouma, V, Miura, F, Lebon, G & Yagome, Y, A Simplified 3D Model for SoilStructure Interaction with Radiation Damping and Free Field Input, Bulletin of Earthquake Engineering, Vol 9, Issue 5, pp1387-1402 (2011). [16] Chen, D H, Du, C B, Yuan, J W & Hong, Y W, An Investigation into the Influence of Damping on the Earthquake Response Analysis of a High Arch Dam, Journal of Earthquake Engineering, Vol 16, Issue 3, pp329-349 (2012). [17] Hariri-Ardebili, M A & Mirzabozorg, H, A Comparative Study of the Seismic Stability of Coupled Arch Dam-Foundation-Reservoir Systems Using Infinite Elements and Viscous Boundary Models, International Journal of Structural Stability & Dynamic, Vol 13, Issue 6, DOI: 10.1142/S0219455413500326 (2013). [18] Ghanaat, Y, Failure Modes Approach to Safety Evaluation of Dams, Proceedings, 13th World Conference on Earthquake Engineering, Vancouver, Canada (2004). [19] Yamaguchi, Y, Hall, R, Sasaki, T, Matheu, E, Kanenawa, K, Chudgar, A & Yule, D, Seismic Performance Evaluation of Concrete Gravity Dams, Proceedings, 13th World Conference on Earthquake Engineering, Vancouver, Canada (2004). [20] US Army Corps of Engineers (USACE), EM 1110-2-6053: Earthquake Design and Evaluation of Concrete Hydraulic Structures, Washington, US (2007). [21] Hariri-Ardebili, M A & Mirzabozorg, H, Investigation of Endurance Time Method Capability in Seismic Performance Evaluation of Concrete Arch Dams, Dam Engineering, Vol 22, Issue 1, pp35-64 (2011). [22] Chopra, A K, Hydrodynamic Pressures on Dams During Earthquakes, Journal of Engineering Mechanics, ASCE, Vol 93, pp205-223 (1967). [23] Chopra, A K, Earthquake Behavior of Reservoir-dam Systems, Journal of Engineering Mechanics, ACSE, Vol 94, pp1475-1499 (1968).
51
17/12/13
11:38
Page 52
[24] Chakraparti, P & Chopra, A K, Earthquake Analysis of Gravity Dams Including Hydrodynamic Interaction, Earthquake Engineering & Structural Dynamics, Vol 2, Issue 2, pp143-160 (1973). [25] Hall, J F & Chopra, A K, Dynamic Response of Embankment, Concrete-Gravity and Arch Dams Including Hydrodynamic Interaction, Report No. UCB/EERC 80/39 , Earthquake Engineering Research Center, University of California, Berkeley, CA, US (1980). [26] Mirzabozorg, H, Hariri-Ardebili, M A & Nateghi-A, R, Seismic Behavior of Three Dimensional Concrete Rectangular Containers Including Sloshing Effects, Journal of Coupled System Mechanics, Vol 1, Issue 1, pp79-98 (2012). [27] Federal Energy Regulatory Commission (FERC), Engineering Guidelines for the Evaluation of Hydropower Projects, Chapter 11: Arch Dams, Washington DC, US (1999). [28] Lysmer, J & Kuhlemeyer, R, Finite Element Model for Infinite Media, Journal of Engineering Mechanics, ASCE, Vol 95, Issue 4, pp859-877 (1969). [29] Hessami, K & Jamali, F, Explanatory Notes to the Map of Major Active Faults of Iran, Journal of Seismology & Earthquake Engineering, Vol 8, Issue 1, pp1-11 (2006). [30] Seismotectonic Map of the Central Alborz, http://seismo.gsi.ir, Iran (2005). [31] Ritz, J F, Balescu, S, Soleymani, S, Abbasi, M, Nazari, H, Feghhi, K, Shabanian, E, Tabassi, H, Farbod, Y, Lamothe, Y, Michelot, J L, Massault, M, Chery, J & Vernant, P, Determining the Long Term Slip Rate Along the Mosha Fault, Central Alborz, Iran: Implications in Terms of Seismic Activity, Proceedings, 4th Conference on Seismology & Earthquake Engineering, IIEES, Tehran, Iran (2003). [32] Applied Technology Council (ATC), ATC-58: Guidelines for Seismic Performance Assessment of Buildings, [100% Draft], Redwood City, CA, US (2012). [33] Saouma, V E, Porter, K, Nuss, L K & Hariri-Ardebili, M A, Performance Based Seismic Design Guidelines for Concrete Dams (Part 1), Report, Submitted to Enerjisa Co, University of Colorado, Boulder, CO, US (2012).
52