A GPU-accelerated compressible RANS solver for Fluid-Structure Interaction simulations in turbomachinery

The typical approach in Computational Fluid Dynamics (CFD) relies on the negligible structural deformation due to aerodynamic loads. However when geometry optimization for performance and power efficiency is the ultimate goal, an accurate Fluid-Structure Interaction (FSI) analysis could improve results accuracy. This is particularly important for turbomachinery applications, where usually only centrifugal and thermal effects are considered to deform the original geometry, through a FEM solver, before computing the pure aerodynamic solution. Even if turbomachinery blades are less susceptible to be deformed by aerodynamic loads than other aeronautical components (e.g. helicopter blades, entire planes) trim solutions could lead to better results. Alongside with results accuracy, computational efficiency is mandatory and today GPUs can be exploited to accelerate simulations. In this article we present and describe a GPU-accelerated FSI solver for turbomachinery applications. The solver is validated with a typical literature case