MODEL AND HYBRID DISCRETE PARTICLE SWARM OPTIMIZATION ALGORITHM FOR VEHICLE ROUTING PROBLEM WITH MULTI PRODUCTS, MULTI COMPARTMENTS, MULTIPLE TRIPS, SPLIT DELIVERY, AND MULTIPLE TIME WINDOWS

• Main Author: Marcellina, Sarah
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/43411
• Institution: Institut Teknologi Bandung

Determining optimum number of vehicles and route planning are needed to minimize transportation costs and improve service quality. This problem is known as a Vehicle Route Problems (VRP). This research develops a mathematical model and algorithm for the VRP case with the characteristics of multiple products, multiple compartments, multiple trips, split delivery, multiple time windows, and homogeneous vehicles (VRP-MPMCMTSDMTW). The model developed in this study is based on a real system of fuel distribution by PT. Pertamina in the NTT and Timor Leste region. Mathematical models and algorithms in this study have an objective function to minimize the total transportation cost during planning horizon by minimizing the total number of vehicles and the total travel distance to serve all customer demands. The model solution was solved with the help of LINGO 18. Also, the hybrid discrete particle swarm optimization algorithm with variable neighborhood descent (hDPSO) was developed to tackling the computational time problem in the analytic methods. This algorithm was able to reach the optimal solution with a difference in analytical solutions of 0.79%. Also, based on the computational results, this algorithm performs better than the DPSO algorithm and has succeeded in improving the quality of the DPSO algorithm solution by combining the DPSO algorithm with a variable neighborhood descent (VND). This can be seen from the solution in the form of a better total transportation cost than the DPSO algorithm because of the use of a smaller number of vehicles and a smaller total travel distance obtained.