Integrated production system optimization using global optimization techniques

Abstract

Many optimization problems related to integrated oil and gas production systems are nonconvex and multimodal. Additionally, apart from the innate nonsmoothness of many optimization problems, nonsmooth functions such as minimum and maximum functions may be used to model flow/pressure controllers and cascade mass in the gas gathering and blending networks. In this paper we study the application of different versions of the derivative free Discrete Gradient Method (DGM) as well as the Lipschitz Global Optimizer (LGO) suite to production optimization in integrated oil and gas production systems and their comparison with various local and global solvers used with the General Algebraic Modeling System (GAMS). Four nonconvex and nonsmooth test cases were constructed from a small but realistic integrated gas production system optimization problem. The derivation of the system of equations for the various test cases is also presented. Results demonstrate that DGM is especially effective for solving nonsmooth optimization problems and its two versions are capable global optimization algorithms. We also demonstrate that LGO solves successfully the presented test (as well as other related real-world) problems.

Global Optimization of Gas Lifting Operations: A Comparative Study of Piecewise Linear Formulations

“…To improve the results of the Buitrago et al stochastic algorithm, Mason et al introduced a “discrete gradient method” (DGM) that adapts standard descent methods for the case of piecewise linear functions. The DGM has been used to solve a variety of optimization problems . Mason et al employ four realizations of DGM to solve the problems posed by Kanu et al and Buitrago et al and their own 58-well test case.…”