A comparative seismic soil-structure-interaction (SSI) study was performed for the reactor building of a nuclear power plant in Switzerland in order to gain insights into the relative significance of nonlinear response effects which may be triggered during beyond-design-basis seismic events. The case study reactor building is founded on 20m of gravel underlain by bedrock and with an embedment depth of 9m. The analysis program was designed as a sequential study to investigate both the independent and combined effects of nonlinear response of soil, soil-structure interface, and structure on the characteristics of in-structure response
spectra (ISRS) throughout the reactor building at a hazard level greater than the design basis earthquake.

An integrated finite element (FE) model of the soil-structure system, including soil continuum and detailed structure, was built to perform nonlinear time domain analysis (NLTD). For comparison to the commonly employed equivalent linear approach in frequency domain (ELFD), the modeling assumptions in time domain were made consistent with the ELFD approach in an equivalent linear time domain analysis (ELTD). Subsequently, the nonlinear effects were added to the ELTD model, one at a time, to create various NLTD models. Finally, different combinations of nonlinear effects were considered. The interface nonlinearity was introduced through contact surfaces. Soil nonlinearity was incorporated through a hysteretic plasticity model whose shear response is dependent on soil effective pressure. To calibrate the plasticity model, gravel’s shear stiffness degradation curve was modified to produce shear strength values consistent with the laboratory-measured friction angle. Composite layered shell finite elements with nonlinear material properties were used to model key structural shear walls.

The reasonable match between ELFD and ELTD results confirmed the general viability of time domain approaches for seismic SSI analysis in nuclear industry. The comparison of ISRS obtained from NLTD models (with single and combined nonlinearity) with those obtained from ELTD indicated a significant effect on response due to energy dissipation through interface sliding and soil nonlinearity, neither of which are captured in typical practice using ELFD approaches. In general, the relative importance of site and interface nonlinearities is a function of contact friction coefficient, soil stiffness, and excitation intensity. Details of this case study further demonstrate that the relative importance of nonlinear effects is site- and case-specific.