Multi-Echelon Mix and Blend Optimization in Grains Supply Chain

This work focuses on designing an agile supply chain model to optimize the flow of soybean and corn grains through a multi-echelon network of facilities in Brazil. The key objective of this model is multifold starting from maximizing the throughput utilization of facility assets, gains from mixing and blending to minimizing the demurrage of river barges, while proactively tackling business constraints such as limited storage capacity, drying capacity and higher production volumes during the peak harvest seasons. While the problem statement seems straight-forward, the main challenge arises from blending equations being quadratic in nature rendering the problem computationally intractable. We employ novel approximation techniques to replace the large MINLP formulation and the proposed MILP is solved using Gurobi, to derive a 15-day tactical plan of good solution quality within reasonable computational effort.