Samara Polytech PhD student improves oil production efficiency using machine learning

The project is based on the new method of data analysis

A young scientist of Samara Polytech has designed a computer appliance for the necessary prediction of process parameters of oil wells with the help of the unique methods of machine learning, that is «Prognoz-N».

Yuri Shtyrlov, a postgraduate student of Samara Polytech, believes that neural networks, when analyzing big databases collected during the preparation of a well for operation, can give accurate results, but do not allow to understand what key indicators influence the efficiency of the well operating parameters. Therefore, the young scientist found an original mathematical approach, developed by the scientists of the Computing Center of the Russian Academy of Sciences, and adapted it to the oil and gas sector. The results of the latest research are published in the collection of abstracts of the X International Scientific and Practical Conference (https://clck.ru/QzPp8).

This approach is based on the most effective Data Science machine learning methods: the method of optimal reliable partitions and the method of statistically weighted syndromes. These methods reveal reliable patterns in a large amount of real data collected during the operation of a large number of objects over a long time. The machine learns from this database, that is the training sample, and then makes a prediction algorithm for the new object. Then you can optimize the key parameters and get the desired result.

«The software algorithm «Prognoz-N» is designed to work directly on the rig. Thanks to it, the specialists identify key indicators that directly affect the well profitability (the volume of the produced oil minus the operation costs)», Yuri Shtyrlov, a postgraduate student of the Institute of Automation and Information Technologies of Samara Polytech explains. «In addition to the determining of the key indicators, the boundary for each of them is calculated that should be eliminated in order to transfer the investigated well to the group of profitable wells, if the earlier prediction was unfavorable. Thus, we create a step-by-step plan of actions necessary to make the well operating parameters suitable to achieve the maximum target values».

Thus, «Prognoz-N» is a software package that is configured for a specific task of a specific customer for the unrestricted work with any well.

For reference:

Samara State Technical University (SSTU, Samara Polytech) as a flagship university offers a wide range of education and research programs and aims at development and transfer of high-quality and practically-oriented knowledge. The university has an established reputation in technical developments and focuses on quality education, scientific and pragmatic research, combining theory and practice in the leading regional businesses and enterprises. Education is conducted in 30 integrated groups of specialties and areas of training (about 200 degree programs including bachelor, master programs and 55 PhD programs) such as oil and gas, chemistry and petrochemistry, mechanics and energy, transportation, food production, defense, IT, mechanical and automotive engineering, engineering systems administration and automation, material science and metallurgy, biotechnology, industrial ecology, architecture, civil engineering and design, etc.