AN EFFICIENT ORDER REDUCTION STRATEGY IN EARTHQUAKE NONLINEAR RESPONSE ANALYSIS OF STRUCTURES

Since earthquake dynamic response analysis of large and complex structures are computationally time demanding, efficient methods that can reduce the system order are of high interest. In this sense, there are different methods available, which try to provide a proper equivalent model. However, in the presence of nonlinearities in the structural elements, most of those methods are ruled out due to their linear assumptions. Therefore, this contribution aims at providing an efficient strategy, which can reduce the order of the nonlinear structural model while retaining important structural characteristics for further earthquake dynamic response analysis. The model order reduction (MOR) strategy is developed based on the proper orthogonal decomposition (POD) method to derive a set of nonlinear deterministic POD modes according to the information of the response history (snapshots) of the full order structure under one or a set of representative earthquake excitations. Subsequently, the POD modes are utilized to create the reduced-order models of the structure subjected to different earthquake excitations. Then, the reduced order models need substantially less amount of computational time in comparison to the full order models. This study presents the application results of the introduced new strategy to a realistic building structure, which is base-isolated by means of frictional bearing elements for better seismic performance. The results demonstrate accurate approximations of the physical (full) responses by means of this new MOR strategy if the probable behavior of the structure has already been captured in the POD snapshots.