IMPROVED RESERVOIR CHARACTERIZATION USING AUTOMATIC HISTORY MATCHING PROCEDURES

The primary task in reservoir history matching is the estimation of re servoir parameters such as porosities and permeabilities so that the m odel predictions match as closely as possible the field measurements. In actual practice, prior estimates of the most probable values of the parameters are usually available. Bayesian estimation was used to inc orporate this prior information into a 3-D 3-Phase reservoir simulator with automatic history matching capabilities. In addition, minimum an d maximum constraints were imposed on the parameter values. A case stu dy is presented where the postulated reservoir model is quite differen t from the actual reservoir. It is shown how parameter estimates obtai ned by automatic history matching may be used to help modify the postu lated model, These included changes to reservoir boundaries and detect ion of impermeable barriers within the reservoir. Estimates of predict ed in-place hydrocarbon volumes were also compared with actual values. The final reservoir description is shown to be a significant improvem ent over the initial postulated model.