EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS Earthquake Engng Struct. Dyn.

2009; 38:307329 Published online 1 September 2008 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/eqe.851

Seismic response of intake towers including damtower interaction

SUMMARY The seismic response of the intakeoutlet towers has been widely analyzed in recent years. The usual models consider the hydrodynamic effects produced by the surrounding water and the interior water, characterizing the dynamic response of the towerwaterfoundationsoil system. As a result of these works, simplied added mass models have been developed. However, in all previous models, the surrounding water is assumed to be of uniform depth and to have innite extension. Consequently, the considered added mass is associated with only the pressures created by the displacements of the tower itself. For a real system, the intake tower is usually located in proximity to the dam and the dam pressures may inuence the equivalent added mass. The objective of this paper is to investigate how the response of the tower is affected by the presence of the dam. A coupled three-dimensional boundary element-nite element model in the frequency domain is employed to analyze the towerdamreservoir interaction problem. In all cases, the system response is assumed to be linear, and the effect of the internal uid and the soilstructure interaction effects are not considered. The results suggest that unexpected resonance amplications can occur due to changes in the added mass for the tower as a result of the towerdamreservoir interaction. Copyright 2008 John Wiley & Sons, Ltd.
Received 3 August 2007; Revised 21 July 2008; Accepted 22 July 2008 KEY WORDS:

CE database subject headings: intake structures; seismic analysis; dams; hydrodynamic pressures; uidstructure interaction; structural dynamics

1. INTRODUCTION The safety of the intakeoutlet towers after an earthquake is of great importance because it is related to the continuity of the water supply. In some cases, these towers are freestanding and are founded on an enlarged base on the reservoir bottom, whereas in other cases they are structurally
Correspondence

to: M. A. Mill an, Dep. Estructuras, Escuela Superior de Ingenieros, Camino de los Descubrimientos s/n, Isla de la Cartuja, 41092 Sevilla, Spain. E-mail: mmillan@princeton.edu, mangelmm@us.es Visiting researcher. Assistant Professor. Professor.

Copyright

2008 John Wiley & Sons, Ltd.

308

M. A. MILLAN, Y. L. YOUNG AND J. H. PREVOST

water
Infinite reservoir Infinite reservoir

tower

Tower excitation

Figure 1. Towerwater system in typical models. The surrounding water is assumed to be of uniform depth and to have an innite extension.

Plane of symmetry

reservoir

dam

Intake-outlet tower

z x y

Figure 2. Symmetric damtowerreservoir model considered in this paper.

connected to the surrounding land or to the upstream concrete dam. It is recognized that the response of the tower can be signicantly inuenced by the uid structure interaction effects. There are several solutions for earthquake-induced hydrodynamic pressures on exible, submerged, cylindrical structures surrounded by a compressible uid as described hereafter. The most extended procedure [14] accounts in the calculation the effects of added mass due to the internal and external water, as well as the inuence of uid compressibility and hydrodynamic radiation damping. In a series of publications, Goyal and Chopra [14] presented frequency functions and earthquake responses, as well as simplied procedures, to determine the maximum seismic forces and the added mass due to the presence of the interior and exterior uid. In all previous models, the surrounding water is assumed to be of uniform depth and to have an innite extent (see Figure 1). Consequently, the added mass effect is associated with the pressures
Copyright 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

SEISMIC RESPONSE OF INTAKE TOWERS

309

Figure 3. Damtowerreservoir BEM model. Upper part: concrete dam; bottom part: earth dam.

created by the displacements of the tower only. However, for a real system, the intake tower is located in proximity to the dam, where wave reections from the dam may change the resultant added mass, which in turn may change the natural frequencies and the seismic response of the tower. The objective of this paper is to investigate the inuence of the dam on the tower during an earthquake. Two simplied problems are considered: (1) a concrete intake tower in the presence of a concrete dam and (2) an earth dam. A schematic representation of the problem is presented in Figure 2. A coupled three-dimensional boundary element-nite element method in frequency domain is employed to analyze the dynamic damwatertower interaction problem. The boundary element method (BEM) model used in this research has been described earlier in [57]. The BEM models for the concrete dam and earth dam are shown in Figure 3. In all cases, symmetry about the x z plane is exploited.
Copyright 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

310

M. A. MILLAN, Y. L. YOUNG AND J. H. PREVOST

The effects of the interior uid and the soilstructure interaction are not considered in order to focus on the inuence of the damtower interaction.

