ABSTRACT

Managing agri-food supply systems (AFSS) is a complex task due to all the factors that are typically present in these systems, for example, perishability, long lead time from cultivation to consumption, seasonality, weather and yield uncertainty, fragmented structure of the industry, high demand- and price variability, contamination concerns, fragility, and high weight-to-value ratios. This research focuses specifically on crop-based perishable AFSS systems with independent farmers for which it is difficult to coordinate cultivation activities with upstream operations of the supply chain, and where it is not possible to decouple processing or distribution from harvesting operations by building on-farm inventory after harvest. Two main industries are considered as case studies: sugarcane and vegetables for the fresh market. In the sugarcane industry, two main problems are addressed in this research, the farms clustering and harvest sequencing problem, and the truck release control problem. The first problem is modeled as a bi-objective non-linear program that aims to maximize sugar output and minimize the maximum cluster diameter when clustering farms for harvesting, and a heuristic algorithm is developed to find a good set of non-dominated solutions. For the second problem, a stochastic model is developed to determine the optimal number of trucks to be released into the sugarcane harvest system considering randomness in key inputs. Finally, the third problem deals with the development of a demand and capacity sharing protocol to support horizontal collaboration among private distribution networks in the fresh vegetables industry based on a real-time truck-sharing system for the long-haul transport stage. The effectiveness of the protocol is tested by simulating its implementation to a collaborative network of six distributors, and the results show significant improvement in delivery times, long-haul transport costs, and total traveled distance.