OPTIMASI RUTE DAN PENGEMBANGAN SKEMA KOMPENSASI KURIR PADA SISTEM DISTRIBUSI LAST MILE MEMPERTIMBANGKAN TIME WINDOWS KENDARAAN HETEROGEN, DAN MULTI-TRIP DENGAN ALGORITMA GENETIKA

• Main Author: Mulyo, Gabrella
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/63740
• Institution: Institut Teknologi Bandung

The rapid growth of e-commerce due to the COVID-19 pandemic has impacted the flow of last mile goods delivery which is characterized by high volume of small packages and business-to-consumer (B2C) shipments. However, last mile is the delivery phase with the highest cost contribution (40%) so there is an opportunity to reduce costs on last mile. PT X is a food ingredients e-commerce that is facing last mile distribution problems marked by the KPI targets for last mile cost and on-time arrival have not been achieved. This study aims to solve problems related to last mile costs through the development of models and algorithms to determine the optimal daily distribution route from one warehouse in Jakarta to 363 power users/new customers (PC/NC) to minimize the total cost of last mile distribution and develop driver payment scheme by considering various constraints. Constraints includes customer time windows, duration of work, multiple vehicle types, and the ability of vehicles to run multiple trips to increase vehicle utilization. This problem is known as MTHVRPTW and refers to a mathematical model (Seixas & Mendes, 2013) and (Anaya-Arenas, Chabot, Renaud, & Ruiz, 2016). The mathematical model can produce a global optimum for a data size of 10 customers. MTHVRPTW is basically an NP-hard problem so that the computation time using the branch-and-bound algorithm increases exponentially as the number of customers increases. A two-stage algorithm with a route-first cluster-second approach is proposed to solve the real case. In the first stage, a solution construction with a Split algorithm based on dynamic programming is used and a genetic algorithm is employed for the improvement stage. In the second stage, a greedy algorithm is used to generate multi-trips. Implementation of the algorithm results in a last mile cost savings of 14,5% with a computation time of 102 minutes. Further improvement of 21,7% savings can be achieved by employing combined compensation scenario that results in synergy between each of the cost structures. The parameters that are sensitive to cost are the working duration limit and the speed of two-wheel vehicle.