Planning an Integrated Stockyard–Port System for Smart Iron Ore Supply Chains via VND Optimization

Abstract

Stockyard–port planning is a complex combinatorial problem that has been studied primarily through simulation or optimization techniques. However, due to its classification as non-deterministic polynomial-time hard (NP-hard), the generation of optimal or near-optimal solutions in real time requires optimization techniques based on heuristics or metaheuristics. This paper proposes a deterministic simulation and a meta-heuristic algorithm to address the stockyard–port planning problem, with the aim of reducing the time that ships spend in berths. The proposed algorithm is based on the ore handling operations in a real stockyard–port terminal, considering the interaction of large physical equipment and information about the production processes. The stockyard–port system is represented by a graph in order to define ship priorities for planning and generation of an initial solution through a deterministic simulation. Subsequently, the Variable Neighborhood Descent (VND) meta-heuristic is used to improve the initial solution. The convergence time of VND ranged from 1 to 190 s, with the total number of ships served in the berths varying from 10 to 1000 units, and the number of stockyards and berths varying from 11 to 15 and 3 to 5, respectively. Simulation results demonstrate the efficiency of the proposed algorithm in determining the best allocation of stockpiles, berths, car-dumpers, and conveyor belts. The results also show that increasing the number of conveyor belts is an important strategy that decreases environmental impacts due to exposure of the raw material to the atmosphere, while also increasing the stockyard–port productivity. This positive impact is greater when the number yards and ship berths increases. The proposed algorithm enables real-time decision-making from small and large instances, and its implementation in an iron ore stockyard–port that uses Industry 4.0 principles is suitable.