Solving Container Stowage Problem using Particle Swarm Optimization Algorithm with Multiple Social Learning Structures

The growth of international trade resulted in the raising of the total containers transported annually. In Indonesia, the number of container throughput has reached 12.85 million TEUs in 2018. In providing fast and efficient port services, container terminals (CT) constantly face complex problems on its everyday operation, one of the operational level problems occurred at CT is Container Stowage Problem (CSP). CSP is an NP-Hard Problem, therefore metaheuristic algorithm is used to solve the problem. This study proposed the use of Global, Local, and Neighborhood – Particle Swarm Optimization (GLN-PSO) algorithm to solve CSP. The basic idea of GLN-PSO is emphasize on the particle movement that considers multiple social learning structures. GLN-PSO algorithm is a development of the PSO algorithm that has been used in previous studies and is proven to outperform Bee Swarm algorithm. In this study, GLN-PSO is used to solve two types of data, namely small instances (consist of 5-27 containers) and medium instances (consist of 100-140 containers). The result showed that for small instances, GLN-PSO can outperform the PSO algorithm in terms of objective function value by 0.72% and in terms of computational time by 0,172 seconds. For medium instances, GLN-PSO has not been able to outperform the PSO algorithm by producing a gap of 27% for the objective function and 8,2 seconds for computational time