Middle East Technical University Middle East Technical University Middle East ..... Earthquake Engineering Research Center, University of California, Berkeley,.
Effect of Base Roughness on Seismic Response of Concrete Gravity Dams Baris BINICI Middle East Technical University Ankara Turkey
Alper ALDEMIR Middle East Technical University Ankara Turkey
Ali GHARIBDOUST Middle East Technical University Ankara Turkey
Introduction Seismic response of concrete gravity dams is dominated by two types of deformation modes: body deformations with cracks and base sliding due the damage at the dam-foundation interface. Rigid block models are commonly used to estimate base sliding displacements which are extremely crucial to design water stoppers, whereas deformable body nonlinear numerical models are preferred to estimate crack patterns during earthquakes. It is extremely challenging to estimate the switch between two modes and their interaction depending on the crack amounts and surface roughness at the dam-foundation interface. In this study, two scaled dam experiments were conducted using the pseudo dynamic test method devised specifically for dam testing. The only difference between the two specimens was only the surface roughness condition at the dam base. After briefly explaining the employed test procedure, results are presented for the two simulated dynamic tests. It was found that significant sliding displacement can take place if sufficient shear locking is not provided at the interface. Upon providing sufficient shear friction resistance through improved roughness conditions, the damage was in the form of severe cracking both at the dam base and as inclined cracks on the dam body.
1. Background Mankind have struggled to find water resources to maintain their lives since the beginning of civilizations. This effort has recently become more challenging due to the rise in human population and global warming, which necessitates more water storage for both agricultural and energy generation purposes. This fact obligates the proper and efficient usage of natural resources to supply more water and energy. In addition, the fossil fuel resources have been lost their popularities due to their threat to the environment as they release greenhouse gases in the energy production process. Consequently, the clean energy technologies like hydroelectric power plants come into prominence in developing countries. Evidently, dams are one of the most suitable structures that could serve for both water storage and energy production. From the beginning of 1900’s, excessive amount of researches have been carried out to identify the behavior of gravity dams especially under the effect of ground excitations. The charm of these special structures among the researchers are i- explaining an extremely complex interaction problem of dam-reservoir-foundation, ii- ensuring the functionality of dams after disasters, iii- estimating complete or partial collapse scenarios of dams. Although there are numerous researches on the analytical examination of dams (Fenves and Chopra 1984, Medina and Dominguez 1989, Slowik and Saouma 2000, Lim et al 2012, Banerjee et al 2014, etc.), the experimental research is very limited due to the tremendous sizes of dam bodies. This problem could be overcome by utilizing small scales in laboratory specimens or by instrumenting existing dams. While performing laboratory experiments, the simulation of hydrodynamic effects and the acquirement of equivalent stress distribution over the dam body with the unscaled prototype are the important challenges. Although the former issue could be solved by placing a reservoir separated by a plastic film from the upstream dam face (Niwa and Clough 1980), the behaviour of dam bodies with empty reservoir cases was mostly investigated experimentally in literature. The latter problem was dealt with the utilization of extra external forces, with the scaling of gravitational acceleration (or density of the material) or with decreasing both modulus of elasticity and compressive strength (Niwa and Clough 1980, Donlon and Hall 1991 and Harris et al 2000). In literature, it was favored to scale the gravitational effects by utilizing centrifuge machines as this solution did not change the material characteristics of concrete (Uchita et al 2005).
