Multi-agent model for an Urban Distribution Centre

In a 6 month research at Kyoto University and Delft University of Technology, the urban distribution centre (UDC) was investigated with a multi-agent model to find situations where the UDC can be financial viable. From a literature study the main modeling components were found: commodity estimation, vehicle routing, traffic environment, agent interaction and various policy measurements. Policies to test the impact on the business case of the UDC are: different congestion and toll rates, subsidy and a dynamic fee. Hereby a gap in literature is addressed by dynamic modeling with a multi-agent model with vehicle routing and incorporating dynamic usage. Dynamics are included with variable traffic conditions (resulting in dynamic travel times) and dynamic demand locations. Dynamic UDC usage was tested by various delivery schemes and a dynamic fee, which changes hourly. After generating the multi-agent model with a genetic algorithm for vehicle routing, it has to be checked that an increasing UDC usage rate corresponds to decreasing both the NOx emission and km count (NOx emission reduction of 19.0% ± 13.7 % and km count reduction of 18.8% ± 12.9% over 2318 runs). The model tested different UDC delivery schemes (early delivery, fixed time delivery or full truck delivery), congestion rates (low, normal or high), toll rates (none, low or high) and subsidy levels (none, 25%, 33% or 50% subsidized). The only positive business case for the UDC is with a subsidy rate of 50% (rate 4), which shows the complexity of generating a positive business case for an UDC. Per setting 360 runs were performed. To enhace the UDC financial performance the dynamic UDC fee was tested. This fee varies during the different hours of the day. Five cases were tested by increasing the general dynamic fee curve with an additional global shift of -20, -10, 0, 10 or 20 Yen per parcel. The impact of NOx emission or km count reduction correcsponds to the fixed fee scenario. Per setting 50 runs were performed. Significant differences between various dynamic fee settings are present. The fact that the UDC income is close to being significantly higher than the -300.000 Yen per day, compared to similar fixed fee conditions is highly encouraging. However further and full evaluation of all possible dynamic fee settings is desired to find potential better dynamic fee settings than the currently used settings. At the end of this research three cases are compared for dynamic fee usage: a reference case, highly congested city case and the involved municipality case. Comparing the three scenarios, there is no significant difference between the NOx emission or km count reduction and the freight carriers costs. For the UDC income there is a significant income increase in the third scenario with respect to the other scenarios. Also for the municipality there is a significant increase in costs in the third scenario. Since there is no significant decrease in NOx emission or km count between the second and third scenario, the municipality will need other social or financial incentives, before the municipality will participate in providing such high and long term subsidies. From this research it is clear that generating a positive business for a UDC is highly challenging. For a UDC to be profitable, it seems to be necessary to receive subsidy permanently. This analysis of the UDC analysis relates to the low UDC success rate of the research of Browne, Sweet et al (2005). The conditions for generating a successful business case for the UDC are hard and limited. Further research of dynamic fee settings and investigation of the possibility to implement this in a real-life case can contribute to future prospective and the success rate for urban distribution centres.