Abstract
Previously scheduled production plans frequently need to be updated because of demand uncertainty. After making a comprehensive definition of nervousness which includes costs for changes in production schedule and quantity, we suggest three methodologies. Two methods are modified versions of very well-known methods: the Wagner–Whitin algorithm and the Silver–Meal heuristic. However, our definition of nervousness and its consequences for altering predetermined production volumes make the well-known property of producing either zero or a sum of several periods’ demand suboptimal. Therefore a third method, a new mixed integer linear programming formulation, is proposed which is shown to be more effective in some cases. Numerical analyses are carried out for a wide range of possible cases, through which we provide insights to the most appropriate algorithm in a parameterized space.
Scope and purpose Uncertainty in demand forecasts and a rolling horizon create volatility in lot-sizing results. This volatility is characterized by frequent changes in predetermined production schedules and is highly undesirable for production managers. It causes nervousness in the system in terms of canceling existing setups, introducing new setups, and altering the production volumes. In this paper, we propose new cost structures for these changes, and offer several models that identify less nervous production schedules in a rolling horizon basis. For practitioners, this work identifies the most preferable algorithm for a variety of system parameters.