Cavity radiation

DaveStupple

Not sure if this warrants a fresh thread or if it is a continuation of my previous nonsense, but here I am so let's give it a try. I've got a thermal steady that solves OK in both internal and CCX solvers, but the radiation is lossy only rather than cavity. I think I have worked out the steps to get cavity working so that radiation is absorbed as well as shed, and I have a trivial example (two adjacent cubes) looking roughly like it works OK. In trying to get a more complex example working, the steady state threw errrors, so I switched to transient. I tried many adjustments (density and specific heat) with no luck. I find it spends ages saying it's calculating viewfactors, and then it just stops with a promising looking green panel, but no actual results to view. I have had it stop at just 'Calculating the viewfactors', but also with a few more promising outputs beyond this but never a final result. Ideally, if this can be made to work we can show that cavity radiation is negligible, but if I don't get it working I won't know. Latest effort attached. If someone can help me move this on I would be ever thankful. Ideally, steady state would be simpler but any progress would be progress. File is too big (at 108K) to upload: I'll try putting a link to it here...
https://dropbox.com/s/od5xk3ycpwh42zk/T7RCcard.liml?dl=0

DaveStupple

Let me know if the link works

disla

Look at the conclusions at the end of this post:

https://mecway.com/forum/discussion/comment/6023/#Comment_6023

If you can live with this 15% I would try:

-Uniform temperature on the spiral area. (2000K-2500K)
-Consider the spiral as very tight (touching spirals) an radiate only through the outer surface. That's a lower bound.
-Put that simple piece inside a closed cavity (sphere) and measure the blocked energy as the difference between input energy and Output energy at the external surface.

Remember to set a negative sink temperature to assure viewfactors add to 1.

I think that can give you an idea of what percentage of radiation can't go out and its role inside the problem.

NOTE: Those view factors must be very difficult to determine properly as it is a surface shadowing itself with many small surfaces each one at a different angle (spiral).

disla

I have downloaded your file and considering only external surface to compute the view factors shouldn’t be a bad hypothesis. The winding is very dense.

DaveStupple

disla,

Thanks for these ideas. I had seen your previous post, I wasn't sure if you concluded that it was inaccurate or if it was OK.

Regarding viewfactors, I found this (sounds like hours of fun):
"The calculation of viewfactors involves the solution of a four-fold integral. By using analytical formulas derived by Lambert this integral can be reduced to a two-fold integral. This is applied in CalculiX right now: the interacting surfaces are triangulated and the viewfactor between two triangles is calculated by taking a one-point integration for the base triangle (in the center of gravity) and the analytical formula for the integration over the other triangles covering a hemisphere about the base triangle. One can switch to a more accurate integration over the base triangle by increasing the variable “factor” in subroutine radmatrix, look at the comments in that subroutine. This, however, will increase the computational time."

How does setting the sink negative ensure viewfactor adds to 1?

disla

Manual: "Sometimes, it is useful to specify that the radiation is closed. This is done by specifying a value of the environment temperature which is negative if expressed on the absolute scale (Kelvin). Then, the view factors are scaled to one exactly."

If I understood it correctly, it means that once computed, no matter what, the view factors are scaled to sum one. That warranty there is no energy lost due to computational inaccuracies. That is valid if you are balancing a system enclosed in a sphere or a box for example as I mentioned before.

Additionally:

dhondt
Aug '22

"Hi,

did you check that the viewfactors add up to 1? You can force that by specifying a negative environment temperature on the Kelvin scale (cf. Documentation on *RADITATE). Due to numerical roundoff viewfactors are not exact."