oa Generation of Low-Order Reservoir Models Using POD, Emperical Grammians and Subspace Identification

In this paper we present a number of data-driven approaches to obtain non-linear low-order models of heterogeneous reservoirs. Relying on the `proper orthogonal decomposition' (POD) method of snapshots, the proposed approaches scale-up an existing high-order dynamic reservoir model to a model of lower order.