CANDIDATE WELL SELECTION FOR INTERVENTION AND WORKOVER: AN ARTIFICIAL INTELLIGENCE APPROACH A PROJECT (original) (raw)

Changes such as ruptured tubing and/or casing, plugged perforations or tubing etc. might occur in the reservoir, or near wellbore zone. Therefore, it becomes necessary to service the well so as to maintain and improve oil and gas production or injection performance. Candidate well selection is the process of identifying and selecting wells for remedial operations which have the capacity for higher production and better economic return. Not all wells are considered suitable for workover operations. The conventional approach of candidate well selection makes use of engineering and geological analysis which requires a large amount of manual work and this process is very time consuming. Fuzzy Logic as one of the emerging techniques of artificial intelligence has as its main advantage the capacity to embed human knowledge into useful algorithms. This technique performs well when no mathematical model is obtainable for a problem as is the case with workover candidate wells selection. This research project presents a Mamdani fuzzy logic evaluator where rules for acidizing candidate-well selection were derived from existing literatures. Ten wells were screened against six criteria. From the results it was observed that three wells were excellent candidates. Another three wells were good candidates; two wells were possible candidates while two wells were not candidates for stimulation by acidizing operation. Error analysis was carried out to ascertain the accuracy of the evaluator, R-squared value of 0.98 was obtained indicating a good performance by the evaluator. The results of the fuzzy logic evaluator were reasonable based on the fact that good wells are good candidates for stimulation. Based on the results obtained and the ability of fuzzy logic models to handle high uncertainty, this AI technique is highly recommended for use in inference modelling where high imprecision and uncertainty is involved.