Sammendrag
High fidelity simulations of flow might be quite demanding, because they involve up to O(106 – 109) degrees of freedom and several hours (or even days) of computational time, also on powerful hardware parallel architectures. Thus, high-fidelity techniques can become prohibitive when we expect them to deal quickly and efficiently with the repetitive solution of partial differential equations. One set of partial differential equation that we encounter on a regular basis is the Navier Stokes Equation which is used to simulate flow around complex geometries like sub-sea structures. To address the issues associated with computational efficiency, a field of Reduced Order Modelling is evolving fast. In this paper we investigate Proper Orthogonal Decomposition as a potential method for constructing reduced bases for Reduced Order Models. In the case of flows around cylindrical bodies we found that only a few modes were sufficient to represent the dominant flow structures and energies associated with them making POD to be an attractive candidate for bases construction.
Vis fullstendig beskrivelse
