Term 252

AI-Enabled Methane Leak Detection for Pipelines Using Infrared Spectroscopy

Project Type: Self-Initiated

Project Description

Methane leakage from pipeline systems poses significant safety, environmental, and economic risks, requiring reliable and timely detection methods. This project presents the design and implementation of a bench-top methane leak detection system based on infrared absorption principles. Infrared light is transmitted across controlled leak regions of a pipeline section, where the presence of methane is detected through attenuation of specific wavelengths caused by gas absorption. A laboratory pipeline test rig with controlled methane leakage points was constructed to simulate realistic leak scenarios. An infrared emitter–detector pair was installed at selected locations along the pipeline to monitor variations in received signal intensity. The detected analog signal is conditioned and digitized using a microcontroller, then transmitted to a cloud platform via a cellular communication module for remote monitoring. Artificial intelligence techniques are implemented to analyze the infrared sensor data and distinguish between normal operating conditions and methane leakage events. Machine learning models are trained using labeled experimental data to classify leak presence and estimate leak severity based on signal patterns and temporal variations. A software application retrieves data from the cloud and provides real-time leak alerts, signal visualization, and event logging. System performance is evaluated through controlled experiments by varying leak rates and operating conditions. Key performance metrics include detection accuracy, response time, false alarm rate, and communication reliability. The results demonstrate the feasibility of combining infrared sensing, cellular communication, and AI-based analysis for effective methane leak detection in pipeline systems.