The work analyzes the results of the implementation of the Production Management System (hereinafter System), based on the concept of the Deming Cycle. The system allows automatic selection of Well Interventions, integrated planning of production activities, as well as calculate the production forecast using an integrated field model.
The system uses Big Data technologies, optimization algorithms for various target functions, machine learning methods. The system implements automation of typical tasks, solved by such specialists as geologist, production technologist, reservoir engineers, economist, risk manager. This allows to issue effective recommendations for various production activities on the asset. These recommendations pass the stage of approval by real experts of the Company, after which, based on optimization algorithms, an integrated plan of operational activities is formed. Based on the received integrated plan, the production profile is calculated taking into account interference effects from each of the activities. Then the fact of the planned activities is monitored, on the basis of which the expert system is self-learning in order to improve the accuracy of planning in future.
A significant number and not always satisfactory quality of the data required for the operation of the System, available at the Company, led to the need to develop additional algorithms that are able to work in the case of a limited amount of data. The results of the work in this case require a deeper analysis by field specialists before making a decision. To understand the reliability of the results obtained, the Risk parameter was introduced into the System, which clearly demonstrates which of the recommendations are based on less reliable data and use less accurate methods for evaluating the effects from the activity.
Thanks to the implementation of the Production Management System, significant results have been achieved in terms of reducing bottlenecks and optimizing operating costs. In particular, due to better integrated planning with regard to the selection of measures, taking into account their compatibility, the shortfalls were reduced by more than 10%. By improving the quality of monitoring the effectiveness of planned activities and the corresponding automatic System self-learning, it was possible to achieve a reduction in operating costs by more than 2%.