This paper addresses the stochastic permutation flow shop problem (SPFSP) in which the stochastic parameters are the processing times. This allows the modeling of setups and machine breakdowns. | A simheuristic for bi-objective stochastic permutation flow shop scheduling problem Journal of Project Management 4 2019 57 80 Contents lists available at GrowingScience Journal of Project Management homepage A simheuristic for bi-objective stochastic permutation flow shop scheduling problem Eliana María González-Neiraa and Jairo Rafael Montoya-Torresb a Department of Industrial Engineering Faculty of Engineering Pontificia Universidad Javeriana Carrera 7 No. 40-62 Bogotá . Colombia b Research Group in Logistics Systems Faculty of Engineering University of La Sabana. Km 7 north highway of Bogotá . Chía Cundina- marca Colombia CHRONICLE ABSTRACT Article history This paper addresses the stochastic permutation flow shop problem SPFSP in which the Received October 14 2018 stochastic parameters are the processing times. This allows the modeling of setups and ma- Received in revised format No- chine breakdowns. Likewise it is proposed a multi-objective greedy randomized adaptive vember 19 2018 search procedure GRASP coupled with Monte-Carlo Simulation to obtain expected Accepted January 7 2019 Available online makespan and expected tardiness. To manage the bi-objective function a sequential com- January 8 2019 bined method is considered in the construction phase of the meta-heuristic. Moreover the Keywords local Search combines 2-optimal interchanges with a Pareto Archived Evolution Strategy Stochastic permutation flow shop PAES to obtain the Pareto front. Also some Taillard benchmark instances of deterministic Bi-objective permutation flow shop problem were adapted in order to include the variation in processing GRASP times. Accordingly two coefficients of variation CVs were tested one depending on ex- Tardiness pected processing times values defined as twice the expected processing time of a job and Makespan a fixed value of . Thus the computational results on benchmark instances show that the variable CV provided lower values of the
