Simulation of reordering policies for perishable products under service level constraints: sensitivity analysis based on design of experiments and response surface methodology

A multi-period inventory management model for perishable products is developed with simulation approach. A single-product scenario is modeled under continuous and periodic review policies, considering Weibull demand, non-null deterministic lead time, a combination of lost sales and backorders, fixed product shelf life, and service constraints. Eight inputs, including managerial aspects and product characteristics, and six outputs, i.e. the minimum cost and the optimal operating leverages of the two policies, are evaluated using Design of Experiments. Discrete-time simulation is used to determine, for each configuration, the optimal operating leverages, namely target stock level, reorder period, and base stock, which minimize the total cost under service constraints. The outcomes were then analyzed through Analysis of Variance and Response Surface Methodology, to assess the impact of each input parameter on the responses. Results show that the optimal operating leverages mainly depend on demand features, some unitary costs (order issuing in particular), shelf life, and lost sales fraction. The shelf life turns out to be significant only if lower than a threshold value, which varies with the shape factor of the demand distribution: for shelf life values greater than the threshold, the products could be managed as non-perishable from an inventory point of view.