Dendrite growth of slm

[dingling312]

Hi，
Now I am trying to simulate the dendrite growth of slm. When I set the temperature gradient within 10000k/cm（the greater the temperature gradient, the more difficult the dendrite growth is. ）, the dendrite can be obtained. But when the temperature gradient and cooling rate reach the order of slm（cooling rates of the order of millions of Kelvin per second and thermal gradients of the order of tenths of millions of Kelvin per meter）, no dendrite can be obtained, and the calculation speed is very slow. Do you have a solution?
In addition, do you have the example of slm or articles about Additive manufacturing on the homepage?

Best，
Ling

[Bernd]

Hi Ling,

we still do not have a standard example for SLM, but there are publications on simulation of SLM using MICRESS which are also listed on our webpage:

G. Boussinot, M. Apel, J. Zielinski, U. Hecht, J.H. Schleifenbaum
Strongly Out-of-Equilibrium Columnar Solidification During Laser Powder-Bed Fusion in Additive Manufacturing
Physical Review Applied. 11. (2019)
https://doi.org/10.1103/PhysRevApplied.11.014025

C. Kumara, A. Segerstark, F. Hanning, N. Dixit, S. Joshi, P. Nylen and J. J. Moverare
Microstructure modelling of laser metal powder directed energy deposition of Alloy 718
Additive Manufacturing; Volume 25, January 2019, Pages 357-364
https://doi.org/10.1016/j.addma.2018.11.024

I personally have doubts about your "typical" SLM conditions. I would say rather G=1,000,000 K/cm and 100,000 K/s. Apart from that I cannot know what goes wrong in your simulations, but if the total undercooling gets too big I would expect a massive transformation to occur.

Bernd

[dingling312]

Hi，Bernd
Thank you for your reply。
When I try to adjust the time step to "linear_step 5e-5 5e-4", I can calculate the result, but the picture is very rough and the calculation only takes a few seconds. The time step limit is "1.E-7 1E-4".

Ling

[Bernd]

Hi Ling,

It seems that spatial resolution is not high enough and/or parameters are not optimally set. I would need to see the driving file to be able to tell you more...

Bernd

[dingling312]

Hi，Bernd
This is my driving file。As it is showed from the articles you presented, the grid spacing ranges from 0.02 to 0.05 and the number of grids ranges from 300 to 500.
When I extend the time to ensure grain growth, however, because of the very large cooling rate, too long growth time will lead to a very large temperature difference in the growth process（more than 200）. So I have to reduce the interface energy and improve the Kinetic coefficients to ensure grain growth.

Ling

[Bernd]

Dear Ling,

There are several things which I found when going through your input file:

1) "tab_log 0.002": You definitively should write tab_log outputs much more often to get important information in the .TabL, .TabP, .TabT and .TabD text output files

2) "# Limits: (real) min./s, [max./s], [phase-field factor], [segregation factor]
1.E-8 1e-5"
Please take care that you don't cut off interface mobility (see here)

3) I don't understand why you start from a single small initial grain - why not from a flat surface?

4) "avg 1 max 20": The automatic mobility correction ("mob_corr") works correctly only for "avg 0.5". You should use that if possible. "max 20" cuts the driving force at a too low value for SLM - you should increase this value well abov e the solidification interval (for solidification ΔG~ΔT).

5.) The interfacial energy appears to be quite low. In other alloys typical values are 10 times larger. This may lead to too small cell structures in your case.

6.) The interface mobility denotes a "physical" value because you use "mob_corr". It should be high enough (>10). Furthermore, you should use interface stabilitsation, e.g. by a factor of 10:
"1.0E-5 1.0E-4"

7.) The anisotropy of the interface stiffness and mobility are too high. The values should not exceed ~0.3, otherwise facetted-like structures could be formed.

8.) Updating of thermodynamic data is deactivated in your case. You should have relinearisations at least at intervals which correspond to 1-10K of cooling. You can also use "automatic" mode and enter directly the temperature interval.

9.) "normal mob_corr": Automatic mobility correction prevents too fast interface kinetics due to poor resolution by reducing the interface mobility. This increases undercooling up to high values, so that switch to "massive mode" with flat front may occur too early. Please use antitrapping correction (atc) to prevent or at least reduce this phenomenon: "atc mob_corr"

10.) Nevertheless, your grid resolution could be too low...

Bernd

[dingling312]

Hi，Bernd
Thanks for your reply. I still have some questions on your reply start with #6. When you mentioned the interface mobility, is it means Kinetic coefficient mu? Meanwhile, I don't donw which parameter can be used to set the interface stabilitsation.
When I try to creat a flat surface, I replace the grain geometry from round to rectangular like this:

# Input data for grain number 1:
# Geometry?
# Options: round rectangular elliptic
rectangular
# Center x,z coordinates [micrometers], grain number 1?
7.50000
0.00000
# Length along x-axis [micrometers]
15.0000
# Length along z-axis [micrometers]
0.300000

However，I can't get any dendrites.
Ling

[Bernd]

Hi Ling,

Yes, the interface mobility is the same as the kinetic coefficient mu. The stabilisation parameter can be set in the same line after the interface energy like I denoted:

# Interfacial energy between phases 0 and 1? [J/cm**2]
# [max. value for num. interface stabilisation [J/cm**2]]
1.0E-5 1.0E-4

When you start from a flat front it may take some time to get the required instability for a breaking of the front into dendrites. This is much easier if there is some noise or asymmetry (like e.g. a grain boundary or precipitates). In reality, as the base metal is not structureless, you automatically have some reasons for breakup of the front...

Bernd

[dingling312]

Dear Bernd，

I try to modify the input file as you suggest, but I can't get dendrites either。

No matter how I adjust the "avg 0.5 max 100" and "Kinetic coefficient mu " , I can only get such a result.

Ling

[Bernd]

Hi Ling,

Please try with a higher resolution (e.g. Δx=0.01 µm).

Apart from that, you still do not do any updating of the thermodynamic data. Try something like

# Which global relinearisation mode shall be used?
# Options: manual from_file none
manual 1.E-5

Bernd