2. BASIC THEORETICAL ANALYSIS OF THE PROBLEM As explained before, the main difference between the real behavior of the system and that of the typical models is the presence of the dam. A simplied analysis of the problem can be done by decomposing the coupled system as a sum of simpler models as discussed hereafter. As represented in Figure 4, the complex behavior of the coupled damtowerwater system subject to bottom seismic excitation can be decomposed as the sum of two simpler cases of excitation. Case 1 includes the excitation of the dam base but not of the tower base. Case 2 includes the excitation of the tower base but not of the dam base. Both the dam and the tower are exible in all cases and may be excited by pressure waves propagating through the reservoir water. From a very simple analysis of the two problems, some important conclusions can be obtained about the behavior of the coupled model. For the rst simplied case with dam excitation only, the main component of the pressure eld is due to the dam oscillation, which creates a propagating wave from the dam to the innite reservoir. This wave will produce resonance of the tower if the excitation frequency is close to the tower natural frequencies, or if the excitation is related to the natural frequency of the damreservoir system (even if this frequency does not correspond to the tower natural frequency). As a consequence of this behavior, the pressure eld effect on the tower cannot be represented simply as an added mass to the tower mass. For the second simplied case with tower excitation only, the pressure eld is created by the tower vibration and thus can be properly represented by an added mass. When the distance between the tower and the dam is large, the results obtained from Case 2 are almost identical to those obtained from the analysis of a tower reservoir model without the dam, as described in Figure 1.

CASE 1

Dam excitation

Fixed base

Dam excitation

Tower excitation

CASE 2
Fixed base Tower excitation

Figure 4. Damtowerwater system subject to bottom seismic excitation can be decomposed as the sum of two simpler cases of excitation. Case 1 includes the excitation of the dam base but not of the tower base. Case 2 includes the excitation of the tower base but not of the dam base.
Copyright 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

SEISMIC RESPONSE OF INTAKE TOWERS

311

3. CONCRETE DAM MODEL 3.1. Description of the BEM model The dam is idealized as a triangle with a vertical upstream face and a downstream face with a slope of 0.8:1. The transversal dimension of the dam is assumed to be equal to the reservoir width, which is 300 m. Free slip boundary conditions are applied at the ends of the dam on the lateral face of the block, and zero displacement condition is applied in the transversal direction. The water depth H at the upstream face of the dam is the same as the dam and the tower height, which is 100 m. The dam is assumed to be symmetric about the x z plane. The half width of the dam is b = 150 m (the full width is 2b = 300 m). The intake tower is assumed to be a cylinder with uniform cross section. The outside diameter is taken to be 2 R = 13.6 m and the wall thickness is taken to be 0.2 R . To simplify the mesh, the tower is modeled as a solid cylinder without the hollowed core. The effect of the hollowed core is considered by changing the material properties (Youngs modulus and solid density) of the solid cylindrical tower to match the natural frequencies of the hollowed cylindrical tower. The reservoir is discretized to a distance of 600 m. At this point, a transmitting boundary of the Robin type [8] is used to simulate an innite channel of constant rectangular shape. Quadratic rectangular and triangular boundary elements are used to model the dam, the tower and the reservoir. The reservoir lateral boundaries are assumed to be rigid. The assumed parameter values for the dam are: modulus of elasticity E = 3.45 1010 N/m2 , Poissons ratio = 0.17, density = 2480.0 kg/m3 , damping ratio = 0.05. The water wave velocity c = 1440 m/s, water density w = 1000 kg/m3 and reservoir depth H = 100 m. For the selected values, the natural frequency of the reservoir is r = c/(2 H ) = 22.6 rad/s = 3.6 Hz. The material properties of the actual hollowed core tower are assumed to be the same as the dam. To allow the tower to be simplied as a solid cylinder, the tower modulus of elasticity is E T = 2.0 1010 N/m2 and the tower density is T = 893.5 kg/m3 , which gives the same fundamental frequency (5.7 rad/s or 0.91 Hz) as the actual hollowed core concrete tower. Two different models have been dened in order to analyze the effect of the damtower separation. A separation distance S = 50 m is used in the rst model and S = 100 m in the second model. The wave reection at the bottom boundary of the reservoir, accounting for the exibility of the bottom rock and of the possible bottom sediments, is represented in a simplied way using a bottom absorption coefcient ( = 0 implies a non-reective boundary and = 1 implies a fully reective boundary). It is dened as (see [9] for details) = 1 qc 1 + qc (1)

