Author | Jared Crean, Kinshuk Panda, Jason E. Hicken |
---|---|
Title | Adjoint-based Local Reanalysis of Nonlinear PDEs |
Year | 2018 |
Journal | Proceedings of AIAA Multidisciplinary Analysis and Optimimzation Conference |
Abstract | During the design process, an engineer may need to predict the performance of many alternative geometries. A reanalysis method is attractive in this context, because it uses information at a baseline geometry to accelerate predictions of other geometries. In this work, we use an adjoint-based indicator to identify regions of the domain where perturbations to the solution will have the largest effect on an output of interest. The governing equations can then be re-solved on only the most influential regions, reducing computational cost. We refer to this method as adjoint-based local reanalysis. Computational results compare the output computed by reanalysis against re-solving the governing equation on the entire domain. |
PDF File | Download |