Managing Perishables with Stochastic Supply

While festivals bring a reason to cheer for everyone, businesses dealing with a spike in demand for perishables may have to live with the misery of lost sales and/or expired items. In the case of the dairy industry that deals with liquid milk, both raw material, and finished goods are perishable, which implies that merely stockpiling inventory of either item, without paying attention to potential inventory losses, cannot be an optimal strategy. In developing countries, the supplier base for perishables like milk, fruits, vegetables, flowers, etc. mostly comprise of small farmers instead of corporate/professional agencies, thus leading to supply variability. During special occasions like festivals, as individuals set aside more of the raw material for their own consumption, we encounter a reduction in supply. Around the same time, we notice a spike in customer demand, leading to a demand-supply mismatch. Companies dealing with perishables need an analytical approach to manage this.

In this thesis, we present a framework to address this problem of intermittent demand-supply mismatch using a 3-stage stochastic optimization model. We decide on the sourcing targets, the production plans based on supply realized, and finally, the dispatch plan based on orders received. As a case study, we analyze the operations and data from a private dairy company in eastern India, to understand the research problem and the applicability of the resulting model. We notice the impact of demand spikes and supply reduction in two areas: we increase supply targets in the periods preceding the demand spike; and we increase supply targets in periods when supply is expected to decrease, while demand is as usual. When there are multiple festival days within the time series, the compounding of impact depends on the sequencing of the events. Finally, when we introduce the realistic constraint that the supply target needs to be constant throughout the time series, we see a degradation in the profitability, as we need to tradeoff between lost sales and wasted products. While the focus of this case study is the dairy industry, the conclusions from this research are broadly applicable to other industries dealing with perishables.