PAPER 0204
Page 1/11
Gravity dams under seismic excitation exhibit two major failure types, failure due to body crack propagation and dam base sliding. Previous researches provided simplified method to assess sliding of dams. Chopra and Zhang (1991) studied earthquake induced base sliding response of concrete gravity dams, with no bond, by developing a semi-analytical procedure. In their research, the possibility of estimating the sliding response of flexible dam by analysing the response of a rigid dam model is investigated. The analyses results indicate that, in presence of full reservoir, the acceleration required to cause downstream sliding of dam is much lower than that of other failure modes that is due to hydrostatic and hydrodynamic loads. However, considering empty reservoir upstream tipping is possible, too. Danay and Adeghe (1993) studied weak base rock joints to develop an empirical formula for predicting the seismic-induced slip of the concrete. They performed series of parametric studies on gravity dams based on the “sliding block” concept and the equivalent single degree of freedom system criteria. Proposed formula was validated by finite element analysis, and provides a fair prediction of the seismic-induced slip for all typical sections of concrete gravity dams. In a study carried out by Basili and Nuti (2011) a nonlinear single degree of freedom model in which dam-water-foundation interaction is considered was developed to obtain concrete gravity dam base sliding induced by an earthquake. Considering threshold value for the sliding foundation resistance a cohesive model was utilized and the nonlinearity was assumed in the foundation rock. The study concluded that dam seismic residual displacement can be approximated by performing equivalent SDOF system parameters instead of nonlinear dynamic analyses while it depends on dam height, impounded water level and foundation rock type. In this study, pseudo dynamic tests are conducted on two different 1/75 scaled dam models of Melen Dam, designed to supply fresh water and to generate electricity to Istanbul, will be utilized. The geometrical properties of the deepest section of scaled and unscaled dam are summarized in Figure 1.a. All of the specimens are made of roller compacted concrete (RCC) but the base roughness conditions of the two specimens are different. The foundation of the first specimen is intentionally roughened (Roughened Base, RB) to have enough shear resistance according to ACI 318-11 corresponding to a roughness amplitude of 6mm whereas the foundation of the second specimen (Flat Base, FB) is smoothened. Therefore, the effect of shear resistance at the foundation level of dam specimen will be investigated. The mechanical properties of dam specimens are determined from standard cylinder compression tests at the experiment days, which come out to be 15.58 MPa and 20.40 MPa for the RB and FB specimens, respectively.
2. Seismicity and testing methodology 2.1. Summary of local seismicity conditions The prototype dam, Melen Dam, is located in Sakarya Province, which lies on the most active seismic zone in Turkey. Therefore, site specific design spectrums for three different seismic hazard levels are generated and then site specific synthetic ground motions are obtained from those site specific design spectrums (Akkar 2010). The specimens are tested under the effect of three different earthquake scenarios, i.e. Operational Based Earthquake (OBE), Maximum Design Earthquake (MDE) and Maximum Credible Earthquake (MCE), corresponding to 144yr, 475yr and 2475yr return periods, respectively. Those ground motion histories are depicted in Figure 1.b-c. In this figure, the response spectrums for those scenarios are also plotted. 2.2. Summary of testing methodology The pseudo dynamic testing procedures are appropriate for lumped mass systems as they are implemented by utilizing hydraulic jacks at a finite number of joints. Therefore, for distributed mass systems like dam monoliths, the application of the pseudo-dynamic testing procedures requires some simplifications on the test procedure. In literature, Fenves and Chopra (1985) proposed a simplified single degree of freedom (SDOF) approach to estimate the seismic demands on the dam monoliths approximately and, nearly thirty years later, Basili and Nuti (2011) claimed that the single-degree-of-freedom approach was sufficiently accurate in estimating the base shear and overturning moment demands at the dam base. Inspired from these ideas, a pseudo-dynamic testing procedure was proposed in a recent study (Aldemir et al 2015). For this purpose, the scaled cross-section of a dam is idealized as a distributed stiffness system up to its critical height (hp) where a concentrated mass (m) is located as shown in Figure 2. With a proper selection of m and hp, earthquake induced stress demands at the base of the dam can match those obtained from a rigorous procedure, hence a SDOF idealization is shown to be a viable option for conducting pseudo dynamic tests of dam monoliths under seismic loading. In this study, the program EAGD, developed by Fenves and Chopra (1984), was utilized to obtain the exact stress and force distributions. In this program, dam-water interaction, wave absorption at the reservoir boundary, water compressibility, and dam-foundation rock interaction are considered by utilizing frequency domain analysis.
