Postal Item Delivery Route Optimization Based On Close-Open Mix Vehicle Routing Problem

Abstract

Postal delivery is the world's longest-running service business process. Every city on the planet has this service, which is supported by a network of various depots serving as collection and delivery points. In distributing postal item delivery, Pos Indonesia through the Postal Processing Centre (PPC) is a distribution streamline that connects cities and regions with delivery centres closer to the final customer. Pos Indonesia has previously established a distribution pattern based on a zone-based system comprising a primary distribution centre and various delivery centres, delivery schedules, internal fleets, and predetermined routes. However, with reduced production capacity, as demonstrated by reducing traditional mail and less than truckload operations and anticipating customer needs for timely delivery, delivery patterns with fixed schedules and the internal fleet are unable to address this issue. Close-Open Mixed Vehicle Routing Problem (COMVRP) is proposed to optimize delivery by involving external fleets, and this is to reduce transportation costs where external fleets do not have fixed costs and do not need to return to PPC after delivery. The Genetic Algorithm is used to find heuristic solutions for each delivery history after Nearest Neighbor (NN) implementation to validate vehicle routes. Combined COMVRP-NN-GA produces a set of solutions, which will be used in the Monte Carlo Simulation (MCS). The results demonstrate that COMVRP outperforms existing scenarios, from optimizing six-vehicle routes to estimating five routes using a single external vehicle route, significantly reducing total mileage. The simulation findings based on the historical delivery dataset may be used to develop future delivery patterns based on the number of vehicles and total route distance and reduce transportation costs.