[HectorRedal]

Hi,

I would like to know if two simulations that have the same Reynolds and Grassof number are thermally equivalent?

I am trying to simulate the problem described in the following article:
http://heattransfer.asmedigitalcolle...icleid=1438948

I am supposing air with Pr = 0.71.
When I double the viscosity, doubling the inlet velocity (keeping reynolds number), and doubling the temperature difference in the formula for calculating the Grashof number, the results I am obtaining are different.
I have always understood that non-dimensional number allow comparison between different simulations/experiments that show same behaviour.

May I am missing something?

As thermal coefficient expansion (beta) for the Grashof number I am using 1/T, being T the low temperature in the problem.

Should I use any other value for the temperature (i.e., the high temperature)? The middle temperature in the range?

Any comment is welcome.

Best regards,
Hector.

[FMDenaro]

You can easily answer by looking at the governing equations in non-dimensional form. If you have the same non dimensional parameters the two non-dimensional solutions must be the same.

[HectorRedal]

Originally Posted by FMDenaro

You can easily answer by looking at the governing equations in non-dimensional form. If you have the same non dimensional parameters the two non-dimensional solutions must be the same.

This was also my understanding, but what I am observing is not this.
The non-dimensional solution differ by almost 5%.

[FMDenaro]

There is a difference in the two simulation, maybe different iteration accuracy

[HectorRedal]

Originally Posted by FMDenaro

There is a difference in the two simulation, maybe different iteration accuracy

If I run a simulation without solving the energy equation, the difference between two solution having the same non-dimensional numbers (but using different values for velocity, or density, or viscosity, etc.) is almost zero.

I only observe this discrepancy when trying to solve a problem with energy equation coupled. The difference between both solutions is more ore less 5%. No more.
For me, it's more than can be justified by the accuracy iteration. Maybe I am wrong.