Risk Assessment: Theory, Methods, and Applications

CHAPTER 16

UNCERTAINTY AND SENSITIVITY ANALYSIS

…in this world, nothing can be said to be certain, except death and taxes.

—Benjamin Franklin

16.1 INTRODUCTION

All results from a quantitative risk analysis are uncertain to some degree. In some cases, the uncertainty may be large and the conclusions based on the risk analysis may therefore be questionable. Several guidelines for risk analysis require that uncertainty analyses be conducted as part of the risk analysis, such that it can be demonstrated that the conclusions reached using quantitative risk analysis have taken uncertainty into account (e.g., see HSE, 1989, 2003a).

The uncertainty may be due to many different causes, ranging from use of inadequate models and data to misinterpretation of system functions and failure to identify potential accident scenarios.

To take uncertainty into account does not necessarily require a complex and formal uncertainty analysis. In many cases, a conservative approach to risk analysis using conservative model approximations and conservative values of input parameters may be sufficient for the decision-maker to have confidence in the results from the analysis. The requirement for a careful uncertainty analysis is obviously also dependent on the importance of the decision to be made. In any case, the study team should be aware of problems related to uncertainty and do their best to avoid or reduce the uncertainty in all steps of the risk analysis process.

In most risk analyses, it will not ...