Global Optimization of Oil Production Systems, A Unified Operational View

Abstract

Abstract Oil production systems (OPSs) are complex and heterogeneous. They comprise several equipment and their wells flow naturally or using a variety of artificial lift methods. To optimize these systems, integrated multidisciplinary skilled teams managing a considerable volume of information are required. In practice, OPSs are not optimized globally; instead some subset of the system is optimized locally or many simplified assumptions are used which may not be consistent with the OPS conditions. For example, artificial lift wells are tuned according to their completion and surface facilities without considering the rest of the system. Traditional optimization techniques such Linear and Dynamic Programming, available from the Operation Research (OR) community, fail to find a global optimum solution for this problem or demand a considerable amount of computational resources making impractical their application. Under these techniques, an objective function has to be stated precisely and has to cover all the variables to optimize, being necessary a strong mathematical and OPS knowledge. This paper presents a new methodology for global optimization of OPSs. We developed an optimization procedure merging Artificial Intelligence (AI) techniques with OR techniques to deal with this problem. Combining the power of evaluating alternative solutions and flexibility of Genetic Algorithms (GAs) with the local search abilities of the Tabu Search (TS) meta-heuristic, a robust solution is found requiring a reasonable amount of computational resources. The OPS manager can easily change the objective function, only the optimization target is requested, and the performance of our technique is not affected. Through this objective function and our OPS model, the OPS is unified under economical or operational criteria. Our method is being tested on 25 wells OPS combining SRP, PCP and ESP wells. A 5% of increase in oil production and a 10% of reduction in energy consumption is expected.