Airline Crew Scheduling with Random Delays

Flight crew scheduling often involves complex optimization procedures to produce crew assignments that minimize a function of all assignments. Problems frequently occur when disruptions (equipment, weather, illness, etc.) make the original schedule invalid and necessitate some form of schedule repair to ensure that flights are covered.

These disruption effects are compounded by the nature of the optimization problem in seeking extreme point results. Small changes beyond built-in buffer times on a single flight can cascade throughout a system causing multiple delays. By considering such possibilities beforehand, a schedule can adapt quickly to disruptions by isolating them within the system and minimizing ensuing complications.

A stochastic programming model can obtain this form of robustness by minimizing the sum of costs without disruptions and the expected value (perhaps with some risk aversion factor) of additional costs due to disruptions and schedule repair.

Here is a presentation on the crew scheduling problem with random delays.

Send questions to jrbirge@northwestern.edu.