Optimal Spatiotemporal Reduced Order Modeling for Nonlinear Structural Dynamics

University Of Oklahoma Norman Campus, Norman OK

Investigators

Abstract

The overall goal of this project is to formulate optimal, data driven, reduced order modeling (ROM) frameworks for significantly faster and more accurate computational predictions of nonlinear dynamical systems. The modeling framework is based upon the optimal prediction formalism and includes information in the form of relevant spatiotemporal correlations. To achieve the overall project goal, the work will a) apply the method to variational (weak) forms and b) investigate, using the ROM methodology and fully resolved simulation, two problems in nonlinear structural dynamics which are expected to exhibit complex spatial and temporal (scale) characteristics. The two specific problems to be investigated are: 1) turbulent behavior of isotropic plate vibration and 2) vibration characteristics of three-dimensional continua with anisotropic constitutive relationships. Reduced order modeling is an integral part of the engineering design process as fully resolved simulation is still too expensive to be practical for parametric and optimization studies. The advances in reduced order simulation provided by the work will have significant impact on the ability to provide fast and reliable simulation to aid in the innovation of new engineering products. Results and ideas generated from the project will be disseminated through peer-reviewed publication and conference attendance. In order to introduce students to the important concept of reduced order simulation, a new graduate level course on the subject will be developed at The University of Oklahoma. To further broaden the impact of the work, underrepresented groups (undergraduates in general, women and minority students) will be actively recruited to participate both directly in the research activities and indirectly through a new course.