Where does numerical simulation go? Even during the past decades, the details of the simulation models were following increasing computer speed. The faster the computers became, the bigger the models became. It is great that we can cover more and more details and more and more physics, but in doing so, we always stay at the same solver runtime/turnaround time. This leaves me asking myself, “Don’t we need two directions?”
One direction is to cover more details and more physics for detailed analysis. The other direction should be for quick models with enough details to cover the main effects but remain fast enough to leverage optimization technology, robustness studies or on-demand result generation. How much additional impact could be made on the development process if you can judge the robustness of a design overnight by running many simulations with different parameters? To do this, the solutions and the processes need to be robust and repeatable. At Altair, robustness and repeatability of solvers and processes will continue to be a top priority to make this happen. Together with high scalability for modern high-performance computing infrastructure, we can reduce the simulation time to minutes.
