Gas Lift Optimization Using Proxy Functions in Reservoir Simulation

Abstract

Summary Artificial lift by means of gas injection into production wells or risers is frequently used to increase hydrocarbon production, especially when reservoir pressure declines. We propose an efficient optimization scheme that finds the optimal distribution of the available gas lift gas to maximize an objective function subject to surface-pipeline-network rate and pressure constraints. This procedure is a nonlinearly constrained optimization problem solved by the generalized reduced-gradient (GRG) method. The values of objective function, constraint functions, and derivatives needed for optimization can be evaluated through two methods. The first method repeatedly solves the full-network equations using Newton iteration, which takes into account the flow interactions among wells; however, this method can be computationally expensive. The second and more efficient method is a new approach proposed in this paper. It constructs a set of proxy functions that approximates the objective function and constraints as functions of gas lift rates. The proxy functions are obtained by solving part of the network that consists of a gas lifted well or riser, assuming a stable pressure at the terminal node where the partial network is decoupled from the rest of the network, and are used to inexpensively evaluate the objective function, constraints, and necessary derivatives for the optimizer. A procedure to predict the proxy functions on the basis of previous values can be used to reduce the number of partial-network solves, and the partial-network solution has been parallelized for faster simulation. These two methods can be applied at different timesteps during the course of the simulation. The proposed methods are implemented within a general-purpose black-oil and compositional reservoir simulator and have been applied to real-field cases.