Cr-Ni eutectic alloy simulation

[Bernd]

Dear lihaoge,

one often overlooked reason for an empty simulation domain is the comoving frame. In the "moving frame" definition in the section "Boundary conditions" you have the choice to define either a temperature or a distance from the top as criterion for movement domain. If a too high temperature or a too big distance is defined, the frame will move, and everything vanishes...

Apart from that, you should not use periodic boundary conditions in z direction if you define a temperature gradient - this does not make any sense and leads to numerical issues. You should use isolated at the bottom and top for the phase-field parameter, and "fixed" (with fixed concentration set to the initial concentration) for concentration field at the top:

# Boundary conditions for phase field in each direction
# Options: i (insulation) s (symmetric) p (periodic/wrap-around)
# g (gradient) f (fixed) w (wetting)
# Sequence: W E (S N, if 3D) B T borders
ppii
#
# Boundary conditions for concentration field in each direction
# Options: i (insulation) s (symmetric) p (periodic/wrap-around) g (gradient) f (fixed)
# Sequence: W E (S N, if 3D) B T borders
ppif
...


Bernd

[lihaoge]

Dear Bernd,

Many thanks for your constructive reply, I have decreased the distance from the top as criterion for movement domain from 200 to 30 and even smaller like 0.1 and changed the boundary conditions like you said. However, there is still no results. I changed the time input data to see if there is some results in the 0.55s later, and there is no results. But after I changed these data, the simulation time increased a little.
Do you have any ideas to deal with this problem? Thanks!

lihaoge

[Bernd]

Dear lihaoge,

the next frequent reason why initial grain vanish is that temperature is too high so that they just vanish. It is instructive and also important to check the initial equilibrium which has been calculated in the .log file:

# The linearisation parameters of the phases LIQUID/BCC_A2 are:
# -------------------------------------------------------------
1636.0000 ! T0 [K]
41.080576 ! dG [J/cm**3]
0.50846838 ! dSf+ [J/cm**3K]
0.50027131 ! dSf- [J/cm**3K]
756.01853 ! dH [J/cm3]
44.005130 ! c0(CR)/LIQUID
47.682397 ! c0(CR)/BCC_A2
4.0136958 ! m(CR)/LIQUID
3.1066241 ! m(CR)/BCC_A2
4.01652445E-03 ! dcdT(CR)/LIQUID
-5.27429134E-03 ! dcdT(CR)/BCC_A2

A positive sign of the driving force dG means that the second phase (i.e. in interaction 0/1 it is phase 1) is shrinking (at least if dS is positive which is typically the case in solidification).
Thus, you need to decrease the initial temperature until dG is slightly negative (enough to overcome curvature).

Please note:
- you can estimate the needed temperature change using the equation dG=dS*deltaT (where dS is the average of dSf+ and dSf-). It tells you that you need to lower temperature by around 80K in order to get BCC stable.
- please generally check also for a reasonable initial equilibrium in this place: The phase compositions c0 should be reasonable, m should also be in a moderate order of magnitude. If the solidus slope (m(CR)/BCC_A2 in this case) is too high, it means that the solubility range of this element in the liquid is low, and you should think about using the stoichiometric approximation (in your case, everything is OK here).

Bernd

[lihaoge]

Dear Bernd,

Thank you very much! I have changed both "Initial temperature at the bottom" and "temperature at which the initial equilibrium will be calculated" from 1636K to 1530K and the dG is -12. Is that means the undercooling is around 100K? Do I have to change the "max. nucleation temperature for seed type 1" from 1636K to 1530K? And after I changed the temperature to 1530K, the grain began to grow up. But as the results showed that the behavior of solidification is strange, could you please have a look on both the attached photos and the .dri file to find where the error is?

Lihaoge

[lihaoge]

Dear Bernd,

I think the microstructure of lamellar eutectic is that one phase1 nucleates first, then the other phase nucleates and grows on nucleus 1, but the results only showed one phase and the grain grow up together, is that because the error input from "grain input"?

lihaoge

[Bernd]

Dear lihaoge,

you should consider the time scale: your whole simulation is scheduled to be finished after 0.55 seconds (time input data), however, the frequency for checking for nucleation is all 1 second - that means you never check for nucleation...

When you will check for nucleation the first time, you will get a similar output in the .log file about the initial equilibrium and the driving force which you should check.

Apart from that, with your current cooling rate of 1K/s and gradient of 1K/cm, and given your time and length scales, your simulation is practically isothermal...

Bernd

[lihaoge]

Dear Bernd,

Thank you very much! I have decreased the frequency for checking for nucleation to 5E-04 and changed the time input as:
linear_step 5E-04 5E-03
linear_step 2E-03 2E-02
linear_step 2E-02 0.55
end_of_simulation

And I have increased the cooling rate and gradient. This time the simulation shows a well results!

