Determining and quantifying hazard chains and their contribution to incident probabilities in flight operation

Currently, the safety level defined within an airline is heavily influenced by the perception of safety managers or expert advice. Furthermore, statistical and probabilistic methods are only marginally applied in the evaluation of Flight Operation Data (FOD). Our research focuses on calculating nominal incident probabilities based on the FOD of a single airline. This paper describes improvement to the calculation of incident probabilities by incorporating incident-specific hazards and hazard chains, a combination of hazards. This method takes hazards such as system failures, human error, or adverse weather conditions into consideration, without which the incident probabilities do not accurately reflect real flight operation. Such events also contribute to incidents and increase the probability of their occurrence. First, we determine the probability of these events using FOD or maintenance data. Based on those events, we build hazard chains in order to identify the most likely error paths (which might consist of only one single hazard) leading to the incident being evaluated. Then, we assess the impact of these hazards on the incident probability by using our simulation model. The probabilities we obtain are conditional probabilities. By applying Bayes's theorem on the conditional probabilities, we are then able to quantify incident probabilities, which include both extreme non-hazardous values of contributing factors and possible hazards and hazard chains. We apply our proposed method to the example of an incident type runway overrun.