Search

Process safety incidents have large consequences on people, community and environment; hence, it is important to have a learning organization to build on lessons learned from incidents both within and outside the organization. In this study, a web-based collection of process safety incidents is categorized and analyzed in a two-tiered manner to identify proximate causes. Further, a risk-based framework is used to determine deficiencies in the safety management systems. 96 incidents were collected and subjected to the analysis. Even though these incidents are from across the world, they are not statistically representative due to the lack of a universal database of process safety incidents. During the analysis, it was observed that 60% of the incidents resulted in explosions and 16% had equipment malfunction as a proximate cause. In addition, the analysis using the risk-based framework showed that incidents were influenced by a deficiency of safe work practices, operating procedures and conduct of operations. Future work in this area includes the development of tools for continuous monitoring of safety critical systems by leveraging lessons learned from previous incidents.

Many incidents have helped to define and develop process safety. Each has provided valuable learning opportunities. However, it is even more important to identify insights that can be obtained from an analysis of a large set of incidents that represents those that typically occur. This larger picture illuminates trends and commonalities and provides learning opportunities that are even more important than the causes of any one individual incident. The Chemical Safety Board has published the results of over 60 investigations of process safety incidents. These data have been analyzed to identify commonalities and trends so that measures to help protect against future incidents can be developed. Recommendations are made to address key issues identified.

A dataset of 975 incidents investigated by the Health and Safety Executive (HSE) from 1991 has been analysed using a loss of containment (LOC) model in Storybuilder, a software tool for incident analysis developed as part of the Occupational Risk Model (ORM) project, sponsored and developed by the Dutch Ministry of Social Affairs and Employment ( Ale et al., 2008 ) to identify patterns of occurrence. The LOC model breaks down main and underlying causes of incidents in terms of barriers, tasks and safety management systems that failed in the course of an incident. Failure of containment due to substandard condition/material, with a significant contribution to structural failures including corrosion was the most frequent cause of incidents. The number of failures during indication, detection, diagnosis and response of a substandard containment was high which suggests that inspection and maintenance of pipework and process equipment focused on detection of mechanical damage and its precursors could significantly reduce incident occurrence. Failures to provide adequate means of measuring and indicating process parameters such as temperature, level or pressure was next in frequency of occurrence. Level detection had the largest share in the failure of the barrier. Direct intervention by operators failed in a large number of incidents thus discouraging reliance on this barrier as an effective means to stop a LOC incident. It is recommended that an analysis of barrier failures, task failures and the underlying management delivery failures of all incidents, including dangerous occurrences, investigated by HSE would provide quantitative evidence on the evolution of underlying causes of incidents and their associated safety management delivery failures. This evidence would allow prioritisation of preventative intervention policies, and could help develop metrics that would quantitatively show the effect of HSE’s work on patterns of incident occurrence, the barriers that commonly fail and their associated tasks and management delivery failures.