Avoiding Engineering Catastrophe: New Insights from Data

摘要

High-consequence, low-frequency events, such as toxic releases, fire, explosion, maritime losses and railway collisions are prevented and mitigated by good engineering practice throughout the lifecycle of an installation. However, all engineered barriers can degrade and subsystems can interact in a complex manner, which can make it challenging to assess how close a system is to a dangerous failure overall. Analyses after major incidents often show - with the benefit of hindsight - that a number of 'weak signals' of impending disaster were present, but these were only useful when analysed holistically. In early 2018, the "Discovering Safety" programme was launched. This programme is being delivered through the Thomas Ashton Institute and combines the strengths of the Health & Safety Executive and the University of Manchester and is supported and funded by Lloyd's Register Foundation. The programme addresses occupational safety issues together with process safety and allied technical safety topics. The focus of this programme is on using new data analysis techniques to extract valuable intelligence that will enable the prevention and mitigation of accidents, and improve 'plateaued' safety performance. A key project within the programme is the "Loss of Containment Insights" project. The project aims to create a source of intelligence by using data analytical tools on HSE's existing datasets relating to onshore loss of containment events and their precursors. This paper outlines the stakeholder interactions which led to the development of the project, the approach taken, strengths and weaknesses of RAMS intelligence in the process sector and elsewhere and then proposals for next steps together with summary findings from data analysis carried out in 2018/19. A key goal surrounding the publication of the paper is to gain constructive feedback from process industry specialists to ensure that the project is successful and the outputs useful to industry.