This paper combines Bayesian networks (BN) and information theory to model the likelihood of severe loss of separation (LOS) near accidents, which are considered mid-air collision (MAC) precursors. BN is used to analyze LOS contributing factors and the multi-dependent relationship of causal factors, while Information Theory is used to identify the LOS precursors that provide the most information. The combination of the two techniques allows us to use data on LOS causes and precursors to define warning scenarios that could forecast a major LOS with severity A or a near accident, and consequently the likelihood of a MAC. The methodology is illustrated with a case study that encompasses the analysis of LOS that have taken place within the Spanish airspace during a period of four years.

This paper aims to present the application of a fishbone sequential diagram in air traffic management (ATM) incident investigations performing as a key connection between safety occurrence analysis methodology (SOAM) and accident/incident data reporting (ADREP) approaches. SOAM analysis is focused on organizational cause detection; nevertheless, this detection of individual causes from a complete incident scenario presents a complex analysis, and even more, the chronological relationship between causes, which is lacking in SOAM, should be tracked for post-investigation analysis. The conventional fishbone diagram is useful for failure cause classification; however, we consider that this technique can also show its potential to establish temporal dependencies between causes, which are categorized and registered with ADREP taxonomy for future database creation. A loss of separation incident that occurred in the Edmonton area (Canada) is used as a case study to illustrate this methodology as well as the whole analysis process.

A Mid-Air Collision (MAC) is a fatal event with tragic consequences. To reduce the risk of a MAC, it is imperative to understand the precursors that trigger it. A primary precursor to a MAC is a loss of separation (LOS) or a separation infringement. This study develops a model to identify the factors contributing to a LOS between aircraft pairs. A Bayesian Network (BN) model is used to estimate the conditional dependencies of the factors affecting criticality, that is, how close the LOS has come to becoming a collision. This probabilistic model is built using GeNIe software from data (based on a database created from incident analysis) and expert judgment. The results of the model allow identification of how factors related to the scenario, the human factor (ATC and flight crew) or the technical systems, affect the criticality of the LOS. Based on this information, it is possible to exclude irrelevant elements that do not contribute or whose influence could be neglected, and to prioritize work on the most important ones, in order to increase ATM safety.

Heinrich’s pyramid theory is one of the most influential theories in accident and incident prevention, especially for industries with high safety requirements. Originally, this theory established a quantitative correlation between major injury accidents, minor injury accidents and no-injury accidents. Nowadays, researchers from different fields of engineering also apply this theory in establishing quantitatively the correlation between accidents and incidents. In this work, on the one hand, we have detected the applicability of this theory by studying incident reports of different severities occurred in air traffic management. On the other hand, we have deepened the analysis of this theory from a qualitative perspective. For this purpose, we have applied the convolution operator in identifying correlations between contributing causes to different incident severities, also known as precursors to accidents, and system failures. The results suggested that system failures are mechanisms by which the causes are manifested. In particular, the same underlying cause can be manifested through different failures which contribute to incidents with different severities. Finally, deriving from this result, an artificial neuronal network model is proposed to recognize future causes and their possible associated incident severities.