Abstract
Corrosion of ferrous metals remains a major challenge in civil and industrial engineering because it can progressively reduce structural integrity before visible damage becomes apparent. Early, non-destructive detection methods are therefore needed to support timely maintenance and reduce the risk of failure. This study investigated whether gas-phase sensing can be used to detect corrosion-related volatile emissions from rusting steel. A bespoke electronic nose (eNose), incorporating a custom gas sensor array, was developed and evaluated alongside Fourier transform infrared spectroscopy (FTIR) using the same corroding samples. The sensor array was trained using headspace collected from non-corroded steel, corroded mild steel, and progressive corrosion samples, and was subsequently deployed within a simulated industrial environment consisting of a 10 m mild-steel pipe. Both the eNose and FTIR differentiated corroded from non-corroded conditions, supporting the feasibility of corrosion detection through gas analysis. Machine-learning models showed strong classification performance, with tree-based models obtaining perfect separation, and logistic regression having a 0.997 accuracy. The sensor array was able to identify corrosion-related signatures under deployment conditions. These findings demonstrate the potential of a compact, low-power eNose as a practical non-destructive tool for early corrosion screening in enclosed industrial environments.
