Scheduling a production line using heuristics and constraint programming
Academic Article in Scopus
-
- Overview
-
- Additional document info
-
- View All
-
Overview
abstract
-
© 2017 Proceedings - 22nd International Congress on Modelling and Simulation, MODSIM 2017. All rights reserved.Sequencing and scheduling are activities that are of paramount importance in the man-ufacturing industries. In today¿s economic environment, the efficiency gains produced by effective scheduling represent not only a competitive advantage, but also are crucial to keep customers happy through adequate service levels and to ensure adequate use of resources. In this paper, we minimise the makespan in a single-machine environment with release dates and sequence-dependent setup times, or a 1|sjk, rj |Cmax scheduling problem in the notation introduced by Lawler et al. [1982]. This problem is based on a deodorant manufacturing facility where a number of products are produced, and a sequence must be determined such that the costs of switching from one family of products to another are minimised, subject to raw material availability and delivery dates. Schedules must be produced for a horizon of two weeks, which normally requires planning the production of around forty products (from a total of around three hundred) which may belong to any one of nine product families. The major component of the cost associated with changing from one family of products to another is due to cleaning the line when products from different families are produced in sequence. We present results obtained with two solution methodologies: first, we use the common strategy of producing a good initial solution via dispatching rules and then refining this solution through Tabu Search, and second, we develop a Constraint Programming model. Our results show that finely-tuned heuristic and Constraint Programming models can produce a satisfactory result with realistic data sets in a short execution time.
status
publication date
Additional document info
has global citation frequency
start page
end page