Eventos Anais de eventos
COBEM 2021
26th International Congress of Mechanical Engineering
Optimum well location and rates using genetic algorithms considering surrogates
Submission Author:
Eduarda de França Andrade , PE
Co-Authors:
Eduarda de França Andrade, Leonardo Oliveira, Silvana Maria Bastos Afonso da Silva, Jefferson Wellano Oliveira Pinto
Presenter: Eduarda de França Andrade
doi://10.26678/ABCM.COBEM2021.COB2021-1864
Abstract
During the exploitation of a reservoir, some of the steps that most impact the economic return of the field are the definitions of the location of the well and their flow-related controls. The definition of an optimal design for the development of the field is a complex problem, due to its non-linear nature, involving several decision variables, constraints and multiple scenarios. In face of such complexity, the approach based only on the operators' technical knowledge can ignore profitable production scenarios. This limitation motivated the development of this work, which aimed to develop an automatic tool to optimize the location and controls of the wells sequentially, to maximize the Net Present Value (NPV) of the field, taking into account the costs production and drilling. In the sequential approach, the location of the wells is defined using the genetic algorithm (GA), considering an updated adaptive substitute model. Then, the production and injection flows are optimized through a local search through the approximate sequential optimization (SAO) strategy. Due to the stochastic nature of GA, the optimization of the proposed problems is repeated twenty times. There is an improvement in NPV, compared to the base case, mainly due to the reduction in water production and the increase in the recovery factor. The optimized scenario presented different positions for all wells, from those initially proposed by the operator, with the production wells located, preferably, in the areas of lower permeability and the injectors in the regions where this property is high. It is observed that the local adaptation of the surrogate models enables a better approximation and a lower computational cost. The model used showed good calibration, given the parameters of the GA used, since the repeatability of the rounds presented low standard deviation, emphasizing the robustness of the GA in the decision-making aid for the location problem and optimal management of wells in reservoir. Concerning the previous studies, in this work, a approach and a recent methodology of adaptation of the substitute model are developed and applied to the problem of location and management of reservoirs. The developed tool is applied to a practical case reported in the literature.
Keywords
Reservoir Engineering, Optimization, Combined Strategies, Surrogate models, Well Placement and Flow Rate
