Nonlinear risk optimization approach to water drive gas reservoir production optimization using DOE and artificial intelligence

The risk of investment in petroleum industry is usually high due to uncertainties associated with performance prediction of hydrocarbon reservoir. To reduce the investment risk, the uncertainties should be estimated accurately. Statistical methods of risk and uncertainty assessment in oil and gas industry have limitations because of underlying principles. This study proposes an improved strategy to reduce uncertainty associated with different engineering and geologic factors. The method optimizes conventional uncertainty assessment methods using evolutionary computing algorithm. In this regard, first; gas recovery factor from water drive gas reservoirs is predicted using five different methods including Box-Behnken design (BBD), Full Factorial design (FFD), Central Composite design (CCD), Uniform design (UD) and relative variation factor (RVF). Next, genetic algorithm (GA) is applied to optimally integrate all conventional uncertainty assessment methods in one equation as a new approach. Part of water drive gas reservoir based on actual data is simulated to extract necessary information in order to be able to quantify the uncertainty of ultimate recovery factor. The effect of average reservoir permeability (K-G), permeability anisotropy (K-v/K-h), aquifer size (V-aq), production rate (Q(g)), perforated thickness (H-p) and lower limit of tubing head pressure (THP) on reservoir performance has been studied. The results show that average reservoir permeability, tubing head pressure, permeability anisotropy, aquifer size, perforated thickness, and production rate have the greatest impact on recovery factor uncertainty, respectively. The results also indicate the higher performance of proposed method over the conventional uncertainty assessment methods to predict the most likely recovery factor. The prediction error of most likely recovery factor by applying GA, BBD, FFD, CCD, UD, and RVF is 0.11, 2.2, 3.15, 2.78, 7.22, and 11.4% respectively. (C) 2016 Elsevier B.V. All rights reserved.