A first MILP model for the parameterization of Demand-Driven MRP

Abstract

Demand-Driven Material Requirements Planning (DDMRP) is a recent production planning method firstly introduced in 2011. The method is often described as a Push-And-Pull strategy. Indeed, the method combines both real-time observation of stock and costumers demands (Pull) and anticipation of future demands (Push). To prevent against shortages caused by peaks in demand, DDMRP introduces buffer stocks that are replenished when the flow falls below a given level. DDMRP has proven its competitiveness compared to classical production planning methods like MRP or Kanban in terms of service level and inventory costs. However, it has many parameters to be fixed by the manager. Inappropriate parameterization of DDMRP can lead to poor performance of the method. For this reason, efficient methods to compute the optimal parameters can be of crucial importance. In this paper, three parameters intervening in the sizing and replenishment of buffer stocks are considered. We suggest a first exact optimization method for the parameterization of DDMRP. The average on-hand stock is minimized while the service level, measured by the percentage of orders delivered on-time to costumers (OTD: On-Time Delivery), is forced to 100% by means of a constraint. The suggested MILP (Mixed Integer Linear Programming) approach is tested thanks to a commercial solver (CPLEX). A set of 24 data instances spanning over a planning horizon of 100 days has been generated. For all these instances, optimal solutions are found within a few seconds. The suggested approach can be used as a decision-support tool that helps the manager fixing the parameters of DDMRP