Version 0.6.0 · 12 October 2026 · Hawk & Mason Studios · hello@hawkmason.com
This card states what Crewops Insight computes, from which published model, under which assumptions, and what it cannot tell you. The report edition of this card, delivered with every assessment, adds the equations and parameter values so that a compliance team or an authority can follow each figure back to its source.
Insight reads an operator's roster (planned or flown) and, for each flight duty, predicts the crew member's alertness through the duty and at each top of descent, states the flight duty period against the applicable limit, and summarises the pattern over a month. It supports a fatigue risk assessment, the FRM documentation that ORO.FTL asks for, and the scientific annex of a request to deviate from a certification specification.
It is not a fatigue risk management system by itself, it is not a go/no-go device, and it makes no statement about any individual's fitness to fly. Its sleep is predicted from the roster, not measured. The operator's systems and records prevail.
Insight implements the Three-Process Model of alertness (Åkerstedt and Folkard; Åkerstedt, Folkard and Portin, Aviation, Space, and Environmental Medicine, 2004) with the parameters re-estimated on aircrew by Ingre, Van Leeuwen, Klemets, Ullvetter, Hough, Kecklund, Karlsson and Åkerstedt, "Validating and extending the Three Process Model of alertness in airline operations", PLoS ONE 9(10): e108679, 2014. This is the model family behind the Boeing Alertness Model.
Alertness is the sum of the model's three processes (homeostatic sleep pressure, the circadian rhythm and the ultradian afternoon dip), evaluated on a 5-minute grid and converted to the Karolinska Sleepiness Scale (KSS, 1–9) with the published mapping; Samn-Perelli (1–7) is printed beside it with EASA's pairing of KSS 7 with SP 5. The parameter set is the aircrew fit of Ingre et al. 2014, used as published; the equations and values are in the papers cited below and are listed in full in the report edition of this card.
Sleep generator. The roster gives sleep opportunities, not sleep. Outside duty the model puts the person to sleep and wakes them by the published thresholds, never within one hour before a duty, never inside the duty and its margins, and not within two hours of the last waking. Before the analysed window the model settles for a week of free days.
Time zones. The body clock starts on home-base time and absorbs a daily fraction of the remaining offset to the current zone (the aircrew fit; adjustable as a study assumption).
In-flight rest (augmented crews). When the roster lists three or more pilots on a duty, each leg long enough yields a rest share per pilot, computed from the leg's block time by the CS FTL.1.205(c) minimums. The modelled pilot's slot ends two hours before on-blocks (the landing pilot's rest) unless the study assumption says the relief slot. The slot is taken as sleep; by default the whole slot is credited (the favourable bound), and the study assumption can credit a smaller fraction.
Top of descent. 30 minutes before on-blocks by default (20 or 45 selectable). KSS at that instant is the per-landing figure; the per-duty figure is the highest landing.
State: Crewops Insight version; the model (Three-Process Model, Ingre 2014 parameters); the assumptions the run printed (in-flight rest credited or not, facility class, rest slot, fraction credited, sleep before the rotation, adaptation rate, top-of-descent lead); the rule set; the file(s) analysed and their dates. The CSV export carries the assumptions in every row.