PhD Dissertation Defense

Zahra Sadat Seyedzadeh Qomi, a PhD candidate in Industrial Engineering, will defend her doctoral dissertation entitled: “Design of a Sustainable Supply Chain Network for Fresh Grapes and Their Processed Products under Uncertainty: A Case Study” under the supervision of Dr. Mohammad Saeed Jabal-Ameli.

Defense Details
    • Date: September 29, 2026

    • Time: 1:00 PM

     • Venue: Room 208, Second Floor, School of Industrial Engineering

Dissertation Defense Committee:
Dr. Mohammad Saeed Jabal-Ameli, Dr. Ehsan Dehghani, Dr. Farnaz Barzinpour, Dr. Ahmad Makui, Dr. Reza Tavakkoli-Moghaddam, Dr. Hamidreza Pasandideh

Student ID: 400871034
Email: seyedzadehzahragmail.com

Abstract:

 The growth of the global population in recent years has increased the demand for food and created numerous challenges in ensuring access to safe and sufficient food. Agricultural and horticultural supply chains, as one of the primary sources of food provision, are responsible for transferring products from farms and orchards to final consumers. Among these products, grapes are considered one of the strategic horticultural crops in Iran due to their varietal diversity, high nutritional value, and wide range of applications in various industries.

In this dissertation, an integrated framework for designing a sustainable grape supply chain is proposed. In the first proposed framework, a hybrid approach based on data envelopment analysis and artificial neural networks is developed to select and prioritize grape growers. This approach eliminates reliance on subjective judgments while enabling the prediction of efficiency for new decision-making units and improving the computational efficiency of the model in large-scale settings. Subsequently, a multi-objective model is formulated considering the three sustainability dimensions economic, social, and environmental while also incorporating the limited shelf life of the product. To address demand uncertainty, a robust convex optimization approach is employed, and the multi-objective model is solved using the augmented ε‑constraint method.

In the second proposed framework, reinforcement learning based on the Thompson sampling algorithm is integrated with the mathematical optimization model to support dynamic decision-making for allocating products between fresh and processed markets. Furthermore, to enhance computational efficiency in large-scale problems, Benders decomposition is employed to obtain the global optimal solution.

The results of the case study conducted in Iran indicate significant Pareto trade-offs among the economic, social, and environmental objectives. Specifically, reducing total costs leads to increased environmental impacts, while improvements in social indicators result in higher operational costs. The uncertainty analysis shows that higher demand variability increases the overall cost of the network, whereas adopting a robust approach, despite slightly increasing costs, significantly improves system stability. The reinforcement learning algorithm converges to a stable allocation policy in which a substantial portion of the products is directed to the fresh market, reflecting the higher profitability of this market. Moreover, the optimal network design demonstrates that the simultaneous establishment of a juice-processing plant and a compost production facility creates industrial synergy, reduces reverse logistics costs, and generates both economic and environmental value.

Keywords: Viticultural supply chain, Sustainable supply chain, Supplier selection, Reinforcement learning, Artificial neural network, Uncertainty, Benders decomposition