Expert Systems With Applications 165 (2021) 113837 Available online 8 August 2020 0957-4174/© 2020 Elsevier Ltd. All rights reserved. Multi-temperature simulated annealing for optimizing mixed-blocking permutation fowshop scheduling problems Shih-Wei Lin a, b, c , Chen-Yang Cheng d , Pourya Pourhejazy d , Kuo-Ching Ying d, * a Department of Information Management, Chang Gung University, Taoyuan 333, Taiwan b Department of Neurology, Linkou Chang Gung Memorial Hospital, Taoyuan 333, Taiwan c Department of Industrial Engineering and Management, Ming Chi University of Technology, New Taipei 243, Taiwan d Department of Industrial Engineering and Management, National Taipei University of Technology, Taipei 106, Taiwan A R T I C L E INFO Keywords: Scheduling Mixed-blocking fowshop Makespan Metaheuristics ABSTRACT Scheduling problems play an increasingly signifcant role in the design and optimization of highly computerized and automated production systems. Given the importance of just-in-time production in advanced manufacturing, scheduling methods should enable the users to consider various blocking situations in a zero work-in-process scheme. In this situation, Permutation Flowshop Scheduling Problem with Mixed-Blocking Constraints (MBPFSP) is a much-needed scheduling extension that allows for heterogeneous blocking criteria between successive machines. Considering the scale of integrated production systems, and the inherent complexities involved in this type of scheduling problems, effcient and robust solution algorithms are necessary to facilitate industry applications of this emerging scheduling problem. This study extends to develop an improved meta- heuristic, the Multiple Temperature Simulated Annealing (MTSA) algorithm, to provide high-quality solutions to MBPFSPs, considering makespan. Using extensive benchmark experiments, it is shown that the developed al- gorithm outperforms the state-of-the-art existing approaches applied to solve the MBPFSP. Overall, this research sets the stage for MBPFSP’s industry scale applications, narrowing the gap between the scheduling theory and practice. 1. Introduction The Permutation Flowshop Scheduling Problem (PFSP) has received wide recognition in the scheduling feld such that it is now among the most studied combinatorial optimization problem (Fernandez-Viagas et al., 2017). Belonging to the NP-Complete category, PFSPs fnd the best permutation of n jobs, to be processed by m machines, in a production system that is characterized by identical processing orders (Ruiz et al., 2019). Given industry-specifc characteristics, different extensions may be needed to apply PFSPs in real-world practices. There is a wealth of mathematical extensions to the extant PFSP that addresses practical needs and real-world production situations, i.e. through adding new constraints to the original model (Cheng et al., 2019) or considering various objective functions (Birgin et al., 2019; Ramezanian et al., 2019). Permutation fowshop scheduling problems with Blocking con- straints (BPFSP) are prime examples that help address limited, or zero, buffer space before each machine, where the work-in-process (WIP) items can proceed to the next machine only when certain conditions are met. Blocking situations can be caused by various reasons (see (Miyata & Nagano, 2019)); it arises in many industries from chemical and phar- maceutical (Hall & Sriskandarajah, 1996) to robotic cells (Ribas & Companys, 2015), and cider production (Riahi et al., 2017; Trabelsi et al., 2012). BPFSP assumes that the blocking conditions are identical all over the production line. The most prevalent blocking examples are, but not limited to: Release when Starting blocking (RSb; Levner, 1969), in which a new job i +1 on machine j can be processed once the next machine, j + 1, starts processing the current job i; Release when Completing blocking (RCb; Martinez, 2005), where the new job i +1 can proceed to machine j when the next machine, j + 1, releases the current job i for processing on machine j + 2; Release when Completing blocking* (RCb*; Trabelsi et al., 2012), where machine j can start processing job i +1 immediately after machine j +1 releases job i, regardless of its next status; and Without blocking (Wb), whenever the current job can proceed to the * Corresponding author. E-mail addresses: swlin@mail.cgu.edu.tw (S.-W. Lin), cycheng@ntut.edu.tw (C.-Y. Cheng), pourya@ntut.edu.tw (P. Pourhejazy), kcying@ntut.edu.tw (K.-C. Ying). Contents lists available at ScienceDirect Expert Systems With Applications journal homepage: www.elsevier.com/locate/eswa https://doi.org/10.1016/j.eswa.2020.113837 Received 26 February 2020; Received in revised form 10 July 2020; Accepted 1 August 2020