Loading...

Data-Driven Blood Supply Chain Network Design under Uncertainty: Robust Bi-Objective Optimization

Zarei, Mohammad Hassan | 2025

58 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 58638 (01)
  4. University: Sharif University of Technology
  5. Department: Mathematical Sciences
  6. Advisor(s): Varmazyar, Mohsen
  7. Abstract:
  8. The blood supply chain faces substantial challenges due to fluctuations in donation and demand, as well as the perishability of blood products. Consequently, decision makers require an integrated analytical framework to ensure resilient and efficient operations. This study develops a multi-product, bi-objective robust optimization model for designing a blood supply chain capable of withstanding demand variability. From the perspective of blood bank managers, the model simultaneously minimizes operational costs and transportation time while satisfying hospital demand. The decision scope includes locating and configuring permanent and temporary facilities, determining shipment policies, and managing substitution among compatible blood groups. The model is applied to a case study of Tehran Province. In the forecasting phase, after data cleaning and feature engineering, several machine learning algorithms were evaluated. Owing to its superior accuracy and stability, the CatBoost model was selected as the final demand forecasting method. In the optimization phase, the bi-objective problem was converted into a single-objective one using the epsilon-constraint method, and solved in CPLEX under real, predicted, and simulated datasets. Results indicate that the forecasts exhibit higher consistency with real data compared to simulated series, enabling meaningful scenario analysis and identification of efficient operating points. At the network design layer, the optimal configuration of permanent centers is determined; at the operational layer, blood group O emerges as the core driver of the system and requires safety stock. Policy comparison shows that under normal conditions, non-substitution despite increased time and cost offers lower wastage and is therefore preferred. Under disruptive or crisis scenarios, however, substitution becomes advantageous for improving service levels. The Pareto frontier analysis reveals practical trade-offs among cost, service, and time, offering actionable insights for designing targeted blood donation campaigns and improving system preparedness
  9. Keywords:
  10. Blood Supply Chain ; Network Design ; Demand Forecasting ; Robust Optimization ; Crises ; Augmented Epsilon-Constraint

 Digital Object List

 Bookmark

...see more