Unphysical results for same thermal diffusivity

Hello everyone,

I am currently working on my master’s thesis. I simulate scenarios of a geothermal doublet over multiple sandstone formations in OGS5.8. Therefore, I have set up an easy model with one sandstone layer surrounded by an almost impermeable mudstone above and below to do a sensitivity analysis.
As the mudstone is assumed to be impermeable, heat transport should only depend on thermal diffusivity in that area.
Changing the specific thermal conductivity and the thermal capacity of the mudstone counterwise to achieve the same thermal diffusivity, yields different simulation results for temperature.
No other parameters were changed in the simulations. Therefore, results were expected to be exactly the same.
I added three figures that show the results. Thermal diffusivity was 0.983 m²/s in all cases.
I already checked if the very low (1e-19 m²) permeability of the mudstone still has an influence, but it does not.
Now I don’t know how to fix this problem anymore.

I added the input files of the base case and would appreciate any help!
Thanks in advance.


Neuer Ordner.zip (5.1 MB)

1 775
1 2.09

1 1150
1 2.974668

1 516.6666667
1 1.480561

Hi Jonas
You use Heat Transport equation and in this equation temperature distrubution in the domain directly is related with thermal conductivity.
If you change the thermal conductivity normally you will see different temp. maps. It is not
related with thermal diffusivity.
look the examples: Heatconduction (Dirichlet)

Hi Murat,
thank you for your answer.

The analytical solution of the example is dependent on thermal diffusivity (as written on the page of your link, for transient heat transport).
I expected my results to be the same, if the non permeable material has the same thermal diffusivity.
Or does inhomogenity have an effect, so the equation cannot be simplified?

Hi Jonas,
thermal diffusivity (α ) isn’t an alone entity, it is combination of thermal conductivity (λ ) and volumetric heat capacity (ρ *C p). To get same results from two different model runs, all entities must be same. You can work also in 1D to check your results.