Question about Wind Speed in the UWG

Kia ora katoa

I notice that I am having this issue with local New Zealand weather files and the most recent (v1.5 ) version of LBT and the associated calculator. I am puzzled: surely UHI temperature changes would be mitigated by wind, so the speed of the wind should not be an issue, but should be part of the model?

I say this because I am from a very windy climate:

The above graphic is admittedly from our windiest city with the average daily wind speed profile in red, and the average for each day of the month in blue. The lowest average across a month is in April: 6m/s or roughly 22km/hr or 13mph.

Or doe the UHI model not deal with wind?

M

Hey @MichaelDonn ,

The people who would know that most about this are @josephyang , who worked with the UWG for his Masters thesis or @SaeranVasanthakumar , who translated the UWG from MATLAB to Python.

I have spent a little bit of time with the UWG source code and I seem to remember the wind speed being used in some places. Is the impact of the wind just not as large as you expected? Or did you run some test with and without wind and didn’t see a significant change?

@chris, @MichaelDonn

As I recall, UWG’s finite difference approximation has previously struggled with instability coming from the wind speed’s effect on the convective heat transfer coefficients of surfaces.

I think this mitigation should be accounted for in the simulation of the vertically stratified air, and heat transfer of mixed air at the urban boundary layer, although I don’t have a good grasp of this part of the UWG. However, it would be contradicted by higher wind speeds lowering construction performance by disrupting the stagnant air alongside surfaces.

There should be an error message reporting extreme temperatures/flux if this is the problem. What error message are you getting when the UWG fails?

If it is a finite difference problem, then you can try and smooth out the instability by increasing thermal mass in the roof, reducing the simulation timestep (get better slope approximations), increasing the simulation timestep (remove possible floating point precision errors), or removing any large spikes in timeseries parameter that could cause a large increase in heat flux, like wind speed, solar, or drybulb temperatures.