A method to select failure conditions for further human factors scrutiny based on the prediction of human errors using a Bayesian network and elicited error probabilities

The assumed flight crew response to Failure Conditions (FC) is used in System Functional Hazard Assessments (SFHA) to mitigate the effects of FCs on the aircraft, the crew, and the occupants, and to justify their resulting severity classifications (i.e., Minor, Major, or Hazardous). SFHAs help to show compliance with FAR/CS 25.1309(b). However, the flight crew response in the SFHA is assumed to be in a timely manner and free of human error. Human errors are assessed in a separate Human Error Analysis (HEA), which helps to show compliance with FAR/CS 25.1302 and FAR/CS 25.1309(c). A clear link between 25.1302 and 25.1309 does not exist and therefore it is challenging to determine which FCs need further scrutiny. This paper proposes a metric to aid in this decision-making process. The metric is the output of a Bayesian network which uses elicited quantitative probabilities from experienced pilots as inputs. Further study to validate some aspects of the proposed method is recommended. A training script, questionnaire, and an R script to generate the proposed metric is provided. In addition, independent validation of the qualitative aspects in a HEA, validation of the uncertainties in the three levels of Situational Awareness, and a quantitative assessment of human reliability for use in identifying continued operational safety (COS) issues are also briefly mentioned.