PAPER 0204
Page 2/11
(a)
Melen Dam
Scaled Dam Specimen Fv m Fh
(b)
(c)
Figure 1 – (a) Geometrical properties of the deepest section of scaled and unscaled dam, (b) Spectrums for OBE, MDE and MCE and (c) Ground motion histories As mentioned above, the objective was to obtain analogous stress distributions over the dam base, which was the most vulnerable to cracking portion of gravity dams. At first, the additional physical loads (both vertical and lateral) should be calculated in order to obtain similar stress distributions on scaled dam base. To determine the extra static loads (dead+hydrostatic), a trial-and-error procedure was carried out. In this procedure, the aim was to obtain equal stress distributions over dam base from both scaled (ANSYS) and unscaled (EAGD) models. After this trial-anderror procedure, the closest stress results were obtained with a 400 kN of vertical and 174 kN of lateral loads. It is apparent from Figure 2.a that the principal stresses over the dam base from both scaled (ANSYS) and unscaled (EAGD) analysis are very similar. After that, unscaled dam was analysed using EAGD for the selected ground motions. The resulting overturning moment-base shear response is shown in Figure 2.b. The slope of the M-V response curve results in hp=70m, which corresponds hp=0.95m if a scale factor of 1/75 is applied. Scaled dam model was then analysed with the same ground motions for different values of concentrated mass (m) in order to determine the optimal equivalent mass to use in the analyses. The m-values that minimize the difference between the demand parameters obtained from scaled and unscaled models were identified as m1=37.5 ton, m2=40.0 ton, m3=55.0 ton for the OBE, MDE and MCE ground motions, respectively. The comparisons of the base shear, moment and stress demands obtained from the dynamic analyses of scaled and unscaled models of RB specimen for OBE and MCE motions are presented in Figure 3 and 4. Scaled model estimates the time history of the stress with a reasonable accuracy (less than 30% error for the vertical stresses). These comparisons for FB specimen result in similar matches but those graphs are not shown due to the size limitations. More explanation on the test procedure can be found in Aldemir et al (2015).
3. Instrumentation and testing After curing of the test specimens, the instrumentation was installed. Linear variable differential transducers (LVDTs) were installed to record the lateral and vertical movements of dam base in both upstream and downstream directions (Figure 5.a and 5.b). In addition, three different LVDTs were installed at the top of the specimen. One of the LVDTs was used to check the feedback information supplied by a high precision displacement transducer (accuracy of ±10 µm) to the pseudo dynamic system (Figure 5.c). Another LVDT was placed to measure the relative displacement of the top of test specimen with respect to its base (Figure 5.d). The third one was set up on directly concrete just below the transfer plate in order to detect the slip at the interface of transfer plate and concrete portion (Figure 5.d).
PAPER 0204
Page 3/11
(a)
(b)
Figure 2 – (a) Comparisons of maximum principal stress due to statical loading and (b) Effective height calculations The vertical loading system was built with tie rods and hydraulic cylinders (Figure 5.e). The prestressing forces on the tie rods were transferred to the dam as compressive forces through a built-up steel section. Keeping this vertical load constant, the hydrostatic load (174 kN) was applied to the specimen using an hydraulic jack. The execution of the pseudo dynamic testing was initiated by assuming at rest conditions with zero lateral force. Detailed information on the instrumentation could be found in Aldemir et al (2015).
Figure 3. Comparison of the analysis results of RB specimen for the OBE ground motion: (a) Base Shear, (b) Overturning Moment and (c) Vertical Stress 3.1. RB specimen After the hydrostatic loading, the first level of earthquake, the OBE ground motion, was applied to the specimen. During the OBE motion, some minor cracks at the base were observed. The maximum crack length was recorded as 200mm with the maximum crack width reaching 0.2 mm (Figure 6). The maximum base shear and the tip displacement demands were measured as 48 kN and 0.33 mm, respectively. Furthermore, the deformation at the base was less than 0.05 mm during the full duration of the test (Figure 7). Afterwards, the same specimen was tested at the second hazard level, MDE. During this stage of the testing, the maximum tip displacement was obtained as 0.66 mm, corresponding to a maximum base shear demand of 132 kN. The increase in the displacement and force demands with respect to the OBE level were calculated as 100.7% and 173.1%, respectively. Consequently, MDE level testing resulted in increased cracks lengths and widths. The maximum crack width and length reached 0.5mm and 450mm, respectively (Figure 6). Unlike the first specimen, no
PAPER 0204
Page 4/11
body crack was observed at the upstream face during this level. The base deformation was less than 0.10mm during this level as well (Figure 7).
