Abstract

Engineers and analysts typically design products using engineering simulation software developed for solving one type of physics problems, although most of the product applications involve multiple physics. There is a growing demand to improve product development processes by creating more realistic Multiphysics simulations. One approach is to use all-encompassing Multiphysics simulation software which can simulate Multiphysics scenarios, but this approach becomes challenging when solving problems with large domain and increasing complexity. In this presentation, we will demonstrate how we use a coupling engine to connect different physics solvers and map the interfaces which interacts between different solvers and transfer the data between them. The challenge with mapping interfaces is that each physics solver requires a unique mesh which is optimized for individual solver and the data transferred is both intensive (Ex. Displacement) and extensive (Ex. Force). To map both types of data between the source and target meshes on the interfaces, profile preserving, and conservation-based algorithms are developed. The profile preserving algorithm generates mapping weights for the target mesh using linear shape and radial basis functions. If there are non-overlapping interface regions, they are mapped by using nearest nodes or by extrapolation. The mapping weights for the conservation-based algorithm are generated by area/volume fraction-based scatter followed by target-side gather. Another challenge with this solver agnostic approach is stabilizing the simulation and reaching convergence. To address stabilization issues with tightly coupled systems, algorithms have been developed for ramping and relaxation of data transfers between the interfaces. Additionally, a python module extension is developed to enable parsing of results from the solvers and dynamically adjust the time step size for transient analyses. For example, adjusting the time step size based on flow courant number and mesh courant numbers have helped to stabilize the simulations. Although coupling multiple independent physics solvers brings different challenges, a highly customizable approach enables the solution of complex Multiphysics systems.