Data-driven investigation of thermal and flow parameters variation in gas pipelines under corrosion, dent, and leakage faults using experiments, machine learning, and scalable numerical simulation

• Studied common pipeline faults: corrosion, leakage, and dents. • Combined experiments, machine learning, and simulation for fault analysis. • Enabled smart monitoring using sensors and wireless data systems. • Supports the development of advanced condition monitoring systems. • Simulated fault behavior with a scalable multi-node pipeline model. Pipelines are a critical component of energy infrastructure for the continuous transportation of fluids and gases. Each pipeline has at least one inlet where energy enters and a flow driven and maintained by pumps or a compressor. Chemical reactions, mechanical damage, or material deterioration often cause corrosion, dents, and leakage. These can disrupt flow, reduce efficiency, and raise safety concerns. This study investigates the impact of these faults through a combination of experimental testing, machine learning analysis, and transient numerical simulation. Experiments were conducted using a controlled pipeline setup, where thermal and flow parameter variations were recorded under simulated fault conditions. Faults were introduced by creating dents, a leakage point, and adding real pipe debris to replicate corrosion byproducts. The experimental setup integrated low-power microsensors, wireless data transmission, and a cost-effective data acquisition unit. A machine learning model was trained on the collected data to classify fault types. A scalable numerical simulation was developed to extend the analysis and address experimental limitations. The pipeline was modeled as a multi-node system, with fault-specific modifications such as increased friction for corrosion, localized diameter reduction for dents, and mass outflow for leakage. The framework supports early fault detection and improves condition monitoring systems.