| Rolling horizon procedures, where an infinite horizon problem is approximated by the solution to a sequence of finite horizon problems, are common in production planning practice and research. However, changes in planned quantities, referred to as planned changes, occur in each period as new updated demand information becomes available, which may disrupt supporting activities that have been initiated based on plans developed in earlier periods. We examine the planned changes of production plans in a rolling horizon environment with Clearing Ftmctions (CF) by examining the average total realized costs over the entire horizon when the production plan is implemented. We try to reduce the total planned release changes using two different Chance Constraint ( CC) formulations that can represent several production strategies previously proposed to make production more stable, i.e. extending the planning horizon by forecasting, setting ending conditions and safety stock.;In the first part of the study, we compare different chance constraint formulations with a deterministic LP model on the issue of total release changes reduction. Results show that including IP Chance Constraints reduces the total planned release changes significantly under most scenarios while maintaining total realized costs and achieving target service level comparable to other production strategies. SF chance constraint formulations also help to reduce overall release changes but only when utilization is high.;In the second part, we introduce change costs into the objective function in order to freeze the schedule. Firstly, we derive bounds on the values of the unit change costs that will guarantee freezing of the schedule in a single-stage single-product system with fixed lead time and limited capacity without congestion. We find that positive and negative release changes have different origins and require different responses. Thus, we need to introduce different costs for positive and negative release changes. We then extend the analysis to multiple products model in a single stage with different lead times and limited capacity without congestion. We find that when two products compete for the same capacity, we have the problem of capacity switching. Thus, the negative change costs to freeze schedule derived from single product analysis do not work. We need to increase the negative change costs since freezing schedule for one product increases the chance of another product. Furthermore, when multiple products use the same capacity, we need to set the change cost by considering all the unfrozen products to guarantee elimination of negative changes. Numerical results show that setting release change costs to freeze schedule based on our analysis, we can successfully freeze schedule by introducing change costs on the objective function. |
