Production optimization using a modified net present value

摘要

This paper aims to present the initial concept of modified net present value (MNPV) and its first results. The main goal of this study is to propose a production strategy aimed at maximizing the revenue of a reservoir using the best choice of reservoir inputs. The MNPV was tested with two widely used reservoir benchmarks in the oil industry: Egg Model and SAIGUP. This new strategy modifies the objective function in order to take into account not only the financial results of well's production but also the well's data. Thus, it is possible to compare different production strategies by just using medium-term reservoir simulations. This fact drastically reduces computational costs and increases optimization efficiency, anticipating production without significant degradation of the optimal NPV in comparison with similar approaches in the literature. The findings on this study indicate an equivalent undiscounted NPV to those found in literature, a reduction of 60% of the reservoir life-cycle time, and at least 95% of the reduction in computational costs. This number was obtained considering that each interaction of ss-StoSAG (singly-smoothed stochastic simplex approximate gradient) represents a computation cost of at least a dozen of complete reservoir simulations. Twenty iterations were needed to reach the final solution. MNPV only uses a small group of simulations with computation cost equivalent to one interaction ss-StoSAG. Thus, the 95% reduction does not take into account the sweep time of the reservoir for each optimization strategy. In other words, the obtained results show an average profit increase of 250% at the beginning of the reservoir production without sacrificing the total reservoir profit. Although the production strategy is still being developed for injector wells, its initial results for production wells are promising.