Influence of Forecast Error and Forecast Bias on Safety Stock on a MRP System with Rolling Horizon Forecast Updates

This paper investigates the effects of rolling horizon forecast updates on a production system relying on material requirements planning (MRP). The underlying demand model is the MMFE (martingale model of forecast evolution) model extended by forecast biases revealed certain periods before delivery, i.e. information quality is not strictly increasing as assumed in MMFE. Simulation is applied to model the MRP planning method and the shop floor behavior of a two stage production system including a two level bill-of-materials with 8 finished goods and 4 semi-finished materials. Several scenarios on the demand model parameterization are tested and a finite solution space for the MRP planning parameter safety stock is enumerated to minimize overall costs. In this numerical study, preliminary results to identify the influence of forecast uncertainty on MRP planning parameter safety stock are identified when rolling horizon forecast updates occur.