where c is the water wave velocity and q = w / r Cr . r and w are the soil and water densities, respectively. Cr = E r / r is the transversal wave velocity in the soil. E r is Youngs modulus for the soil. An absorption coefcient of = 0.75 has been considered in the model with the 50 m. separation distance between the dam and the tower. In order to study the damtower interaction effects, the results are compared with those obtained using a model without the dam, which are obtained by applying transmitting boundary conditions at the two axial extremes of the reservoir. One of the BEM models is shown in Figure 3 (upper part). In all cases, symmetry about the x z plane is exploited.
Copyright 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

312

M. A. MILLAN, Y. L. YOUNG AND J. H. PREVOST

Figure 5. Natural modes of vibration and associated displacement and pressure distributions of the intake tower. Horizontal excitation. The pressures have been normalized by the corresponding peak pressure for the rst mode of the tower (without the dam), p 1.

3.2. Frequency-domain seismic response Results for the models with and without the dam subject to horizontal and vertical seismic excitations are presented in this section. The rst two mode shapes and pressure distributions for the coupled damtower case (100 m separation) are presented in Figure 5. The results correspond to the rst and second resonance frequencies. The pressures p are normalized by the peak pressure for the rst modal response of the tower (without the dam) under horizontal excitation, p 1. As shown in the gure, for the rst mode, the results with the dam (tower + dam + reservoir) are similar to the results without the dam (tower + reservoir). However, a signicant amplication of the mode 2 pressures can be observed for the case with the dam.
Copyright 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

SEISMIC RESPONSE OF INTAKE TOWERS

313

Table I. Peak frequencies for different elements of the system (damtower distance = 100 m).
Element Tower only Dam + reservoir Reservoir only Tower + innite reservoir Tower + reservoir + dam Mode First Second First First Second First New peak Second (rad/s) 5.70 35.75 21.50 22.62 4.25 25.25 4.25 21.50 25.50 / 1 1.000 6.267 3.769 3.965 0.745 4.426 0.745 3.769 4.470

To understand the signicant amplications in the mode 2 response, the natural frequencies of the different elements in the system are compared in Table I. As expected, the second natural frequency of the tower in the presence of the dam (tower + reservoir + dam) is slightly higher than the natural frequencies of the tower without the presence of the dam (tower + reservoir) due to the decrease in added mass effect, since the reservoir is blocked in the downstream end by the dam. More interestingly, as shown in Table I, the presence of the dam introduces a new resonance peak at 21.50 rad/s, which is the same as the fundamental frequency of the dam with the effect of the added mass due to the reservoir, 21.50 rad/s, and is near the fundamental frequency of the reservoir itself, 22.62 rad/s. Hence, it can be concluded that this new additional peak on the tower response is caused by the damreservoir system resonance. The results imply a potentially dangerous situation of simultaneous tower, dam, reservoir resonance if the excitation frequency is between 21.5 and 25.5 rad/s, which explains the difference in the pressure response of the tower with and without the dam for the second mode. The predicted displacements at the top of the tower and the dam due to horizontal and vertical seismic excitations at the base are shown in Figures 6 and 10, respectively. The displacements, u , in both the gures are normalized by the peak displacement in the rst modal response of the tower + reservoir without the presence of the dam under horizontal excitation, u 1. The circular frequency, , is normalized with the rst natural frequency of the tower alone without the reservoir, 1 . For the case of horizontal excitation, there is a small decrease in the displacement amplitude of the tower + dam + reservoir case, but no change in frequency at the rst resonance peak ( / 1 = 0.745). However, compared with the case of the tower without the presence of the dam (i.e. tower + reservoir), a noticeable increase can be observed in the displacement at the second resonance peak, along with a slight shift in frequency. More importantly, a new peak can be observed between the rst and the second resonance peaks in the tower + dam + reservoir case. The frequency of this new peak corresponds to the fundamental frequency of the damreservoir system ( / 1 = 3.77), and is near the fundamental frequency of the reservoir ( / 1 = 3.96), as well as the second natural frequency of the tower with added mass due to the reservoir in the presence of the dam, / 1 = 4.47. Thus, it can be concluded that the new additional peak is due to resonance as a result of towerdamreservoir interactions, which signicantly modied the global response of the tower in the presence of the dam. A noticeable inuence of the damtower distance is observed in Figure 6 where the new peak displacement amplication ( / 1 = 3.96) changes from 1.75 for the 50 m distance case to 1.27 for
Copyright 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

