[MikeITA]

Hi all,

I'm using FLUENT to simulate the solidification process of water. The geometry is a cylinder chilled through the external surface. When I set up the properties of material like a polynomial I obtain perfect results but if I set up the properties of material like a piecewise-linear, for a better accuracy, I obtain not smooth and irregular results. What could be? The boundary conditions are invariated...
Thanks

Mike

[Simbelmynë]

Hi,

I guess you mean piecewise polynomial, not piecewise linear?
This depends on the experimental range of validity and should probably be more valid. However, if the transition between different ranges is not smooth then this may be the cause of your non-smooth results. You should check this by a simple hand calculation.

Secondly, why do you wish to change something that gives perfect results?

[MikeITA]

Hi Symbelmyne,

no sorry, I wrote well. I know piecewise-polynomial would be better but in my case some water properties have a gap at the transition temperature so a piecewise-linear gives to me a good accuracy as well. First I want change the properties even if I have smooth results because a polynomial function don't fit very well the real trend of the properties. Second that results are smooth but different from my experimental results.

Anyway you probably are right, that irregular results are caused by the gap that, for instance, the thermal conductivity has at the transition temperature. Is there something I could do to regularize the results?

Thanks very much for your help.

Mike

[Simbelmynë]

I am not sure what you mean by regularize the results.

Have you compared the results to experimental data?

and in that case

Is there a large discrepancy compared to experimental data?

Your solidification front might need a very fine mesh to resolve it properly. Have you done a mesh sensitivity study to rule out possible effects of the mesh size?

[MikeITA]

During the transition the temperature should be constant but I obtain an oscillation, so with regularize I mean see clear results and linear during the phase changed. I compared the results with the experimental data and they have a large discrepancy, the temperature during the solid-liquid transition can swing around from -2 to +2 degrees when should be a constant value (0 degrees).

The mesh is very fine and it's ok, considering that I have a 2D geometry and I'm doing a axisimmetric simulation. I've done a mesh analysis and all the parameters are ok.

[Simbelmynë]

Not sure what it can be in this case. You have a very low latent heat I guess if there is no temperature change during the phase transition. Are the experimental measurements published, and if so do you have a reference?

Anyways, with the polynomial expression your simulation fits the experimental data perfectly? In this case the gaps in the piecewise expression seems to be the problem (although I do not understand why you wish to use a piecewise expression when the polynomial expression fits the experiments perfectly).

[MikeITA]

firstly thanks for all the time you are dedicating to me. Secondly the experimental data are obtained by the experiments I'm doing.

With the polynomial expression the results are smooth and clear but not like the experimental data. I wish to use a piecewise expression because a unique polynomial can't fit perfectly the property trend. The thermal conductivity, like density or specific heat of water, has a big gap in corrispondence of 0 degrees.

Here my email, contact me if you are interested to receive some pictures. I can't upload them on the website:
n1705195j@e.ntu.edu.sg

Thanks again

Mike