Integrating operations and marketing decisions to manage perishability risks with target minimum remaining shelf-life available to consumers

Abstract We study an aggregate level planning problem at the bottleneck packaging stage of a Make-and-Pack production system of perishable products and integrate sequencing decisions. The perishability risks are accounted for by producing to order with remaining shelf-life (RSL) dependent demand and age-dependent expected product holding cost till dispatch. Products dispatch time determines the Maximum-RSL delivered. Unlike past studies, we examine a target Minimum-RSL that marketing aims to provide. The product demand decreases from its Maximum-RSL to the Minimum-RSL. The problem is modeled as a two-stage stochastic 0-1 integer non-linear programming problem. The first-stage decisions are the inventory of intermediate products, packaging sequence, capacity allocation, Maximum-RSL, and Minimum-RSL. In the second stage, market demand is realized and has a stochastic error component. The stochastic problem is converted into its deterministic equivalent using the Sample-Average-Approximation method. Representing the demand scenarios using Low Discrepancy Sequences and reducing variance using the Antithetic Variates technique provide stable solutions. Sensitivity analysis demonstrates that the integrated-firm improves freshness criteria (Maximum-RSL and Minimum-RSL) by reducing setup time/demand variability/demand satisfaction requirement (shortage cost) or increasing importance of customer service in product freshness (product holding cost). Maximum-RSL increases with increased packaging rate. High demand variability, shortage cost, product holding cost, and low packaging rate promote integrated sequencing-capacity allocation. An efficient heuristic based on the Generalized Benders’ Decomposition method is proposed to solve large-size problems. The heuristic provides a better solution than the DICOPT solver, solves far fewer non-linear programming problems than the Simple Branch-and-Bound solver, with less computational effort.