Document Type


Publication Date




Flow shop scheduling deals with the determination of the optimal sequence of jobs processing on machines in a fixed order with the main objective consisting of minimizing the completion time of all jobs (makespan). This type of scheduling problem appears in many industrial and production planning applications. This study proposes a new bi-objective mixed-integer programming model for solving the synchronous flow shop scheduling problems with completion time. The objective functions are the total makespan and the sum of tardiness and earliness cost of blocks. At the same time, jobs are moved among machines through a synchronous transportation system with synchronized processing cycles. In each cycle, the existing jobs begin simultaneously, each on one of the machines, and after completion, wait until the last job is completed. Subsequently, all the jobs are moved concurrently to the next machine. Four algorithms, including non-dominated sorting genetic algorithm (NSGA II), multi-objective simulated annealing (MOSA), multi-objective particle swarm optimization (MOPSO), and multi-objective hybrid vibration-damping optimization (MOHVDO), are used to find a near-optimal solution for this NP-hard problem. In particular, the proposed hybrid VDO algorithm is based on the imperialist competitive algorithm (ICA) and the integration of a neighborhood creation technique. MOHVDO and MOSA show the best performance among the other algorithms regarding objective functions and CPU Time, respectively. Thus, the results from running small-scale and medium-scale problems in MOHVDO and MOSA are compared with the solutions obtained from the epsilon-constraint method. In particular, the error percentage of MOHVDO’s objective functions is less than 2% compared to the epsilon-constraint method for all solved problems. Besides the specific results obtained in terms of performance and, hence, practical applicability, the proposed approach fills a considerable gap in the literature. Indeed, even though variants of the aforementioned meta-heuristic algorithms have been largely introduced in multi-objective environments, a simultaneous implementation of these algorithms as well as a compared study of their performance when solving flow shop scheduling problems has been so far overlooked.




This article is the authors' final published version in Machine Learning with Applications, Volume 11, March 2023, Article number 100445.

The published version is available at Copyright © Tavana et al.

Included in

Business Commons