Publications

The need for pipeline leak detection systems arises from the critical importance of maintaining the integrity and safety of pipeline networks across various industries, including oil and gas, water distribution, and chemical transportation. Pipelines serve as vital infrastructure for transporting valuable resources over long distances, but they are susceptible to leaks due to corrosion, mechanical damage, or operational failures.

Pipeline leaks can have severe consequences, including environmental contamination, loss of product, financial liabilities, and safety hazards to nearby communities. In response to these risks, pipeline operators implement leak detection systems to promptly identify and mitigate leaks, minimizing their impact and ensuring regulatory compliance.

Effective leak detection systems utilize advanced technologies such as acoustic sensors, pressure monitoring, and flow analysis to continuously monitor pipelines and provide early detection of leaks. By proactively identifying and addressing leaks, these systems help safeguard the environment, protect public health and safety, and preserve the integrity of critical infrastructure networks.

Institut Teknologi Bandung (ITB) and PT Scada Prima Cipta (SPC) have developed leak detection system software based on advanced mathematical techniques, using Deep learning and artificial intelligence, to improve sensitivity while delivering superior reliability and robustness. This system can be customized and implemented cost-effectively according to the operating conditions of the client's existing pipeline transmission

The leak detection model is developed using the deep learning method. The mathematical approach needs actual leak data to train the leak detection model, however such data could not be obtained from oil fields. Therefore, for training purposes hypothetical data are developed using the transmission pipeline model, by applying various physical configurations of the pipeline and applying oil properties correlations to estimate the value of oil density and viscosity. The various leak locations and leak rates are also represented in this model. The accuracy of this approach is determined by the quality of training data