Dendritic solidification

dendritic solidification, eutectics, peritectics,....
Post Reply
sheuli
Posts: 1
Joined: Fri Jan 04, 2019 8:41 am
anti_bot: 333

Dendritic solidification

Post by sheuli » Mon Jan 14, 2019 12:50 pm

1. The phasefield solution and dendrite morphology should be grid independent. However, in our solution of dendritic solidification, the morphology changes with grid spacing. This indicates grid dependent solution of phase field equation.
How this issue can be resolve to obtain a grid independent solution?

2. No SI units are found to be included in the example problems. The units are found to be CGS units.

3. The software shows units of enthalpy and specific heat as J/cm3 and J/cm3xK respectively. Where in principle it should be J/gm and J/gm °K. It is not understood why mass units were replaced by volumetric units and the data for volumetric units for enthalpy and specific heat are not available.

Bernd
Posts: 1074
Joined: Mon Jun 23, 2008 9:29 pm

Re: Dendritic solidification

Post by Bernd » Mon Jan 14, 2019 6:01 pm

Dear sheuli,

Welcome to the MICRESS forum!

The phase-field solution is grid independent if resolution is sufficiently high. In this respect, MICRESS is quite good as it provides stable and correct solutions already at a relatively coarse resolution. However, dendritic growth is a chaotic process, and as such it will never produce a microstructure which is independent from numerical details even at very high resolution. By the way, the same is observed in experiments where you never will get the same structure twice...

It is true that in MICRESS we CGS units in many places. The reason is that this is often more convenient - of course it also depends what you are used to.

The phase-field equation is formulated in terms of volume. In case of using mass units you would additionally need the density or molar volume in order to calculate the heat release per volume which is needed for the phase-field equation. Therefore, when working without thermodynamic database it is more convenient to enter the volumetric units.
If the simulation is coupled to a database which also contains volume data, it is no problem anyway...

Bernd

Post Reply