Using dogs to detect hidden corrosion

Dogs used as detectors in remote scent tracing (RST) technology usually detect the presence of explosives or contraband in scent samples collected by sucking air from containers or air freight. In this study, five dogs were trained to detect corrosion under the insulation (CUI) of pipes in scent samples collected at a gas processing plants. CUI is a major problem in oil and gas processing plants, causing safety risks and leading to production loss. Scent samples were made by sucking air through drain plugs in the insulation material surrounding the pipes onto filters. During a two year project, dogs trained to detect corrosion using insulation material collected earlier from other corroded locations at the plant were able to detect corrosion on the filters collected from intact insulated pipes at that plant at the same level of proficiency, detecting corrosion at around 59% while producing on average less than 3% false alarms. The systematic training approach, the integration of field samples into training runs and the use of several dogs to improve the reliability of the system are described. Preliminary results on double blind samples were promising: the sensitivity of the detection of field samples was 92%, and the selectivity 93%. The application of such a system as a tool in a preventive maintenance program at oil and gas processing plants could be useful to determine timing of maintenance, thus allowing a more efficient allocation of costly resources necessary for the customary visual inspection.