Dynamic risk assessment in industry: implications and open questions

Risk analysis is about to enter into an era of larger and more complex data sets (big data), where the main challenges are represented by the ability to provide continuous acquisition, effective process and meaningful... [ view full abstract ]

Risk analysis is about to enter into an era of larger and more complex data sets (big data), where the main challenges are represented by the ability to provide continuous acquisition, effective process and meaningful communication of such information. The new generation of wireless sensors, increasingly powerful computers and optical fibres have the potential to be used for such purpose. "Google Trends" shows that the number of Google searches for the term "big data" has increased about 100 times since 2011 and today it has reached its peak. On the contrary, the term "dynamic risk" had its peak in 2009 and today its popularity on the search engine has decreased of about one third. Despite the fact that many factors may affect such trends and they do not represent the actual applications, this may reflect the challenges of dynamic risk assessment to find its place in industry standard approaches. While dynamic risk management is common practice in finance as a response to the financial crisis in 2008 (this explains its popularity peak one year later), most of the methods for quantitative risk assessment in a safety-critical sector such as chemical and process industry only allow for static evaluations of risk in a frozen instant of the system life. Standards (e.g. ISO 31000 on risk management and NORSOK z013 on risk and emergency preparedness analysis) and relevant regulations (e.g. EU Seveso directives on the control of major-accident hazards involving dangerous substances) suggest updates of risk analysis, but mostly in conjunction with major changes in the plant or organization. Research on how to dynamically assess risk has been carried out in several oil and gas companies, but no real implementation has been attempted. Some open questions are still undermining this kind of approach and should be clearly addressed in order to provide reliable models and exploit the opportunities of new technology developments. Some of the main questions are: i) which strategy should be adopted – proactive and based on early deviations, reactive and learning from past incidents, or a combination of the two? ii) How early warnings and information from occurred events should be continuously assessed and connected to system risk? iii) How to systematically implement such evolutionary process in the organization? This presentation aims to give an overview on some of the answers to such issues from the scientific community and highlight the related uncertainties to support future developments and applications.