An integrated epidemiological supply chain management approach for the COVID-19 vaccine distribution and allocation problem

Abstract

This thesis proposes an integrated model for vaccine distribution and allocation that combines an SEIR compartmental model with a transhipment model. Given the NP-hardness of the problem, eight solution methods are explored and evaluated in terms of solution quality and run time. The best-performing method was found to be a greedy marginal benefit heuristic combined with a genetic algorithm while a logic-based Benders Decomposition approach provided provably optimal solutions to homogeneous vaccine allocation problems, albeit in prohibitively long solve times. Applied to a case study based on data from Ontario during the COVID-19 pandemic, we identify the best practices for allocating and distributing a limited supply of vaccines to minimize both total cases and deaths. The best identified vaccination strategies were able to reduce the total number of cases by 25% compared to a Pro-rata allocation of vaccines while saving $6,800,000 in vaccine acquisition and transportation costs.