Figure 4. Comparison of the analysis results of RB specimen for the MCE ground motion: (a) Base Shear, (b) Overturning Moment and (c) Vertical Stress
(c) (d)
(b) (a)
(a)
(e)
(e)
(c)
(b)
(d)
Figure 5. Test setup and instrumentation : (a) LVDT’s at upstream dam base, (b) LVDT’s at downstream dam base, (c) High precision displacement transducer, (d) LVDT’s at the dam tip and (e) Static axial load application setup During the MCE level testing, the top displacement demand reached to a maximum of 1.65mm, which was approximately 2.5 times (150 % increase) the displacement demand observed in the MDE level. Similarly, the base
PAPER 0204
Page 5/11
shear demand was obtained as 222 kN, nearly 1.7 times (68 % increase) the maximum demand measured during the MDE level testing (Figure 7). Two parallel body cracks in the upstream face were formed on the dam body reaching to a length of 300mm (Figure 6). During this earthquake motion, base cracking at the downstream face was not observed. After the end of the MCE testing, a static pushover test was also conducted in order to determine the reserve capacity in the specimen as no base sliding or stability problems were observed during previous tests (Figure 8.a). During the pushover loading, the specimen reached its maximum capacity at a base shear value of around 330 kN corresponding to a tip displacement of 2.5 mm. A yield plateau was obtained until a tip displacement of nearly 4.3 mm was reached. After that point, the capacity of the specimen suddenly dropped to 260 kN due to the enhanced body cracks and their propagations through the dam body (Figure 8.b). The specimen had lost its load carrying capacity immediately after the inclined cracks converged to the downstream toe of the dam specimen. The failure was caused by shear – compression mechanism (Figure 8.c).
Crack Width 0.1 mm
OBE
Crack Length 200 mm Upstream Base Cracks
Crack Width 0.5 mm
Crack Length 200 mm MDE
Upstream Base Cracks
Crack Width 0.4 mm
Crack Length 200 mm
MCE
Body Crack
Figure 6 – Observed Cracks of RB Specimen during OBE, MDE and MCE experiments
PAPER 0204
Page 6/11
Figure 7 – Force and displacement demands of RB Specimen during OBE, MDE and MCE experiments
(a)
(b)
(c) Figure 8 – Results of Pushover experiment of RB Specimen 3.2. FB specimen experiment The first level of earthquake (OBE) was applied to the specimen after the application of hydrostatic loading. During the OBE motion, some minor cracks on the dam body were observed. The maximum crack length was recorded as 110mm with the maximum crack width reaching 0.1 mm (Figure 9). The maximum base shear and the tip displacement demands were measured as 88 kN and 0.16 mm, respectively. Furthermore, the deformation at the base was less than 0.05 mm during the full duration of the test (Figure 10). Afterwards, the same specimen was tested at the second hazard level, MDE. During this stage of the testing, the maximum tip displacement was obtained as 0.32 mm, corresponding to a maximum base shear demand of 217 kN. The increase in the displacement and force
PAPER 0204
Page 7/11
demands with respect to the OBE level were calculated as 100% and 146.6%, respectively. Another body crack was observed at the upstream face during this level of earthquake. The base deformation was less than 0.10mm during this level as well (Figure 10). During the MCE level testing, the top displacement demand reached to a maximum of 1.70mm, which was approximately 5.3 times (431% increase) the displacement demand observed in the MDE level. Similarly, the base shear demand was obtained as 269 kN, nearly 1.24 times (24% increase) the maximum demand measured during the MDE level testing (Figure 10). Consequently, MCE level testing resulted in increased cracks lengths and widths. During this earthquake motion, body cracking at the downstream face was also observed. During this scenario, base sliding was also detected. The maximum base sliding value was 1.16mm, which was nearly 70% of the total tip displacement. Therefore, it could be stated that the tip displacement was dominated by the excessive base sliding not the body deformations contrary to the RB specimen. After the end of the MCE testing, a static pushover test was also conducted in order to determine the reserve capacity in the specimen (Figure 11.a). During the pushover loading, the specimen reached its maximum capacity at a base shear value of around 286 kN corresponding to a tip displacement of 2.5 mm. A yield plateau was obtained until a tip displacement of nearly 14.1 mm was reached. No capacity drop was observed due to the base sliding (Figure 11.b). The specimen had displaced more than 14mm due to the excessive base sliding. This amount of base sliding displacement could cause the functionality loss of the dam body. Therefore, the failure was due to the violation of serviceability criteria (Figure 11.c).