314

M. A. MILLAN, Y. L. YOUNG AND J. H. PREVOST

Figure 6. Top of the tower displacement response for the coupled damtowerreservoir case. Concrete dam. Horizontal excitation.

the 100 m distance case. Considering that this new peak is related to the reservoir resonance, which appears to be due to the presence of the dam, the decrease can be expected since the water pressure is reduced when the damtower distance is increased. However, the rst peak ( / 1 = 0.745) is increased from 0.71 for the S = 50 m case to 0.89 for the S = 100 m case and the second peak ( / 1 = 4.470) is increased from 1.0 for the S = 50 m case to 1.21 for the S = 100 m case. This contradictory behavior is explained in the following using the two simple case decompositions described in Section 2. The resonance at the natural frequencies of the tower is determined by the addition of the tower responses for Case 1 (dam excitation only) and Case 2 (tower excitation only) described in Section 2. As explained before, the inuence of the dam on the tower response for large separation distance between the dam and the tower for Case 2 is negligible. Thus, the changes in the peak should be explained using Case 1. When the dam is excited with the tower base xed, the resultant pressure eld does not vary uniformly with distance, as represented in Figure 7. The response of the tower for this case depends on the damtower distance and may or may not increase the global response of the tower, depending on the value and phase of the pressure wave. In Figure 8, the real and imaginary components of the frequency displacement response at the top of the tower for Case 1 and Case 2 are represented. The displacements at the rst natural frequency of the tower for Case 1 are in opposite phase with the displacements for the tower + reservoir case (Case 2 in Section 2). Since in Case 1 the pressures decrease with distance away from the dam, but not in the Case 2 pressures, and since the two pressure responses are in opposite phase, its total value increases when the separation distance increases from 50 to 100 m. Similarly, the displacements at the second natural frequency for Case 1 change when the damtower distance increases (due to the change in the pressure eld with distance) and present a small shift in frequency. This produces an increase in the nal tower displacement at this frequency. A similar consideration can be applied to the analysis of the change in the new resonance peak. From Figure 8, the displacements corresponding to case 1 are reduced with distance away from
Copyright 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

SEISMIC RESPONSE OF INTAKE TOWERS

315

Figure 7. Real and imaginary part of the water pressure distributions for Case 1 described in Section 2. Second tower natural frequency. Concrete dam. Horizontal excitation.

the dam and those corresponding to Case 2 (tower + reservoir) do not change. Thus, a reduction in the nal displacements of the tower is expected when the distance changes from 50 to 100 m. The natural modes of vibration and the associated water pressure distribution for different damtower distances are shown in Figure 9. A very important change can also be observed for the vertical excitation case shown in Figure 10. For the tower without the dam, when an innite reservoir is considered, there is no horizontal displacement due to the geometric symmetry, and hence it is not shown in Figure 10. When the presence of the dam is considered, however, a vertical excitation produces waves propagating from the dam to the innite, which leads to an unexpected vibration of the tower around the new
Copyright 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

316

M. A. MILLAN, Y. L. YOUNG AND J. H. PREVOST

Figure 8. Real and imaginary part of the horizontal displacement of the top of the tower for Cases 1 and 2 in Section 2. Second tower natural frequency. Concrete dam. Horizontal excitation. As shown, the pressure eld varies with distance.

Copyright

2008 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

SEISMIC RESPONSE OF INTAKE TOWERS

317

Figure 9. Natural modes of vibration and associated water pressure distribution for different damtower distances. Horizontal excitation.

