Variable neighborhood search for optimizing the transportation of agricultural raw materials

Abstract

Low price of raw materials in sugar industry and characteristics of production method lead to the specific transport organization problem. A new variant of Vehicle Scheduling Problem (VSP) that arises from transportation of sugar beet is considered. The problem is formulated as a Mixed Integer Quadratically Constraint Programming (MIQCP) model, which reflects the objective and specific constrains from practice. Computational experiments are conducted on real-life instances obtained from a sugar company in Serbia and the set of generated instances of larger dimensions. The proposed MIQCP model is used within the framework of Extended Lingo 15 solver, providing optimal solutions on small-size instances only. In order to find solutions on larger problem instances, a metaheuristic method based on Variable Neighborhood Search (VNS) is designed. Obtained computational results show that the proposed VNS quickly reaches all known optimal solutions on small-size instances. On larger problem instances, for which Lingo 15 solver could not find even a feasible solution, VNS provides best solutions in relatively short running times.