Simulation and Prediction in Safety Case Evidence

The use of simulation in safety analysis is often criticised. We are faced, however, with safety critical systems of ever-greater complexity, and with the demand to extend safety engineering to cover large-scale sociotechnical systems such as hospitals and military forces. Simulation methods offer ways to analyse aspects of these systems that other, ‘traditional’ methods cannot, because of size, complexity or non-linear behaviour. It would be foolish to claim that simulation techniques aren’t frequently misused, or that there haven’t been serious errors influenced by poor simulation. Every method of analysis and prediction, however, has its own risks, flaws and limitations. The solution is twofold; first, an engineer must know the specific capabilities and limits of the methods that they use, and hence know what information they can give and how they complement each other. Second, an operator must update their predictions (from both accidents and near-miss incidents) as part of their safety management system, thereby evaluating both the correctness of their predictions and the effectiveness of their tools. In this paper, we review the general strengths and weaknesses of simulation methods, and compare these to the capabilities of other several other methods. We raise the common vulnerability of all the methods to changes in the world, and draw some conclusions for safety analysis in general.