Dear Bernd
Today, i have a question about the Solidification of Zn-Al.
I make a GES file. And i did Sheil simulation to see which phase i should consider for phase field simulation. Unfortunately, mob2 database in the Thermo_Calc. does not have mobility information of Al and Zn in the specific phase such as HCP_Zn and etc. I arbitrary put diffusivity values for each phase that i considered for the simulation. Therefore, in terms of diffusivity, i think we have to further optimize. But currently diffusivity is not our major concern.
The question is as follow,
As you can see from the attachment, i want to simulate lamella structure. In our experimental observation, there is large size and equiaxed HCP_Zn phase and also fine lamella structure between HCP grains consisting of Al-rich FCC and Zn-rich HCP. How can i simulate such a kind of microstructure with MICRESS? I want to simulate similar shape with our experimental value. Is it possible?
For your information i put the driving file and ppt file containing some explanation.
And the Al contents is 1 wt%
Best regard
Zn-Al solidification
-
betleenkim
- Posts: 41
- Joined: Tue May 07, 2013 1:01 pm
- anti_bot: 333
Zn-Al solidification
- Attachments
-
- ZnAl.txt
- driving file
- (20.82 KiB) Downloaded 314 times
-
- Question about lamella structure.pptx
- Question
- (42.78 KiB) Downloaded 296 times
Re: Zn-Al solidification
Hi betleenkim,
Yes, I think it is possible to simulate that! Did you already try to run the input file which you appended?
After checking the driving file I have the following remarks/thoughts:
1.) For the initial microstructure, you define 9 random seeds which are equally distributed in the domain. First of all, by using "voronoi", they will fill up all the domain and no liquid will remain!
Apart from that detail, I would expect that the hcp-Zn phase nucleates on the substrate surface. I personally would start with two small initial grains at the surface at positions 0,0 and 0,20µm. 20µm should correspond to the thickness of the liquid layer. The eutectic phase will form once they have grown sufficiently and only few liquid has remained between them.
In a later step, you could increase the simulation domain or even try 3D simulation.
2.) The anisotropy of the phases seems not to have any importance on this small scale as the eutectic seems to be regular. I guess that the size of a dendrite/grain is much bigger. On that scale, anisotropy would be important. But on the small length scale, anisotropy probably (especially of the 2/2-interface) is not needed and only unnecessarily complicates things...
3.) You apply symmetric boundary conditions to the east and west side of the domain. This only makes sense if you assume symmetry planes there! I my proposed setup (see 1.)), symmetric conditions would make sense, because the initial grains are put exactly into the symmetry planes. In a random initial structure, you should use isolating or periodic boundary conditions!
4.) I do not see why you propose to use particle pinning for the 2/2 interface - do you intend to simulate grain growth afterwards? This is at a completely different time scale as solidification!
5.) You intend to include nucleation of phases 1 and 2 from the melt. From my understanding, the solid phases should nucleate on the solid-liquid interface of the respective other phase, i.e. phase 1 on 0/2 and phase 2 on 0/1. You defined the second seed type as phase 2 on 0/2...
6.) The definition of the relinearisation scheme should be either for all interfaces, or for each phase interaction separately. It makes no sense to do both!
(Remark: If the intervals are exactly identical, nothing bad would happen. If they are not, or there is a numerical shift between the exact time steps, relinearisation could happen twice, with a corresponding performance loss!)
Please keep us informed about your progress!
Bernd
Yes, I think it is possible to simulate that! Did you already try to run the input file which you appended?
After checking the driving file I have the following remarks/thoughts:
1.) For the initial microstructure, you define 9 random seeds which are equally distributed in the domain. First of all, by using "voronoi", they will fill up all the domain and no liquid will remain!
Apart from that detail, I would expect that the hcp-Zn phase nucleates on the substrate surface. I personally would start with two small initial grains at the surface at positions 0,0 and 0,20µm. 20µm should correspond to the thickness of the liquid layer. The eutectic phase will form once they have grown sufficiently and only few liquid has remained between them.
In a later step, you could increase the simulation domain or even try 3D simulation.
2.) The anisotropy of the phases seems not to have any importance on this small scale as the eutectic seems to be regular. I guess that the size of a dendrite/grain is much bigger. On that scale, anisotropy would be important. But on the small length scale, anisotropy probably (especially of the 2/2-interface) is not needed and only unnecessarily complicates things...
3.) You apply symmetric boundary conditions to the east and west side of the domain. This only makes sense if you assume symmetry planes there! I my proposed setup (see 1.)), symmetric conditions would make sense, because the initial grains are put exactly into the symmetry planes. In a random initial structure, you should use isolating or periodic boundary conditions!
4.) I do not see why you propose to use particle pinning for the 2/2 interface - do you intend to simulate grain growth afterwards? This is at a completely different time scale as solidification!
5.) You intend to include nucleation of phases 1 and 2 from the melt. From my understanding, the solid phases should nucleate on the solid-liquid interface of the respective other phase, i.e. phase 1 on 0/2 and phase 2 on 0/1. You defined the second seed type as phase 2 on 0/2...
6.) The definition of the relinearisation scheme should be either for all interfaces, or for each phase interaction separately. It makes no sense to do both!
(Remark: If the intervals are exactly identical, nothing bad would happen. If they are not, or there is a numerical shift between the exact time steps, relinearisation could happen twice, with a corresponding performance loss!)
Please keep us informed about your progress!
Bernd
-
betleenkim
- Posts: 41
- Joined: Tue May 07, 2013 1:01 pm
- anti_bot: 333
Re: Zn-Al solidification
Dear Bernd
I am really appreciate your answer.
I will try following your instruction and check the result.
Thank you again
Best regard
I am really appreciate your answer.
I will try following your instruction and check the result.
Thank you again
Best regard