Figure 10. Top of the tower displacement response for the coupled damtowerreservoir case. Concrete dam. Vertical excitation.

resonance mode ( / 1 = 3.77) and the second resonance mode ( / 1 = 4.47), which resulted in two closely spaced peaks in the displacement response. When the damtower distance increases, the vertical tower response decreases at both the peaks. As pressures propagating from the dam are the main cause of this resonance, an increased
Copyright 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

318

M. A. MILLAN, Y. L. YOUNG AND J. H. PREVOST

Figure 11. Reservoir bottom absorption inuence. Horizontal and vertical excitation. The displacements, u , are normalized by the maximum rst peak displacement of the tower with reservoir (Case 2) with the corresponding coefcient, u (Case 2, ).

Copyright

2008 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

SEISMIC RESPONSE OF INTAKE TOWERS

319

damtower distance results in a decreased displacement response (as for the horizontal excitation case). Similar conclusions can be obtained by analyzing pressure results but they are not included in the present paper due to length limitations. The inuence of the reservoir bottom absorption is analyzed in Figure 11 for horizontal and vertical excitations. The displacements, u , are normalized for this case by the maximum rst mode displacement of the tower with the reservoir (Case 2) with the corresponding coefcient, u (Case 2, ). For example, the displacement for the tower placed 50 m from the dam with = 0.75 (denoted by tower S = 50 m, = 0.75) is normalized by the displacement for the tower in an innite reservoir with = 0.75, u (Case 2, = 0.75) The wave reection at the bottom boundary is represented by the absorption coefcient , which can vary from 0 to 1. = 0 implies a non-reective boundary and = 1 implies a fully reective boundary. In this paper, an = 0.75 is assumed for the bottom absorption case and = 1 for the non-absorptive bottom case. A similar analysis considering the simple cases in Section 2 is used to explain the behavior of the system. In Figure 12, the real and the imaginary part of the displacement responses are compared for the cases with and without the bottom absorption. A small amplication of the rst resonance peak is observed for the horizontal excitation case when changes from 1 to 0.75. As shown in Figure 12, the Case 1 pressures remain almost unchanged with except for the higher frequencies, but the tower response for the tower + reservoir case is reduced from = 1 to = 0.75, as can be expected. Since they are in opposite phase, the nal peak is increased for the = 0.75 case. The most important change due to the bottom absorption is that the peak related to the reservoir resonance is signicantly reduced, as shown in Figure 11. However, the second resonance peak of the tower increases for the horizontal excitation case when the bottom absorption is considered. For the tower + reservoir case (Case 2 in Section 2) a clear reduction in both the real and the imaginary part of the displacement response can be observed for all the frequency ranges. For Case 1, very close results for majority of the frequency range can be observed for = 1 and = 0.75 cases, except a small increase in the response for = 0.75 and a small shift in frequency around the second natural frequency of the tower. This case 1 behavior can be explained by considering the change in the pressure eld caused by the reservoir bottom absorption. Both the effects lead to a higher second-peak response. The contrary takes place for the vertical excitation case because, for this case, the resonance is related mainly to the water pressures created by the dam vertical excitation, which are reduced when the bottom absorption is considered. The results suggest that dangerous simultaneous resonance of the tower, dam, reservoir can occur for the selected conguration due to towerdamreservoir interaction. This effect cannot be realized with the previous analysis method due to the assumption of the innite reservoir without the dam. 3.3. Time-domain seismic response Due to the drastic differences observed in the frequency-domain response, the time-domain response of the tower with and without the presence of the dam is analyzed in this section. The time-domain seismic response is obtained from the frequency-domain response using the inverse FFT. The earthquake data used in the analysis correspond to the El Centro earthquake (1940) and the Taft earthquake (1952). The selected earthquake has some xed characteristics that can heavily inuence the nal response. To examine the sensitivity to the frequency content, the
Copyright 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

320

M. A. MILLAN, Y. L. YOUNG AND J. H. PREVOST

Figure 12. Real and imaginary part of the horizontal displacement of the top of the tower for Cases 1 and 2 in Section 2. Second tower natural frequency. Concrete dam. Reservoir bottom absorption = 0.75. Horizontal excitation.

