Skip to Main Content
The size of the output originating from high performance scientific simulations is a major bottleneck in parallel computing. Submodelling allows for the local refinement of the results without the necessity to load and store the larger, global simulation data. This enables a user to focus in several regions of interest and disregard the rest of the analysis domain, and by doing so to heavily reduce the storage requirments. However, submodelling is only possible whenever the local region of interest is loosely coupled with the rest of the simulation. In this paper, the author introduces a method for submodelling that gives good results for non-stationary, turbulent flows, where the refinement window captures real-time, swirling vortexes, that travel throughout the domain.