Change phase model

[GastónGarcía]

Hi, someone can tell me which are the differences between the three models proposed by Elmer (time, spatial 1, spatial2)
Thanks
Gastón

[raback]

Hi

I'll try to give a rough explanation with the help of some copy-paste from previous topics. Now the basic problem is to extract effective heat capacity from the enthalpy curve.

In the spatial versions the effective heat capacity is approximated from
c=dE/dT
where in "spatial 1" the differential is analytical based on the polyline, while in "spatial 2" dT related to temperature difference between nodal values of element.

The temporal uses the chain rule,
c=(dE/dt)/(DT/dt)
where the differentials are taken with respect of time.

The inherent problem is that the melting heat results basicly to a delta peak in the heat capacity that must still be somehow integrated accurately enough with the given gaussian points. Therefore some smearing is advicable. The 'spatial 2' won't jump over melting point since no matter how small the bandwidth it will be located in some elements. The other versions are to my understanding more needy of smearing.

-Peter

[GastónGarcía]

Thank you

[GastónGarcía]

I have a problem, after reading, I thought that spatial 2 is the best option. I had the whole model probed without the phase change option and it worked perfectly, then I set a table with the values of the enthalpy depending on temperature and set the phase change option. After that my model lose coherence. Someone can tell which was the problem?
Thanks

[raback]

Hi, It's an explicit method i.e. only conditionally stable. Try with smaller timestep. Also remember that the normal heat capacity must be incorporated in the enthalpy, and that the units are J/m^3. -Peter

[GastónGarcía]

Excuse me, but I make a step 10 times smaller and the problem is still that the final temperature is bigger with phase change than without it.

[GastónGarcía]

I found the problem, I was failing in the enthalpy values.
Thanks