DEVELOPING A REDUCED-ORDER MODEL FOR FLOOD SIMULATION USING PROPER ORTHOGONAL DECOMPOSITION AND DISCRETE EMPIRICAL INTERPOLATION METHOD

Abstract

 A reduced-order model for flood simulation was developed by using the proper orthogonal decomposition (POD) and the discrete empirical interpolation method (DEIM). The primitive reduced-order equations were first derived by substiting linear combinations of the POD basis vectors for the state variables into the original model equations. The DEIM was then employed to approximate the nonlinear terms of the primitive equations efficiently. The reduced-order model was applied to an imaginary computation domain with several conditions. The numerical experiments revealed that (1) the reduced-order model was capable of reproducing the results of the original model, while the simulation errors highly depended on both simulation conditions and the degree of model reduction, and (2) the computational cost of the reduced-order model was less than half that of the original model for the given conditions.