Copyright

2008 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

SEISMIC RESPONSE OF INTAKE TOWERS

321

original earthquake spectrum is shifted slightly to determine how the nal response can change for a similar earthquake with a slightly different frequency content. The inuence of the frequency shift on the nal time-domain horizontal displacements at the top of the tower with and without the presence of the dam is compared in Figure 13 considering both the horizontal and vertical seismic inputs. The displacement amplications shown in the gures are relatives to the horizontal displacement in the tower + reservoir (without dam) case under horizontal excitation only. Three different lines are presented in the gures, corresponding to the tower + dam + reservoir case under horizontal excitation only, the tower + dam + reservoir case under vertical excitation only, and the tower + dam + reservoir case under horizontal and vertical excitation combined. The last one is obtained by adding the two previous responses, since the analysis is linear. The responses for horizontal excitation only and vertical excitation only considering the dam are clearly lower than the response for horizontal excitation without the dam, since the rst natural mode dominates the response and there is no rst mode resonance for the vertical excitation case or its peak is reduced for the horizontal excitation case. However, when the horizontal and vertical excitations are considered simultaneously, the addition of both responses may produce a nal response higher than the horizontal excitation case without the dam (usual models). For example, consider the case of the response of the concrete damtowerreservoir system to the El Centro earthquake. Figure 13 (upper part) shows that for zero frequency shift and horizontal excitation, the displacement at the tower top is roughly 27% less than that produced in the tower without the dam under horizontal excitation. For vertical excitation and no frequency shift, the displacement at the top of the tower in the damtowerreservoir system is roughly 25% less than that produced in the tower without the dam under horizontal excitation. Assuming the following values for the response: Tower displacement in the towerreservoir system under horizontal excitation = 100% Tower displacement in the towerdamreservoir system under horizontal excitation = 73% Tower displacement in the towerdamreservoir system under vertical excitation = 75% Then, for the case of combined horizontal and vertical excitations, we can add the time-domain responses for the horizontal excitation case and the vertical excitation case. If the maximum amplications for the horizontal case and the vertical case occur at the same time, then the maximum tower displacement for the towerdamreservoir system for the above example would be 148 or 48% greater than the response for the case with tower without the dam. However, the maximum tower displacement would probably be less than 148%, since the maximum amplications for the horizontal case and the vertical case would occur at different times (in fact, for this particular example, the maximum tower displacement of the towerdamreservoir system subject to combined horizontal and vertical excitations is 110%). Nevertheless, for the combined horizontal and vertical excitation case, the maximum response of the towerdam reservoir system is still likely to be signicantly higher than that for the case of the tower without the dam. The implication, for this particular example, is that under a realistic excitation, which has contributions from a broad range of frequencies, the presence of the dam in fact leads to a reduction (not amplication) in the response if the excitation is purely horizontal. However, under vertical excitation, which would not produce any response in the towerreservoir system, the presence of the dam causes signicant amplication in the response to be produced.
Copyright 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

322

M. A. MILLAN, Y. L. YOUNG AND J. H. PREVOST

Figure 13. Concrete dam. Horizontal displacement amplication at the top of the tower in the time domain for the damtower coupled case relative to the tower + reservoir case. El Centro and Taft earthquakes. Damtower distance 50 m. Damtower H: horizontal excitation only; damtower V: vertical excitation only; damtower H + V: horizontal and vertical excitation combined.

Copyright

2008 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

SEISMIC RESPONSE OF INTAKE TOWERS

323

Figure 14. Concrete dam. Reservoir bottom absorption = 0.75. Horizontal displacement amplication at the top of the tower in the time domain for the damtower coupled case relative to the tower + reservoir case. El Centro and Taft earthquakes. Damtower distance 50 m. Damtower H: horizontal excitation only; damtower V: vertical excitation only; damtower H + V: horizontal and vertical excitation combined.

Copyright

2008 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

324

M. A. MILLAN, Y. L. YOUNG AND J. H. PREVOST

The same analysis is made for the reservoir bottom absorption case and is presented in Figure 14, showing increased amplications compared with the non-absorptive case. This result is expected since the rst resonance, which dominates the nal response, is increased when the bottom absorption is considered.

