There is a saying that assumptions are the mother of all mistakes. Not surprisingly, the world of risk and safety analyses is full of assumptions.
Assumptions are a fundamental part of decision-making since complete information is seldom available at the time of decision-making.
Assumptions play an essential part in mathematical modelings, such as economic models, weather forecasts, epidemics modeling, etc. This is undoubtedly true for quantitative risk analysis (QRA), the focus area of this article.
The picture is from the Buncefield accident in 2005, where a significant explosion resulted in massive destruction to plant and residential areas. The explosion was caused by overfilling of a storage tank with petrol. Luckily, there were only 43 reported injuries and zero fatalities. Before this accident, it was an industry practice to assume, as part of risk analyses, that spillage of petrol or other stable HC liquids would not cause explosions if ignited but “only” result in liquid pool fires. This example illustrates both the importance of proper assumptions and the challenge of defining them based on the limited information available.
So how do we avoid assumptions that could lead to wrong output from Quantitative Risk Analysis (QRA)? At ORS Consulting, we achieve this by following these steps:
Formulate upfront detailed assumptions used as the basis for calculations before initiating any modeling or analyses.
Validate assumptions together with the Client and other relevant stakeholders and disciplines, preferably through a workshop/work meeting. Operations must be part of the process.
Perform sensitivity studies to investigate the impact of critical assumptions.
STEP 1 – FORMULATION OF ASSUMPTIONS
It is vital that assumptions are carefully reviewed and that the project-specific context is considered. Hence, a blind copy of assumptions from past projects should be prevented.
Lessons learned from accidents, near misses, and operational experiences must be taken into account when formulating assumptions.
Oversimplification of assumptions should be avoided for handling complicated situations. Assumptions that at first sight may seem reasonable and concise may imply that several unstated derivative assumptions are implicitly made, which may not make sense when probed in detail.
On the other hand, simplified assumptions may be excellent for multiple applications to simplify subsequent analysis/calculation. Such examples could be applications where inaccuracies in model input data do not justify more detailed mathematical analysis.
Advanced modeling should not be justified just because it is possible. Advanced modeling should only be made if the additional gained insight justifies the extra spending compared to more straightforward methods.
STEP 2 – VALIDATION OF ASSUMPTIONS
Formulated assumptions must be validated together with the Client. Stakeholders with operational experience and knowledge of the assets are essential to consult in this process.
STEP 3 – SENSITIVITY STUDIES
After the mathematical modeling has been completed based on the formulated assumptions, the criticality of premises must be assessed to see if changing certain assumptions may significantly affect the study outcome. It is recommended to identify such sensitive cases in cooperation with the Client. Some of the sensitivity cases may even be identified already as part of Step 2.
