Algoritma Ant Colony System Untuk Menyelesaikan Multi Depot Vehicle Routing Problem Dengan Variabel Travel Time

Seiring dengan berkembangnya dunia industri yang menjadikan masalah distribusi produk bertambah kompleks, perusahaan dituntut untuk dapat melakukan distribusi secara efisien. Penelitian ini membahas tentang optimasi pada distribusi hasil produksi yang mencakup sistem perancangan rute dan penentuan urutan pelayanan konsumen. Permasalahan tersebut dimodelkan sebagai Vehicle Routing Problem (VRP) dengan time window. Seiring dengan meningkatnya permintaan konsumen, suatu perusahaan dapat memiliki lebih dari satu depo untuk memenuhi permintaan. Kasus seperti ini selanjutnya dimodelkan sebagai Multi Depot Vehicle Routing Problem (MDVRP). Terdapat dua tahapan untuk menyelesaikan MDVRP yaitu clustering dan assignment. Dalam clustering digunakan metode Simplified Parallel Assignment, sedangkan untuk assignment digunakan metode Ant Colony System. Pada akhir penelitian, metode Ant Colony System dibandingkan dengan metode Simulated Annealing dan Particle Swarm Optimization. Dari hasil simulasi metode Particle Swarm Optimization mampu menghasilkan travel time minimum. Sedangkan hasil algoritma ant colony system terbukti tidak mengalami perubahan yang signifikan saat mengalami perubahan parameter, termasuk perubahan permintaan konsumen dan persediaan depo. ======================================================================================================== With the growing business and hence increase in the complexity of products distribution, minimizing the cost of logistics becomes a significant factor in reducing the overall cost. This research discuss about optimization of products distribution including route planning-system and assignment of the daily products delivery. This problem can be modelled as Vehicle Routing Problem (VRP) with time window. As the increasing of consumer’s demands, the company may have more than one storehouse / depot in a city. Thus, it can be modelled as Multi Depot Vehicle Routing Problem (MDVRP). There are two step to solve MDVRP, those are clustering and assignment. In the clustering problem, Simplified Parallel Assignment was used while to solve the assignment problem, Ant Colony System was used in this research. At the end of this research, two others method were used for comparison, those are are Simulated Annealing and Particle Swarm Optimization. As the result of simulation, Particle Swarm Optimization methods has proven give the best result with minimum travel time, while the excellence of ACS itself is not showing a major change of travel time in the change of parameter including consumer demand and depot capacity