Wednesday, 27 July 2016
The following is an extract from a research piece I did recently, which examined the factors involved in 11 controlled flight into terrain (CFIT) fatal accidents. This extract summarises the human performance factors and has some interesting findings:
Human performance deficiencies were identified in the investigations of all 11 accidents and human factors overwhelmingly dominated the lists of causes and contributory factors. The study set out to analyze these deficiencies against two separate models, the Dupont ‘Dirty Dozen’ and the Pilot Competencies Model, as described at the outset. All of the accidents had identifiable factors from both models; indeed some accidents appeared to include deficiencies in virtually all of the Pilot Competencies together with many of the ‘Dozen’.
The graph below shows the number of accidents from the original 11, in which each of the markers from the Pilot Competencies Model was found to be deficient:
It is perhaps not surprising that the pilots in every accident exhibited deficient Situational Awareness – had they been more aware of the situation it seems likely that they would have taken more appropriate action to avoid terrain proximity. It is disappointing to see that the pilots in all accidents were also deficient in the application of their procedures but on the other hand this may be encouraging, in that procedural compliance is possibly indicated as a significant factor in the mitigation of CFIT risk. Poor communication was also identified in most (8) of the accidents, supporting a view that CRM (crew resource management) in general and communication in particular, are vital for the avoidance of CFIT.
The next graph illustrates the same data for the Dupont ‘Dirty Dozen’ markers:
Once again Lack of Awareness tops the table, appearing in all of the 11 accidents and reflecting the deficient Pilot Competency of Situational Awareness above. There is no Dupont marker that corresponds to the Pilot Competency of Application of Procedures so we can’t see a correlation but Lack of Communication again appears in the same 8 accidents. Lack of Teamwork and Lack of Assertiveness were identified in over half of the accidents and these two markers are in some ways related. If the Captain in particular is not a ‘team player’ and fails to respect colleagues and their opinions, the FO may become isolated and feel unable to intervene, even to save their own life. This risk is exacerbated by a steep authority gradient in the cockpit. Norms featured as a marker in almost half (5) of the accidents, when pilots developed and employed their own processes, either when they found that the promulgated process was inefficient, ineffective or difficult, or when there was no applicable process for them to employ.
Stress, Pressure, Fatigue, Complacency and Distraction each appeared in only 3 or fewer of the accidents but these conditions are sometimes difficult to identify from investigation reports. Unless the CVR records specific and attributable voice characteristics, or the individual mentions that they are affected by a condition, it may be that the condition goes unnoticed in the investigation.
It is worthy of note that neither model specifically addresses markers for Monitoring and/or Cross-checking, although Lack of Awareness and Situational Awareness respectively could be taken to include those competencies. Deficiencies in monitoring and cross-checking were apparent in several of the accidents. It is precisely these functions, functions we know humans perform quite poorly, that the EGPWS/GPWS/TAWS seeks to augment in CFIT prevention.
Finally, the graph below combines the two models to show the number of markers associated with each of the 11 accidents:
Accidents with a greater number of markers from one model also appear to have a similarly greater number from the other model – total markers per accident varied from 7 to 14 but the variation between models was never greater than 2. This may indicate that the two models identify similar human performance deficiencies but it may also be a reflection of the amount of information available from the accident reports upon which the study was based. These varied from a few pages to well over one hundred.