Well Placement Optimization Using Simulated Annealing and Genetic Algorithm

The general success ratio of wells drilled lies at 1:4, which highlights the difficulty in properly ascertaining sweetspots. well placement location selection is one of the most important processes to ensure optimal recovery of hydrocarbons. Conventionally, a subjective decision is based on the visualization of the HUPHISO (a product of net-to-gross, porosity and oil saturation) map. While this approach identifies regions of high HUPHISO regarded as sweetspots in the reservoir; it lacks consideration for neighbouring regions of the sweetspot. This sometimes lead to placement of wells in a sweetspot but near an adjoining aquifer; giving rise to early water breakthrough - low hydrocarbon recovery. Recently, heuristic optimization techniques. Genetic algorithm (GA) and simulated annealing (SA) has received attention as methods of selection of well-placement locations. This project developed and implemented GA and SA well-placement algorithms and compared the reservoir performance outputs to that of conventional method. Firstly, a reservoir performance model was built using a reservoir flow simulator. In the base case, the wells were placed based on a subjective selection of gridblocks upon the visualization of the HUPHISO map. Thereafter, JAVA routines of GA and SA well-placement algorithms were developed. The numeric data (ASCII format) underlying the map were then exported to the routines. Finally, the performance model was updated with new well locations as selected based on the GA and SA-based approach and the results were compared to the base case. The Comparison of the results showed that both GA and SA-based approach resulted to an increased recovery and time before water breakthrough.