Evaluating the root causes of fatigue and associated risk factors in the Brazilian regular aviation industry

TL;DR

This study uses a biomathematical model to analyze fatigue risk factors in Brazilian aviation, revealing how night shifts and early morning flights significantly increase fatigue and suggesting key indicators for crew scheduling.

Contribution

It introduces a detailed analysis of fatigue risk factors using SAFTE-FAST in Brazilian aviation, highlighting the impact of night shifts and proposing key performance indicators for rostering.

Findings

Fatigue risk increases by 23.3% with more night shifts.

Maximum wakefulness exceeds 24 hours in certain rosters.

Fatigue hazard correlates with number of early morning departures.

Abstract

This work evaluates the potential root causes of fatigue using a biomathematical model and a robust sample of aircrew rosters from the Brazilian regular aviation. The fatigue outcomes derive from the software Sleep, Activity, Fatigue, and Task Effectiveness Fatigue Avoidance Scheduling Tool (SAFTE-FAST). The average minimum SAFTE-FAST effectiveness during critical phases of flight decreases cubically with the number of shifts that elapse totally or partially between mid-night and 6 a.m. within a 30-day period ($N_{NS}$). As a consequence, the relative fatigue risk increases by 23.3% (95% CI, 20.4-26.2%) when increasing $N_{NS}$ from 1 to 13. The average maximum equivalent wakefulness in critical phases also increases cubically with the number of night shifts and exceeds 24 hours for rosters with $N_{NS}$ above 10. The average fatigue hazard area in critical phases of flight varies quadratically with the number of departures and landings within 2 and 6 a.m. ($N_{Wocl}$). These findings demonstrate that both $N_{NS}$ and $ N_{Wocl}$ should be considered as key performance indicators and be kept as low as reasonably practical when building aircrew rosters. The effectiveness scores at 30 minute time intervals allowed a model estimate for the relative fatigue risk as a function of the time of the day, whose averaged values show reasonable qualitative agreement with previous measurements of pilot errors. Tailored analyses of the SAFTE-FAST inputs for afternoon naps before night shifts, commuting from home to station and vice-versa, and bedtime before early-start shifts show relevant group effects ($p < 0.001$) comparing the groups with and without afternoon naps, with one or two hours of commuting and with or without the advanced bedtime feature of the SAFTE-FAST software, evidencing the need of a better and more accurate understanding of these parameters when modelling fatigue risk factors.