Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of GraphSearch.
Today private vehicles in the cities is a high problem, due to the congestion they create, their pollution and their use of public space. Various alternatives exist, as public transportation and other modes, and still the part of people using cars remain too important. What makes the individual car so indispensable, and what alternative mode could replace them, assuring lower negative impact? The main advantage of the car is that it is the most convenient point-to-point mode: it can reach high speed and allows carrying voluminous or heavy personal equipment and multiple people. It is in our habits, and losing these features is hard to accept. It is especially the case for people that need to pick or drop somebody, carry luggage, shopping bags, or some furniture, or have to reach places where there is not any other alternative transport (as train or bus). In term of offered advantages, the closest alternative transport mode to the private car is the ride-sourcing service. Just as private vehicles, it allows point-to-point trips and has similar vehicle-related characteristics. However, ride-sourcing also presents some inconvenience, as the less trivial trip-organizing process, some inevitable waiting time, and the service cost. Still, there are some fields in which ride-sourcing could offer an important improvement – most of all it could rationalize and optimize vehicle’s use. In fact, cars can generally offer up to four passenger places and still mostly they are carrying only one person – the driver. Also, when the private vehicles aren’t driven, they occupy a lot of space on the streets, staying unused more then 90% of the time. A transfer from the private vehicle’s mode (PV) on the ride-sourcing service mode would surely reduce the number of vehicles in the streets. Also, this mode could be designed in a way that uses the ride-sharing potential, that would reduce the required number of vehicles. Gaining space on the streets and reducing the required number of vehicles could permit to reduce consequently city’s congestion and improve urbanistic aspects. There are already some Transport Network Companies (TNC) that offer similar services, as for example Uber Pool. However, the demand for these services isn’t high enough to evaluate their impact on the traffic. Also, the performance of such a service could most likely be related to the number of users, increasing the probability of ride-sharing trips. Various research have already been performed to put in evidence the positive effect of the ride-sourcing using ride-sharing systems (Santi, 2014), (Alonso-Mora, 2017). They show that ride-sharing is actually possible, can be performed in a dynamic model, and can be implemented in a manner that the additional waiting and travel times remain in the acceptable range for the passenger. However, these researches do not consider the speed variation, due to the city congestion. With the ride-sharing, TNC system should allow to reduce the effective number of vehicles, which should help to resolve the congestion, but in the same time, TNC system also involve additional operational travels during which vehicles do not transport anyone and still circulate on the streets. The congestion is a very important aspect that is also a huge problem for a lot of cities. For these reasons, it’s interesting to figure out what would be the ride-sourcing and ride-sharing implementing TNC mode’s impact on it. Also, such a mode could be subject to a dynamic controlling strategy to reduce vehicles’ number in high-congested areas, by smartly deploying them among various regions. This strategy is described in M. Ramezani’s (Ramezani, 2018) and N. Geroliminis’ works (Geroliminis, 2012). Building and testing a dynamic model with a described TNC mode would permit to evaluate the congestion-solving abilities of this type of system. However, creating an MFD based Dynamic Model would require knowing in advance some specific parameters that would realistically describe TNC vehicles’ behaviour. To estimate these parameters, as there are not enough existing data for this type of TNC, a Microscopic Simulation Model is necessary. The aim of this thesis is to design and build a Microscopic Simulation Model, that reproduces the described TNC vehicles behaviour. The goal is to obtain valuable insight on TNC mode’s effect on traffic and its characteristics permitting future parametrization of a Macroscopic Dynamic Model.
Nikolaos Geroliminis, Caio Vitor Beojone