Mohammad Aghabeigi - National Iranian South Oil Company, Ahwaz, Iran.
Alireza Tabatabaei Nejad - Sahand University of Technology, Tabriz, Iran

Optimization of well placement is a complex problem in reservoir engineering because of the nature and uncertainty in reservoir rocks properties, fluid properties, well specifications, production or injection strategies, and economic considerations. In addition, optimal well placement is essential in success of future infill drilling programs. Several optimization methods have been used for well placement problem. Among those, Genetic Algorithms (GAs) have shown potential capability for optimization of such a complex problem. However, GA procedures are problem-specific and need to be adapted, tuned and enhanced for well placement problem. There are several parameters that can be adjusted for enhancing the speed and efficiency of GAs. In this work, we investigated the effect of initial population, population size, crossover probability, and mutation probability. We found that tuning GA can significantly increase the speed of convergence and also reduce the number of required simulation. Also, we found that selecting initial population based on random selection will result in more efficiency.

Optimization of well placement, Genetic Algorithms, Numerical Simulation