Simulating Dynamic Vehicle Routing Problem using Agent-Based Modeling and Simulation

Changing demand trend (e-commerce business, timely delivery) has contributed to dynamism and complexity of routing problems. Despite its importance, Dynamic Vehicle Routing Problem (DVRP) has received little attention. Most of VRP works have been characterized by static and deterministic problem. To fill the gap, the present study aims at developing Agent-Based Modeling and Simulation (ABMS) to simulate DVRP by capturing dynamism and stochasticity. Urban freight transport serving sixty-two retailers at Yogyakarta city (Indonesia) is taken as a studied case. GIS data and Particle Swarm Optimization are embedded in ABMS to mimic retailers' location, road infrastructure in the real system and to evaluate optimal routes respectively. Experiments were conducted to evaluate the effect of the degree of dynamism (dod) on logistics performances. Results indicate a linear relationship between the dod and both travelled distance and lateness. For every 1% increase of the dod, the travelled distance would increase,on average, by 7%.