Computers & Industrial Engineering 169 (2022) 108238 Available online 13 May 2022 0360-8352/© 2022 Elsevier Ltd. All rights reserved. Minimizing total completion time in the no-wait jobshop scheduling problem using a backtracking metaheuristic Kuo-Ching Ying a , Shih-Wei Lin b, c, d, * a Department of Industrial Engineering and Management, National Taipei University of Technology, Taipei 106, Taiwan b Department of Information Management, Chang Gung University, Taoyuan 333, Taiwan c Department of Emergency Medicine, Keelung Chang Gung Memorial Hospital, Keelung 241, Taiwan d Ming Chi University of Technology, New Taipei City 243, Taiwan A R T I C L E INFO Keywords: Scheduling No-wait jobshop Total completion time Metaheuristics ABSTRACT This study focused on the no-wait jobshop scheduling problem (NWJSP), which is NP-hard in the strong sense. A new benchmarking algorithm, named backtracking multi-start simulated annealing (BMSA), is presented in this study for minimizing the total completion time in NWJSPs. To effectively and effciently fnd (near-) optimal schedules, the simulated annealing algorithm is strengthened by integrating the multi-start and backtracking mechanisms that enable the search processes to escape from the local optimum. The performance of the BMSA algorithm is assessed by comparing its experimental results to those of the best available metaheuristics on three well-known benchmark problem sets. Computational results and statistical analyses confrm that the BMSA al- gorithm signifcantly outperforms these benchmark metaheuristics. This study mainly contributes to providing an innovative approach for solving this highly intractable scheduling problem, which is also worth applying to solve a practical NWJSP. 1. Introduction As one of the famous shop types, many extensions of the traditional jobshop scheduling problem (JSP) have been developed to satisfy the various conditions and requirements of real-world manufacturing sys- tems. This work addresses the JSP with an additional no-wait constraint, which usually occurs when the shop lacks storage capacity or has pro- cess technology limitations that require consecutive operations in any jobs that must be processed without waiting. The no-wait jobshop scheduling problem (NWJSP) is not only theoretically interesting but also an applicable problem with numerous industrial applications, such as in steel industries (Wismer, 1972), chemical industries (Rajendran, 1994), food industries (Gershwin, 2018; Hall & Sriskandarajah, 1996), and pharmaceutical industries (Raaymakers & Hoogeveen, 2000). For a thorough review of the real-world applications of NWJSPs, we refer the readers to the comprehensive survey presented by Hall and Sris- kandarajah (1996) and Samarghandi (2013). The frst study that imposed a no-wait constraint in JSPs was pre- sented by Kubiak (1989) who presented a pseudo-polynomial time method to solve the NWJSP with two machines. Given the applicability of large-size NWJSPs in various industries, numerous attempts have been made to tackle this highly intractable scheduling problem. Regarding the scheduling objective, minimizing makespan among all jobs has been the most studied and reported performance measure; for example, Mascis and Pacciarelli (2002) and AitZai et al. (2016) pro- posed branch-and-bound algorithms, Vermeulen et al. (2011) presented an integer linear programming model and constraint programming, and Ozolins (2020) applied dynamic programming to minimize makespan in NWJSPs. Since the complexity of this problem is NP-hard in the strong sense (Bansal et al., 2005), these exact methods can only solve very small problems to the optimums. Therefore, most researchers focus on proposing effective and effcient heuristic algorithms to achieve an acceptable balance between the computation time and the quality of the solution. According to the literature, many researchers have attempted to develop various heuristic algorithms that can provide approximate so- lutions for the NWJSP with the makespan criterion. The most famous heuristic algorithms include: Genetic Algorithm (GA) (Bozejki & Makuchowski, 2011; Brizuela et al., 2001; Samarghandi, 2019), Vari- able Neighborhood Search (VNS) algorithm (Schuster & Framinan, 2003), Fast Tabu Search (FTS) algorithm (Schuster, 2006), Complete Local Search with Memory (CLM) algorithm (Framinan & Schuster, * Corresponding author at: Department of Information Management, Chang Gung University, Taoyuan 333, Taiwan. E-mail addresses: kcying@ntut.edu.tw (K.-C. Ying), swlin@mail.cgu.edu.tw (S.-W. Lin). Contents lists available at ScienceDirect Computers & Industrial Engineering journal homepage: www.elsevier.com/locate/caie https://doi.org/10.1016/j.cie.2022.108238 Received 23 October 2021; Received in revised form 1 April 2022; Accepted 8 May 2022