Diffusion Problem

[materialmoop]

Hi,

I am trying to model carbon partitioning after quenching/heating. However, my carbon is partitioning the wrong way (from FCC to BCC). Why might this be? I have tried changing the databases in the Ges5 file, and have trying adding in cementite, but I still experience this problem. I have attached my driving file and my Ges5 file. Thank you!

[Bernd]

Hi,

I cannot see an obvious reason for the behaviour which you describe. As you are reading the initial microstructure from a restart file, I cannot run the simulation to see myself what happens. Could it be that you changed the phase identity between the first simulation which created the restart file and the current input file, so that fcc regions in the initial microstructure read from the .rest-file will now be treated as bcc and vice versa?

Bernd

[materialmoop]

Hello,

I was able to solve the problem by inputting the parameters for the diffusion/phase diagram directly

However, I am trying to do a second quench (as part of quench and partitioning), and when I do so, all of the austenite disappears. I know there is a minimum amount of carbon needed in the austenite in order to stabilize it, but regardless of what the amount is (I have tried changing it using the "limits" option), it still disappears a few seconds into the second quench. I have trying using the driving force offset at various values, but it does not seem to help; it simply causes the grain boundaries to grow very large.

The only thing that seems to help is changing the kinetic coefficient between FCC/BCC, but I assume I must keep that constant between the first and second quench? I am not sure what else the issue could be. I have attached all the files below (T0015relin is the first quench,= input for heatrelin. heatrelin=input for cpartrelin. cpartrelin= input for quench2) . Many thanks!

[Bernd]

Hi materialmoop,

To be honest, I do not understand what you are doing and which type of process you want to model. I also do not understand, why you use coupling to a (now binary) .ges5-file, while you do not use it neither for thermodynamics nor for diffusion coefficients . On top of that, you manually defined diffusion data which always gives you values <1.E-19 cm2/s which is practically equivalent to 0 (i.e. you only consider massive transformations everywhere! ) (You can check that the .conc1 output is not changing during all 4 steps).

I would advise you to use thermodynamic data from database, because with a single linearised phase diagram description you cannot even approximately describe the fcc-->bcc transformation over a large temperature interval. Furthermore, you should use realistic diffusion coefficents, preferably from the database.

After that, your simulation results will look very different. Now, you need to decide which type of phase transformation you are aiming at. If it is a "normal" γ/α-transformation, I would use "redistribution_control" for the 1/2-interface with "normal mobcorr" for carbon. This would give you diffusion-controlled growth kinetics. You probably would have to increase grid resolution in this case to resolve the diffusion profiles properly.
Otherwise, if e.g. you assume a martensitic transformation without diffusion control (massive transformation) for the quenching part, you should assume "para" instead of "normal mobcorr" to allow for smooth overrunning of the concentration field without numerical trouble and with well-defined phase transformation kinetics. But during slow heating or holding, you always should assume diffusion-control to allow for proper partitioning.

By the way, you are "twice" reading the restart file in your current input files. The second definition under "Initial microstructure" is pointless, because it is overwritten by the direct restart function. I personally would prefer to use only the second one (under "Initial microstructure"), because it gives you more flexibility in changing parameters between the individual steps. Please note that you would have to remove the initial grain setting in order to make it work, as you have set to read the restart file into grain 0.

Best wishes
Bernd

[materialmoop]

Hello,

Thank you for your response and for your suggestions. I implemented them, but unfortunately am still facing the problem. I am trying to model quench and partitioning. In the first driving file, I am starting from a fully austenitic microstructure and am quenching to the first quenching temperature. In the second file, I am just heating up to the partitioning temperature. The third file is the carbon partitioning, and then a fourth file involves quenching to room temperature. The preceding file's microstructure is the input for the next file.

However, even when using a ThermoCalc database, the carbon somehow partitions from FCC to BCC. I have tried refining the resolution as much as possible following these steps: https://docs.micress.de/7.2/micress/top ... mobilities but still experience the same issue when using the ThermoCalc database only... could this be an issue with my database?

Many Thanks

[Bernd]

Hi materialmoop,

Sorry for my late answer. I am on holidays right now and not able to regularly check the forum.

I cannot tell you easily which problems could still be present. Did you use realistic diffusion coefficients now? This should change everything completely.

Can you send again your updated input files together with the .GES5-file? Then I could check myself what may be the problem. Please excuse if my answer will again be retarded because of my travel.

Bernd

[materialmoop]

Hi,

No problem The diffusion coefficients are more realistic now, but unfortunately the problem remains. I have attached all my files with the .GEs5 file.

Thank you so much!

[Bernd]

Hi materialmoop,

The problem seems to be located in your .GES5-File. How did you create it? If I use yours, it segregates in the wrong direction as you say. However, If I create an own one like that:

@@ LOGFILE GENERATED ON UNIX / KTH DATE 2013- 9-11
@@
@@ GES5 file generation for Gamma Alpha (FeCMn)

set_ges_version 5
go dat
sw tcfe13
d-e fe c mn si
rej phases *
rest phases liq bcc_a2 fcc_a1
get-data
app mobfe8
d-e fe c mn si
rej phases *
rest phases liq fcc bcc
get data
go gibbs
save TC2024a_tcfe13_mobfe8_FeCMnSi
exit

it works correctly.

I still would like to better understand what you intend to achieve in the different simulation steps. I think that your DG_offset for the fcc/bcc-interface has the wrong sign (bcc is stabilized), and do not understand why you do not allow for this phase transformation during heating and partitioning while at the same time applying extremely fast ripening of bcc.

Best wishes
Bernd

[materialmoop]

Hi Bernd,

Thank you for your reply Unfortunately when I tried to paste that script for the .Ges5 file into thermocalc and then into MICRESS, it tells me that the database doesn't contain any elements...I have tried changing the mobility databases (6, 5 ect) and still get "database doesn't contain any elements". I used the original script from the Micress challenge "creating a .Ges5 file" but added in other elements (Mn, Si):


set-echo
go dat
sw TCFe11
d-sys fe c mn si
l-sy CONSTITUENT
l-sy phases
rej phases *
rest phases liq fcc bcc
get
app mobfe5
d-sys fe c mn si
l-sy CONSTITUENT
l-sy phases
rej phases *
rest phases fcc bcc
get
go gibbs
save FeCMnSi

My goal is to model quench and partitioning, and I wanted to create one input file for the first quench and have that microstructure be the input for heating. That microstructure would be the input for the partitioning step, and that microstructure would be the input for the final quench.
As for the first file, I wanted to start with a fully austenitic microstructure and quench to the expected amount of martensite (about 76% for this quenching temperature). I used a dG offset to try to model martensite as acicular ferrite, and held the interface as stationary during the heating and partitioning steps so that the martensite would not continue to grow and overtake the austenite.

Many thanks!

[ralph]

Hi materialmoop,

you should get error messages in the Thermo-Calc console if you check Bernd's script command by command. Do you have a TCFE13 licensed at all?

I think in the script from the MICRESS challenge, you should be more specific when restoring the phases. There are multiple FCC and BCC phases defined in the database. You might get the wrong one restored or get an error message (TC2025b).

Best,
Ralph