Optimization of Multi-Echelon Distribution System with Time Window Using Evolutionary Algorithm

Abstract

Distribution efficiency plays a critical role in modern supply chain management, particularly when facing strict delivery schedules and high operational costs. This research addresses the optimization of a multi-echelon distribution system for a logistics company, PT. POS INDONESIA, which currently faces challenges in route planning and fleet utilization across its network of 1 main warehouse, 9 overflow warehouses, and 40 retailers. The complexity of the problem is compounded by the presence of strict Time Window constraints, where failure to deliver within specified hours results in service level violations. This study proposes a mathematical model for the Multi-Echelon Distribution System with Time Windows and utilizes an Evolutionary Algorithm to solve it. The algorithm was implemented to minimize total travel distance while strictly adhering to vehicle capacity and customer service time limits. The optimization process involved replacing the traditional "nearest-neighbor" manual planning approach with a global search heuristic capable of generating efficient route clusters. The results demonstrate significant improvements in system performance. The implementation of the Evolutionary Algorithm reduced the total travel distance by 33.3%, decreasing from 255.8 km in the baseline scenario to 170.5 km in the optimized model. Furthermore, the optimized routing strategy allowed for a 25% reduction in the required fleet size, lowering the number of active vehicles from 12 to 9 units. Most importantly, the model achieved a 100% service level by eliminating all time window violations, compared to 12 violations observed in the baseline schedule. These findings confirm that Evolutionary Algorithms are highly effective tools for solving complex logistics problems, providing tangible benefits in terms of cost reduction and operational reliability.