Recursive Covariance of Structural Responses

Abstract

This paper describes an effective method of obtaining covariance responses in recursive form for a multi‐degree‐of‐freedom linear structural system subjected to nonstationary random excitations. The earthquake excitations are expressed as a multiply‐correlated nonstationary autoregressive process model. By utilizing white noise characteristics of the AR model, this covariance response matrix equation is derived from state space matrix equation with mixed excitations and response components. Then, a torsional vibration model, subjected to two horizontal components of earthquake excitation, is analyzed. A BWR type nuclear power plant building is used as an example to show how recursive covariances are applied to evaluate statistical values of nonstationary maximum responses.