Date of Graduation

7-2021

Document Type

Thesis

Degree Name

Master of Science in Industrial Engineering (MSIE)

Degree Level

Graduate

Department

Industrial Engineering

Advisor/Mentor

Nurre Pinkley, Sarah G.

Committee Member

Eksigolu, Sandra D.

Second Committee Member

Proano, Ruben

Keywords

Drones; Optimization; Vaccines

Abstract

Despite tremendous efforts from governments and humanitarian organizations, millions of children in low- and low-middle-income countries (LICs and LMICs) are still excluded from the benefits of immunization. The vaccine distribution in LICs and LMICs is challenging for several reasons, such as limited cold chain capacities, vaccine wastage, uncertain demand, and lack of access to immunization services. A promising avenue to address these issues is the utilization of drones for vaccine delivery. Drones can fly at high speed on direct paths and could enable on-demand deliveries to mitigate limited storage capacities. Further, their independence of road networks could allow them reaching unserved rural areas. Despite initial implementations of drone delivery networks in the field, the literature on the benefits and design of such networks is limited. Thus, in this work, we develop a two-stage stochastic supply chain model to design a drone delivery network for vaccines that maximize the number of fully and partially immunized children by maximizing vaccine availability. The first stage considers the establishment of a drone fleet and drone hubs. The second stage, after demand uncertainty reveals, considers the operational decisions including inventory and transportation for the distribution of vaccines. Thereby, our model captures unique challenges inherent to the distribution of vaccines in LICs and LMICs, such as limited cold chain capacities, vaccine wastage, uncertain demand, and limited access to services. To capture the accessibility to immunization services, we consider the spatial distribution of locations and model the access of a community to immunization services as a function of the distance to the closest point of vaccination. To solve the problem, we deal with the uncertainty using sample average approximation and use Bender's decomposition for solving. Furthermore, we propose a preprocessing method to reduce the large problem size that arise for real-life datasets. As a case study, we test the proposed model for Niger's vaccine supply chain to examine the benefits of vaccine drone delivery. Through a rigorous experimentation we provide managerial insights to help decision-makers to improve vaccination rates.

Share

COinS