Construction of adaptive reduced-order reservoir models based on POD-DEIM approach

This paper introduces a method for constructing adaptive reduced-order reservoir simulation models based on the POD-DEIM approach for field development optimization and assisted history matching problems. The approach is based on adapting the orthogonal decompositions bases to the varying model configuration. The method utilizes information contained in the bases of the original model and supplements them with new components instead of constructing a new model from scratch. Adapting the bases significantly reduces the computational costs of building reduced-order models and allows the application of such models to tasks requiring multiple simulations with different configurations. The paper presents an implementation of the POD-DEIM model for a two-phase flow problem and discusses examples of adapting this model to changes in well configuration and geological properties of the reservoir. We propose a generalized approach using POD-DEIM models in combination with the bases adaptation technique to solve optimization problems, such as field development optimization, selection of the optimal well locations, geometries, and well regimes, as well as history matching.