Otimização ágil do posicionamento de poços em reservatórios de óleo e gás através de um novo metamodelo e metaheurísticas de busca local

Abstract

To address the challenge of well placement optimization problem (WPO), this thesis introduces some new well placement optimization methods, for the estimation of the best location for vertical and directional production wells, including the best trajectory of each well and the number of them. The goal is to maximize the net present value (NPV) of production from an oil and gas reservoir with a low cost computational method. The methods try to run the simulator as few as possible by means of metaheuristics and search space reduction techniques. Some totally new populationbased algorithms are the core of each of the proposed methods. Furthermore, the approach also uses a surrogate model, it was specially designed to guide the optimization cycle with less effort. The IMEX-CMG black-oil reservoir flow simulator yields the oil and gas production values and the PUNQ-S3 benchmark model was chosen to be the reservoir model of all experiments. Experiments demonstrate the approach reach superior results over the produced benchmarks and over some solutions selected from the literature. The proposed technique delivers outcomes in a faster way (more than twice faster), better NPV values and better robustness (much lower standard deviation). It’s able to be adopted as a complete low cost WPO solution or even as an early stage to a long-running method (providing good seeds).Também disponível on-line.