In the beginning, SAFE was created as a one-off regulatory tool for the UK Civil Aviation Authority by developing the existing general QinetiQ Alertness Model and validating it against actual duties performed by aircrew across a range of airlines and their operations.
As the tool’s benefits emerged, it became obvious that airlines should have access to it to help them manage their alertness issues.
Initial Research at The Institute of Aviation Medicine
The science base for the underlying fatigue model was developed by The UK Ministry of Defence’s (MoD) Institute of Aviation Medicine (IAM) which through time became merged with other human factors groups from the Royal Air Force, Royal Navy and the British Army to be absorbed into the newly created Defence Evaluation Research Agency (DERA). After the privatisation of this part of the UK government research laboratories in 2001, it was transferred into a new company, QinetiQ Limited, and became known as QinetiQ’s Centre of Human Science (CHS) group.
In 1975, IAM Sleep Scientist and FRMSc Director, Dr Barbara Stone, led a project to build and equip a world class sleep laboratory that included generously proportioned soundproofed rooms with human performance test beds and full spectrum lighting that mimicked real daylight, permitting subjects to be convinced that they were in different time zones. The latest EEG and other analysis equipment was procured and used to conduct a series of controlled studies for MoD and others that eventually became the fundamental research base for SAFE. These studies were designed to measure performance and fatigue under varying sleep conditions at all times of the day and night. The measures included time since sleep, time of day and sleep inertia.
The CHS Fatigue Model
IAM statistician Mr Mick Spencer led the overall SAFE development team to design the programme collect the data and use it to create the mathematics of a fundamental fatigue model which eventually became known as the CHS Fatigue Model. This model forms the basis of SAFE.
Robust Data Collection and Validation of the Model
The original aircrew data underlying the application was largely collected from the 1980’s until the early 2000 when the first version of SAFE was released for comment by the airline industry as version 2.09. This data was supplemented with the results from further, more recent studies as part of the continuous improvement of the model that was released as version 5.5. The current version 6.2 incorporates a cabin crew module (CARE) that is suitable for assessment of fatigue in cabin crew flying short, long and ultra-long haul routes.
During the earlier period, short and long haul crew and cargo operations from a variety of airlines operating worldwide were studied. Diaries were analysed and both sleep and performance were measured with many instruments including a battery of performance measures, actiwatches, thermal probes and EEG equipment using not only the controlled conditions of a sleep laboratory with associated EEG equipment to measure brain activity but also and predominantly, from flight crew during their operational duties. Consequently, SAFE is particularly suited to airline operations because it has been constructed from and validated with a huge amount of data collected from operational aircrew across a portfolio of short haul and international airlines over a period of over 30 years. This has been recognised by the Federal Aviation Authority who have published in footnote 18 on page 28 of their September 2010 NPRM, a note that SAFE is the only model that has been validated for use by aircrew.
One such study, conducted over a period of three years, comprised aircrew carrying pilgrims between Indonesia and Saudi Arabia. The pilots were instrumented and their performance measured for the same operation departing at every hour around the clock throughout the programme. The aircrew data was supplemented by field studies with volunteers when the body temperature measurements allowed a more complete measure of circadian adaptation.
Airline Specific Features
As well as measuring the sleep and fatigue and the effects of time zone changes, early starts, late finishes the model also includes calculates the amount and benefit of sleep on board aircraft when relief crew are carried. The model even allows for the difference between sleeping in a number of classes of seat as well as in a special bunk. The beta version of the model was freely distributed and validated by more than fifty airlines world-wide.
With the advent of ultra-long haul operations the model was extrapolated in 2002 to calculate the crew composition and schedules possible with such flights. The models predictions were then tested with the first ULR flights between Singapore and the US in 2003 showing a close correlation between prediction of likely fatigue levels and that experienced by aircrew.
A similar result was obtained in a large study carried out in 2014 covering air taxis and emergency medical services operators. This demonstrated a very close correlation between the SAFE predictions and the actual data collected proving the algorithms are very robust.
Mr Spencer and Dr Stone have published numerous parts of their work in a number of publications throughout the development. The CAA PAPER 2005/04. Aircrew Fatigue: A Review of Research Undertaken on Behalf of the UK Civil Aviation Authority provides a useful summary of the studies up to 2003 from which SAFE was constructed. Although not completely up to date, the paper gives a good coverage of the studies and science that have been incorporated into the model. Further papers published by Dr Stone during her career are available here.
FRMSc are indebted to both QinetiQ and CAA in agreeing license terms to be able to further develop SAFE and make it available commercially. It is very pertinent to also acknowledge the innovative work on this project by Mr Mick Spencer in collaboration with his team that include Dr Barbara Stone, Dr Karen Robertson, Ms Anne Melia and others over the long period of its construction.
Further, we are indebted to the numerous volunteers and aircrew who participated in the studies, without whom none of this work would have been possible. We would also like to thank those individuals who helped in the co-ordination and running of the research studies.