Quantification of Uncertainty of Reserves with High Quality History Matching Models in a Mature Field

Abstract Development decisions of oil and gas fields are based with many uncertainties within the supporting data. Uncertainties are larger in the beginning which gets reduced with time as more data is acquired and understanding of the field is improved. However, significant uncertainties persist even in mature fields which are case for redevelopment. Therefore, enormous inherent uncertainties need to be characterized, analyzed and incorporated in the simulation study to the extent possible in order to improve the decision making process. This paper discusses the relevance of reviewing and incorporating the field uncertainties related to geological and/or dynamic properties in estimation of incremental volumes in the redevelopment of a mature oilfield. Reservoir models need to be updated through history matching process to reflect actual measurements from reservoir. In this study a workflow to quantify the uncertainty of a mature oilfield of offshore Malaysia and to estimate the incremental oil potential from infill wells has been designed. During history matching process, dynamic and static parameters of uncertainty, relative permeability curve, local permeability around wells, the aquifers? volume and productivity, and contacts between gas-oil, and oil-water have been considered. An objective function is applied for the history matching which consists of the error between simulated and observed data from wells. An "Evolutionary strategy optimization", proprietary to MEPO software is used to obtain multiple equivalent models with high quality. While running an optimization for history matching, the ranges of uncertainty parameters have been narrowed and adjusted the ranges to be used for quantification of uncertainty of Estimated Ultimate Recovery (EUR). In the prediction stage, both filtered models from actual runs and sampled models from Latin Hypercube have been used. This approach not only allows to evaluate intrinsic uncertainty, but also to keep the variety and risk of prediction. Monte-Carlo simulation technique then makes it possible to acquire the distribution of EUR.