But the simulation takes very long time as:

Intermediate output for t = 5.00000E-04 s
CPU-time: 1 s
Current phase-field solver time step = 1.10E-06 s
Average conc. of comp. 1 = 44.0000000 at%
Temperature at the bottom = 1530.0 K
Temperature gradient = 200.000 K/cm
Simulation area shifted by 0 layers
Velocity of the comoving frame = 0.000 cm/s

Grain 2 reaches full size at t = 6.0599E-04 s

Intermediate output for t = 1.00000E-03 s
CPU-time: 650 s
Current phase-field solver time step = 4.00E-10 s
Average conc. of comp. 1 = 44.0000000 at%
Temperature at the bottom = 1529.9 K
Temperature gradient = 200.000 K/cm
Simulation area shifted by 0 layers
Velocity of the comoving frame = 0.000 cm/s

Intermediate output for t = 1.50000E-03 s
CPU-time: 5005 s
Current phase-field solver time step = 3.67E-10 s
Average conc. of comp. 1 = 44.0000000 at%
Temperature at the bottom = 1529.8 K
Temperature gradient = 200.000 K/cm
Simulation area shifted by 0 layers
Velocity of the comoving frame = 0.000 cm/s

Intermediate output for t = 2.00000E-03 s
CPU-time: 20557 s
Current phase-field solver time step = 3.31E-10 s
Average conc. of comp. 1 = 44.0000000 at%
Temperature at the bottom = 1529.8 K
Temperature gradient = 200.000 K/cm
Simulation area shifted by 0 layers
Velocity of the comoving frame = 0.000 cm/s

Intermediate output for t = 2.50000E-03 s
CPU-time: 57026 s
Current phase-field solver time step = 1.97E-10 s
Average conc. of comp. 1 = 44.0000000 at%
Temperature at the bottom = 1529.8 K
Temperature gradient = 200.000 K/cm
Simulation area shifted by 0 layers
Velocity of the comoving frame = 0.000 cm/s

Intermediate output for t = 3.00000E-03 s
CPU-time: 125889 s
Current phase-field solver time step = 3.57E-10 s
Average conc. of comp. 1 = 44.0006349 at%
Temperature at the bottom = 1529.7 K
Temperature gradient = 200.000 K/cm
Simulation area shifted by 23 layers
Velocity of the comoving frame = 0.4600 cm/s

Phase 1 disappeared at 3.4903676E-03 s


Intermediate output for t = 3.50000E-03 s
CPU-time: 160962 s
Current phase-field solver time step = 3.56E-10 s
Average conc. of comp. 1 = 44.0029224 at%
Temperature at the bottom = 1529.8 K
Temperature gradient = 200.000 K/cm
Simulation area shifted by 70 layers
Velocity of the comoving frame = 0.9400 cm/s

Is that because the frequency for checking for nucleation is too small or the time input is too small? Should I increase them? But if I increase them, the results are strange like before.

Thank you very much!

lihaoge

[Bernd]

Dear lihaoge,

you can easily check in the .TabP text output file where all the time is spent. TabP shows the cumulative wallclock time, so you should check the numbers at the bottom. There is an extra column for nucleation, if this column shows the highest values nucleation is the reason.

Bernd

[lihaoge]

Dear Bernd

I am trying to model a new ternary eutectic solidification Cr-Ni-B, the ratio is about Cr55Ni44B. Based on your advises I know the basic knowledge of eutectic solidification, but I still have bunch of questions about my new ternary model. Firstly, I calculate the SCHEIL from thermo-calc and I found that there are 4 phases appear: Liq, Bcc, Fcc and Cr2B-ORTH. But when I creat the .GES5 file I found that the MOBFE2 database does not contain the data of Cr2B-ORTH. I noticed that in the example Delta_Gamma, the cementite is added as a further phase into the simulation. Why the cementite is added as a further phase instead of the same as bcc or fcc? Can I add Cr2B-ORTH as a further phase? If the answer is yes, where should I modify in the .dri-file? If the answer is no, what should I do if I do not have the data of Cr2B-ORTH?

Thank you very much!

lihaoge

[Bernd]

Dear lihaoge,

it is important to distinguish between thermodynamic data and diffusion data. If the Scheil simulation gives you Cr2B-ORTH it means that this phase is included in the thermodynamic database which you use. If you say it is not included in the MOBFE2 then there is obviously no information about diffusion available for Cr2B-ORTH.

You can either neglect diffusion in this phase or search in literature or specify an estimated value manually in the diffusion data input of MICRESS. In many cases, diffusion in intermetallic phases can be neglected in MICRESS simulations because firstly they are typically small particles and secondly often have a reduced solubility range of its elements (so that no relevant gradients exist).

I do not understand what you mean by the term "is added as a further phase into the simulation". All phases which you define in MICRESS are "added into simulation". Or do you mean when creating the .GES file? Or do you mean when defining diffusion data in MICRESS? Here, also all phases are treated equally: They are either included or not included. Please, help me to get the point.

In MICRESS, you need to define all phases which you want to use (phase data input), link them to the phases in the .ges file (phase diagram input) and define their diffusion data (diffusion data input). Furthermore, you need either to set initial grains of this phase (grain input) or nucleate them during runtime (nucleation input), otherwise it will not appear.

Bernd