Efficient generation of rotating workforce schedules

• Published in: Discrete Applied Mathematics Vol. 118; no. 1; pp. 85 - 98
• Main Authors: Musliu, Nysret, Gärtner, Johannes, Slany, Wolfgang
• Format: Journal Article Conference Proceeding
• Language: English
• Published: Lausanne Elsevier B.V 01.09.2002; Amsterdam Elsevier; New York, NY
• Subjects: Backtracking; Constraint satisfaction; Exact sciences and technology; Mathematics; Sciences and techniques of general use; Search; Workforce scheduling; Constraint satisfaction; Backtracking; Workforce scheduling; Search
• ISSN: 0166-218X; 1872-6771
• DOI: 10.1016/S0166-218X(01)00258-X

Generating high-quality schedules for a rotating workforce is a critical task in all situations where a certain staffing level must be guaranteed, such as in industrial plants or police departments. Results from ergonomics (BEST, Guidelines for shiftworkers, Bulletin of European Time Studies No. 3, European Foundation for the Improvement of Living and Working Conditions, 1991) indicate that rotating workforce schedules have a profound impact on the health and satisfaction of employees as well as on their performance at work. Moreover, rotating workforce schedules must satisfy legal requirements and should also meet the objectives of the employing organization. In this paper, our description of a solution to this problem is being stated. One of the basic design decisions was to aim at high-quality schedules for realistically sized problems obtained rather quickly, while maintaining human control. The interaction between the decision-maker and the algorithm therefore consists of four steps: (1) choosing a set of lengths of work blocks (a work block is a sequence of consecutive days of work), (2) choosing a particular sequence of blocks of work and days-off blocks amongst these that have optimal weekend characteristics, (3) enumerating possible shift sequences for the chosen work blocks subject to shift change constraints and bounds on sequences of shifts, and (4) assignment of shift sequences to work blocks while fulfilling the staffing requirements. The combination of constraint satisfaction and problem-oriented intelligent backtracking algorithms in each of the four steps allows for finding good solutions for real-world problems in acceptable time. Computational results from a benchmark example found in the literature confirmed the viability of our approach. The algorithms have been implemented in commercial shift scheduling software.