Question-and-Answer Resource for the Building Energy Modeling Community
Get started with the Help page
Ask Your Question

Calculating 3D Thermal Bridges

asked 2014-09-09 02:39:19 -0500

updated 2015-07-10 21:18:09 -0500

In order to calculate the 3D thermal bridges (such as any "point" thermal bridges to calculate Chi-value), THERM, my go-to program for 2D ones, cannot be used.

Such a 3D thermal bridges would be for example steel rebars in a concrete floor protruding to form a balcony. Or steel pillars for a raised-floor system above a built-up roof.

I was wondering what tool people use for that and what are the pros and cons of the tools they've used.

Especially, if anyone has an open source or very cheap alternative, I'll take it. It doesn't make sense to buy very powerful and expensive finite element method tools if you're going to be using them only a handful of times for simple problems like this. For example I've tried Comsol which is great, but out of my budget by a factor of 10 at least.

If you are aware of any library of common 3D thermal bridge values, I'm interested as well.

edit retag flag offensive close merge delete


I don't know of any good freeware solutions. If someone created a plugin that could turn Sketchup geometry into a matrix of material properties, it would be pretty easy to implement something in MATLAB. Interested?

mbrusic's avatar mbrusic  ( 2014-09-09 17:41:01 -0500 )edit

See Morrison Hershfield report for ASHRAE Technical Committee 4.4 "Thermal Performance of Building Envelope Details for Mid- and High-Rise Buildings (1365-RP)".

jblake's avatar jblake  ( 2014-09-10 14:02:59 -0500 )edit

Mike, while this is definitely doable, the problem is that Maltab isn't freeware either. Thanks jblake for posting ashrae 1365-RP, I already had a copy and it's definitely a good resource!

Julien Marrec's avatar Julien Marrec  ( 2014-09-11 03:00:35 -0500 )edit

We use HEAT 3 by Building Physics. Hope that helps.

Shahizal's avatar Shahizal  ( 2014-09-11 09:43:55 -0500 )edit

Thanks for pointing out the software. But since it is not detailed at all, it should be considered as a comment and not an answer.

Julien Marrec's avatar Julien Marrec  ( 2014-09-11 12:20:55 -0500 )edit

1 Answer

Sort by ยป oldest newest most voted

answered 2014-09-11 15:36:47 -0500

The community of modelers I work with typically will take the 3D problem and do their best at turning it into a combination of 2D problems and use THERM. One reason is the error introduce by reducing the 3D into a series of 2D problems is far smaller than what you typically get when you are taking either and converting to a 1D problem.

I think you should think about this as a 3D-2D version of the ASHRAE Zone method or ISO 6946 where you calculate and upper and lower limit based on assumption of 1D heat flow (upper limit) and isothermal planes (lower limit). You could do the same assuming for 3-D assuming the 2D sections do not interact (upper limit) or interact completely in one plane (lower limit).

Unfortunately I cannot point you to an example.

For software, others mentioned comsol multiphysics, MATLAB, and heat3 in the comments, none of which is free. Instead of using MATLAB, one could do any similar programming in Python or GNU-Octave for free. As a free replacement of heat3 or comsol, you could look to elmer, an open source multiphysics, FEM software

I have not used elmer since I have been a comsol user for a number of years, but I wanted to suggest something that might be used as a comsol alternative.

You might also check to see if anyone has developed libraries for 3-D bridging in the Modelica buildings library. Again I'm not a user (yet) but I know that it is gaining a lot of attention.

edit flag offensive delete link more


Ralph, I'm familiar with the 1D reduction of 2D problems as described by ASHRAE, and often the gap between the lower & upper limit is so big it's somewhat a pointless endeavor (at least that's what I found, and one of the reasons why I use THERM... Though the libraries of Psi-values I have are usually pretty decent. You use Comsol. I'm familiar with the program but I haven't used it on many occasions. You say it's closer to reality to reduce 3D into 2D... Do you have a quantitative idea about that? (I've been doing that so far, using THERM, but not knowing if it was nowhere close to reality)

Julien Marrec's avatar Julien Marrec  ( 2014-09-11 16:14:38 -0500 )edit

To clarify, I was saying the 3d to 2d conversion was more accurate than the 2d to 1d conversion. You are reducing your degrees of freedom from 3 to 2 which is a 33% reduction whereas when you do 2d to 1d it is a 50% reduction. I kind of think of it as bounding the error. In 2d to 1d I do the upper and lower limits and take the average, I should be within 50% of the right answer. Similarly, if I do that in 3D to 2D I should be within 33% of the right answer.

That said, if you can exploit symmetry in problems when you break it up (just like in 2d to 1d) you know your answer will be closer to the parallel path answer and you can take a guess closer to the high limit you get from the parallel path method.

Ralph Muehleisen's avatar Ralph Muehleisen  ( 2014-09-13 09:00:27 -0500 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer


Question Tools



Asked: 2014-09-09 02:39:19 -0500

Seen: 2,560 times

Last updated: Sep 11 '14