Optimal producer well placement and multiperiod production scheduling using surrogate modeling

Abstract

The main objective of this paper is to determine the optimal placement of oilfield producer wells while accounting for surface infrastructure constraints. To accomplish this objective, we replace a first-principles optimization model with a hybrid approach, which draws the correlations for oil production from data obtained by rigorous simulations of grid discretized reservoir models. We use these correlations to build a Multiperiod Mixed-Integer Linear Programming (MILP) problem with the production constraints and maximization of the Net Present Value (NPV) as the objective function. A computationally efficient multiperiod MILP model is proposed to evaluate the performance of this methodology with a surrogate correlation for the oil production rate built in ALAMO. This set of equations takes into account the distance of the producer well from the injection well, the depth of the producer well, and the time of operation. This model allows higher levels of grid discretization and longer time horizons of interest. In order to address systems where high spatial resolution is necessary and computational expense is high, we introduce a methodology of Spatial Aggregation and Disaggregation. In such method, we first simplify the problem through aggregation to then decompose the obtained solution, producing highly resolved well placement strategies. With this approach, it is shown that the optimal well placement locations can be obtained with order of magnitude reductions in computational expense.