[Wolfgang57]

Hi all,


I am trying to simulate a floor heating system with chtMultiRegionŚimpleFoam with Openfoam 4.1. I chosed a cube as a model, the setup is illustrated in the attached picture caseSetup_completeFloorHeating.png. The model consists of an interior room (1 m x 1m x 1m) with air (fluid region) and some solid regions representing the ceiling, walls with several layers and the floor setup, several layers, too. The floor setup from exterior to interior is the follwing: the floor layer with contact to the environment (which has a temperature of 271,15K), an insulation (thicknes 5 mm), a heating (thickness 5mm) and a floor layer (thickness 10mm). All the layers completely cover the floor (that means they have dimensions of 1m x 1m). The heating layer is heated up with e. g. 100 W with fvOptions untility. The simulation is steady-state and runs without difficulties. The temperatrue distribution on the floor layer is physically realistic and the temperature values are in agreement with my practical experiences, see attached picture OF4_Temperature_fullFloorHeating_topView.png. From the results I have calculated the convective heat transfer coefficient and compared it with experimentally measured results. The comparison showed a rather good agreement as long as y+ of the floor boundary layer was below 5.

Normally, a floor heating system does not cover the whole floor area but only a part of it. Due to the encouraging results of the complete floor simulation i set up a partial floor heating simulation, see the attached picture caseSeup_partialFloorHeating.png. Here you can see, that the insulation and the heating layers cover only a part of the complete floor area. Again, the simulation runs without any difficulties.

When I visualized the results with paraview I found a very strange temperature distribution for the surface of the floor layer, see attached picture OF4_Temperature_partialFloorHeating_minTemp_topVie w.png. You can see that on the surface of the floor layer (which is in the interior room) the simulation shows temperatures of 264,4 K which is below the temperature of the environment (271,15K) though the heating power was as high as to reach a maximum temperature of about 390 K. This seems to me physically impossible.

Lookiing for a reason for this strange result I changed a lot of things (fvSchemes, fvSolution, thermophysicalProperties, mesh,...) but the phenomenon of "under-temperature" did not disappear, though it changes its value. It became lower if I refined the mesh and if Ichanged enlarged thermal conductivity of the floor set-up. But it was always present. I made up to 100000 iterations in order to be sure to be in the vicinity of a convergent state. I have the feeling that that I get a mathematically possible solution but it is not a physically possible solution.

Up to now I have no idea what I can do to get physically more realistic results and I hope that someone of you can help me solving this problem. Maybe, my interpretation is completely wrong or my set-up has a basic mistake. I have attached the case for the partial floor simulation which is not exactly the setup to reproduce the attached images but the "under-temperature" can be reproduced.

Many thanks for your help in advance.
Best regards
Wolfgang

[Wolfgang57]

I have simulated the partial floor heating system again with and without radiation in order to see what the influence of the radiation might be, see attached pictures which show the temperature distribution of the surface of the floor layer with and without radiation.
The temperature distribution without radiation shows the effect of "under-temperature" I have mentioned in my previous post: it is -to my opinion - a physically not possible result. The temperature distribution of the case without radiation shows physically meaningful results, see attached picture OF4_noRadiation_partialFloorHeating.png. All values of the temperature are above the value of the environment and the plotted temperature along a line in the lower part of the picture (non-heated part of the floor layer) shows the expected behaviour.

Then I looked at the distribution of the radiation Qr of the interface between the interior region and the floor layer, see attached picture OF4_Radiation_partialFloorHeating_Qr.png.

As far as I know the interpretation of Qr is the following: If a considered surface has a higher value of temperature than the surrounding surfaces, then the considered surface emits a net amount of radiation and the surrounding surfaces absorb this amount, e. g. the heated part of the floor layer surface has a higher temperature than the ceiling and the interior plaster. Therefore, the heated part of the floor layer emits a net amount of radiative flux, whereas the others absorb it. In Openfoam 4 it is the convention that the radiative heat flux is negative if a surface emits a net amount (is my assumption okay?).



Now, looking at the Qr-distribution of the interface between interior region and floor layer it can be seen the following:

1. The upper part of the picture shows that Qr is positive. This means the floor layer absorbs radiation from its surroundings which is to my opinion okay because the surface temperatures of the floor layer are lower than that of the interior plaster and the ceiling.
2. The middle part which contains the heated part of the floor layer shows high negative values of Qr which means they emit net radiation. This is clear because the heated part has a much higher temperature than the interior plaster and the ceiling. The most left and right part show positive values of Qr because they are not heated, compare the correspondig picture of the temperature distribution.
3. The lower part shows in its middle as high negative values as the heated part, though the temperatures are the lowest of all. In this part the temperatures are much lower than the temperatures of the interior plaster and the ceiling (even lower than in the outer environment) but they emit the same net amount of radiation as the heated part! This is -to my opinion- clearly not possible.

The above mentioned results show that turning the radiation off simulation with chtMultiRegionSimpleFoam gave physically meaningful results. Turning radiation on yields unphysical results. Could the reason for this behaviour be that there is a severe bug in the solver if radiation is included?
If I carry out the corresponding transient simulations the results are the same.

Please, let me know if anyone of you made the same experiences or even has a solution to this problem. Maybe,my case setup or my interpretation is wrong, then,please tell me.


Best regards
Wolfgang

[Wolfgang57]

Does nobody have an idea? I am grateful for every hint.

[peterhess]

The radiation seams to have difficultiues in the simulation.

I made the folowing changes:

Innenraum_to_Innenputz
{
nFacesInCoarsestLevel 600;
featureAngle 1;
}
Innenraum_to_Bodenbelag
{
nFacesInCoarsestLevel 500;
featureAngle 1;
}
Innenraum_to_Decke
{
nFacesInCoarsestLevel 500;
featureAngle 1;
}

To increase the resolutin of the radiation.

Regards

Peter

[Thomas Baumann]

Hello,

I know it is a bit late. But have you already found a solution?

Could it be due to very low kappa values of the materials (the density and the cacapcity of the solid has no influence)?

At the moment I have similar problems with a case with a thick layer of insulation that has a lot of radiation by nearby installed hot rods.

Regards Thomas