Probabilistic Seismic Soil Structure Interaction Analysis of the Mühleberg Nuclear Power Plant Reactor and SUSAN Buildings

November 24, 2010
Publication: ASME 2010 Pressure Vessels and Piping Conference
Author(s): Philip Hashimoto David Nakaki James J. Johnson Yahya Bayraktarli Olivier Zuchuat
Abstract: Deterministic and probabilistic seismic soil structure interaction (SSI) analyses were performed for the Mühleberg Nuclear Power Plant SUSAN Building in support of the seismic probabilistic risk assessment of the plant. An efficient hybrid method, employing computer programs SASSI2000 and CLASSI was used in this analysis. This method takes advantage of the capability of SASSI2000 to analyze embedded structures with irregular geometry and the computational efficiency of CLASSI to rapidly perform the SSI response analysis of large structure models. The structure embedment was modeled using SASSI2000 in which the layered site was represented by the median-centered earthquake strain-compatible soil profile. Impedance functions and scattering vectors were calculated by imposing rigid body constraints to the embedded foundation. The fixed base structure dynamic properties, foundation impedances, and scattering functions were input to the SSI response analysis. Median-centered deterministic SSI analysis was first performed with CLASSI using a single set of acceleration time histories compatible with the site surface uniform hazard spectrum. In structure response spectra (ISRS) were calculated at selected locations in the building. The subsequent probabilistic analysis was performed using computer program SMACS which implements the Latin Hypercube Simulation (LHS) approach documented in NUREG/CR-2015. Variables in the LHS included the earthquake ground acceleration time histories, structure stiffness and damping, and soil stiffness and damping. Thirty response simulations were performed using SMACS in which the variable values were randomly selected. The use of SMACS has the advantage that the response analysis simulations can be executed in a fraction of the time that would be required with SASSI2000 alone. For each simulation, ISRS were calculated. Probability distributions, described by the median and 84th percentile response spectra, were calculated from the thirty simulations. The deterministic median-centered ISRS and probabilistic ISRS were compared. The results demonstrated consistency between the two methods. Also, sharp spectral acceleration peaks in the deterministic ISRS were reduced in the probabilistic analysis.
Markets: Nuclear