A multi-stage stochastic programming approach for blood supply chain planning

Perishability of blood products and uncertainty in donation and demand sizes complicate the blood supply chain planning. This paper presents a novel bi-objective mixed-integer model for integrated collection, production/screening, distribution and routing planning of blood products, and seeks to simultaneously optimize the total cost and freshness of transported blood products to hospitals. To cope with inherent uncertainty of input data, a multi-stage stochastic programming approach with a combined scenario tree is presented. Due to the high complexity of the problem, a novel hybrid multi-objective self-adaptive differential evolution algorithm is developed, which benefits from the variable neighborhood search with fuzzy dominance sorting (hereafter it is briefly called MSDV). MSDV is validated through comparing its performance with two of the most common multi-objective evolutionary algorithms (i.e. MOICA and NSGA-II). Applicability of the proposed decision model is also tested through a real case study. Our results show that the solution efficiency of a network can be balanced with its effectiveness through customer satisfaction. Further, several sensitivity analyses are carried out to provide valuable managerial insights.