History Matching with Ensemble Kalman Filter Using Fast Marching Method in Shale Gas Reservoir

Abstract

Abstract In case of shale reservoirs, there are lots of uncertainty parameters: matrix permeability, fracture permeability by hydraulic fracturing, enhanced permeability surrounding fractures, fracture half-length, vertical enhanced ratio, horizontal enhanced ratio, and porosity. Reasonable uncertainty quantification enables us to predict shale reservoir performance. Ensemble Kalman filter (EnKF) is one of the most reliable history matching method. However, EnKF has the limitation of simulation time because each prediction step requires forward simulation as the number of ensemble members. Larger size of a field is simulated, much simulation time is consumed. Fast marching method (FMM) is able to reduce simulation time and shows reliable result of shale gas fields compared with other simulators. Therefore, we ropose the concept of unification of EnKF and FMM for reliable and fast. To validate the advantage of the proposed method is compared with EnKF using a conventional forward simulaotor. Also, the proposed method is coupled with distance-based generalized sensitivity analysis (DGSA) to select sensitive model parameters. FMM-EnKF using selected parameters is useful when it is difficult to obtain reservoir uncertainty information. The three methods, the standard EnKF, FMM-EnKF using all parameters, and FMM-EnKF using selected parameters, are compared by history matching time, distribution of uncertainties, and cumulative responses. The standard method and the proposed method shows decline in width of uncertainty after history matching. However, the proposed method completes history matching faster than the standard EnKF. Most of all, the proposed method using selected parameters decreases the width of cumulative responses compared with the proposed method using all parameters. Therefore, this method is going to contribute the section of history matching and prediction of future production.