A rigorous approach to uncertain production optimization using a hybrid algorithm: Combination of a meta-heuristic and a Quasi-Newton method

Abstract

Integrated production optimization from a typical large scale hydrocarbon field is a very complicated task because of geological and economic uncertainty, existence of several linear and nonlinear constraints, and time consuming nature of reservoir simulation. Therefore, it is crucial to develop an efficient framework to perform full field optimization under uncertainty to maximize the returns.This paper, first tries to bridge the gap between reservoir and wells with the surface facilities such as pipelines, separators, injection manifolds. This integrated modeling maximizes the reliability of simulation output. Then, the Broyden–Fletcher–Goldfarb–Shanno (BFGS) coupled with meta-heuristic bat algorithm (BA) is applied for production optimization under geological and economic uncertainty. Cholesky decomposition is used to construct reservoir realizations. The bi-objective function is expected net-present-value (NPV) and its standard deviation.Results reveal that the expected NPV increases over the base case NPV between 2.35% and 9.27% for various time steps. Also, the uncertainty assessment shows that the most likely cumulative oil production (P50%) could be improved over the current value at least by 2.23% under geological and economic uncertainty. Therefore, the optimization framework based on integrated modeling, and combining gradient-based with meta-heuristic algorithms could efficiently solve the full-field worldwide problem.