Tarım tedarik zincirlerinde talep, hasat ve verim risklerini göz önüne alan bir planlama yaklaşımı

In this study, we focus on an agricultural planning problem under demand, yield and harvest uncertainties. We present a model that captures variations of the harvest period, the yield, and the demand. By using the proposed model, we consider finding both the optimal seeding areas and seeding times for annual plants and only the optimal seeding areas for perennial plants.We study two versions of the problem for annual and perennial plants, depending on the number of periods in the planning horizon: single-period case, and multi-period case. We establish optimal solution for both annual and perennial plants for single-period case. When the number of suppliers increases, the computational complexity of the objective function also increases, and finding the optimal solution becomes computationally demanding. Hence, we propose a normal approximation for the supply. The approximation provides results very close to the optimal solution and the deviation from the optimal solution becomes negligible as the number of suppliers increases. In the multi-period case the optimal solution is found for the planning problem of perennial plants. However, like in the single period problem, the computational complexity of objective function increases as the number of suppliers increases. Thus, we develop efficient solution procedures to solve large-sized problem instances. Numerical experiments show that these procedures are quite accurate. For solving the multi-period planning problem of annual plants, we proposed several heuristics. We present numerical analysis that compares different approaches with each other and with the optimal solution. The best performing two approaches are selected and compared based on different criteria.Finally, we study a case from the industry. The problem is modelled by using the presented model and solved by using two introduced approaches. We observed that the proposed solution methodology yields significant improvements in the objective function compared to the case where a deterministic planning approach is used. We compare results of both approaches and explore the effect of system parameters in detail.