Evaluation of a Reservoir Simulation Study: West Civit Dykeman Sand Unit

Abstract

Reservoir simulation studies were performed on this Oklahoma waterflood in 1973 and 1977. The first study's recommendations proved to be economically successful, but reservoir performance was poorer than predicted. Four additional years of performance history led to a better match of the unit's water production and a more conservative estimate of upstructure reserves. Introduction Reservoir simulation has become an accepted and almost routine procedure for optimizing many types of recovery processes. Simulation results have appeared in the literature for a wide variety of recovery mechanisms ranging from simple primary drives to very exotic enhanced oil-recovery techniques. Many of these studies have recommended operating programs costing millions of dollars. The large programs costing millions of dollars. The large incremental oil and gas recoveries predicted are justification for these expenditures.The literature, unfortunately, lacks extensive after-the-fact evaluations of results of reservoir simulation studies. Most important is how the reservoir actually performed compared with the predicted behavior. performed compared with the predicted behavior. The significance of follow-up studies, generally, is not that we have great uncertainty in the underlying physics or numerical techniques but rather because physics or numerical techniques but rather because we know there are severe limitations on the data describing the reservoir. It is well-known, for example, that almost all reservoir simulation studies proceed in two phases:the history match andproceed in two phases:the history match andthe performance projections. The history-match phase is an attempt to overcome the limitations of the phase is an attempt to overcome the limitations of the data, which could include unknown information such as reservoir limits or unreliable data such as estimated gas flared. During this phase, the reservoir parameters in the model are "adjusted" until the parameters in the model are "adjusted" until the model's performance matches the observed reservoir behavior. More often than not, the history-matching phase of a study consumes more man and computer phase of a study consumes more man and computer time than do the performance projections. Perhaps more significant is the fact that a good history match does not guarantee accurate performance projections. The problems here are the accuracy of the projections. The problems here are the accuracy of the history and the fact that multiple sets of reservoir parameters may yield equally accurate history parameters may yield equally accurate history matches but may provide substantially divergent performance projections. Consequently, a complete performance projections. Consequently, a complete follow-up would include not only a comparison between the projected incremental recovery and that actually realized but also would evaluate quantitatively the history match.The best means of accomplishing this latter evaluation would be to rematch the old history with any new well information. This second history match then would be compared with the first match to determine if both were achieved with approximately the same reservoir parameter values. The West Civit Simulation Studies The West Civit Dykeman Sand Unit is the portion of the West Civit field producing from the Dykeman sand at an average depth of 3,859 A (1176 m). It covers approximately 996 acres (4×10 m) in Garvin County, OK (Fig. 1). The reservoir was discovered in Oct. 1967. JPT P. 1599