The definition of a production strategy is one of the most important tasks in reservoir management, since it influences projects attractiveness. However, the process to define variables such as well placement, number and type of wells and operational conditions is time-consuming and it demands high computational effort. The use of an optimization algorithm to achieve a good solution can be very valuable to the process but it can also lead to an exhaustive search, demanding a great number of simulations to test many possibilities.
In order to minimize the number of combinations (number of wells and positions), it is proposed the use of the quality map, that is a two dimensional representation of regions with production potential in a reservoir, as a criteria to allocate the wells.
The optimization algorithm used in this work is the genetic algorithm (GA) which is a method based on natural evolution process. The main characteristic of GA is the ability to work in a solution space with non-smooth and non-linear topology, where the traditional methods generally fail.
The methodology proposed in this work is used to optimize production strategies in a realistic reservoir model, defining the number and position of production and injection wells, and the production/injection flow rates. The number of individuals in a population and the number of generations were also varied to evaluate the efficiency of the algorithm.
Results showing the performance of several optimization processes are presented. The influence of the GA parameters control are analyzed and compared.
Introduction
The main activity in reservoir engineering is the planning of strategies for the development and management of petroleum fields. The wells location is one of the most important aspects in production strategy definition. It determines the number of wells, the production/injection patterns, operational conditions and facilities specifications, i.e., all parameters that affect the reservoir behavior, economic analysis and, consequently, projects attractiveness.
The process of choosing the best location for wells is basically trial and error. It's a time-consuming routine and demands high computational efforts, since the productivity depends on many variables related to well characteristics, reservoir and fluid properties, which can only be understood trough numerical simulation.
The use of an optimization algorithm to find a good position for the wells can be very useful to the process but it can also lead to an exhaustive search, demanding a great number of simulations to test many possibilities, most of them disposable. Therefore, criteria to restrict the number of alternatives can be required to make an automated process viable. In this work, the Quality Map (QM) is used to delimit the possible locations for wells to minimize the number of combinations (number of wells and positions).
Background
Optimum reservoir management is an important theme in petroleum industry. Most of the studies related to reservoir performance optimization focus the well placement.
Aanonsen et al (1995) proposed a method to optimize well locations under geological uncertainties based on response surfaces and experimental design. Multiple regression and kriging were used to reduce the number of simulation runs. A methodology to optimize the number and location of producer well in new fields was developed by Pedroso and Schiozer (2000). It was applied in primary recovery stage developed with vertical wells. The work utilizes parallel computing with intention to accelerate the process. Mezzomo and Schiozer (2002) proposed an optimization procedure based on reservoir simulation that evaluates both individual wells and field performance. The methodology helps managers to make decisions that lead to an adequate recovery for the reservoirs, maximizing profits and minimizing risks associated to the investments.