Software may put to rest the crew-fatigue debate
The Centre for Human Sciences at British aerospace and defense research group Qinetiq has developed software that it says can assess and predict crew fatigue for any given set of flight operations. The system for aircrew fatigue evaluation (Safe) has been produced with the support of the UK Civil Aviation Authority and is now undergoing operational evaluation by the agency and several leading airlines.
The hope is that the program, which is still being refined by Qinetiq scientists based at Farnborough in southern England, will provide an objective model for evaluating crew fatigue. Eventually, the intention is to enhance the model so that it can also quantify the effect of fatigue on accident risk.
Safe may be able to settle long-running air-transport industry disputes over the correct parameters for flight-duty and rest time limits. In Europe, pilot unions, airlines and European Commission officials have been squabbling for years over proposed new limits, as they have in the U.S. The discord has, in part, been caused by an inability to agree on common models for assessing fatigue.
The Qinetiq model is based on extensive research with crews from carriers operating a wide variety of different services, including long-haul (covering several different time zones), short-haul and high-frequency charter. Crewmembers kept detailed diaries, over a prolonged period, of their duty and sleep habits, and self-assessments of their level of alertness. Several thousand pilots have participated in the studies, which have incorporated data from more than 10,000 flights.
The data was corroborated using wristband monitors to measure levels of wakefulness and by having the crew complete various tasks on a handheld computer at different stages of each duty period. The tasks were designed to evaluate reaction time and general cognitive impairment.
The research has analyzed patterns of fatigue according to factors such as the number of time zones crossed, the time of day or night that a duty period begins, how many sectors are operated during a duty period, the duration and quality of in-flight rest, closely consecutive duty periods, and the timing of sleep before early starts. Early stages of the work entailed recording electrical brain activity during layover sleep and the use of thermometers to measure changes in the extent and phases of the body’s circadian rhythms throughout duty schedules.
The Safe model predicts levels of alertness and fatigue as the sum of two factors: first, the time of day for the duty period, or more specifically, the prevailing circadian rhythm of the individual; and second, the pattern of sleep and wakefulness preceding the duty period.
The user can enter, modify, store and retrieve individual duty schedules. The program color-codes the duty period according to predicted levels of fatigue, with green representing the highest level of alertness, and yellow, orange and red indicating rising levels of fatigue.
Safe can also estimate the extent to which a pilot’s circadian rhythm has adapted to a different time zone. The system provides flexibility so that, for example, the benefits of additional sleep before a duty period can be assessed rather than simply assuming standard sleep patterns.
To validate its research on adaptation of circadian rhythms, Qinetiq has used its Farnborough laboratory (fitted with full spectrum lighting to change circadian patterns) to test the effects of various east- and westbound time changes on volunteers. The results showed, for example, that it takes at least six days–and in some cases seven–for people to adapt to a 10-hour time-zone transition.
The sleep component of the Safe model is based on data related to recovery of alertness after waking up (the so-called “sleep inertia” phase). It also draws on data related to the exponential reduction in alertness following sleep and the corresponding increase in alertness generated during sleep.
As the program has been developed, changes have been made to the sleep component of the model to incorporate the findings of research with regard to the timing, duration and quality of sleep and rest.
For example, data from Lufthansa crews flying between Germany and the Seychelles islands in the Indian Ocean revealed a discrepancy between reported and predicted levels of fatigue. This raised questions about the recuperative value of daytime sleep during a 12-hour layover between two successive night duty periods.
Typically, the Lufthansa crews reported sleeping for around six hours during the layover, but the anticipated benefit of this rest was not generally reflected in alertness levels during the return flight.
According to Qinetiq researcher Mick Spencer, this suggested that daytime sleep in an environment outside the closely controlled conditions of the laboratory (such as a hotel bedroom) could not be guaranteed to provide the same benefits. The tests revealed a substantial reduction in “slow wave” (deep) sleep even when the total duration of sleep was constant.
The Qinetiq research also revealed that disruption to sleep during layovers is “considerably greater” after a long eastbound flight than after the same degree of time-zone transition to the west. After long eastbound legs, crew sleeps tended to be short and fragmented. For at least the first 48 hours after the outward flight, crews tended to take sleep at all sorts of different times. This challenged the validity of the Safe program’s sleep model and so for periods after eastbound flights it has been adapted to take partial account of both local times and individual circadian rhythms.
The studies confirmed the well known difficulties associated with getting sufficient sleep before an early-morning duty period start. Data from short-haul crews showed that as duty reporting times got earlier, they were waking up ever earlier but were unable to fall asleep sufficiently earlier to compensate for this.
Accounting for Sleep Patterns
Similarly, the sleep component of the Safe model was adapted to take into account observed sleep patterns after duty periods finish late at night. Crews’ inability to sleep much beyond their normal wake-up time validated laboratory research indicating that wake-up time depends more on the circadian clock than on the time people fall asleep. Studies showed that crew sleep times reduced from an average of 7.4 hours after duty periods ending between 1 p.m. and midnight to an average of just 5.3 hours following duty periods ending between 3 a.m. and 5 a.m.
The Qinetiq data also shows that the benefits of in-flight rest vary considerably according to when it is taken and how long it is available. During the middle of the night on long-haul flights crews averaged more than 3.5 hours of sleep, compared with less than an hour in the least favorable conditions. Unsurprisingly, there was a significant increase in the amount of sleep achieved when crew bunks were available.
The Qinetiq team also identified how varying levels of pilot workload between different flights and the duration of commuting times can influence levels of alertness. These have not yet been incorporated into the Safe model but could be if sufficient information were available.
The software has been adapted to take into account variable factors such as multiple sectors, time on task, early starts and cumulative fatigue. For example, the studies showed clearly that each additional sector following the second sector flown results in an increase in fatigue–even allowing for overall flight and duty durations. Research also demonstrated consistently reduced levels of alertness for crews starting work early in the morning, with a further increase in fatigue experienced on subsequent duty periods that could not be entirely explained by the shorter period of sleep before the early start.
As it seeks to further refine the Safe program, the Qinetiq team is now looking to incorporate more complex parameters relating to the effects of time on task (the time each crewmember spends doing specific tasks) and cumulative fatigue. It is now looking at the effect of consecutive overnight duty periods and rosters, including a mix of early starts, late finishes and overnights. Further research will also take closer consideration of pilot lifestyles, including how they spend their off-duty hours.
Drawing on Wider Expertise
Also envisioned for future versions of the software is an accident risk-assessment factor. For the time being, Qinetiq has conceded that this could provide only approximate estimates based on a combination of simulator studies and analysis of crew-performance data recorded during flights.
With counterparts from other European research agencies, Qinetiq has formed a European Committee for Air Crew Scheduling and Safety and is seeking to advance the Safe model by drawing on wider expertise. So far, however, this program has failed to secure the necessary European Union funding. Several years ago the European Commission’s transport directorate had expressed an interest in using Safe to help resolve the dispute of flight-duty time limits, but this initiative was dropped following a wholesale reshuffle of the agency’s top personnel.
According to Dr. Barbara Stone, principal consultant with Qinetiq’s Centre for Human Sciences, Safe could be used by operators to develop the safest and most cost-effective crew rosters. She said the software will allow a much more flexible and thorough approach to assessing crew-fatigue issues, freeing operators and pilots from the rigid confines of existing flight-time limits, which have had to be based on somewhat sweeping assumptions.