High Performance Computing for Simulation of the Urban Microclimate

In light of the increasing urbanization and climate change, there is great need to advance methods for the simulation of the urban microclimate. Applications of such tools include planning and design of urban spaces for healthy levels of exposure, and forecasting of the urban microclimate for extreme exposure mitigation. Example metrics includes the physiological equivalent temperature, requiring the derivation of air temperature, mean radiant temperature, and wind speed at high resolution.

The work presented here employs municipal datasets including 3D City Surface Models, building typology and vegetation information, in order to generate a high-resolution input model for a transient Multiphysics simulation of radiative heat transfer, using downtown New York City as the testbed. Additional input parameters such as albedo and emissivity were derived using ground-based imagery, as well as estimated thermodynamic properties of building facades, using a constant indoor temperature. The model consists of 400 buildings composed of approximately 24 million elements.

The size of the computational domain and the complexity of the thermodynamic processes combined within the created input model, necessitates the use of a high-performance computing cluster to calculate a range of climate parameters over a period of 24 hours with the total computation time being 60 hours. The results of the simulation will be used to develop reduced-order models that provide rapid predictions of the microclimate across the same area.

The efficacy of the simulation was studied using surface temperature measurement by contact sensors and broad-band, airborne and spaceborne thermal imaging, as well as weather stations for measurement of wind speed, and trajectories.