Analysis of propagation of stochastic errors through an MRP system, and an industrial case study

This work arose out of a case study in production planning in a switchgear factory. The aspect considered here concerns variability of load on a component manufacturing shop which happened to be a brazing shop. The component shop manager wanted to know what variability in his shop load could be attributed to propagation of sales forecasting errors, assembly production errors and component production errors, through the Materials Requirements Planning system which was the basic mechanism loading his shop. Knowing this would help him to determine whether some of the load variability faced by his shop was due to any other causes in the production planning system; such as manual adjusting of loads to match shop capacities, subjective decisions on lot sizing, lack of coordination between production planning and materials procurement etc. The primary interest here is in showing how exactly deviations in stocks and in monthly requirements in the relevant MRP system can be represented as moving sums of the basic monthly stochastic errors impinging on the system at all levels. In obtaining such a linear representation of the system capacity limitations and stock-outs are not considered; i.e. capacities and base stocks are assumed sufficiently large. As the production shop made a large number of components, with essentially independent demands, it was assumed that the requirement modifications on account of minor capacity problems would not overwhelm the basic effects of propagation of errors through the MRP system. A very limited amount of data was available for the case study. However, it did indicate scope for improvement. Towards the end this is briefly indicated.