Crack Width 0.1 mm Crack Length 630 mm
OBE
Loading Plate Cracks Crack Width 0.1 mm
Crack Length 195 mm MDE
Body Crack Crack Width 0.1 mm Crack Length 110 mm
MCE
Body Crack
Figure 9 – Observed Cracks of FB Specimen during OBE, MDE and MCE experiments
PAPER 0204
Page 8/11
Figure 10 – Force and displacement demands of FB Specimen during OBE, MDE and MCE experiments
(a)
(b)
(c) Figure 11 – Results of Pushover experiment of FB Specimen
4. Discussion of results and conclusions The effect of different base applications on the lateral response of concrete gravity dams is investigated in this study. Two different foundation conditions are utilized, i.e. roughened and flat. In the RB specimen, no base sliding is observed during neither earthquake scenarios nor pushover experiment. The observed damages are all base cracks and some body cracking during OBE, MDE and MCE levels. However, the body cracks cause the failure of the dam body due to the loss of load carrying capacity. The observed body crack would apparently cause water leakage
PAPER 0204
Page 9/11
through dam body. Unlike the RB specimen, very limited base and body cracks are observed in FB specimen. However, some base sliding is detected even during MCE level. Base sliding during pushover experiment is measured as about 12mm, which results in stability loss of the dam body. Also, the base sliding would tear the waterstoppers at the foundation interface and cause leakage. Therefore, the base application has an important effect on the seismic behaviour of concrete gravity dams. The mode of failure is totally dependent on the base treatment of the dam body. For example,
Acknowledgements This study was supported by the Turkish National Sciences Foundation TUBITAK under the grant number 111M712. The first author acknowledges the support from TUBA-GEBIP for continuing research endeavors. The contributions of Salim Azak and Hasan Metin during the laboratory work are also gratefully acknowledged. References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
11. 12. 13. 14. 15.
16. 17. 18. 19.
