PF simulation of Eutectic solidification

[Bernd]

Dear Omid,

the Publication of Y. Yang and Y. A. Chang contains full information on the Gibbs energies of the ternary with all its phases. Therefore the only step which is missing is to put it in form of a .tdb database file. In principle, these .tdb files are simple text files, but you need to know the exact syntax how to write the information from the paper.
The easiest way, I think, is to ask the authors of this paper whether they can send you such a file. I am sure they have one on their computer and will send it to you - the data are public anyway and you will cite their paper in your work...

Bernd

[omid]

Dear Bernd,

Regarding to this question: "# How shall diffusion of component 1 in phase 0 (or 1) be solved?" what are the possibilities rather than "diagonal d" and "diff" ? Could you please advise me for example when the component 1 is "Si" and the target phase is Mo3Si, which one is most suitable choice?

Regards,

[Bernd]

Dear omid,

here is a sketch how to obtain linearisation data in a ternary system.

1.) First, one makes an isothermal section of the phase diagram which shows the tie-lines (equilibrium lines) between the solid phases and liquid. The temperature should be the reference temperature for the pair-wise phase diagram description.

WP_20151103_001.jpg (67.27 KiB) Viewed 16716 times

2.) Select a tie-line for the given phase pair, in this case phase 2 with liquid. It should be chosen such that the phase planes are well approximated by the tangent planes in the reference points. The compositions of the reference points can be taken from the intersection of the tie-line with the phase planes.

WP_20151103_002.jpg (71.84 KiB) Viewed 16716 times

3.) Intersections of the 3D-phase diagram should be drawn through each of the reference points along the direction of the two dissolved elements (K1, K2). In these 2D-phase diagrams, the slopes m of the solidus and liquidus lines can be measured.

WP_20151103_003.jpg (161.02 KiB) Viewed 16716 times

Bernd

[omid]

Dear Bernd,

Now I see what I should look for!
Many thanks for your constructive helps!
Regards,

[omid]

Dear Bernd

Could you please have a look on both the attached photos and the .dri file to find the OBVIOUS ERRORS ? as you see in the photos, there is some thing strange at the bottom of domain (strange shape of nucleis which appears after the first time step) and a clearly strange behavior of solidification!

Regards,

[Bernd]

Dear Omid,

One obvious error is the definition of diffusion coefficients. You set the diffusivity of component 2 to 1.E-8 cm²/s which is typical for a solid but not for a liquid phase. This leads to a "spreading" of the interface which can be observed in your results.

Apart from that, you will need to calibrate the interface mobility values (see here).

When I check MICRESS results for numerical issues, I typically focus on the .conc* and the .driv outputs. The .phas output is good for showing growth of different phases, but not for a check whether the simuation runs properly. The .driv output is also essential for calibration purposes (see above).

Often, it turns out to be difficult to find stable growth conditions for invariant eutectics. In such cases it may be helpful to start from an initial structure which is already close to steady state. In our case, you could already start with three alternating lamellar phases. You could set them by deterministic grain input using rectangular grains.

Bernd

[omid]

Dear Bernd,

In the numerical verification descriptions you introduced a link which seems to be no longer available, could you please upload it again?
link of numerical parameters: http://board.micress.de/viewtopic.php?f ... on+limited
deactivated link:
a) Qualitative work:

Choose a resolution for which you have the feeling that you can afford (in terms of calculation time). Start with a guessed mobility value. If the simulation is crashing or shows instabilities, lower the mobility value until you get a stable run. Look at the .driv output to see whether the movement of the interface is strongly slowed down by the choice of your mobility ( viewtopic.php?f=18&t=25 )! If that is the case, the grid resolution is not fine enough! Try again with a smaller value. On the other hand, if the simulation is running properly with your first guess of the mobility and resolution value, try to reduce the mobiltiy or increase grid spacing to make your simulation even faster.

Moreover, I see another strange behavior, when I am trying to run my test simulation and am visualizing the results, I see that from initiation till some time steps, the growth and colors of both nuclei and matrix are constant, but from one time step the color of background changes suddenly! Do you now what is the reason behind that?

