Event trees are often used in the assessment of aviation safety. Since catastrophic accidents are rare, the number of observed events to quantify the model may be small or even zero. In particular, some sequences of an event tree may have never occurred. This has the potential to create large uncertainty ranges for the parameters estimated. The objective of this paper is to identify the most critical nodes in an event tree with respect to their potential for reducing the overall uncertainty of the model. To do this, an uncertainty important metric from the literature is applied. The metric answers the question: If one were able to know the exact value of a parameter, say, through a large data collection effort, how much would the overall uncertainty of the model improve and which parameter would lead to the best uncertainty improvementƒ We provide several case studies of the method to existing event sequence diagrams in the Integrated Safety Assessment Model. Finally, we consider the method applied to hypothetical event trees with no observed events, such as those that might apply to future airspace concepts. Results show that nodes that are closest in proximity (within the model structure) to a catastrophic end event tend to be the most critical. One application of these results is to prioritize nodes for quantification via expert elicitation exercises.
Uncertainty Importance Analysis for Aviation Event Trees
2019-04-01
325360 byte
Conference paper
Electronic Resource
English
Importance of aerodynamic safety in aviation
Engineering Index Backfile | 1927
|The Importance of Aviation R&D
Online Contents | 2001
|Coordinated aviation network resource allocation under uncertainty
Online Contents | 2012
|