4. EARTH DAM MODEL 4.1. Description of the BEM model Although the dynamic behavior of earth dams is known to be nonlinear, linear behavior is assumed in this work in order to analyze the inuence of the dam presence on the dynamic response of the tower. This assumption can be accepted considering small amplitude vibrations and taking into account that the dam response is not the main objective of the research so there is no need to do the analysis with a more complicated model. The earth dam is idealized as a triangle with an upstream and a downstream face with a slope of 1.5H:1V. The transversal dimension of the dam is assumed to be equal to the reservoir width, which is 300 m. Free slip boundary conditions are applied at the ends of the dam on the lateral face of the block, and zero displacement condition is applied in the transversal direction. The water depth H at the upstream face of the dam is the same as the dam and the tower height, is 100 m. The dam is assumed to be symmetric about the x z plane. The half width of the dam is b = 150 m (the full width is 2b = 300 m). The BEM model for the earth dam and intake tower is shown in Figure 3 (bottom part). As for the concrete dam case, symmetry about the x z plane is exploited. Others parameters like reservoir discretization length, transmitting boundary and intake tower representation are assumed to be identical to those considered for the concrete dam analysis. In this case, the damtower distance is assumed to be 50 m at the ground level. The assumed parameter values for the dam are: dam modulus of elasticity E = 150 106 N/m2 , dam Poissons ratio = 0.364, dam density = 2000.0 kg/m3 , dam damping ratio = 0.05, water wave velocity c = 1440 m/s, water density w = 1000 kg/m3 and reservoir depth H = 100 m. 4.2. Frequency-domain seismic response Results for the models with and without the dam subject to horizontal and vertical seismic excitations are presented in Figure 15, where the displacements, u , are normalized for this case by the maximum rst mode displacement of the tower with the reservoir (Case 2) with the corresponding coefcient, u (Case 2, ). As expected, the dam resonance is less clear than for the concrete dam case, since it is a massive structure and it is much more exible than the concrete dam. This produces a minor inuence of the reservoir resonance on the tower response. This behavior can be observed in Figure 15 (upper part) for horizontal excitation, where the tower response shows little difference with or without the dam. However, the dam inuence on the tower behavior is signicant for the vertical excitation case, as shown in Figure 15 (bottom part). Due to the dam presence, the vertical excitation produces horizontal pressure waves that excite the dams second mode response since it is very close to the reservoir resonance frequency. As for the concrete dam case, there is also a small shift in frequency for the second tower mode. The inuence of bottom absorption can also be observed in Figure 15. For the horizontal excitation case, a small
Copyright 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

SEISMIC RESPONSE OF INTAKE TOWERS

325

Figure 15. Top of the tower displacement response for the coupled damtowerreservoir case. Earth dam. Horizontal and vertical excitation. The displacements, u , are normalized by the maximum rst peak displacement of the tower with reservoir (Case 2) with the corresponding coefcient, u (Case 2, ).

Copyright

2008 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

326

M. A. MILLAN, Y. L. YOUNG AND J. H. PREVOST

amplication on the rst and second peaks is shown, similar to the concrete dam case. For the vertical excitation case, a reduction in the resonance peak from 1.64 ( = 1 case) to 1.09 ( = 0.75 case) is observed. 4.3. Time-domain seismic response Similar to the concrete dam case, the time-domain response of the tower with and without the presence of the earth dam is analyzed in this section, using the same earthquake data, considering both the horizontal and the vertical seismic inputs. The inuence of the frequency shift on the nal time-domain displacements at the top of the tower with and without the presence of dam is compared in Figure 16 considering both the horizontal and the vertical seismic inputs. The resultant behavior is very similar to that observed for the concrete dam case. Similar analysis was also conducted to study the inuence of the reservoir bottom absorption. Figure 17 shows increased amplications for = 0.75 compared with = 1, as expected, since the rst resonance, which controls the nal response, is increased when the bottom absorption is considered.

