This study compares inventory system management methods for non-perishable retail goods considering the lost sales case. Several methods, based upon models with different assumptions, are tested in environments with characteristics defined on the basis of demand pattern, relations among inventory costs and others. The analysis was made using a simulation software that evaluates the system performance at each period of time in a given time span. The software contains procedures for demand forecasting based upon exponential smoothing, for physical and economic accounting and for determination of the parameters of each management method. The methods based upon periodic review models and probabilistic demand outperformed, respectively, continuous review and deterministic demand models. Besides, the methods that adopt maximun lead-times outperformed those that adopt mean lead-times. The comparative performances of the methods were not affected by changes in demand randomness. The methods that forecast the demand of various periods ahead outperformed the others when there are peaks in demand, but in environments with other demand patterns (steady, increasing or decreasing demand, without seasonality or with smooth seasonality in the short time) similar results were yielded by these methods and those based upon periodic review and steady demand.
inventory; simulation; materials management; retail; stock systems
