A well-functioning FRMS should 'merry' the two perspectives of both established science and operational experience. One mechanism, based on those, implemented for limiting fatigue risk may be hard 'rules' prohibiting certain scheduling patterns. The drawback with these rules is that they are digital in nature (true/false) - and absolutely prohibiting something can become very costly to the airline. This is why these rules rarely gets implemented, which of course results in no improvement at all in flight safety.
An alternative, and better option, is to reduce and distribute fatigue risk using penalties in the objective function, guiding the optimizer constructing the original work patterns. This solution does not guarantee that a certain scheduling combination doesn't appear now and then, but it enables a signifcant reduction of the total fatigue risk. And at a much lower cost.
So how can established science about our physiology for an 'average individual', be blended in with the specific operational experience of an operator? The most straightforward way is simply to apply experience-based 'weights' on the operational conditions to factor up/down the penalties guiding the roster construction. The table below illustrates some typical conditions that can be configured that makes sure that alertness is 'positioned' well within the total planning solution. By increasing the penalty of operating with low alertness into a category C airport for example, the optimizer will produce solutions where pilots are predicted to have, in general, higher alertness for those flights (by picking another pilot and/or a better roster context preceding the flight). By adding these types of elements, and associate them with a weight (examples below in the table), alertness is positioned where it matters the most, at a minimal cost.
Please find more explanations of the mechanisms at play in this document: Best Practice Control of Fatigue Risk Throughout the Crew Management Process
