Assisted process for design optimization of oil exploitation strategy

Abstract

The decision-making processes to select an oil exploitation strategy can be complex due to many variables to be optimized and, it may be unfeasible to simultaneously evaluate multiple variables. In these cases, assisted methods involving engineering analyses and mathematical optimization algorithms may address the problem. In the literature, while several methodologies optimize specific parts of the design infrastructure of the exploitation strategy, methodologies treating the whole process are less common. Since scientific attention focuses on the solution to specific parts of this problem, methodologies that deal with the whole problem remain less clearly developed and require more tests and studies. This paper presents an assisted method to optimize a set of variables of an oil exploitation strategy in a deterministic approach. The methodology proposes to hierarchically order variables, group them according to their nature and importance and combine different optimization procedures with practical engineering analysis. The optimization variables are separated into three groups: (1) design variables group; to be decided before field development, representing the choice of configuration and equipment, (2) control variables group; these determine the operation of the oil field and (3) design future variables group, applicable at later stages such as infill drilling. It is required the estimation of a maximum number of simulation runs based on the available time, also considering interconnections between the groups. We applied the methodology to a reservoir model based on a Brazilian offshore oil field in the pre-development phase, the period before the well development drilling, when little information is available. Results indicate that this is an efficient procedure to evaluate deterministic scenarios, suggesting optimization procedures for each decision variable and achieving good quality solutions within a reasonable number of simulation runs. This is useful in many practical cases, mainly those that require long simulation time runs. Although this work considers one scenario, a deterministic approach is the first stage for optimizing uncertain scenarios in the probabilistic optimization process. In conclusion, applying an adequate assisted process can significantly reduce the number of required simulations. We also observed that the order of variables was an important aspect to be analyzed before starting the process.