Reg. Modeling of alpha-gamma transformation

[sachn]

Hi all,
I'm new to Micress. I wanted to model the formation of Gamma from alpha, during heating in steel. I found one example in micress, which shows gamma to alpha transformation during cooling, but not for the reverse. Do you have any standard example of a heating simulation? In particular, I'm confused of how micress is going to take the 'undercooling' parameter, as in this case, its heating and not cooling,

Thanks,
Sachin

[Bernd]

Dear sachn,

Welcome to the MICRESS Forum.

Essentially, there is no principal difference between heating and cooling, apart from the fact that now bcc phase nucleates. The "critical undercooling" for nucleation should always be positive. Essentially, you should think of it rather as a driving force for nucleation, given in units of a temperature.
We do not have a standard example for that, but you can easily start from the normal Gamma_Alpha_dri example. You just change the temperature and the relation between phase numbers in MICRESS and the database. Then, phase 2 is bcc and should nucleate and grow instead of fcc in the original example.

Bernd

[sachn]

Dear Bernd,
Thank you for your reply,the problem was resolved by making alterations you specified.The output file of the problem is attached.(as private message)
After successfully running the simulation for alpha to gamma heating problem, I wanted to modify it as heating problem for dual phase steel DP600 which I defined as a 6 component system with C,Mn,Si,P,and S as dissolved components in Fe.But when running the result I got demixing errors for P ,Si and S at the bcc/fcc interface.So I defined these three components as stoichiometric (in both fcc and bcc phases),then the simulation was running but the round nuclei at the interfaces instead of increasing in size was spreading out linearly along the interface:suspecting it to be some problem associated with interface mobility between fcc and bcc I decreased the kinetic coeffecient between fcc and bcc from 1E-5 cm**4/(Js) to 7E-9 cm**4/(Js) with the surface energy between the phases set at 2E-5 J/cm**2.But still I am getting some "trying hard errors" in between even though the simulation was completed .Please let me know whether the changes I made by defining stoichiometry and changing the mobility value sre appropriate in this context, if not what really are the parameters I should edit to make the simulation correct.The output file for this problem is also attached. as private message.


Thanks,
Sachin

[Bernd]

Dear Sachin,

you cannot use stoichiometric conditions in both phases at the same time. To avoid the "demixing" problem you should rather switch to diagonal extrapolation method for redistribution which you can do by using the keyword "interaction" in the Phase diagram input data without further parameters. This avoids the demixing artifact of the somewhat more exact multi-binary extrapolation method. The loss of exactness can be compensated by a (slightly) lower relinearisation interval.

For other people to follow our discussion it would be nicer if you attach screenshots instead of sending result files as PM. Thanks!

Bernd

[sachn]

Dear Bernd,
Thanks for your reply.As you suggested I changed the previously defined stoichiometric relationship between various elements to interaction mode(as you can see in the screen shot attached with this).But regarding reducing the relinearisation interval,I have some doubt because it was previously set as 'manual 1' and # Relinearisation mode for interface 1 / 2 was set as 'none'.When I was reducing any of these intervals together or separately (within the 'interaction' mode),I was getting errors at earlier stages(like 2.02s,3.03s etc).The simulation ran smoothly upto 280seconds when both the relinearisation mode were set as 'none' with a mobility value of 9.5E-9 cm**4/(Js)(please comment on this mobilty value selected)at a surface energy of 2E-5 J/cm**2 between fcc and bcc,then it showed 'trying hard' error.But when the simulations (phase and driving force)were displayed I am getting some instabilities (screenshots attached).Please point out where I went wrong.

Kind regards
Sachin

[Bernd]

Dear Sachin,

With the keyword "interaction" one can switch to the "diagonal" redistribution model and, optionally, furthermore include single off-diagonal terms of the partition matrix ( additional "interaction" between two dissolved elements). If you just put the keyword "interaction" without further parameters, and this was what I proposed for getting rid of the artificial demixing, only the switch to diagonal mode is done. Further including of off-diagonal terms can help to increase extrapolation accuracy, but only if there are strong interactions between these specific elements (you can find out by checking the reduced partition matrix in the .log file). But this expert approach may lead to further complications and should only be done wisely.

From what you show, it is difficult to judge why you run into numerical problems. The value of the interface mobility seems low, but this depends on the time and length scales which I do not know. Can you please paste your complete input file to show these details?
Avoiding more frequent relinearisation because it creates error messages is a bad idea because this behavior demonstrates that the simulation is not stable. You should check for e.g. for unphysical (negative) compositions in the two phases, which would lead to errors always when you try to relinearize.

Bernd

[sachn]

Dear Bernd,
Thank you for your valuable comments.I will make the changes you suggested.Screen shots of my entire input file are attached below.Please go through them and help me locate my errors.

Kind regards,
Sachin

[sachn]

continued....

[sachn]

Please don't forget to comment on the relinearisation interval

[Bernd]

Dear Sachin,

wouldn't it have been much more easy just to past the text of the input file or to append it?

I cannot see anything odd in your input, apart from the interface mobility of the 1/2 interface which is definitively too low. I guess it should be about 3 orders of magnitude higher.

I have no objections against the relinearisation interval.

Bernd