Erteleme ve envanter paylaşımı yoluyla insani ön konumlandırmada verimliliğin artırılması

Prepositioning is a critical disaster preparedness mechanism for humanitarian organizations (HOs) but requires significant investment. Improving the efficiency of prepositioned stocks is a primary concern within the humanitarian community. This thesis, conducted in collaboration with the Emergency Supply Prepositioning Strategy (ESUPS) Working Group, examines the impact of implementing postponement and stock-sharing strategies in regional and country-level humanitarian warehouses. We focus on a setting where multiple HOs preposition supplies within the same (regional and/or country) warehouse to serve disaster-affected countries. Traditionally, these prepositioned supplies are branded with the respective HOs' logos, hindering the sharing of surplus stock during disaster response. Our proposed system defers the branding process for a portion of the stockpile in the regional warehouse until after a disaster, facilitating the sharing of unbranded stock among HOs. We limit stocks to branded form at country warehouses for faster response. This thesis aims to achieve two primary objectives: (i) evaluate the impact of proposed collaboration strategies based on postponement and stock sharing on the post-disaster response performance of HOs, and (ii) devise effective methods for implementing these strategies. To accomplish these goals, we develop two analytical frameworks, each forming a chapter of the thesis. Together, these frameworks provide a comprehensive approach for enhancing the efficiency and effectiveness of disaster preparedness efforts in humanitarian operations management. In Chapter 2, we focus on a setting where HOs in a single regional warehouse serve a specific region (e.g., the Caribbean), with each HO responding to disasters within its designated area in this region. To assesses the effectiveness of postponement and stock sharing, we develop a two-phase inventory allocation model for distributing branded and unbranded stocks to disaster-affected countries and for sharing unbranded stocks among HOs. We incorporate this model into a Monte Carlo simulation algorithm that accounts for uncertainties in disaster occurrence and impact. In Chapter 3, we consider a setting where HOs in both regional and country warehouses respond to disasters within their designated areas in a specific region. To operationalize postponement and stock sharing, we develop a two-stage stochastic optimization model that minimizes expected response time weighted by disaster severity under fixed total base stock constraints. This model optimizes the levels of branded and unbranded items for each HO across warehouses, facilitates stock sharing in the regional warehouse, and devises optimal stock mobilization strategies to affected countries. We apply the approaches developed in this thesis to case studies on hurricane preparedness and response in the Caribbean, using data from ESUPS, its partner HOs, and the United Nations Humanitarian Response Depot (UNHRD) Panama warehouse. Our extensive numerical analyses generate practical insights and examine the sensitivity of our results to various modeling choices and parameters. The simulation study in Chapter 2 reveals a U-shaped relationship in response time as the postponement rate increases, with consistent improvements in fill rate and inventory utilization. The optimization model in Chapter 3 shows substantial key performance indicator (KPI) improvements through postponement and stock sharing and provides insights into the impact of disaster severity levels and base stock levels on network design and distribution strategies. Overall, this thesis demonstrates significant benefits of postponement and stock-sharing strategies in improving the efficiency and effectiveness of prepositioning efforts. Our findings offer valuable insights for improving decision-making in disaster preparedness within humanitarian networks.