This paper presents an application of the EPRI methodology for determining the probabilistic seismic structural response analysis (displacements and acceleration spectra). The objective is to explicitly account for the variability in response due to randomness of the ground motion and uncertainty of the soil data and the building dynamic behavior. The main challenge is to handle complex information and modeling in an industrial type study. This study has been performed as a part of a SPRA of a power plant situated in a moderate seismicity area. First the ground motion of the free field is determined on the basis of a Probabilistic Seismic Hazard analysis. The median and the 84th percentile Uniform Hazard Spectra of 100,000 years return period are considered and artificial earthquake time histories are generated. The material variability is considered for the Young's moduli of the materials, the damping of the structure and the soil shear modulus and material damping. A Latin Hypercube sampling method was used in order to generate a relatively small but statistically representative number of samples of material parameters and time histories. A statistical work has been done to check out the relevance of the initial LHS estimates. Best estimate mathematical building models were developed in order to perform the seismic response analysis. The safety-related buildings were represented by 3D finite element models. The soil structure interaction was accounted by coupling the FE code (Code_Aster) with a boundary element code (Miss3D) which allows a realistic representation of the soil profile. The influence of the building weight on the soil state of stress was taken into account by a correction factor. The maximum soil shear strains expected during an earthquake that matches the selected UHS level were determined by a non linear model (Cyberquake) and the shear modules of the soil layers were reduced accordingly. The ground motion is imposed on the free surface. The deconvolution of the UHS ground motion between the surface and the raft depth is determined by the resolution of the SSI problem. The floor response spectra were generated on the basis of the time histories applied to the best estimate model. Then the median and the 84th fractile were determined for further use in the SPRA analysis. For the studied case, the results demonstrated the beneficial effect of the SSI for massive buildings founded on soft soil. Our feedback is that the EPRI methodology provides a good balance between mathematical accuracy and practicability required by complex industrial studies.
展开▼