كنفرانس بين المللی مهندسی صنايع و سيستم ها

صفحه اصلی / دوازدهمین كنفرانس بين المللی مهندسی صنايع و سيستم ها

Risk-Averse Agricultural Supply Chain Optimization Under Climatic Uncertainty: A Two-Stage Stochastic Programming Model with CVaR Integration

نویسندگان :

Ali Gholami¹ Seyed Omid Hassanpour²

1- دانشگاه تهران 2- دانشگاه تهران

کلمات کلیدی :

Agricultural supply chain،Two stage stochastic programming،Conditional value at risk (CVaR)،Rice cultivation optimization،Drought resilience،Scenario based optimization

چکیده :

Agricultural supply chains are inherently susceptible to climatic variability and the resulting stochastic fluctuations in crop yield, particularly for strategically vital staple crops such as rice. Conventional deterministic planning models are structurally incapable of capturing drought risk, thereby exposing farmers and agri-food operators to severe economic losses during adverse weather events. This paper proposes a novel two-stage scenario-based stochastic programming (SP) model for the simultaneous optimization of crop cultivation area allocation and multi-channel distribution planning across two Iranian rice varieties (Hashemi and Shiroudi) under three climatologically grounded scenarios (dry, normal, and wet). The principal methodological innovation lies in the explicit integration of Conditional Value-at-Risk (CVaR) into the composite objective function, enabling direct and parametric control of downside risk in extreme tail scenarios. The model is formulated as a Mixed-Integer Linear Program (MILP) and solved to provable global optimality using an exact solver. Computational results demonstrate that for a risk-averse decision-maker (λ = 0.5, α = 0.90), the optimal land allocation comprises approximately 80.00 ha of high-quality Hashemi and 51.43 ha of the drought-resilient Shiroudi variety. This diversified portfolio strategy guarantees a minimum profit margin of 15,700 million Tomans at the 90th percentile confidence level, substantially enhancing supply chain resilience. A comprehensive comparative analysis against ten representative studies confirms the multidimensional novelty of the proposed framework relative to deterministic, robust optimization, and chance-constrained alternatives.