Numerical simulations of urban heat islands and transport and dispersion of airborne materials around building clusters

Numerical simulations of air flows in and around urban areas are in considerable interest to understand and mitigate discomforts associated with urban heat islands and protection of residences from accidental and intentional releases of toxic pollutants. Such simulations, however, encounter a considerable challenge: Air flows from building to terrain scales coexist so that horizontal grid spacing of a few m to a few km are required to represent air flows in both scales.

To address building scales we have added CFD (Computational Fluid Dynamics) capabilities to a three-dimensional mesoscale model HOTMAC. The new model is referred to as A2Cflow where “A2C” stands for “Atmosphere to CFD.” The model capabilities of A2Cflow have been greatly enhanced from those for HOTMAC.

We also upgraded a companion transport dispersion model RAPTAD to A2Ct&d, where “t&d “stands for “transport and diffusion”. A2Ct&d uses wind and turbulence distributions predicted by A2Cflow.

A2Cflow simulated successfully separation, recirculation, and reattachment of air flows around obstacles placed in a wind tunnel where a horizontal grid spacing of 2 cm was used in simulations. The same model was used to simulate air flows around a cluster of buildings under the influence of diurnal variations of mesoscale weather conditions such as sea- and land-breezes. A nesting method was applied to capture a large scale terrain and a small scale building variations. Horizontal grid spacing of 10 m for the inner domain and 40 m to 160 m for the outer domains were used.

Temperatures of building walls and roofs were computed by solving a one-dimensional heat conduction equation in the direction perpendicular to the walls and roofs. The boundary conditions were the heat balance equations at the outer sides of walls and roofs, and the room temperatures specified at the inner sides of the walls.

Air flows and turbulence distributions were considerably different from those obtained in neutral conditions such as in a wind tunnel. Air circulations between the two cubic buildings were in clock wise directions in early morning and counter clock wise directions in late evening. There were no organized circulations in late afternoon and in early morning hours prior to the sunrise. The time variations of air circulations were resulted from differential heating and cooling of walls by the sun.

Since larger horizontal grid spacing was used in outer computational domains, individual buildings can no longer be resolved in a model. We proposed to use an urban canopy parameterization to capture gross effects of buildings on air flows. We will present preliminary results on how to smoothly transit from an urban canopy parameterization to detailed simulations of air flows around buildings.