OPTIMIZATION OF THE PRODUCTION OF AN OIL FIELD SUBMITTED TO WATER INJECTION USING THE Q-LEARNING ALGORITHM
Intelligent System. Reinforcement Learning. Q-Learning Algorithm. Optimization of Oil Production.
For the development of an oil field the operational solution is quite complex due to the high amounts of variables involved in the process, such as: spacing between wells, well numbers, fluid injection system for supplementary recovery, among others. With this, an optimization method that allows the evaluation of several production profiles for different representations, to establish the optimal alternative from an economic perspective, becomes relevant. This work presents an intelligent decision support system that aims to optimize the search for alternatives for the development of an oil field, submitted to the water injection process, implemented by the Q-Learning algorithm of the reinforcement learning method. Each alternative concerns the distance between the injector and producer wells, as well as the injected water flow in different stages of BSW (Basic Sediments and Water). The implementation of the algorithm consists in finding the optimal (or near-optimal) alternative, in a timely manner, to provide the highest Net Present Value (NPV) obtained from the initial investment cost, oil price, oil production and during production time, that is, the most accessible operational condition in economic terms.