FAILURE PROBABILITY MODELING FOR PIPING SYSTEMS SUBJECT TO CORROSION UNDER INSULATION

Corrosion under insulation (CUI) is found to be a major problem for insulated piping systems in refineries, petrochemical and gas processing plants. Since those pipes carry hydrocarbons or other dangerous process fluids, gradual thinning due to CUI may cause the pipes to leak, leading to a hazardous situation. Due to the nature of CUI which is hidden, the challenge is in the monitoring, detection and, hence, prediction of CUI. Also, due to scarcity of data, the current CUI inspection and maintenance strategy adopts the risk-based inspection (RBI) approach where the assessment of the probability of failure for CUI adopts either the qualitative or semi-quantitative methods. These approaches were highly subjective and to overcome this drawback, the quantitative approach is usually employed where this approach bases the failure probability estimates on historical failure data. This study presents a methodology for quantitatively estimating the probability of failure of piping systems subject to CUI based on the type of data available. In the absence of failure data and wall thickness data, logistic regression model was proposed by considering the inspection data as a binary data. When the wall thickness data is available, the probabilistic models, namely degradation analysis, structural reliability analysis and Markov chain model, were proposed.
The study recommended that for the case where wall thickness data is minimal, a good model that can be used for quantitative risk assessment is the structural reliability analysis. If more wall thickness data is available, degradation analysis and Markov chain model are the potential models. This study also demonstrated that the logistic regression model is not applicable for quantitative risk assessment. In summary, the quantitative approach is necessary as a means for quantitatively establishing future reliability for piping systems subject to CUI. Even though applying the quantitative method is optional in the current RBI analysis, quantitative risk assessment is, in fact, now a required element of the maintenance optimization methodology.