Applying Flow-Based Reduced-Physics Framework for Selection of Initial Model Ensemble Under Subsurface Uncertainty

Summary In this paper, we present a unified model ensemble screening workflow that utilizes a reduced-physics method as the core engine. This approach allows for the computation of various flow metrics, which could then be transferred to the clustering engine to select a subset of representative models amongst the initial model ensemble. The reduced-physics method used in this work is flow diagnostics, which enables the derivation of multiple flow metrics such as dynamic Lorenz coefficients, sweep efficiencies, well allocation factors, etc. from single-phase, steady-state, incompressible-fluid pressure and time-of-flight outputs. The workflow is showcased using an initial ensemble derived from a realistic asset-scale deepwater reservoir.