Simulation Run Time Reduction using Model Order Reduction Techniques

Optimization and real-time control of production scenarios in oil/gas reservoirs is time-consuming and requires powerful computers with highamount of storage capacity. This is becausethese systems are of large scale. The timeconsuming problem of running such models is handled in two ways: Parallel Computing andModel Simplification. Model simplification is implemented in two ways of Black box modeling and white box modeling [1].Recently Model Order Reduction (MOR) techniques are proposed which benefit from strong points of so-called white-box and blackbox traditional modeling approaches. Sincegoverning equations of multi-phase fluid flow in the reservoir are mainly nonlinear; therefore nonlinear model order reduction methods likeProper Orthogonal Decomposition (POD) are employed to reduce the system order and accelerate the simulation process.As it is reported in the literature MOR by POD has resulted in speedup factor of 3-4 when it is applied to system of equation solved by Newtonmethod. By applying Newton method greater speedup factors could not be achieved since full order Jacobian matrix should be computed in every iteration in order to obtain reduced Jacobian matrix (Jr) [2]. In this work POD is applied to a system of equations which is solved by IMPES method and encouraging speedup factor of 6 or more is obtained