Optimization and Performance Evaluation of a Foam Plugging Profile Control Well Selection System.

Abstract

Most of the oilfields are currently experiencing intermediate to late stages of oil recovery by waterflooding. Channels were created between the wells by water injection and its effect on the oil recovery is less. The use of water plugging profile control is required to control excessive water production from an oil reservoir. First, the well selection for profile control using the fuzzy evaluation method (FEM) and improvement by random forest (RF) classification model is investigated. To identify wells for profile control operation, a fuzzy model with four factors is established; then, a machine learning RF algorithm was applied to create the factor weight with high accuracy decision-making. The data source consists of 18 injection wells, with 70% of the well data being utilized for training and 30% for model testing. Following the fitting of the model, the new factor weight is determined and decisions are made. As a consequence, FEM selects 7 out of 18 wells for profile control, and by using the factor weight developed by RF, 4 out of 18 wells are chosen. Then, the profile control is conducted through a foam system proposed by laboratory experiments. A computer molding group numerical simulation model is created to profile the wells being selected by both methods, FEM and RF. The impact of foam system plugging on daily oil production, water cut, and cumulative oil production of both methods are contrasted. According to the study, the reservoir performed better when four wells were chosen by the weighting system developed by RF as opposed to seven wells that were chosen using the FEM model during the effective period. The weighting model developed by RF accurately increased the profile control wells' decision-making skills.