Surface Pipeline Leak Detection Using Machine Learning

Abstract

Petroleum, also known as \"rock oil,\" is a material that has developed over many years beneath the earth\'s surface. The sea (both onshore and offshore) is one of the main suppliers of petroleum products. Transporting oil or petroleum products from the Middle Sea region to the shore region involves a number of steps that are successfully completed via pipeline transportation. One of the biggest problems with oil and gas production is pipeline leakage. Every time there is a leak, they pose a very serious harm to the environment. This study focusses on inspecting pipelines for leaks by looking for cracks or holes in them. This study offers a revolutionary machine learning-based method for pipeline leak detection. This study offers a unique method for detecting pipeline leaks that makes use of machine learning (ML) techniques. The goal of the suggested method is to improve leak detection procedures\' accuracy and efficiency. A large dataset of sensor data gathered from a moving object, calling it as Rover, is used to train the machine learning model. By looking at these data, the model can learn to identify patterns and anomalies that point to leakage. In order to increase the reliability of leak detection, developing a verification framework and an in-pipe inspection robot, mentioned earlier as Rover, that allows the operator to validate their discovery of a leak by passing it via a neural network-based system. The proposed method has several advantages over traditional leak detection methods. This work can be used to compare the leak detection system with sensors on the rover and in the pipeline. These two approaches\' accuracy can be compared to determine which is best.

Introduction

Oil and natural gas are vital global energy sources, with complex and costly extraction and distribution processes. Natural gas, once seen as a nuisance and often flared, has gained importance due to shale gas development and lower emissions than coal. Transporting oil and gas involves pipelines and pump units, but pipeline leaks pose risks including environmental pollution, economic loss, and safety hazards. Leaks can be caused by pressure drops, cracks, or external factors like flooding.

Pipeline leak detection methods are broadly categorized into external and internal inspection techniques. External methods include acoustic sensors, water usage audits, visual inspections, fiber optic sensing, and time-domain reflectometry, often relying on sensor data analysis for anomaly detection. Internal inspection frequently uses Pipeline Inspection Gadgets (PIGs), equipped with sensors to detect metal loss, cracks, leaks, and geometry changes.

Recently, machine learning, especially Artificial Neural Networks (ANNs), has been integrated with traditional sensor data to improve leak detection accuracy. Robotic in-pipe inspectors ("rovers") are also being developed, equipped with sensors and mapping tools.

The project discussed aims to design a pipeline leak detection system using machine learning implemented via a rover. The rover is built around an ESP32 microcontroller, pressure sensors, DC motors controlled by a motor driver, and powered by batteries. It navigates a PVC pipeline segment with a simulated leak point. Commands are sent to the rover via Google Firebase, which also logs sensor data for analysis.

Machine learning models like Random Forest and Long Short-Term Memory (LSTM) networks are applied to collected pressure data to detect leaks. Random Forest is an ensemble decision tree method that reduces overfitting, while LSTM networks, with their gated architecture, can learn long-term dependencies in sequential data, aiding in accurate leak detection.

Conclusion

The research has produced a low-cost leak detection system that can detect low pressure leaks that are located distant from the inlet and outlet pressure sensors and that the conventional system would overlook due to the deterioration of the leak-induced pressure variation. Installing a pressure sensor with remote gearbox capabilities at the pipeline\'s 10 kilometer midpoint completes the system. This sensor reacts more effectively to changes in pressure brought on by leaks close to the inlet. According to experiments, leaks at the intake had the best probability of being discovered because of the high pressure there, but leaks halfway to the outlet have the best chance of going unnoticed because of the low pressure there. Here, the accuracy is low because the simulation only obtains a small number of datasets. with a gradual decline in sensor activity.suggesting a sensor-equipped pipeline robot.Phase 2 of this work will be carried out in order to improve the accuracy. Equipped the pipeline and setup a Rover with pressure sensors that can move through the pipeline. Rovers may move straight through the pipeline, making it possible to locate leaks with extreme precision. This allows for the integration of data from sensors and the analysis of spatial patterns. A fixed sensor on a pipe only provides localized information. Pipeline leak detection accuracy using sensors placed on pipeline 72% and Pipeline leak detection accuracy using sensors placed on rover 98%.

