Optimization of Infill Wells in Heterogenous Reservoirs Using a Genetic Algorithm

Olayiwola, Teslim Olakunle (2018-02-02)

Main theses

Thesis

Optimal positioning of wells has always been the priority of most reservoir engineers in the face of the dwindling global price of crude oil. As it is always the objective to maximize recoverable reserves over the years, much research has been carried out in order to determine the techniques appropriate for estimating the optimal number and location of wells needed to improve the recovery from a given field. In this research, well placement optimization in a highly heterogeneous reservoir involving an executable space-filling design and genetic algorithm workflow was developed for improved investment return. The desired objective function was derived using a surrogate-based modelling approach. The objective of this study is to determine and compare the performance of different surrogate modes. The specific objective is the application of the appropriate surrogates in determining the optimal location and completion properties of horizontal wells using space-filling design and genetic algorithms in a complex multidisciplinary optimization problem. This approach was implemented using MATLAB® and Schlumberger Eclipse® 100. More specifically, surrogates, such as polynomial-based (quadratic, polynomial and multiplicative), geometric-based (kriging and radial basis function) and the integer based optimizations were modelled as a function of completion properties of a horizontal well using the developed surrogates. After the numerical simulations, the most economical values of the NPV were estimated. It was observed that the NPV increases as the number of infill wells increases but attains a constant value at optimal value. In addition, the geometric-based models are an effective tool useful in developing surrogate-based rather than polynomial-based models. The results also demonstrate that this method can significantly accelerate the speed of well placement optimization and can help achieve a significant increase in investment return of an actual field if implemented as demonstrated in this case study.

Collections: