A probabilistic expert system assisting workover candidate selection and increasing the technical and economic success rate of workovers, has been developed in cooperation with a major Central-European operator.
In order to streamline and standardize the workover candidate selection process across 240 fields using a corporate-wide software solution, expert knowledge, data analytics and already existing workflows and methods have been integrated. The underlying logic is capable of detecting a comprehensive set of well integrity and deliverability problems for active and shut-in wells.
A problem detection logic based on historic workover performance provides risk, cost and success estimates. Well production gain through proposed workovers is calculated automatically based on reservoir and neighbourhood potential. Reservoir and well performance KPIs and NPV calculations enable flexible ranking of candidates with regard to technical and economic aspects.
This paper describes the applied procedures and methods, like data analytics, machine learning and reasoning tools like Bayesian Belief Networks. The system additionally integrates petroleum engineering and knowledge discovery techniques. The screening logic works as a repeatable and automated process that can be scheduled or be executed on demand. The screening of several thousands of wells takes less than an hour. Hence the process can be executed more often and much quicker, leading to an intensified monitoring of possible business opportunities.
Over the last 2 years, several studies in more than 30 fields - with a total number of 7,600 wells - representing a wide variety of reservoir and well characteristics have been performed. In the course of these studies and in independent blind tests the correct and precise functionality was successfully tested.
The smart and automated workover candidate selection process leads to reduced costs, helps to remove non value adding work through automation and improves the overall workover success.