[vkrastev]

Hi all,

since the problem I'm posting about involves OpenFOAM-2.2.x (latest available sources) I cannot report it officially on Mantis as a bug of the current version. However, since I guess also that many guys out there (including me) still use OF-2.2.x as their baseline version, I think it could be useful to share this issue here in the forum.
The question is that the externalWallHeatFluxTemperature boundary condition does not work properly when directly applied to a fluid domain boundary (i. e. substituting both conduction through external solid shells and external convection with the above mentioned bc), unless the fluid mesh is VERY fine close to the heat exchanging wall. To check this behavior, just run the attached 2d cavity test case once with OF-2.2.x and subsequently with OF-2.3.x (you can use the Allrun and Allclean scripts): in the first case you will have a strange temperature and wall heat flux distribution (temperature will go far beyond the limits fixed by the convective boundary condition), in the second case the result will be consistent with the problem physics. I can also share that the problem does not depend on turbulence model (it happens even in a laminar configuration), type of wall functions or discretization schemes and the only way to obtain a consistent solution is to refine A LOT the grid, as I said above.

Have a nice day

V.

[wyldckat]

Greetings Vesselin,

I don't have time to test this myself and to figure out what the problem might be, but there was a recent discussion about this here: http://www.cfd-online.com/Forums/ope...mperature.html

From that thread, it looks like you could try this fix: https://github.com/OpenFOAM/OpenFOAM...207c650796e605

Best regards,
Bruno

[vkrastev]

Hi Bruno,

thanks for the fast reply but in fact the problem was not with OF-2.3.x: I've made my tests with OF-2.2.x (the latest up-to-date version before the git-maintained standard switched up to 2.3.x) and a recently built OF-2.3.x version which already includes the official commit you posted above. Among the two, it is OF-2.2.x that returns inconsistent results, unless the near wall spacing is reduced to VERY small values. This is why I said in my initial post that I cannot report this as an official bug case on Mantis, as I don't think that the OpenCFD guys would bother themeselves with something not involving the currently maintained version. Of course, I could try to implement the latest OF-2.3.x-like boundary condition in OF-2.2.x and see if this will fix the issue (I think I will do it as soon as I have the time for), but since I don't know how many people are aware of this kind of behavior I thought also it could be useful to open a discussion here.

Best

V.

Edit: I have checked the current OF-2.3.x sources and noticed that the implementation has further changed since the above mentioned bug fix. Now it looks quite different from both the old OF-2.3.x and OF-2.2.x versions. anyway, this is the implementation that for me worked fine, in contrast with the OF-2.2.x one