A New Generation of Metaheuristics for Production Strategy Optimization

• Grant number: 20/14710-3
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• Start date: January 01, 2021
• End date: December 31, 2023
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Agreement: Equinor (former Statoil)
• Principal Investigator: Denis José Schiozer
• Grantee: Artur Ferreira Brum
• Host Institution: Centro de Estudos de Energia e Petróleo (CEPETRO). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Company: Universidade Estadual de Campinas (UNICAMP). Faculdade de Engenharia Mecânica (FEM)
• Associated research grant: 17/15736-3 - Engineering Research Centre in Reservoir and Production Management, AP.PCPE

Abstract

The main goal of this research project is the development of new metaheuristics capable of efficiently dealing with the particularities of Production Strategy Optimization Problems (PSOP). Although metaheuristics are well-known and widely adopted general-purpose optimization techniques, they often require thousands of objective function evaluations to obtain high-quality solutions for a given problem. Such an aspect may not be a significant drawback for problems with objective functions that require low computational times to be evaluated, but this is not the case of PSOP. To evaluate a single candidate solution proposed by a metaheuristic for PSOP, oil field simulators are traditionally involved, which may require several hours, even days. In this context, the use of standard metaheuristics is often unfeasible. Therefore, this project aims to develop new metaheuristics tailored to deal with PSOP. To do so, four approaches will be considered: (i) the development of new mechanisms that incorporate knowledge from petroleum engineering to better guide the search, so that convergence can be accelerated; (ii) the evaluation of Estimation of Density Algorithms (EDA) to PSOP, which tend to require fewer function evaluations to obtain high-quality solutions; (iii) incorporate Machine Learning techniques to estimate proxies that can partially replace the computational simulations; and (iv) explore diversity maintenance mechanisms to simultaneously obtain multiple optima during a single optimization run. (AU)