Fatigue models need to be extremely fast if they are to be used proactively in the aviation crew management process. We asked Arvid Müllern-Aspegren, Scheduling Safety Specialist at Jeppesen in Gothenburg, to explain this in more detail to us but in layman terms, and this is his reply:
- During pairing and roster optimization, it could be that a fatigue model is asked, well over a thousand times per second, to evaluate fatigue risk over a sequence of activities covering up to some 40 days of work at a time for each crew. These crew planning optimizers, running in parallel on many CPU cores on powerful servers, are used by all larger operators when putting together the crew plan for the upcoming weeks of flying.
- If a fatigue model cannot cope with responding at roughly the same pace, the overall crew planning process would be significantly prolonged. And since this process is almost always on the critical path, that would mean that airlines would need to resort to only affecting the results with traditional rules instead of a direct influence by the science in real-time. This in turn leads to falling back to performing the detailed assessment on an already produced plan, after-the-fact. This allows for only modifying just a few worst cases - thus losing out on the opportunity of making any major impact to the overall risk. This is why fatigue model performance is key.
A good explanation indeed, and in somewhat simple terms - thank you Arvid. Please find more details on this topic in a PDF found here, called ''The Secret Behind Proactive Risk Reduction".
Comments