EVALUATION OF SITE RESPONSE, WITH APPLICATION TO NUCLEAR STRUCTURES

Abstract

Nuclear power plants (NPPs) and other safety-related nuclear facilities offer unique design challenges not encountered with commercial infrastructure. One such challenge is the characterization of the effects of soil-structure interaction on the response of such facilities, which are generally stiff, strong and heavy. Period elongation due to soil flexibility can lead to an increase in seismic demands in these stiff structures.Unlike the commercial building industry, soil-structure-interaction (SSI) analysis is required for the design and qualification of all nuclear power plants and most safety-related nuclear facilities. Legacy analysis methods use equivalent linear procedures but there is a push to adopt non-linear (time-domain) procedures for new build plants. Regardless of the SSI analysis method used, a necessary precursor is site-response analysis. Site-response analysis is used to benchmark results of and to generate input motions to SSI analysis. The numerical models used for analysis of nuclear structures are required to be verified and validated. However, models of soil columns and domains analyzed to date for nuclear-facility applications have not undergone formal verification and validation, and that is the focus of this dissertation. Herein, a controlled dataset of the seismic response of 1D soil columns in the UB geotechnical laminar box is developed to aid in the validation of numerical models for site-response analysis: a first-of-a-kind study. The soil is dry Ottawa F-55 sand. The curated dataset comprises results from 145 tests performed in 2016. The curated data are archived at the NHERI Design Safe-CI website. This dissertation describes the goals of the experimental program, instrumentation used, and key results, all with a focus on 1D site-response analysis. The challenges encountered in building a robust dataset are identified, including conflicting results from different sensors. The dynamic properties of the Ottawa sand, as existing in the laminar box, are identified. The variability in the dynamic properties of the sand is characterized, and its impact on reliable site-response calculations is identified. The results of a benchmarking study on 1D site-response analysis using the open-source code MASTODON is described. Recommendations for the execution of future experiments in the laminar box, including improved instrumentation and its deployment, are presented, based on the lessons learned in the evaluation of results from the 2016 tests.