Production scheduling in the tobacco processing industry – a disaggregation approach

In this paper a disaggregation approach to solving a production scheduling problem in a tobacco processing plant is outlined which allows the multiple objectives, as specified by the plant's management, to be satisfied without the generation of massive mathematical models that would defy solution. The modelling requires the use of daily and monthly time frames. The forecast monthly demand for the year ahead for each product is obtained from the separate forecasting process and placed into the monthly production scheduling model. By solving this month-based model (a staircase formulation with the monthly models linked together by stock constraints for a horizon of a year), a master monthly production schedule is developed. If there does not exist a feasible solution to this problem, then the forecast demand is not able to be met and reductions to appropriate monthly demands for specific products (determined from the analysis of the solution) must be undertaken and the problem re-solved. Having established the master production schedule, the daily production schedule is then determined by using the master monthly production schedules as targets. The time horizon for the monthly model is normally a year while the daily model has a horizon of a month. The daily model has the added complication that it must account for possible reduced machine capacity inflicted by daily scheduling considerations and also ensure that work-in-progress is handled appropriately. Both models are solved using mixed integer linear programming and provide a rather straightforward resolution to the problem facilitating realistic implementation.