Heuristic lot size scheduling on unrelated parallel machines with applications in the textile industry

In this paper, we present an industrial problem found in a company that produces acrylic fibres to be used by the textile industry. The problem is a particular case of the discrete lot sizing and scheduling problem (DLSP). In this problem, lots of similar products must be generated and sequenced in ten unrelated parallel machines, in order to minimize tool changeovers and the quantity of fibre delivered after the required due date. The company problem is original because a changeover can occur between two lots of the same product due to tool wear. We analyse the problem in detail and present an adaptation of a heuristic found in the literature to solve it. Results obtained with the proposed heuristic are compared with results that used to be obtained by the production planner, using historical data.