Shake Table Model of Concrete Gravity Dam for Computer Code Validation in Monolithic Models

Safety of dams under earthquake loading has always been a major concern in seismically active regions because of the high potential hazard that uncontrolled release of the reservoir poses to the population downstream of the dam. The current practice for the evaluation of the dynamic response of concrete dams at the U.S. Bureau of Reclamation is to initially perform a linear elastic finite element analysis of the structure under static loads plus the maximum credible earthquake predicted to occur at the site. If the response of the structure indicates that significant cracking damage would occur based on high tensile stress results, then non-linear analyses would be completed. In recent years a number of finite element programs have become available to structure engineers which provide the capability of modeling vertical contraction joint opening and concrete cracking which may occur in concrete arch dams during a seismic event. Reclamation has recently purchased the ABAQUS finite element program which is capable of performing nonlinear cracking analyses. The ABAQUS program includes a brittle cracking material model for concrete. This is a fairly complex analytical tool and engineers in Reclamation's structure analysis group need to develop expertise in performing and understanding non-linear modeling and analysis techniques. A set of calibration models would provide a valuable data set for those performing the numerical studies to assure that known results can be obtained. Structure engineers and material engineers need to work closely to develop new standards for material testing so that the appropriate material properties are available for these analyses. Therefore a shaking table test of a 1/50 scale cantilever monolith of Koyna Dam [in India] was built at Reclamation's Material Testing Laboratory for use in calibrating linear-elastic and non-linear material properties, as well as to provide a means of evaluating the effect of various analysis parameters available in the ABAQUS program. Two models were tested, a model with a natural but preexisting crack and a monolithic model. The ultimate goal of the shaking table test program was to develop a correlation between the material testing program and material property requirements for the non-linear numerical concrete cracking model. But before this goal can be accomplished, the basic linear elastic material properties needed to be calibrated and the effects of various analysis parameters needed to be understood. These requirements were accomplished through comparisons of analytical results with laboratory measurements. This report is concerned primarily with these preliminary studies, although some non-linear analysis results are also presented.