Sunday, January 4, 2009

CFD Q&A with IES

Computational Fluid Dynamics is a complex area of analysis seeing increased use on innovative projects, as recognized in a recent edition of Architectural Record. I've done my own studies in the past but have had some lingering questions which Dr. Liam Harrison from IES has been kind enough to answer.

Q> I have some questions about the CFD analysis process and wonder if there is reference material anywhere.

At the moment there is only the Microflo manual which can be found in the Help menu of the VE. We are currently putting together some tutorials but they have to wait for some slack periods in our project work to be completed.

Q> I can find wind data from the energy plus web page (tmy2, tmy3), are there better sources that you use?

We tend to use the energy plus data ourselves. Sometimes we use data taken on the site of a proposed development but this is rare as they tend to put the anemometers on top of a building so it is difficult to back out the localized wind effects caused by the buildings and they rarely have a whole years worth of data.

Q> From this wind data I can find the prevailing wind speed and direction which would be measured generally from 10 meters off of the ground. Is this right, 10 meters?

Yes, the standard height to measure wind speed and direction at a weather station is 10m above ground level. At this height the effects of vegetation, buildings and other obstructions on the wind speed and direction is limited. Also the data is usually taken in open areas such as airports to further limit the effect of obstructions.

I have tried to find a statement on the energy plus web page that all the data is taken at 10m height but can't find one so perhaps you will want to check for yourself to make sure.

Q> If my analysis grids extends 10 M then this weather station reading should indicate appropriate values to use in my study?

Microflo asks for the wind speed at 10M height and the terrain type and then produces a wind boundary profile on the inflow boundary of the CFD model equal to the equation from the ASHRAE Handbook - Fundamentals [2005] (section 16.3 “Airflow around Buildings”) - see Appendix D in the Microflo manual.

Q> If my grid extends 100 M for a tall building then I should use the wind profile power law ( to adjust for the height?

No, give Microflo the wind speed at 10M and the terrain type. Microflo will then automatically vary the wind speed over the inflow boundary using the ASHRAE handbook equation.

Q> Using the wind profile power law I’ll assume a neutral stability condition and use: “α is approximately 1/7, or 0.143.” Right?

Microflo will use a different value depending on the terrain type. See Appendix D in the Microflo manual.

Q> For an urban site the value of α might not be 1/7, but I don’t want to try to figure out the log wind profile ( unless I have to. Do the IES analysis settings take into account if the site is rural or urban?

Yes Microflo can take terrain type into account - the user can specify "Country," "Suburban," and "City" in Settings>>CFD_Settings.

Q> Finally, I’ve reduced the full analysis grid size to keep the number of cells under 500,000. Are there rules of thumb you use for spacing above or around a site?

The academic studies on CFD for the built environment gives the following recommendations:
The inlet, the lateral and the top boundary should be 5H away from the building, where H is the building height. For buildings with an extension in lateral direction much larger than the height, the blockage ratio should be below 3%. The outflow boundary should be positioned at least 15H behind the building to allow for flow development. For the same reason this outflow length should also be applied for an urban area with many buildings, where H is to be replaced by Hmax, the height of the tallest building. To prevent an artificial acceleration of the flow over the tallest building, the top of the computational domain should be also at least 5Hmax away from this building. For the blockage ratio the limit of 3% is recommended, although there are no results on whether it is better to include more of the surrounding buildings in the model and reduce the distance of the lateral boundaries from the built area.

In a Microflo model the recommendation above will be difficult to follow as it will tend to produce very large mesh sizes if you resolve the flow around the buildings. You need to have a fine mesh in areas where the air velocity gradient is greatest, i.e. around the buildings. In area where the velocity gradient is small you can have larger cells, i.e. near the boundaries of the computational domain. E.g. this can be achieved by:
  1. going into "Grid" mode at (1) in image below
  2. click into a region in the mesh, at (3) in image
  3. click "Edit mesh region" at (2)
  4. select a power law to vary the mesh and make it coarser at the boundaries and finer at the building

Even using this technique it will be difficult to resolve the flow near the buildings and have a domain as large as the recommendations so you may have to be pragmatic and forgo some of the computational domain in order to have enough cells near to the buildings. The only way to prove if having a smaller domain and more resolution near the building is better than the reverse situation is to do grid dependence studies where you change the grid to find the point where the solution changes insignificantly between meshes. In general though it is more important to resolve the flow near to the buildings, although this can make to solution less stable and harder to obtain a converged solution (nobody said CFD was easy).

No it's not easy but it doable, even for architects. Just be sure if you are working in a program other than IES that you know how to adjust the wind speed based on terrain type and the height of your grid.

1 comment:

jolly said...

"I admire what you have done here. It is good to see your clarity on this important subject can be easily observed. Tremendous post and will look forward to your future update.We start our GATE Coaching Delhi,