Original article Machine learning in the analysis of the electromagnetic field influence on the rate of oilfield equipment’s corrosion and salt deposition

Introduction. The formation of salt deposits and oilfield equipment’s corrosion in most oil fields has become particularly relevant due to the increase in the volume of oil produced and the increase in its water content in recent years. The deposition of salts in the formation and wells leads to a decrease in the permeability of the oil reservoir, the flow rate of wells. The aim of the work is to use machine learning algorithms to simulate the effects of an electromagnetic field on the processes of salt deposition and corrosion. Prediction of experimental results will allow faster and more accurate experiments to establish the influence of electromagnetic fields on the processes of corrosion and salt deposition.

Materials and methods. Three groups of data were used, to train the models, differing in the composition of the studied initial model salt solution: the waters of the Vyngapurovsk’s and Priobsk’s deposits, as well as tap water. The following machine learning models were used: linear regression with Elastic-Net regularization, the k-nearest neighbors algorithm, the decision tree, the random forest and a fully connected neural network.

Results. The processes of electromagnetic field influence on the formation of salt deposits and corrosion of oilfield equipment were simulated with the help of machine learning algorithms. Python program has been developed to predict the output results of experiments. Modeling with various models and their parameters is carried out.

Discussion and conclusions. It was found that the decision tree and the random forest have the best accuracy of predictions, from the experiments conducted. This is due to the fact that there is too little data in the training samples. With the increase in the number of observations, it is worth using neural networks of various architectures.

1. Koledina, K. F. Solving the problem of multi-criteria optimization of the synthesis reaction of benzylalkyl esters by the method of “ideal” point and lexicographic ordering / K. F. Koledina, A. A. Alexandrova // Computational mathematics and information technologies. — 2022. — Vol. 1, no. 1. — P. 12–19. https://doi.org/10.23947/2587-8999-2022-1-1-12-19

2. Shaimardanova, G. F. Genetic algorithm for solving the inverse problem of chemical kinetics / G. F. Shaimardanova, K. F. Koledina // Computational mathematics and information technologies. — 2022. — Vol. 1, no. 1. — P. 41–49.https://doi.org/10.23947/2587-8999-2022-1-1-41-49

3. Влияние электромагнитного поля на коррозию низкоуглеродистой стали в водных минерализованных средах и на кристаллизацию карбоната кальция в присутствии ионов железа(II) / С. Р. Алимбекова, Ф. Г. Ишмуратов, В. В. Носов [и др.] // SOCAR Proceedings Special Issue. — 2021. — № 1. — С. 116−124. http://dx.doi.org/10.5510/OGP2021SI100515

4. Физические методы предупреждения солеотложения при нефтедобыче / С. Р . Алимбекова, Р. Н. Бахтизин, А. И. Волошин [и др.] // Нефтегазовое дело. - 2019. - № 6. - С. 31−38. http://ngdelo.ru/files/ngdelo/2019/6/ngdelo-6-2019-p31-38.pdf

5. Weisberg, S. Applied linear regression / S. Weisberg // New Jersey : John Wiley & Sons. — 2005. — 352 p.

6. Хуснуллин, Ш. Р. Машинное обучение в анализе влияния частоты электромагнитного поля при различных скоростях потока раствора на скорость коррозии / Ш. Р. Хуснуллин, К. Ф. Коледина, С. Р. Алимбекова // В сборнике: Технические и технологические системы. Материалы тринадцатой Международной научной конференции. Краснодар.— 2022. — С. 462–464.

7. Sun, S. An adaptive k-nearest neighbor algorithm / S. Sun, R. Huang // IEEE. — 2010. — Vol. 1. — Р. 91–94.

8. Чистяков, С. П. Случайные леса: обзор // Труды Карельского научного центра Российской академии наук. — 2013. — №. 1. — С. 117–136.

9. Горбачевская, Е. Н. Классификация нейронных сетей //Вестник Волжского университета им. В. Н. Татищева. — 2012. — №. 2 (19). — С. 128–134.

10. Программа моделирования воздействия электромагнитного поля на процессы солеотложения и коррозии на нефтепромысловом оборудовании (Predict) / Хуснуллин Ш. Р., Коледина К. Ф., Алимбекова С. Р. [и др.]. // Свидетельство о регистрации программ для ЭВМ от 19 января 2023. ― № 2023611354.