Mesh adaptation for unsteady problems via reduced order modeling

Mesh adaptation for unsteady problems is a task that may be thwarted by computational cost. In fact, unlike from steady problems, the fixed point at which grid independence is reached after multiple computational cycles of flow solution and mesh adaptation is by definition time-varying. An approach based on Reduced Order Modelling is presented to show that it is possible to reduce the computational cost of achieving grid independence at each time step of the simulation. Based on the ability to obtain solutions at a very low cost, the methodology of surrogate modeling is employed in conjunction with full CFD computations to drive the mesh adaptation process. Numerical tests are presented to show the effectiveness of this approach.