Utilizing high performance supercomputing facilities for interactive thermal comfort assessment

We outline the current state of the development of a computational steering environment (CSE) for the interactive simulation and local assessment of indoor thermal comfort. The system consists of a parallel CFD kernel, a fast 3D mesh generator and a virtual reality-based visualization component. The numerical method is based on a lattice Boltzmann algorithm with extensions for simulations of turbulent convective flows. Utilizing high-performance supercomputing facilities, the CSE allows for modifying both the geometric model and the boundary conditions during runtime coupled with the immediate update of results. This is made possible by a space-tree based partitioning algorithm that facilitates the meshing of arbitrarily shaped, complex facet models in a matter of just a few seconds computing time. Ongoing developments focus on the integration of a radiation solver, a human thermoregulation model and a local thermal comfort model. Our first step was therefore to develop a prototype for computing resultant surface temperatures mapped for the surface of a numerical manikin. Results are compared with measurement data.