Hi Åsa, I heard from Karolin you are deeply involved in the development of 'rosterability' functionality in the Jeppesen Crew Pairing solution. Can you explain what the problem is that you are trying to solve?
-Of course! Airlines’ crew planning processes are, by necessity, divided into two distinct steps: crew pairing construction and roster construction. This approach makes a lot of sense—it helps manage the combinatorial explosion (read more here) and aligns with the goal to keep crews together over a sequence of flights. It also provides regularity, which is beneficial both for crews and for managing disruptions closer to the day of operation.
Building individual rosters for the entire workforce “from scratch” simply isn’t practical—there are better ways to use that computational power. However, there are cases where certain groups of crew need special handling, requiring pairings that are closely tailored to their specific availability and needs.
For example, in a small crew base, detailed knowledge of crew availability might be crucial to creating rosterable pairings—pairings that are fine-tuned to work well in the roster construction phase. This is where our new rosterability feature makes a real difference.
Before we dive into how rosterability works, how have airlines historically dealt with this problem?
-Airlines have addressed this challenge in various ways, but none are without compromises. For instance, in the case of a small crew base, some airlines plan in sequence by first creating a surplus of pairings for that base and rostering them before returning to adjust and re-build pairings for other bases. Others rely on creating mostly single-day pairings, using detailed daily base constraints to match individual crew availability.
Another approach involves building longer trip sequences that include layovers at the home base. These sequences are later broken up before the rostering phase—to know that the pieces will fit. Some airlines create what we call “base-variant pairings,” which are built such that they can efficiently be reassigned between different bases. Alternatively, airlines might take a more manual approach, building pairings and rosters manually before the main optimization process begins.
Each of these methods has its downsides, such as added lead time, increased manual effort, and often sub-optimal solutions. They can work in a pinch, but they’re far from ideal.
I see. So how do you see that this can be better addressed now?
-The Jeppesen Crew Pairing solution now includes additional functionality to address this more effectively. When constructing crew pairings, the optimizer can now incorporate skeleton rosters and rostering rules directly as input.
This goes well beyond simple daily base constraints. It allows the optimizer to account for all essential roster rules during the pairing construction phase, ensuring that pairings are aligned with crew availability and other needs. This means we can create pairings that are already optimized for rostering—reducing manual adjustments and increasing confidence in the results. The pairings are fine-tuned to the true crew availability and needs.
What’s more, this is achieved in a single optimization run. Combined with our other advanced features, this not only improves the quality of the pairings but also shortens the planning process significantly and reduces manual effort.
That sounds like a game-changer. Finally, I heard a rumor that this new capability might be babtized TrueTune. Any comment on that?
-Sorry, I can’t confirm or deny that rumor! But I will say, TrueTune does resonate with me—it captures the essence of what we’re aiming to achieve.
Thank you for your time, Åsa. Tune away to the true needs!
Comments