Campaign planning under sequence dependent family setups and co-production in process industry

We investigate tactical level production planning problem in process industries, with float glass manufacturing being the specific application domain. Process industries are cost intensive, and as a result, efficient usage of capacity through planning is necessary. In the presence of high sequence dependent family setup costs, the need for planning production in batches, or campaigns as named in the float glass industry, arises. Campaign planning is determining timing and duration of each product family, which translates into setups. Moreover, availability of input data in different resolution, i.e. setup times in continuous time whereas customer demand forecast are available in discrete time, increases the complexity. Co-production is a phenomenon that exists in several industries including float glass manufacturing. Usually due to some special characteristic of the manufacturing process some products need to be produced by necessity. This is another challenge for efficient capacity usage as well as inventory management. We study the problem for different complexity levels. We start with single machine instance and develop two formulations. A novel branch-and-price algorithm is proposed for the parallel machine extension. Finally, we extend the problem to multiple product hierarchy levels and network structure including customer locations. We demonstrate the efficiency of our methods through extensive numerical experiments as well as some further tests to analyze the sensitivity of the cost components.