Airline crew scheduling from planning to operations
Section snippets
Background
The airline industry has been a rich source of operations research problems, mainly due to the combinatorial nature of the problems. Typically the planning problem involves creating lines of work referred to as rosters for aircraft and crew. The crew rosters consist of activities that can be flying, ground duties, training, free days, and personal activities. The crew planning process usually decomposes into two processes.
In the first process, known as pairing, the (anonymous) flying activities
The pairing problem
The pairing problem deals with the construction of itineraries consisting of flights such that all flights of a given schedule are completely covered. A flight is completely covered when all of its basic crew resource requirements are met with sufficiently many qualified crew members. A basic flight requirement is modeled with a crew-need vector of the kind [F1/F2/F3//C1/C2/C3/…], where // are used for separating the main categories of cockpit and cabin, and a single / for separating the number
The crew recovery problem
Crew recovery is time critical re-planning subject to minimal change, both in the pairing and in the rostering sense, as a result of a disruption occurring at any of the three levels. Given these disruptions, the objective here is to repair all illegal individual rosters and cover all existing or newly scheduled flights. For example,
- •
a flight delay makes it impossible for crew to make their connecting flight on their scheduled pairing;
- •
a flight cancellation due to aircraft unavailability causes
Results
We show results on instances for single base and multi-base problems where both generation methods were tested for a single recovery time window of 48 hours, within which all disruptions are considered, outside which rosters are kept locked. On all instances, the enumeration approach was always launched on the crew dependent network. It was implemented as a depth-first search (DFS) that was subject to search width parameters that limit the scope difference of the generated rosters with respect
Conclusions and further areas of research
The restrictive time limits imposed on roster maintenance and day of operation necessitates an alternative integrated approach for crew recovery. The crew recovery model (IP3) provides a framework for individualized feasible pairing generation, with additional flight or trip qualification constraints that model requirements imposed on groups of crew contributing directly in dictating pairing feasibility. When (IP3) is embedded within a sweeping time window that sweeps across a planning horizon,
Acknowledgements
The EU funded Descartes Integrated Operations Control project (together with British Airways and the Technical University of Denmark) and the Carmen Roster Maintenance project (in close collaboration with clients) initiated the research. Our special thanks go to Curt Hjorring and Lennart Bengtsson for their support and advice about the pairing reduced cost column generator. We thank Stefan Karisch for reviewing an earlier version of this work and giving useful suggestions, and Erik Andersson
References (12)
- et al.
Crew pairing optimization
- Tomas Gustafsson, A heuristic approach to column generation for airline crew scheduling. Licentiate Thesis, Department...
- Curt A. Hjorring, Jesper Hansen, Column generation with a rule modeling language for airline crew pairing. In:...
- et al.
Airline crew rostering: Problem types, modeling, and optimization
Annals of Operations Research
(2004) - Ladislav Lettovsky, Airline Operations Recovery: An optimization approach. Ph.D. Thesis, Georgia Institute of...
- Ernesto De Queiros Veiros Martins, Jose Luis Esteves Dos Santos, A new shortest paths ranking algorithm, Technical...
Cited by (74)
Optimizing aircrew recovery considering long connections: A column generation based approach
2023, Computers and Industrial EngineeringIndividual scheduling approach for multi-class airline cabin crew with manpower requirement heterogeneity
2022, Transportation Research Part E: Logistics and Transportation ReviewEnabling integration and interaction for decentralized artificial intelligence in airline disruption management
2022, Engineering Applications of Artificial IntelligenceCitation Excerpt :In complement, the PTFM provides an evaluation of the required duration and delay at different phases of flight upon enacting a particular resolution during disruption management. As such, from an airline operations recovery perspective (Bratu and Barnhart, 2005; Medard and Sawhney, 2007; Petersen et al., 2012; Maher, 2015), the PTFM estimates the recovery impact for a specific disruption resolution conceived by a human specialist in a functional role during airline disruption management. Thus, for any functional role (i.e. intelligent agent or domain manager) operating based upon the consensus routine detailed in Fig. 6 for a private decentralized AI platform, the UTFM and PTFM conjointly define a smart contract that measures the reliability of a human-generated disruption resolution and its corresponding effect on the recovery plan for a disrupted airline route network, thereby enabling integration in a multi-agent system for airline disruption management.
Integration of turnaround and aircraft recovery to mitigate delay propagation in airline networks
2022, Computers and Operations ResearchInflight catering services - A comparison of central and decentral galleys inside the aircraft cabin, a concept-based approach
2022, Transportation Research ProcediaAirline crew scheduling: Models and algorithms
2021, Transportation Research Part E: Logistics and Transportation ReviewCitation Excerpt :Detailed discussions for each of these selected studies are presented in the following. Differently, Medard and Sawhney (2007) distinguish the two types of air crew when constructing pairings. Actually, Medard and Sawhney (2007) solve the crew scheduling problem in a single step by integrating the pairing construction and pairing assignment procedures.
- 1
Previously with Carmen Systems AB, Odinsgatan 9, Gothenburg 411 03, Sweden.