Multivariate Optimization of Networked Production Systems

Abstract

Summary Today, many oil and gas production systems have flowline networks with many wells and surface pipelines. In the design of such production systems, the determination of parameters like separator pressure, the diameters of tubing, pipeline, or surface choke, and the length of pipeline is important to the achievement of an optimum production rate. The optimization of such problems, however, has been difficult because of the nonlinearity of the solution caused by the interaction between these parameters. In this work, multiple production parameters were optimized simultaneously in well networks in terms of a profit-based objective function, such as total production rate, net income from the oil product, or present value discounted by interest rates. The techniques applied in this work were Newton-type methods (derivative based), the polytope method (function-value based), and a new technique that uses a genetic algorithm (GA). On the basis of several test calculations on various types of optimization problems, the polytope method turned out to be the most efficient and consistent for low dimension problems with small numbers of wells, while the GA performed well in large systems with many variables to be optimized. The optimization of a pipeline network system was achieved successfully without any limitation on the selection of objective functions and decision variables to be optimized. The optimization technique can be used in the design stage of newly developed fields or in the planning of workovers in existing fields. The combination of this technique with reservoir simulation would be a powerful tool in a project implementation design process.