EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

2524. Exact methods and (meta-)heuristic solution methods for the Hybrid Flexible Flowshop problem

Invited abstract in session WB-49: Flow shop scheduling problems, stream Lot Sizing, Lot Scheduling and Production Planning.

Wednesday, 10:30-12:00
Room: M1 (building: 101)

Authors (first author is the speaker)

1. Stavros Vatikiotis
Department of Management Science and Technology, Athens University of Economics and Business
2. Ilias Mpourdakos
MANAGEMENT SCIENCE AND TECHNOLOGY, ATHENS UNIVERSITY OF ECONOMICS AND BUSINESS
3. Dimitrios Papathanasiou
Department of Management Science and Technology, Athens University of Economics and Business
4. Yiannis Mourtos
Management Science & Technology, Athens University of Economics & Business

Abstract

Modern manufacturing keeps on introducing research challenges in terms of both problem structure and instance size. The Hybrid Flexible Flowshop (HFFS) represents a setting where a set of jobs needs to be processed on a set of sequential stages, each containing a different number of parallel (and typically identical) machines; each job will be processed at one machine at a time and it is possible to skip some stages. Additional structure is introduced by i) machine-dependent transportation times between the consecutive stages, ii) limited capacity buffers on the entry and on the exit of either each machine or each stage and, iii) two different types of renewable resources - the first responsible for the transportation of jobs between stages and the second affecting job processing in specific machines. To handle the above on instances of several hundred of jobs, while sustaining the ability to calculate optimality gaps, we examine both exact methods (MILP and CP) and (meta-)heuristic algorithms. Since exact methods commonly struggle with large instances, we examine also whether MILP-based and CP-based formulations can solve parts of the problem or sequentially handle parts of the instance. These components are combined with constructive heuristics and evolutionary schemes to offer a method versatile in the sense of offering multiple solutions for diverse time horizons. The results of the methods are compared in computational terms on both benchmark and real-life instances.

Keywords

Status: accepted

Back to the list of papers