Candidate wells selection and ranking based on data mining and multi-criteria decision analysis techniques

Different well productivity methodologies have been used by oil companies to improve oil recovery from mature fields. The candidate wells selection and ranking is one of the critical stages of such methodologies, which is used to identify the best candidates among the average performing well stock. In this paper, a data-driven approach for the identification, selection, and ranking of candidate wells for productivity analysis and intervention is proposed. A set of critical variables of the oilfield production system for carbonate oil reservoirs is identified. Data mining techniques are used for data pre-processing and candidate wells selection, while a multi-criteria decision analysis method called Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) is applied for wells ranking. The measures borrowed from information retrieval, cumulative gain, and discounted cumulative gain are applied for the ranking of wells under the premise that the first well from the list of candidates has the highest probability of success. A genetic algorithm is proposed to optimize the weights applied in TOPSIS. It is based on a population of chromosomes, each one serving as a feasible solution to the optimization problem. A real-life illustrative example for a mature field operating in the Mesozoic formation and located in southeastern Mexico is presented. By comparing the proposed methodology with the results obtained by an expert panel, applying conventional productivity analysis techniques, we demonstrate that the obtained wells ranking better fits the selected criteria, offering considerable time savings. © 2021, Saudi Society for Geosciences.