Reservoir Characterization of Non Gaussian Field Using Combined Ensemble Based Method

The proper reservoir characterization was a decade-long challenge for both geoscientists and reservoir engineers because of the complex subsurface structures. Ensemble Kalman filter was studied rigorously as a promising method for quantify the reservoir uncertainty quantification. Because of its computational efficiency and easy implementation on any simulator, petroleum engineers recommended ensemble Kalman filter (EnKF) as a useful history matching technique. It failed because of the Gaussian assumption and the inconsistency that exist between the updated static and dynamics parameters, which violate the field’s material balance. Wang et al., (2012) introduced an improved method called half iterative ensemble Kalman filter (HIEnKF) to overcome the shortcomings of the EnKF. The new method (HIEnKF) appears to be promising in the field of history matching but has its drawback. Mary Wheeler et al., (2013) introduced the ensemble smoother method to overcome the computational cost induced by HIEnKF. Various challenges arise from the existing methods that motivated us to propose new techniques that can give a promising result in petroleum engineering. We designed our first method by considering the advantage of half iterative EnKF and ensemble smoother. The combined half iterative EnKF and ensemble smoother (CoHIEnKFS) characterize a Gaussian field better than the existing methods. Previous work has shown poor characterization on reservoirs with non-Gaussian permeability distribution, which fails to satisfy HIEnKF Gaussian assumption. We apply a normal score transformation on (CoHIEnKFS) to meet this assumption. The new method (normal score combined half iterative EnKF and ES) produces a good result in the petroleum history matching field.