Material degradation: Findings from historical accident analysis in process industries

Abstract

Material degradation represents a source of risk in the process industry, being responsible for 30% of loss of containment events. This investigation focuses on the analysis of historical events related to material degradation to gain awareness and enhance preparedness in the process industry. A database containing 3,772 records was built. Data collection was performed using industrial-accident open access databases, classifying the information according to macro-sector, type of equipment, substance involved, scenario, age of the plant, actions taken after the event and type of losses. Corrosion emerged as the main failure mechanism, followed by vibration and fatigue. This phenomenon occurs predominantly in plants with more than 25 years of operation, where prolonged exposure to chemical and environmental agents accelerates the degradation of materials. In contrast, more recent plants are more prone to failures caused by vibration. Corrosion events were frequently associated with environmental contamination episodes, with Event Tree Analysis showing it as the most likely scenario, representing approximately 50% conditional probability in documented corrosion incidents. To complete the analysis, two representative case studies were chosen for the application of the quantitative risk assessment. Conditional relationships among the variables of the database were found using Bayesian networks after the first frequentist analysis. This method allowed the investigation of uncertain data revealing a notable rise in the frequency of LOC and toxic gas dispersion. The analysis of past events highlighted the critical failure factors, which can be considered for the adoption of more effective preventive measures.