Development and application of a pipeline remaining life prediction tool using machine learning methods

Introduction. This paper addresses the urgent problem of monitoring the condition of field pipelines in the oil and gas industry, where corrosion defects significantly affect operational efficiency. Traditional methods of assessing pipeline networks, such as in-line inspection and corrosion coupons, have technological and economic limitations. This study proposes a novel approach based on machine learning methods to predict defect growth using retrospective data.
Objective. The aim of the research is to develop a pipeline remaining life prediction tool based on machine learning methods to improve pipeline integrity management. The main task is to create an algorithm capable of forecasting the emergence and development of corrosion defects, thereby enhancing the reliability of pipeline transport, reducing operational costs, and optimizing maintenance processes.
Materials and methods. The study utilized data on the technical condition of pipelines, including results from in-line inspection (ILI), ultrasonic thickness measurements (UTM), and operational parameters. The following machine learning methods were applied for analysis and forecasting: gradient boosting (CatBoost), AutoML, LSTM, and Transformer. Data preprocessing included the selection of key parameters using Pearson correlation analysis and principal component analysis (PCA). The data was split into training and test sets, and the effectiveness of the methods was evaluated using the Mean Absolute Error (MAE) metric.
Results. A comparative analysis of various machine learning algorithms was conducted to predict the depth of corrosion defects in pipelines. The best performance was demonstrated by the model based on gradient boosting combined with a Transformer architecture. 
Conclusion. The developed tool enables early defect detection, automated in-depth analysis of large datasets, and decision-making support. Implementing this approach in operational processes helps reduce inspection and repair costs and improves the safety of oilfield pipeline operations. The tool can be integrated into pipeline condition management systems to provide effective forecasting and maintenance planning.

1. М-01.06.06-04 Guidelines for the organization and implementation of diagnostic programs for the Company’s field pipelines. Moscow: Gazprom Neft . 2017; 160.

2. Pipeline Operators Forum (ФОТ) «Specifications and requirements for in-line inspection of pipelines», 2016.

3. GD-23.040.00-КТН-011-16 Determination of the strength and durability of pipes and welded joints with defects. Transne1, 2018г.

4. API 570 Piping Inspection Code Inspection, Repair, Alteration, and Rerating of In-service Piping Systems, 2003.

5. Methodology for determining the risk of damage to the wall of pipes of main oil pipelines based on inspection data from inline flaw detectors / Ed. by Transne1. Moscow.: Transpress, 1997; 32.

6. Rafael Amaya-Gomez Modeling of pipeline corrosion degradation mechanism with a Levy Process based on ILI (In-Line) inspections. International Journal of Pressure Vessels and Piping, Elsevier. 2019; 21.

7. Caleyo F., Velazquez J.C., Valor A., Hallen J.M. Probability distribution of pitting corrosion depth and rate in underground pipelines: A Monte Carlo study. Corrosion Science. 2009;51:1925–1934.

8. DNV-RP-F101 Recommended practice. Corroded pipelines. Norway: Det Norske Veritas, 2017; 123.

9. Vanaei H.R., Eslami A., Egbewande A. A review on pipeline corrosion, in-line inspection (ILI), and corrosion growth rate models. International Journal of Pressure Vessels and Piping. 2017;149:44–55.

10. OS Gazprom 2-2.3-361-2009 Guidelines for assessment and forecast of corrosion condition of linear section of main gas pipelines. Moscow, Gazprom. 2010.

11. Poluyan L.V. Markov model of corrosion defect growth and its application to pipeline integrity management / L.V. Poluyan. Problems of mechanical engineering and reliability of machines. 2009;6:105–111.

12. ASME B31.8-2003 standard. Gas transmission and distribution piping systems. Revision of ASME 831.8-1999. NY: ASME, 2004; 190.

13. IS 153-39.4-010-2002 Methodology for determining the residual life of oil and gas field pipelines and pipelines of head structures. Moscow, 2002.

14. Goncharov A.A., Corrosion condition and service life of equipment and pipelines of hydrogen sulfide-containing oil and gas fields: dissertation for the degree of candidate of technical sciences, Moscow, 2000; 210.

15. Ahammed M. Probabilistic estimation of remaining life of a pipeline in the presence of active corrosion defects. International Journal of Pressure Vessels and Piping. 1998;75:321–329.

16. Polouian L.V. Holistic approach to acquisition and statistical analysis of ILI results / L.V. Polouian, S.A. Timashev. Proceedings of the Biennial International Pipeline Conference, IPC: 2006 6th International Pipeline Conference, IPC 2006, 25–29 Sept. 2006 / sponsors: ASME, International Petroleum Technology Institute, IPIT. Calgary, AB, 2007; 439–446.

17. Guanzhi Li, Aining Zhang, Qizhi Zhang, Di Wu, Choujun Zhan. Pearson correlation coefficient-based rerformance enhancement of broad learning system for stock price prediction. IEEE Transactions on Circuits and Systems II: Express Briefs, 2022.

18. Jerome H. Friedman. Greedy function approximation: A gradient boosting machine. Ann. Statist, 2001.

19. Le T.T., Weixuan Fu, Moore J.H. Scaling tree-based automated machine learning to biomedical big data with a feature set selector. Bioinformatics. January 2020;36(1):250–256. https://doi.org/10.1093/bioinformatics/btz470

20. Sepp Hochreiter, Jürgen Schmidhuber. Long Short-Term Memory, 1997. https://doi.org/10.1162/neco.1997.9.8.1735

21. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, Illia Polosukhin. Advances in Neural Information Processing Systems, 2017.

22. Stone M. Cross-Validatory Choice and Assessment of Statistical Predictions. Journal of the Royal Statistical Society: Series B (Methodological). January 1974;36(2):111–133. https://doi.org/10.1111/j.2517-6161.1974.tb00994.x