International Journal of Innovative
Computing, Information and Control ICIC International c 2010 ISSN 1349-4198
Volume 6, Number 9, September 2010 pp. 3943–3959
OPTIMIZATION OF SHORT-HAUL AIRLINE CREW PAIRING PROBLEMS USING A MULTIOBJECTIVE GENETIC ALGORITHM
Chiu-Hung Chen1, Ta-Yuan Chou2, Tung-Kuan Liu1, Jyh-Horng Chou1
and Chung-Nan Lee2,∗
1Institute of Engineering Science and Technology
National Kaohsiung First University of Science and Technology 1 University Road, Yenchao, Kaohsiung 824, Taiwan
{ u9515905; tkliu; jhchou }@ccms.nkfust.edu.tw
2Department of Computer Science and Engineering
National Sun Yat-sen University Kaohsiung 804, Taiwan [email protected]
∗Corresponding author: [email protected] Received March 2009; revised July 2009
Abstract. Airline crew pairing problems involve optimizing an overall evaluation
func-tion containing various conflicting objectives and constraints originating from cost and safety considerations. Classical approaches based on set partitioning or set covering meth-ods separate the solution into two phases, pairing generation and pairing optimization, and evaluate the cost by a weighted-sum of objective values. This paper proposes a new multiobjective evolutionary approach to improve the classical solution flow by integrat-ing the two-phase steps as a sintegrat-ingle step and reasonintegrat-ing the multiple practical objectives simultaneously.
Furthermore, this paper also examines real-life daily pairing problems in a Taiwanese short-haul airline as case studies. Compared to man-made pairing plans, the positive perimental results demonstrate the more appropriate and effective crew pairing plans ex-plored according to practical considerations. These considerations include objectives such as duty connection, transition time, layover, pairing number, aircraft changing times, fly-ing time, and duty period.
Keywords: Crew pairing, Multiobjective genetic algorithm, Short-haul airline
1. Introduction. Airline crew scheduling is one of the most important operations in airline companies, since it is a major determinant of crew costs, second only to fuel costs. Crew costs easily exceed one billion US dollars annually for larger airlines [1]. Due to the potential for considerable cost savings, this topic has long been the focus of academic attention. Crew scheduling is separated into two sub-problems, crew pairing and crew rostering. Crew pairing combines the flight legs into several groups, and assigns crew members to these groups through crew rostering to obtain the final crew schedule. This paper focuses on the optimization solution for airline crew pairing problems.
Crew pairing involves optimizing an overall evaluation function, composed of various conflicting objectives and constraints originating from limitations imposed by safety reg-ulations, union contract agreements, and other complex working rules and management policies. Previous researches have discussed and surveyed the details of these criteria with regard to safety and costs [1-4]. Due to its complexity, this problem is classically modeled as a set partitioning problem or a set-covering problem solved by a two-phase approach [5]. In the first phase, or pairing generation, planners generate a group of possible pairing
