Decomposition of demand and optimisation for a perpetual inventory system

The design of an effective inventory control system is difficult, particularly if the demand fluctuates widely and in a seasonal manner. Most traditional analytical models seem to be inadequate and over-simplified. An alternative to established mathematical models is suggested and control policies for a specific industrial situation are presented. The case of paint containers for a large paint company is considered. Demand, which is highly seasonal, has been deseasonalised by a Census-II package to produce a demand distribution pattern, used as an input to simulation exercises. Four policies have been tested and analysed. Simulation results show that the (r,Q) system with a variable ROP adjusted each month on three months’ moving average of seasonalities shows the best results.