Volume 52, Number 4, October–December 2018
|Page(s)||1351 - 1376|
|Published online||06 December 2018|
A genetic algorithm for the steel continuous casting with inter-sequence dependent setups and dedicated machines
Université Le Havre Normandie, LMAH, FR CNRS 3335, ISCN,
Le Havre, France.
2 Audencia Business School, 8 route de la Jonelière, 44312 Nantes, France.
* Corresponding author: firstname.lastname@example.org
Accepted: 11 March 2018
The steel continuous casting planning and scheduling problem namely SCC is a particular hybrid (flexible) flowshop that includes stages: (i) the converters (CV), (ii) the refining stands (RS) and (iii) the continuous casting (CC) stages. In this paper we study the SCC with inter-sequence dependent setups and dedicated machines at the last stage. The batch sequences are assumed to be pre-determined for one of the CC devices with a non preemptive scheduling process. The aim is to schedule the batches for each CC machine including the times setup between two successive sequences. We model the problem as a MILP where the objective is to minimize the makespan Cmax that we formulate as the largest completion time taking account of the setup times for each CC. Then, we propose an adapted genetic algorithm that we call Regeneration GA (RGA) to solve the problem. We use a randomly generated instances of several sizes to test the model and for which we do not know an optimal solution. The method is able to solve the problems in an acceptable time for medium and large instances while a commercial solver was able to solve only small size instances.
Mathematics Subject Classification: 90B35 / 90B50 / 90C11 / 90C59
Key words: SCC / GA / scheduling / Cmax / setup
© EDP Sciences, ROADEF, SMAI 2018
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.