Nowadays, electric vehicles (EVs) are becoming one of the most popular traffic modes in urban mobility system. As a new mode, it has its unique characteristics but also have many features in common with conventional transportation tools. One main advantage of the EV is that it uses the clean energy as the power of driving, which can significantly improve the sustainability of the current mobility system. Therefore, this new traffic mode has been widely and strongly supported by all sectors of society. Many governments around the world even plan to totally forbid the sale of conventional fuel vehicles to promote people’s acceptance towards the EV. However, a critical obstacle to continued progress in this technology is the range anxiety and the inconvenience of the current charging schemes, which to a great extent, limit the flexibility of drivers’ trip scheduling. In the case of this anxiety, the concept of Battery Swapping Station (BSS) becomes popular at present. This infrastructure is expected to provide battery swapping services to the EV like a fuel station, which can greatly eliminate the range anxiety and enhance the flexibility of trip scheduling for EV drivers.
This thesis aims to make the construction planning for this future infrastructure. To be specific, it is to find the optimal spatial distribution of BSSs in the urban traffic network. For solving this type of problem, there have already been many studies that developed optimization models with objectives such as minimizing the overall construction and operation costs of the system or maximizing the overall operational profits. There are also a few studies take the routing problem into consideration as well. However, another important stakeholder in this problem is completely missed in existing research, that is the urban resident. Urban residents now are becoming more and more concerned about the living environments in the city, so they should not be neglected when trying to solve the locating problem of BSSs because the construction and operation of this infrastructure will have a significant impact on urban livability. On the basis of this consideration, this research develops a model that simultaneously calculates the interests of three stakeholders in the optimization, which are system investors, urban residents, and system users (EV drivers) so that a more balanced solution to this optimal locating problem can be obtained, which is more realistic and practical in the future.
The developed model is applied to a real traffic network in this research. Delft City in The Netherlands is chosen to be the experiment site. In general, four types of concerns are tested in this study. The first concern is about the uncertainty of the optimal solution brought by the randomness of EV’s initial SOC. Since there are not mature statistics on collecting EV’s initial SOC distribution at present, a probability density function is used to randomly generate initial SOC for vehicles during experimenting. In this case, it is possible that the optimal locations can be quite uncertain in the city under different draws. Therefore, this uncertainty is specifically tested first. Then the second concern is about the future development of EV-related technologies. To be specific, the optimization is conducted under different maximum traveling ranges of the EV to see what will be the case if the traveling range of the EV becomes wider and wider in the future. The third concern is an optimization strategy that might be applied by the decision-maker to ensure a good living environment in the residential area in the city, so a station size limit is set in this area to see how the optimal solution will change under this circumstance. The last concern is about some possible optimization preferences of the decision-maker, for example, the preference for a better travel time in the traffic network, fewer stations in the residential area, less economic expenditure on station construction. The impacts of these three preferences are examined in this study.
A series of performance indicators are designed to comprehensively reflect the impact of a certain concern on the optimization. And it is found that each of the concern will significantly influence the optimal decision-making in this problem. Much information can be inferred from experiment results, which will be specifically exhibited and explained in later parts of this thesis.
At the end of this research, the limitations of the developed model are analyzed in detail, and recommendations for future research on this topic is formulated based on these limitations.