Simulation of dendritic growth in 3D

dendritic solidification, eutectics, peritectics,....
Post Reply
Shenyz
Posts: 23
Joined: Mon Jul 25, 2016 4:33 pm
anti_bot: 333

Simulation of dendritic growth in 3D

Post by Shenyz » Thu Jan 16, 2020 11:26 am

Hi Bernd,
I have performed a 3D simulation of dendrite growth with your help. I put the initial seed of the solid phase at the lower-left corner(x=1,y=1,z=1)of the simulation domain and let it grow freely with a temperature gradient and cooling rate. However, after some time I find the structure is different from what I think it would be.
0_06C_frac 1.png .png
0_06C_frac 1.png .png (19.58 KiB) Viewed 1873 times
From the result, you can see the dendrite grow along Z direction is curved rather than straight. I am wondering what's the reason for that? The boundary conditions are ssssif for WESNBT.
Thanks and best regards,
Yaozu

Bernd
Posts: 1179
Joined: Mon Jun 23, 2008 9:29 pm

Re: Simulation of dendritic growth in 3D

Post by Bernd » Thu Jan 16, 2020 12:58 pm

Dear Youzu,

What you observe essentially is tip splitting (taking into account mirror symmetry in x direction). In principle, tip splitting can be a physical phenomenon, but in your case it is certainly numerical.

The detachment of the dendrite arm which grows horizontally could simply be due to the (artificial) isolating condition at the bottom. However, because this boundary does not exist in reality, this problem is not important.
The important issue is the detachment of the dendrite arm which grows vertically (tip splitting). I would start with trying the following:

- Take care that the initial seed in the corner of the domain is positioned correctly. If it is a "small grain" (r<Δx), then it should be positioned to the center of the corner grid cell (Δx/2,Δx/2,Δx/2), otherwise exactly into the corner (0,0,0).

- It may be helpful to prevent tip splitting in the early growth state by starting from a "big grain" which consists of few grid cells. Even better is using an elliptic shape in z-direction. In worst case it may help to use the "restart structure_only" function for starting with an already developed dendrite tip.

- It is important to set the initial grain as closely as possible to the tip temperature for stationary growth. You can calculate this temperature using e.g. the KGT model and a pseudo-binary phase diagram description.

Bernd

Shenyz
Posts: 23
Joined: Mon Jul 25, 2016 4:33 pm
anti_bot: 333

Re: Simulation of dendritic growth in 3D

Post by Shenyz » Fri Jan 17, 2020 9:59 am

Dear Bernd:
Thanks for your help! I change the position of the initial seed to (0.0.0), then the tip splitting disappears. However, there is a problem with the result. When I observe the output, the .phas and .frac show nothing while .conc can display the simulated result, as shown in the picture. This is the first time I meet this situation. How to deal with that?
And now the result of my 3D simulation is output in 2D. Can I change the results already output to 3D in DP_MICRESS and how to do that?
0_06C_conc2.png
0_06C_conc2.png (73.46 KiB) Viewed 1861 times
Best regards,
Yaozu

Bernd
Posts: 1179
Joined: Mon Jun 23, 2008 9:29 pm

Re: Simulation of dendritic growth in 3D

Post by Bernd » Fri Jan 17, 2020 3:31 pm

Dear Yaozu,

I have never seen what you describe. Obviously, the .phas image which is shown on the right side corresponds to an earlier time step, because you can see the small grain in the lower left corner. However, the annotation text below the image shows the correct time of 1.2 s.
What do yo mean by "And now the result of my 3D simulation is output in 2D"? According to the annotation text you display layer 1 of 192...
Can it be that DP_MICRESS has problems because your computer is running out of memory?
Which file format did you write from MICRESS? I have not so much experiences with reading .vtk with DP_MICRESS. I usually always write results in native format and use DP_MICRESS for converting afterwards to .vtk when viewing with Paraview.

Bernd

Shenyz
Posts: 23
Joined: Mon Jul 25, 2016 4:33 pm
anti_bot: 333

Re: Simulation of dendritic growth in 3D

Post by Shenyz » Sat Jan 18, 2020 12:28 pm

Dear Bernd:
It may because my computer is running out of memory. I will try to display the results on another computer.
As the same to you, I write results in native format, but I don't know how to use DP_MICRESS for converting them to .vtk format to view with Paraview. Could you please tell me how to do that?
With kind regards,
Yaozu

Bernd
Posts: 1179
Joined: Mon Jun 23, 2008 9:29 pm

Re: Simulation of dendritic growth in 3D

Post by Bernd » Mon Jan 20, 2020 5:22 pm

Hi Yaozu,

Just go to "File/Export window content as dataset". Then you can switch format to vtk and select the outputs you want to join in the .vtk files. Further options should be self-explanatory.

Bernd

Post Reply