This paper addresses issues surrounding soil property variability including uncertainties associated with “best estimate” values and searches for practical ways to assess the impact on the seismic response of a facility, such as a nuclear power plant, resting on it. Specifically, it attempts, using a parametric study, to formulate a probabilistic model that enables the enveloping of uncertainties associated with the soil-structure-interaction component of the seismic problem. The effects of most-likely sources of uncertainty, such as variability of “distinct” soil layer profile and variability of controlling soil properties, are to be addressed by generating a probabilistic profile in which randomization of key parameters that appear to have the most impact on the results of deterministic analyses is implemented. The use of stochastic finite elements and the introduction of correlation functions, in conjunction with finite element discretization of the foundation soil, are explored as means of achieving an enveloped structural response. The on-going evaluation of the Armenian nuclear plant site prompted this study. In order to stress the importance and relevance of the stated goal, the soil-structure-interaction of the nuclear power plant, subject to significant variation of the foundation soil, is examined. The conflicting results of two independent studies of the subsurface provide the basis for the variation range used in this study.
Soil–structure interaction analyses to locate nuclear power plant free-field seismic instrumentation
