Banca de DEFESA: EDUARDA DE FRANÇA ANDRADE

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : EDUARDA DE FRANÇA ANDRADE
DATE: 26/08/2022
TIME: 14:00
LOCAL: Programa de Pós-Graduação em Engenharia Civil
TITLE:

Optimization of well Placement and Flow Rates in oil Reservoirs Using Genetic Algorithms and Adaptive Surrogate Models

KEY WORDS:

Reservoir simulation; Well placement; Production optimization; Surrogate models; Genetic algorithm; Sequential approximate optimization

PAGES: 138
BIG AREA: Engenharias
AREA: Engenharia Civil
SUMMARY:

The development of new fields for oil production is increasingly complex and expensive. Moreover, having profitable production in mature fields, where water production rates are generally high, also represents a challenge for the Reservoir Engineering area. In both situations, it cannot be easy to obtain investment returns using only traditional production management techniques. Therefore, there is a growing interest in the Oil and Gas Industry in developing optimization procedures based on numerical simulations that efficiently guide well operation planning. Thus, the present work focuses on maximizing the reservoir's net present value (NPV) through optimizing field management operations such as well placement and flow management with control cycles. The optimization problem of a known reservoir in the literature was solved through three different approaches using the genetic algorithm (AG) of the MATLAB Toolbox Optimization and the technique of sequential approximation optimization (SAO). In two sequential steps, the first approach optimized the location of wells (AG) and their flow rates (SAO). The second approach used AG to simultaneously define the best position for the wells and the best flow rates for each well in the defined control cycles. In the third, the second methodology (integrated AG) was combined with an additional step that optimized the flows through the SAO. During the optimization process, a series of function evaluations were performed using a reservoir simulator. Due to the high cost of this process and aiming to reduce it, a methodology with adaptive substitute models was used. As the optimization problem studied is constrained, techniques such as chromosome repair and an adaptive penalty method were used that allowed the GA to work respecting the constraints imposed on the problem. For all studies, 20 optimization runs were performed. Among all cases and approaches studied, the solution associated with the best NPV had a 50.44% increase concerning the base case. The significant reduction in water production was the most decisive parameter for this increase, reaching about 87.70% in the best result obtained in this work. The methodologies suggested here brought results consistent with significant improvements in NPV, the main objective of this work.

BANKING MEMBERS:
Presidente - 1130915 - BERNARDO HOROWITZ
Externo à Instituição - DENIS JOSE SCHIOZER - UNICAMP
Externo à Instituição - JEFFERSON WELLANO OLIVEIRA PINTO