Production Optimization Based on Integrated Modeling and Gradient Methods

Abstract

Hydrocarbon production from a large-scale field requires optimal strategies in order to sustain returns. However, the determination of optimal plans for production is a challenging task due to high computational cost and various nonlinear constraints. Therefore, an efficient methodology is required to determine the optimum criteria with minimum cost and maximum precision. In this paper, two quasi-newton second-order algorithms of the Broyden–Fletcher–Goldfarb–Shanno (BFGS) and symmetric rank one (SR1) are used to maximize the net-present-value (NPV) over 17 years of production and injection. The framework is applied to an integrated simulation model of reservoir, wells, and surface facilities of a real oil field located in the Middle East. The Lagrangian method is used to combine the objective function and constraints during optimization. The results obtained are compared with those of the current production scheduled plan provided by the oil industry. The performance of algorithms is affected by the production optimization time scale. For the first 12 years of production optimization, SR1 outperform BFGS. For the last 5 years, BFGS outperforms SR1. However, for all years of production, BFGS is superior over SR1 and increases NPV over the base case NPV from 1.605% to 4.92% for various time steps of production.