Akkar, S., “Melen Barajı için Tasarım Spektrumunun Olasılık Hesaplarına Dayalı Sismik Tehlike Analizi”, Report No: 2010-03-03-1-01-04, 2010, METU. (in Turkish) Aldemir, A., Binici, B., Arici, Y., Kurc, O. and Canbay, E., “Pseudo-dynamic Testing of a Concrete Gravity Dam”, Earthquake Engineering & Structural Dynamics, 44(11), 1747-1763, 2015. American Concrete Institute (ACI 318-11), Building code requirements for structural concrete, 2011, Farmington Hills. ANSYS Inc., “Basic analysis guide for ANSYS 11”, 2007, SAS IP Inc. Banerjee, A., Paul, D.K. and Dubey, R.N., “Modelling Issues in the Seismic Analysis of Concrete Gravity Dams”, Dam Engineering, 24(2), 1-23, 2014. Basili, M. and Nuti, C., “A Simplified Procedure for Base Sliding Evaluation of Concrete Gravity Dams under Seismic Action”, ISRN Civil Engineering, Vol. 2011, 2011. Chopra A.K. and Zhang L., “Earthquake-Induced Base Sliding of Concrete Gravity Dams”, Journal of Structural Engineering, Vol. 117, 1991. Danay A. and Adeghe N., "Seismic-Induced Slip of Concrete Gravity Dams", Journal of Structural Engineering, 1993, 119(1): 108-129, 1993. Donlon, W.P. and Hall, J.F., “Shaking Table Study of Concrete Gravity Dam Monoliths”, Earthquake Engineering and Structural Dynamics, 20(8), 769–786, 1991. Fenves, G. and Chopra, A.K., “EAGD-84: A Computer Program for Earthquake Response Analysis of Concrete Gravity Dams”, Report No: UCB/EERC-734, Earthquake Engineering Research Center, University of California, Berkeley, California, 1984.a. Fenves, G. and Chopra, A. K., “Earthquake Analysis and Response of Concrete Gravity Dams”, Report No: UCB/EERC84/10, Earthquake Engineering Research Center, University of California, Berkeley, California, 1984.b. Fenves, G. and Chopra, A.K., “Simplified Earthquake Analysis of Concrete Gravity Dams: Separate Hydrodynamic and Foundation Interaction Effects”, Journal of Engineering Mechanics ASCE, Vol. 111, No. 6, pp. 715-735, 1985.a. Fenves, G. and Chopra, A.K., “Simplified Earthquake Analysis of Concrete Gravity Dams: Combined Hydrodynamic and Foundation Interaction Effects”, Journal of Engineering Mechanics ASCE, vol. Vol. 111, No. 6, pp. 736-756, 1985.b. Harris, D.W. Snorteland, N., Dolen, T. and Travers, F., “Shaking Table 2D Models of a Concrete Gravity Dam”, Earthquake Engineering and Structural Dynamics, 29(6), 769–787, 2000. Lim, W.Z., Xiao, R.Y. and Chin, C.S., “A Comparison of Fluid-Structure Interaction Methods for a Simple Numerical Analysis of Concrete Gravity-Dam”, Proceedings of the 20th UK Conference of the Association for Computational Mechanics in Engineering, The University of Manchester, Manchester, UK, 321-324, 2012. Medina, F., Dominguez, J., and Tassoulas, J., ”Response of Dams to Earthquakes including Effects of Sediments”, Journal of Structural Engineering ASCE, 116(11), 3108–3121, 1990. Niwa, A. and Clough, R.W., “Shaking Table Research on Concrete Dam Models”, Report No: UCB/EERC-80/06, Earthquake Engineering Research Center, University of California, Berkeley, California, 1980. Slowik, V. And Saouma, V.E., “Water Pressures in Propagating Concrete Cracks”, Journal of Structural Engineering ASCE, 126(2), 235-242, 2000. Uchita, Y., Shimpo, T. and Saouma, V., “Dynamic Centrifuge Tests of Concrete Dam”, Earthquake Engineering and Structural Dynamics, 34(12), 1467–1487, 2005.
PAPER 0204
Page 10/11
The Authors B. BINICI is a Professor at the Structural Mechanics Division of Civil Engineering Department in Middle East Technical University. He received his BSc degree from Middle East Technical University, MSc and PhD degrees from The University of Texas at Austin. His research interests include reinforced concrete behavior and modelling, performance based design, experimental and numerical investigation of dam structures and seismic strengthening. A. ALDEMIR is a Research Assistant and a PhD Candidate at the Structural Mechanics Division of Civil Engineering Department in Middle East Technical University. He received his BSc and MSc degrees from Middle East Technical University. His research interests include analytical and experimental investigations of concrete and masonry structures. A. GHARIBDOUST is a Research Assistant and a MSc candidate at the Structural Mechanics Division of Civil Engineering Department in Middle East Technical University. He received his BSc degree from Middle East Technical University. His research interests include analytical and experimental investigations of concrete gravity dams.
PAPER 0204
Page 11/11
