20 September 2001: Change Power at Iberia

You have become very adept at using Rave. Do you have a programming background?
Not really, but I've always been very interested in computers. I have attended a few Rave courses at Carmen, and spent a lot of time studying it when Carmen Crew Pairing and Carmen Crew Assignment were taken into production. I also have the considerable benefit of having been a pilot at Aviaco and Iberia for 29 years.

What is the main use of Rave at Iberia?
We have been using Rave a lot to adjust our fairness model. This model is very complex at Iberia and changes as the agreement, and interpretations of the agreement, changes. We are also constantly modifying and adding rules or quality aspects, and last month we even modified the crew composition in one fleet.

I know that you have recently been using Rave substantially. Can you tell us about it?
Towards the end of June, negotiations broke down between Iberia´s management and the pilots union SEPLA. The pilots decided to strike every Tuesday. However, Iberia wanted to maintain "a minimum operation" which meant that only 30% of flights on Tuesdays would be cancelled. To some extent, Iberia could choose which flights to cancel. With help from Carmen, we were able to quickly retune the pairing system and create more one-day pairings on Tuesdays. Altering some Rave code and using the optimization's tool´s ability to control global properties enabled this. The idea was to plan for a full operation, but always be prepared to cancel these one-day pairings later if the strike was still not resolved. By cancelling one-day pairings, both crew and aircraft would still be in the right location on Wednesday mornings, and the strike´s impact would be limited. The good news was we only needed a few days to implement this process.

Did you have to use the new process?
No, luckily we didn´t. We actually produced new pairings for July, but our existing mainframe system wasn´t flexible enough to replace the old pairing solution with the new one. Coupled with some late additional information from the CAA, we decided to put the new process on hold; and then shortly before it was time to plan for August, the Spanish government decided the conflict had to be resolved via arbitration. An arbitrator was appointed and had four days to present a solution that both Iberia and SEPLA needed to accept. During these negotiations Rave also proved to be very useful.

How?
The problem was that the proposed solution, which included changes in block time regulations and some other rule changes, were open to interpretation. For example, a new rule was added for long haul stating "maximum four pairings in a month". Should this be interpreted as four pairings overlapping the month or four pairings starting in the month and what was a month in this context - four weeks? I used Rave to change the block hour calculations and introduce all the new rules, and all possible interpretations of them. The changes were made and verified within hours and the rules compiled in ten minutes. I then ran simulations for all scenarios and within 24 hours I could show the consequences of different interpretations. These simulations were then used in selecting an interpretation that was acceptable for both SEPLA and Iberia´s management. Neither the process change, which prepared us for a dramatic change to our planning, nor the very short notice for creating a lot of exact scenarios, would have been possible with our old planning system.

Thank you and let's hope that the future will be less dramatic.
Well, it would be nice, but I don't think so. The aviation business is constantly changing. This includes, crew preferences, crew agreements, cost structures, fleet structures, planning process, etc. That's why we are trying to go with the flow of change rather than trying to stop the world from changing.