Preliminary results of CFD simulations for the scenario of a recent field study in an urban area

In our progression of research for a better understanding of winds in an urbanized domain, i.e. proceeding from physical modeling results to atmospheric field measurements and now on to our present study, ultra-high resolution computations with a mature CFD code based upon our recent field study have been completed. A method for running a single CFD model specifically for 6 different wind directions is demonstrated to show how this model simulates the full range of the wind speed and direction at the field measurement locations surrounding a single building within a cluster of 6 other adjacent buildings.

A horizontal model domain of 400 m by 400 m was set surrounding the 7 building cluster. A vertical domain of 200 m was found to be of sufficient depth for simulation of the influences surrounding the building cluster with all buildings less than 10 m high. The CFD simulations used a standard k-e turbulence model that was applied with the solution for the steady-state RANS (Reynolds-Averaged Navier-Stokes) equations. The cell size is 0.25 m surrounding the faces on the main building and 0.5 m surrounding other building faces. Below 20 m the maximum cell size is 1 m, filling most of the domain and ground surface. Above 20 m, the cell size grows from 2 m, to 4 m and finally 8 m. The total cell count is 4.5 million.

For this study, the only field measurements that were seen by the CFD modelers prior to completing the CFD simulations were those collected specifically at the 10m level of the upwind 10m tower – the Reference Tower. We, therefore, consider this to be a “blind test” of the CFD modeling methods. During this study, wind fields were generated for each10 degree increment between 250 to 300 degrees. The running average (smoothed) of 1-minute (measured) directions were categorized into wind direction bins, analyzed, and compared with a steady-state constant direction CFD model simulations, such that the model inlet boundary wind directions were set at 250, 260, 270, 280, 290, and 300 degrees. In that only one inlet wind speed profile was necessary, the single wind speed profile was developed to represent the wind profile upwind of the building cluster and to match the normalized measurements from the Reference Tower. The steady-state, constant direction CFD model simulations were run for a constant wind speed. For comparison to the time series wind measurements, CFD simulation wind velocity for any specific location, V(CFD(x, y, z)), was normalized by its value at the 10 m level {V(CFDREF10)} of the Reference Tower. Then comparisons between CFD velocity and the field measurements were made by taking the location specific ratios, V(CFD(x, y, z))/ V(CFDREF10), and multiplying by the Reference (SW) Tower's 10 m measured value. This produces a time series of CFD wind speed and direction that is comparable with the field measurement. Results will be shown as derived from the present model study demonstrating a method using steady-state CFD model simulations to model real time varying winds. There are future plans for directly simulating time-varying winds with “large eddy simulation” CFD methods.