References

[1] Danial Waleed, Syed Hamdan Mustafa, Shayok Mukhopadhyay, Mamoun Abdel Hafez, Mohammad A. Jaradat, Kevin Rose Dias, Fahad Arif, Jawwad Imtiaz Ahmed, An in-pipe leak detection robot with a neural-network based leak verification system, : DOI 10.1109/JSEN.2018.2879248, IEEE Sensors [2] Adegboye, M.A., Fung, W.K., Karnik, A., 2019. Recent advances in pipeline monitoring and oil leakage detection technologies: principles and approaches. J. Sens. 19 (11), 2548. doi:10.3390/s19112548. [3] Idachaba, F., Rabiei, M., 2021. Current technologies and the applications of data analytics for crude oil leak detection in surface pipelines. J. Pipeline Sci. Eng. 1 (4), 436–451. doi:10.1016/j.jpse.2021.10.001, ISSN 2667-1433. [4] F. Batzias, C. Siontorou, and P.-M. Spanidis, “Designing a reliable leak bio-detection system for natural gas pipelines,” Journal of Hazardous Materials, vol. 186, no. 1, pp. 35–58, Feb 2011. [5] D. Chatzigeorgiou, K. Youcef-Toumi, and R. Ben-Mansour, “Design of a Novel In Pipe Reliable Leak Detector,” IEEE/ASME Transactions on Mechatronics, vol. 20, no. 2, pp. 824–833, Apr 2015. [6] L. Boaz, S. Kaijage, and R. Sinde, “An overview of pipeline leak detection and location systems,” in 2014 Pan African Conference on Science, Computing and Telecommunications (PACT). IEEE, 2014, pp. 133–137. [7] T. Takayama, H. Takeshima, T. Hori, and T. Omata, “A twisted bundled tube locomotive device proposed for in-pipe mobile robot,” IEEE/ASME Transactions on Mechatronics, vol. 20, no. 6, pp. 2915–2923, 2015. [8] Z. Liu and Y. Kleiner, “State-of-the-art review of technologies for pipe structural health monitoring,” IEEE Sensors Journal, vol. 12, no. 6, pp. 1987–1992, 2012. [9] A. Cataldo, G. Cannazza, E. De Benedetto, and N. Giaquinto, “A new method for detecting leaks in underground water pipelines,” IEEE Sensors Journal, vol. 12, no. 6, pp. 1660–1667, 2012. [10] F. Wang, W. Lin, Z. Liu, S. Wu, and X. Qiu, “Pipeline leak detection by using time-domain statistical features,” IEEE Sensors Journal, vol. 17, no. 19, pp. 6431–6442, 2017. [11] W. Mpesha, M. H. Chaudhry, and S. L. Gassman, “Leak detection in pipes by frequency response method using a step excitation,” Journal of Hydraulic Research, vol. 40, no. 1, pp. 55–62, 2002. [12] A. Lay-Ekuakille, A. Trotta, and G. Vendramin, “FFT-based spectral response for smaller pipeline leak detection,” in Instrumentation and Measurement Technology Conference, 2009. I2MTC’09. IEEE, 2009, pp. 328–331. [13] A. Lay-Ekuakille, G. Vendramin, A. Trotta, and P. Vanderbemdem, “STFT based spectral analysis of urban waterworks leakage detection,” in XIX IMEKO World Congress, Fundamental and Applied Metrology, Lisbon, Portugal, 2009. [14] A. Lay-Ekuakille, G. Vendramin, and A. Trotta, “Spectral analysis of leak detection in a zigzag pipeline: A filter diagonalization method-based algorithm application,” Measurement, vol. 42, no. 3, pp. 358–367, 2009. [15] A. Lay-Ekuakille, C. Pariset, and A. Trotta, “Leak detection of complex pipelines based on the filter diagonalization method: robust technique for eigenvalue assessment,” Measurement Science and Technology, vol. 21, no. 11, p. 115403, 2010. [16] A. Lay-Ekuakille and P. Vergallo, “Decimated signal diagonalization method for improved spectral leak detection in pipelines,” IEEE Sensors Journal, vol. 14, no. 6, pp. 1741–1748, 2014 [17] G. Desjardins and R. Sahney, Pipeline Pigging and Integrity Technology, J. Tiratsoo, Ed. Scientific Surveys Limited/Clarion Technical Publishers, 2003. [18] P.-S. Murvay and I. Silea, “A survey on gas leak detection and localization techniques,” Journal of Loss Prevention in the Process Industries, vol. 25, no. 6, pp. 966–973, 2012. [19] R. Santos, E. De Sousa, F. Da Silva, S. Da Cruz, and A. Fileti, “Detection and on-line prediction of leak magnitude in a gas pipeline using an acoustic method and neural network data processing,” Brazilian Journal of Chemical Engineering, vol. 31, no. 1, pp. 145–153, 2014. [20] S. H. Kim, S. J. Lee, and S. W. Kim, “Weaving laser vision system for navigation of mobile robots in pipeline structures,” IEEE Sensors Journal, vol. 18, no. 6, pp. 2585–2591, 2018. [21] Darkhan Zholtayev, Daniyar Dauletiya, Aisulu Tileukulova Manat Nursultan ,DiasAkimbay,YersaiynBushanov ,AskatKuzdeuov , And Azamat Yeshmukhametov, “Smart Pipe Inspection Robot With In-Chassis Motor Actuation Design and IntegratedAI-Powered Defect Detection 10.1109/ACCESS.2024.3450502,2024 [22] System” E Fajardo, D Moctezuma, “Water leak detection through satellite imagery and deep learning”, Sustainable Water Resources Management, springer 2025 [23] H Yuan, Y Liu, L Huang, G Liu, T Chen, G Su, J Dai, “Real-time detection of urban gas pipeline leakage based on machine learning of IoT time-series data”, Measurement, 2025 - Elsevier [24] F Saleem, Z Ahmad, JM Kim, “Real-Time Pipeline Leak Detection: A Hybrid Deep Learning Approach Using Acoustic Emission Signals”, Applied Sciences (2076 3417 …, 2025 - search.ebscohost.com [25] R Wang, F Cui, G Dong, X Zhao, Q Cheng, “A Pipeline Leakage Detection Technique Using Ground Penetrating Radar Polarization Method” Water Resources …, 2025 – Springer

Copyright

Copyright © 2025 Farisa Jabin T, Shyma Muhammed. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.