Question about latent heat

Numerical methods and mathematical models of Elmer
Post Reply
mabr
Posts: 10
Joined: 19 Apr 2018, 15:53
Antispam: Yes

Question about latent heat

Post by mabr »

Hi Elmer users!

I am trying to benchmark Elmer against a 3D Abaqus study. The case is solving transient heat equation with heat flux boundary condition, that is applied for three seconds and after that there is a cooling phase (this heat flux is set to 0) of 10 minutes. Then this cycle is repeated three times. The data specified in Abaqus is density, specific heat and heat conductivity (all temperature dependent). I was able to implement all these settings and the case is converging fine. However, in Abaqus latent heat option was set with 200kJ/kg to be melted if temperature of 1000 degrees is reached. So this is the part that I am struggling with.

I am a total beginner in phase change, so I have a few basic questions to point me in the right direction.

Is it possible to specify heat capacity and latent heat directly like in Abaqus? Going through /elmerfem/fem/tests/ I noticed there is a case where latent heat and melting point are specified as material property. Is this the case I should study in more detail?

I noticed that in the tutorial (heat equation 1d - temperature of an idealized geological intrusion) a specific enthalpy was created which includes heat capacity and latent heat release. Is it possible to convert heat capacity and latent heat into specific enthaply and work with specific enthalpy instead? I have noticed in many other threads here on the forum that mostly talk about enthalpy or specific enthalpy when it comes to phase change.

How would you tackle this?

Thanks for your time in advance,
regards, mabr
raback
Site Admin
Posts: 4801
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Question about latent heat

Post by raback »

Hi

There are methods for phase change that track the interface (lagrangian) and methods where the mesh is fixed (eulerian). The given melting point is typically related to the lagrangian methods where the temperature on the interface may be set. If the interface is not exactly at the element boudary you can not set the temperature, but rather you set the flux via effective heat capacity derived from enthalpy.

I suggest that you study Elmer Models Manual Ch. 1 and test case PhaseChangeB. There the enthalpy is given an a rather broad interval which helps but the "spatial 2" method should be pretty robust also for sharp enthalpy (i.e. almost a step function).

Enthalpy is needed because if you would directly give effective heat capacity at worst the curve would have a infinitely hight delta peak which would could never be accurately capctured by the numerical integration.

-Peter
Post Reply