some qustions of recrystallisation

ripening phenomena, dislocations, grainboundary topology
Bernd
Posts: 1250
Joined: Mon Jun 23, 2008 9:29 pm

Re: some qustions of recrystallisation

Post by Bernd » Mon Apr 12, 2021 9:20 am

Dear Kamran,

I am still struggling to understand the physical background of the process which you describe. What I understand is that you assume that there are two competitive processes: the recrystallisation of martensite (i.e. nucleation and growth of new martensite grains, with a reX-energy lower than the threshold value, inside the old deformed martensite) on one side, and austenite reversion on the other side. Both processes overlap and are incomplete, so that you end up with a mixture of recrystallized martensite, unrecrystallized martensite, and austenite.
In order to have grain boundaries between recrystallized martensite and unrecrystallized martensite in the end, which are not moving any more, you could consider one of the three mechanisms I described above.
But it could also be possible that you do not have these boundaries because there is reversed austenite between the recrystallized and unrecrystallized martensite. In this case, you only need to define the critical reX-energy for nucleation of the new martensite grains such that it is reached only in few of the deformed martensite grains.
Do you have experimental evidence of the partially recrystallized and partially reversed nature of the material?

Bernd

Bernd
Posts: 1250
Joined: Mon Jun 23, 2008 9:29 pm

Re: some qustions of recrystallisation

Post by Bernd » Mon Apr 12, 2021 2:47 pm

Hi Kamran,

Based on the explanations for your process, I propose the following procedure:

1.) You start from a suitable MICRESS standard example, I would suggest A004_Gamma_Alpha_TQ_dri. You can adapt the chemistry of your system as well as the length scale to your needs.

2.) Interchange ferrite and austenite in the phase definition, so that phase 1 is ferrite now. You can use the initial microstructure definition which you showed above for defining the initial martensite (=ferrite) grain structure. You also need to change from holding to heating by changing the "cooling rate" in the Process conditions (to a positive value).

3.) To include recrystallisation, change phase 1 from "no_recrystall" to "recrystall" (in Phase Properties) and define a small energy threshold like in "T33_Rex_Deterministic_dri". For recrystallisation chose nucleation at interfaces (like Seed Type 1 in "T33_Rex_Deterministic_dri").

4.) While the ferrite to austenite transformation is diffusion controlled, recrystallisation is controlled by the 1/1-interface mobility. You should calibrate this value such that recrystallisation and austenitisation proceed in a similar time-scale. You stop the simulation after a certain time, corresponding to your flash heating period. Alternatively, you also can implement a more complex temperature-time curve (heating-holding-cooling) according to your flash heating process. In this case you need to use a T-dependent interface mobility in order to stop the recrystallisation process after cooling.

Kamran
Posts: 12
Joined: Mon Apr 27, 2020 11:51 pm
anti_bot: 333

Re: some qustions of recrystallisation

Post by Kamran » Mon Apr 12, 2021 3:57 pm

Thank you Brend

I have sent the details to your inbox

Kamran
Posts: 12
Joined: Mon Apr 27, 2020 11:51 pm
anti_bot: 333

Re: some qustions of recrystallisation

Post by Kamran » Thu Apr 22, 2021 10:32 am

Dear Brend

I come back with some findings and also other question
I have send for you my code privately as a message...I could find out that in my model, considering the heating rate of 100 K/s and very fast austenite reversion coinciding with delayed recrystallization. mobility of each phase has more stronger effect on prevention of full REX grains rather than REX energy for grains...So, I decreased the mobilities to simulate our experimental works...in this level everything works well and we have rationalized calculating time...but upon introducing Cementite phase (even in min number) the simulation time increases strongly...could you please check my code and lead me to proper way?
thank you for your always supporting manner

Bernd
Posts: 1250
Joined: Mon Jun 23, 2008 9:29 pm

Re: some qustions of recrystallisation

Post by Bernd » Fri Apr 23, 2021 12:31 pm

Dear Kamran,

When you add a new phase to a simulation, and the time consumption goes up drastically, then you probably have either introduced numerical issues, or the simulation time-step got much smaller. The first case you easily recognize by error messages or destroyed interface profiles, the second one is linked to interface mobility.
I can't see from the input file only whether there are numerical instabilities. But if it is the second case, the clear candidate here is the 2/3-interface (fcc-cementite). I strongly advice to use diffusion-limited interface kinetics with the "mob_corr" option, like you do it for the 1/3 (bcc-cementite) interface, which will reduce the interface mobility to the correct value needed for diffusion control.
You can monitor the effect on time-stepping with the .TabT text output.

Bernd

Kamran
Posts: 12
Joined: Mon Apr 27, 2020 11:51 pm
anti_bot: 333

Re: some qustions of recrystallisation

Post by Kamran » Fri Apr 23, 2021 6:18 pm

Dear Bernd

Thank you for your suggestions...actually I just sent the article of my experimental evidences to you...I will adopt your advices to change my code and let you know the results
Regards

Post Reply