A major reason for electro-mobility being considered a key component for sustainable future cities is the much lower environmental footprint of electric vehicles. While there are many studies on the impact of electric vehicles on local CO2 emissions, for example, little attention has been paid so far to another kind of emission: heat. This is rather surprising when considering that transportation accounts for a significant share of energy consumption in a city (e.g., 21.5% in Singapore) which is eventually converted into heat. A transition to electro-mobility could help in reducing the urban heat island effect, i.e., help to reduce ambient temperature in a city by several degrees.
Simulating and analysing the effect of traffic on urban heat would first require a sub-microsopic simulation of road traffic in an urban area as well as the energy consumption and heat generated by individual vehicles. A part of the SEMSim Platform, developed by TUM CREATE, is the agent-based traffic simulation SEMSim Traffic, which not only simulates traffic but also the internal state of individual vehicles (including energy consumption). For urban heat simulation, vehicle component models are used to simulate the energy consumption of individual vehicles as well as the resulting heat generated by the most relevant vehicle components (e.g., engine, breaks, cooling system).
While heat generation is simulated by the agent-based traffic simulation, the actual heat dispersion is simulated using a Cellular Automata (CA) model. A CA-based approach has been chosen in order to overcome limitations of CFD approaches with respect to scale and dynamism by modeling the area as a three-dimensional space, divided into cells with dimensions of 15cm x 15cm x 15cm. Each cell has a state described by its temperature and material. For each time step the temperature of each cell is updated based on the amount of heat that flows in and out of the cell in the form of radiation, convection and conduction from the vehicles and the environment. More details about our approach can be found here.
One of the problems with large high-resolution CAs, such as ours, is the computational complexity. We address this problem by using a quad-tree optimisation. More details on this method will be published soon. The video above illustrates our approach for CA-based heat dispersion modelling.