A heuristic method for perishable inventory management under non-stationary demand

Our study considers a perishable inventory system under a finite planning horizon, periodic review, non-stationary stochastic demand, zero lead time, FIFO (first in, first out) issuing policy, and a fixed shelf life. The inventory system has a fixed setup cost and linear ordering, holding, penalty, and outdating costs per item. We introduce a computationally-efficient heuristic which formulates the problem as a network graph, and then calculates the shortest path in a recursive way and by keeping the average total cost per period at minimum. The heuristic firstly determines the replenishment periods and cycles using the deterministic-equivalent shortest path approach. Taking the replenishment plan constructed in the first step as an input, it calculates the order quantities with respect to the observed inventory states as a second step. We conduct numerical experiments for various scenarios and parameters, and compare them to the optimal stochastic dynamic programming (SDP) results. Our experiments conclude that the computation time is reduced significantly, and the average optimality gap between the expected total cost and the optimal cost is 1.87%.