Workers-constrained shutdown maintenance scheduling with skills flexibility: Models and solution algorithms

ERTEM M., As'ad R., Awad M., Al-Bar A.

COMPUTERS & INDUSTRIAL ENGINEERING, vol.172, 2022 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 172
  • Publication Date: 2022
  • Doi Number: 10.1016/j.cie.2022.108575
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, ABI/INFORM, Aerospace Database, Applied Science & Technology Source, Business Source Elite, Business Source Premier, Communication Abstracts, Computer & Applied Sciences, INSPEC, Metadex, DIALNET, Civil Engineering Abstracts
  • Keywords: Maintenance scheduling, Workers -constrained preventive maintenance, Skills flexibility, Mathematical modeling, Constructive heuristics, RANDOMIZED ADAPTIVE SEARCH, OPTIMIZATION ALGORITHM, DECISION-SUPPORT, PLANNING-MODEL, CLASSIFICATION, HEURISTICS
  • Eskisehir Osmangazi University Affiliated: Yes


Shutdown maintenance (SM) projects are typically of a complex nature, have a predetermined time window, and require large number of limited resources such as multi-skill workers among many others. This paper devises two mixed-integer mathematical models along with efficient solution algorithms for the flexible multi-skill resource -constrained scheduling problem. Given the limited availability of multi-skill maintenance crew and the pressing need toward minimizing SM completion time, the proposed optimization models allow for the possibility of having a higher skilled worker perform a task requiring lower skill levels, if needed. Given the NP-hard nature of this problem, two constructive heuristics are developed, with the second adopting a multi-start approach for an improved exploration of the feasible solution space. The quality of the two heuristics is assessed using 60 randomly generated test instances of various sizes and complexities. The practical relevance of the problem is also illustrated via an industrial case study drawn from the cement industry. Numerical results suggest an average reduction of 16.52% in shutdown completion time when flexible workers are better utilized. The results also demonstrate the efficiency of the heuristics as they quickly yield solutions that are within average deviations of 1.24% and 2.71%, respectively, from the optimally solved instances. Upon developing efficient lower bounds, Heuristic-2 yields solutions that are on average 0.22% and 4.1% away from the lower bound for large instances involving 100 and 200 tasks, respectively, where optimal solutions for such instances are not attainable within reasonable computational time.