Moreover, when I check on of the tables, I see that the concentration of content 2 in all phases gets NAN after some time steps! Do you have any Idea why it happens?
Another problem is that when I check the Tabl file, I see that from the first time step the temperature is equal to temperature of bottom and it increases by time!! is this phenomena logical?
The last but not least is; as you see in the first picture when the nuclei reach together, they collapse and will not grow up and a bridge shape will get created at the interface!

Regards

[Bernd]

Dear omid,

thank you for reporting the link errors. They are remainders from the latest move of the micress forum to a new server. I have repaired them. If you find more of them, please remind me.

The changing background color you describe could be simply due to automatic rescaling in DP_MICRESS. Which output are you looking at in this case, and which are the values between it switches?

The occurance of NAN value is a sign that something is going really wrong. It typically results from a division by zero, eg. if you would have a phase diagram slope with value 0. How is your phase diagram description looking like?

The "bridges" are results of an interface instability. It could result from too high interface mobility values. Which values did you chose for the solid-solid interaction? If solid-solid interactions are activated, the corresponding mobility values should be at least 4 orders of magnitude smaller than those of the solid-liquid interfaces. This is because of the difference in the diffusion coefficients of solid and liquid.

Bernd

[omid]

Dear Bernd,

-Many thanks for the reply. Sure as soon as I find any of those kind of errors I will inform you immediately. You are welcome

-I see the color changing in phase volume fractions output (phas.mcr), the problem happens when the nuclei got together and collapse! And till for example volume fraction of liquid phase equal to 97,6 % the nuclei reach together and the growth will stop and immediately the color of background changes and then the volume fraction of phases will never change and nuclei growth will stop!

-Regarding to the NAN value, I have never defined a 0 slope! So how could be possible to divide to 0? Moreover, when all the compositions are adjusted to exactly the ternary eutectic point and the composition of phases are stoichiometric, how does it matter the value of the slopes? I guess the amount of slopes should never matter. Bin Ich richtig?

-Regarding to the solid-solid interaction, I already have disabled the these interactions, I just have defined the solid-liquid interaction for all phases and all of the numerical parameters for interaction between three solid phases and liquid one are equal to each other! If you see any obvious mistake in this manner kindly let me know!

I have reduced the mobility for 100 times but I still see the problem of bridge!
-Regarding to the delta G, in the phase interaction part, do you have any advise for me? I know that it is of the greatest importance but have now idea about its functionality! I just followed one of the predefined codes and adjusted it as below:
# Data for phase interaction 0 / 1:
# ---------------------------------
# Simulation of interaction between phase 0 and 1?
# Options: phase_interaction no_phase_interaction
# [standard|particle_pinning[_temperature]|solute_drag]
# | [redistribution_control]
phase_interaction
# 'DeltaG' options: default
# avg ... [] max ... [J/cm**3] smooth ... [degrees] noise ... [J/cm**3]
avg 0.5 max 100
# I.e.: avg +0.50 smooth +0.0 max +1.00000E+02
# Type of surface energy definition between phases LIQUID and 1?
# Options: constant temp_dependent
constant

-I also have a big problem that since I am going to model a eutectic solidification, the vertical growth is more desired to me than the horizontal growth, although I have defined the cooling rate and temperature gradient along the Z axis, but I do not see the vertical growth significantly as desired! And in all cases when the nuclei get together, the Tabl or Tabf show that the solidification stops and the volume fraction of phases do not change!
Regards,

[Bernd]

Dear omid,

I mentioned the possibility of slopes with value 0 because I had re-examined a driving file you sent me lately and found them there. But of course there are other sources of NAN values like wrong mobility values. A factor of 100 may be not sufficient.


Anyway, ternary eutectic systems are quite tricky because they are invariant. It could be helpful if you attach or paste your current version of the driving file in this thread so that I can go through and re-check. If there is nothing obviously wrong, I also can try to run the simulation here to find the problem(s).

Bernd