Bi-level Optimization Algorithms for Exploitation Strategy Definition

• Grant number: 25/12421-8
• Support Opportunities: Scholarships in Brazil - Doctorate
• Start date: October 01, 2025
• End date: September 30, 2029
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Guilherme Palermo Coelho
• Grantee: Rebal Zwan
• Host Institution: Faculdade de Tecnologia (FT). Universidade Estadual de Campinas (UNICAMP). Limeira , 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

This research project aims to explore recent advances in the bi-level optimization field to adapt and implement optimization algorithms for Exploitation Strategy Definition (ESD) problems. ESD is one of the key steps in oil field evelopment, as it involves the specification of important characteristics of the infrastructure required to exploit an oil and gas reservoir. To do so, ESD is often modelled as an optimization problem, where performance metrics that can be associated with production, environmental, and economic indicators should be maximized. However, solving such optimization problem is not trivial, given the generally large search space, the number of indicators that can be used as objective functions, the uncertainties involved, and the cost of evaluating each production strategy, which can be high due to the need of computationally expensive simulations (BRUM et al., 2023; DE MORAES & COELHO, 2022). There are different approaches for solving optimization problems. From the sets of variables to be optimized, we can have imultaneous approaches (also known as joint or coupled), where all the variables are optimized at the same time, and multilevel approaches (also known as sequential, hierarchical, or decoupled), where the variables are divided into groups (levels or subproblems) that are optimized separately and iteratively. Multilevel optimization is particularly interesting to several real-world engineering problems, such as ESD, where the high number of variables and the large search space makes the application of simultaneous optimization hard (MIRZAEI-PAIAMAN et al., 2022; SAID et al., 2020). Within the context of ESD, several works in the literature have already proposed optimization approaches that divide the problem into two levels (therefore, bi-level approaches). De Brito & Durlofsky (2021), for example, model the ESD as a hierarchical problem and iteratively optimize well configuration parameters in the upper-level problem (the first level of the hierarchical problem) and, for each upper-level solution, the well control rules are optimized into the lower-level problem (the second level of the hierarchical problem). Mirzaei-Paiaman et al. (2022) adopt a similar approach but optimizing the number of producers and injectors in the upper-level and the location of the wells in the lower-level problem. Although bi-level optimization is well-established in the reservoir engineering literature, several advances have been proposed in the bi-level optimization algorithms literature in the last years, particularly concerning metaheuristics (SAID et al., 2020; WANG et al., 2023), and these advances have not been incorporated into the reservoir engineering field yet. Therefore, in this research project we aim to close the gap between the two fields, exploring the recent advances in the bi-level optimization algorithms to adapt and implement new optimization algorithms for ESD. Metaheuristics are particularly suitable for ESD, as they can be easily adapted to deal with different types of decision variables, objective functions, and even with uncertainties. However, these algorithms require large numbers of candidate solution evaluations to converge, which is unfeasible in most ESD problems based on reservoir simulations. Therefore, they can hardly be directly applied to ESD without adaptations such as the incorporation of surrogate models to approximate the objective functions and partially replace the expensive computational simulations, or the development of specific operators that incorporate knowledge about the problem and reduce the search space (BRUM et al., 2023; DE MORAES & COELHO, 2022). In this work, the proposed bi-level optimization algorithms will be applied and validated within reservoir development workflows, such as the Closed-Loop Field Development and Management (CLFDM) proposed by Schiozer et al. (2019). Besides they will be compared not only with state