Optimasi Masalah Pemuatan Barang ke Kontainer Menggunakan Algoritma Hibrida: A-Star, Beam Search, Simulated Annealing

Abstract

The Container Loading Problem (CLP) is central to logistics efficiency, cost reduction, and sustainability amid growing global transportation demand. Most existing hybrid approaches combine only two algorithms and focus mainly on box dimensions and volume, with limited attention to stacking stability. This study proposes a triple-hybrid algorithm integrating Simulated Annealing (SA), A-Star (A*), and Beam Search (BS) to improve container loading efficiency for boxed furniture, incorporating a minimum contact area parameter to ensure stable stacking. The model operates in three phases: SA generates diverse initial solutions, A* and BS perform a directed, efficient search, and a final SA refinement optimizes the result. The approach was evaluated on the OR-Library benchmark and a real-world dataset, using space utilization, fitness value, and computation time as metrics. The best-performing configuration achieved 92.72% space utilization, a fitness value of 0.477, and a 43.94-second runtime, outperforming a direct sequential approach in both quality and efficiency. A stacking simulation with a 50% minimum attachment threshold achieved 91.512% raw space utilization, successfully loading 23 of 40 items. Results show that integrating SA, A*, and BS balances exploration and exploitation, offering a practical, near-optimal solution for real-world furniture loading that supports cost efficiency and supply chain sustainability

GS Cited Analysis