5. CONCLUSIONS In all previous towerdamreservoir uidstructure interaction models, the surrounding water is assumed to be of uniform depth and to have an innite extension. Hence, the added mass effect of the external uid is associated with the pressures created by the displacements of the tower only. However, for a real system, the intake tower is usually located in proximity to the dam, where wave reections from the dam may change the resultant added mass, which will in turn change the natural frequencies and seismic response of the tower. In this work, the inuence of the dam on the seismic response of the tower is analyzed using a 3D BEM numerical model, which lead to the following conclusions: (i) For the selected conguration, the presence of the dam leads to the introduction of a new resonance mode near the towers second resonance frequency due to the damreservoir excitations. Consequently, a potential amplication of the response under the horizontal excitation may be produced due to the towerdamreservoir interaction if the excitation frequency is near the fundamental frequency of the dam and the reservoir. As time-domain results show, this amplication is usually less than 1020% in most cases (and there is no amplication for many of the cases). (ii) When the dam presence is considered, the tower under vertical excitation does not present symmetry, which leads to resonances that could not be captured using conventional tower analysis methods due to the assumption of an innite reservoir around the tower. (iii) The results show that when considering simultaneous horizontal and vertical excitation, the tower displacements and pressures can be signicantly amplied due to towerdam reservoir interactions. This behavior is more signicant for a concrete dam than for an earth dam. (iv) Additional research is needed to analyze the inuence of other important parameters in the response: e.g. exibility of the dam and tower, variations in water depth, soilstructure interactions, spatial variability of the excitation, etc.
Copyright 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

SEISMIC RESPONSE OF INTAKE TOWERS

327

Figure 16. Earth dam. Horizontal displacement amplication at the top of the tower in the time domain for the damtower coupled case relative to the tower + reservoir case. El Centro and Taft earthquakes. Damtower H: horizontal excitation only; damtower V: vertical excitation only; damtower H + V: horizontal and vertical excitation combined.

Copyright

2008 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

328

M. A. MILLAN, Y. L. YOUNG AND J. H. PREVOST

Figure 17. Earth dam. Reservoir bottom absorption = 0.75. Horizontal displacement amplication at the top of the tower in the time domain for the damtower coupled case relative to the tower + reservoir case. El Centro and Taft earthquakes. Damtower H: horizontal excitation only; damtower V: vertical excitation only; damtower H + V: horizontal and vertical excitation combined.

Copyright

2008 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe

SEISMIC RESPONSE OF INTAKE TOWERS

329

It should be noted that the models analyzed do not include the effects of the interior water and the soilstructure interaction. Future work will address these effects to enhance the predictive capabilities of the current model.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the use of the 3D Boundary Element software developed by J. Dominguez (University of Seville, Spain), O. Maeso and J.J. Aznarez (University of Las Palmas de Gran Canaria, Spain).
REFERENCES 1. Goyal A, Chopra AK. Hydrodynamic and foundation interaction effects in dynamic of intake towers: frequency response functions. Journal of Structural Engineering 1989; 115(6):13711385. 2. Goyal A, Chopra AK. Hydrodynamic and foundation interaction effects in dynamic of intake towers: earthquake responses. Journal of Structural Engineering 1989; 115(6):13861395. 3. Goyal A, Chopra AK. Simplied evaluation of added hydrodynamic mass for intake towers. Journal of Engineering Mechanics 1989; 115(7):13931412. 4. Goyal A, Chopra AK. Simplied evaluation of added hydrodynamic mass for intake towers. Journal of Engineering Mechanics 1989; 115(7):14131433. 5. Maeso O, Aznarez JJ, Dominguez J. Effects of space distribution of excitation on seismic response of arch dams. Journal of Engineering Mechanics 2002; 128(7):759768. 6. Maeso O, Aznarez JJ, Dominguez J. Three-dimensional models of reservoir sediment and effects on the seismic response of arch dams. Earthquake Engineering and Structural Dynamics 2004; 33:11031123. 7. Mill an MA, Young YL, Pr evost JH. The effects of reservoir geometry on the seismic response of gravity dams. Earthquake Engineering and Structural Dynamics 2007; 36:14411459. 8. Brebbia CA, Dominguez J. Boundary Elements, An Introductory Course. Computational Mechanics. McGraw-Hill Book Co.: Southampton, Boston, New York, 1992. 9. Fenves G, Chopra AK. Effects of reservoir bottom absorption on earthquake response of concrete gravity dams. Earthquake Engineering and Structural Dynamics 1983; 11:809829.

Copyright

2008 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2009; 38:307329 DOI: 10.1002/eqe