Modelling Complex Facades - Unusual Window Shapes

asked 2015-07-17 13:00:11 -0600

Annie Marston
4317 ●5 ●32 http://www.resustain.com

updated 2015-11-07 14:15:15 -0600

I have a building with a curtain wall, the window sections on the wall are curved inwards to create an arc. This results in a greater area of glazing than standard flat windows and possibly a different thermal heat transfer due to the window shape. I would like to investigate the effect of this curved facade on the heat gain in the space and compare it to a more standard facade. The curve is gentle but possibly significant. Does anyone have any suggestions of the best software to analyse this in? We are doing a crude attempt in EnergyPlus but I am not sure the angles of the windows will be dramatically different enough for it to make any difference.

edit retag flag offensive close merge delete

add a comment

2

answered 2015-07-17 14:57:30 -0600

ericringold

10637 ●32 ●8 https://www.nrel.gov/b...

Is the curved curtain wall you're describing something like the one at Sea-Tac? image description

source.

In which case, couldn't you model a planar window with the window U-value adjusted by the ratio of the curved area to the planar area? Assuming the opening for solar radiation is the same, and neglecting any complex convective or solar transmission effects.

edit flag offensive delete link

Comments

Yes, kind of... it is smaller curves rather than one big curve, but basically the same problem, I think they have called it fluted. I can adjust the U-value that makes a lot of sense, thanks. I am still interested in if there are any softwares out there that specialize in modelling the more specific thermal effects of moving from a planar facade to a curved or fluted facades.

Annie Marston ( 2015-07-17 15:26:03 -0600 )edit

I don't know of anything specialized, short of moving into the realm of ANSYS Fluent or the like. Maybe you can get a sense of the effect by building a small representative model in EnergyPlus to see how, say, one 'flute' (divided into a number of planar surfaces) compares to an equivalent single plane with increased U-value, but it sounds like you're already going down that road. Interesting problem!

ericringold ( 2015-07-17 15:44:21 -0600 )edit

Yes this is the road we are going down at the moment. I'll see what it brings. Thanks for your help!

Annie Marston ( 2015-07-20 08:19:52 -0600 )edit

add a comment

2

answered 2015-07-18 10:26:29 -0600

Ralph Muehleisen
501 ●8 http://www.anl.gov/

One thing you could do is use therm to model the curvature in small segements (to get the area close to right) and then do runs using standard vertical wall convection coefficients for the whole surface and the other using a convection coefficient appropriate for the largest angle. You can find convection coefficients for angled planes in the ASHRAE handbook of fundamentals in the heat transfer section.

Then you can get a bound on the U value you would have if you used software that computes true convection. If they end up being close, you might want to use the average of the two as your estimate. if they differ wildly (meaning that convection effects dominated heat transfer) then you know you probably need to go to a more complete heat transfer program like Fluent or Comsol Multiphysics. You might still use the average as a guess.