We consider the inventory control problem for quickly perishable items, such as fresh produce, under the assumption of a fixed shelf life. Daily demand is uncertain and affected by seasonality within the week.Other relevant problem features are the delivery lead time, which is also assumed fixed,the potential skewness of the daily demand distribution,as well as the profit margin and the salvage value (if any) of perished items.In a large part of the literature on perishable items, the extreme cases of FIFO and LIFO inventory issuing are considered, whereas some papers assume a fixed and known mix between the two extremes. We should adopt a model taking into account the actual mechanism of inventory issuing and depletion. If inventory issuing is under the control of an inventory manager, as is the case for blood platelets in a hospital, FIFO is a reasonable assumption under most circumstances. In other cases, consumers are allowed to select the preferred item and make their choice according to personal preferences. Some consumers may prefer items with most remaining shelf life, while others may prefer items closer to the expiration date, possibly for ecological reasons. This results in a mix of FIFO and LIFO consumers, but the exact fraction of each segment may be uncertain. Therefore, it is important to assess the impact of the ambiguity in the consumer LIFO/FIFO behavior, as well as its interaction with the other problem features.Given these aims, we apply quite simple parameterized decision policies, like base-stock or constant order, which are learned by simulation-based optimization. One reason for the adoption of such policies, even though they are not optimal for the problem, is the managerial appeal of simple rules. Alternative heuristic decision rules have been proposed, but they may require the optimization of a larger number of parameters if applied to a case featuring seasonal demand and suffer from a potential fragility in an ambiguous setting. By a similar token, in this study we do not consider more elegant approaches based on stochastic modeling or dynamic programming. Leaving computational issues aside, they may require very specific assumptions, whereas we want to address ambiguity in consumer behavior and robustness of decision rules under different market and economic settings
