Pearlite dissolution

solid-solid phase transformations, influence of stresses and strains
Post Reply
Posts: 2
Joined: Tue May 12, 2020 1:42 pm
anti_bot: 333

Pearlite dissolution

Post by vmattosferreir » Fri Mar 12, 2021 3:19 pm

Dear Bernd,

I've been trying to simulate austenite growth in pearlite. I've been trying to match my simulations with what I've observed experimentally. The best fit was obtained with the simulation attached (cb-200Cs-19). As you can see I have two types of growth: at the colony border and inside of the colonies. I have certain questions:
1. The values of interfacial energy are quite difficult to fit and I am not sure about the effect of this value on the morphology. Since my simulations take a very long time it would be great to have an idea of its effect before modifying it. I've seen a lot of literature with quite different values for interfacial energy which makes it harder to choose a suitable value.
2. In these simulations I see the "flat" growth of austenite between the lamella which is also observed experimentally (see attachment "Experimental observation") and undissolved cementite which is expected. However, in experiments we also see some "finger-shaped" growth of the austenite towards the cementite which I am not able to reproduce. In your opinion would it be possible to have some austenite nuclei behave with the "flat" shape and some others with the "finger" shape?
3. Another issue that I have is that the austenite which nucleates inside of the colony grows only along the ferrite without dissolving the cementite on the sides and growing inside the other ferrite lamella.
4. I've tried changing the mobility of alpha/alpha interface to be equal to gamma/gamma and my simulation (cb-200Cs-21) is running for 2 days and nothing happens besides the nucleation, I would like to understand why this is happening.

I thank you in advance for your support.

Kind regards,
cb-19.png (128.83 KiB) Viewed 231 times

Posts: 1221
Joined: Mon Jun 23, 2008 9:29 pm

Re: Pearlite dissolution

Post by Bernd » Mon Mar 15, 2021 11:37 am

Dear Vitoria,

I will try to give you some general answers to your questions as far as possible. You sent me some additional files per PM, but the input file (.in) was not included, so I cannot know your exact input parameters.

1.) The values of the interfacial energy in general determine the amount of curvature undercooling, and thus whether the microstructure will grow with finer or coarser morphology. Secondly, the differences of the interface energy values for the dual phase interactions determine the angle at the triple points which are formed by these interfaces. Here it is worth noting that differences typically should not be too big, given that a factor of 2 already decides between complete wetting and no wetting, and much bigger differences can easily lead to numerical issues or undefined behaviour.
Thirdly, the interfacial energy has an intrinsic numerical stabilizing effect on the interface profile which can be enhanced to a certain extent (max. ~factor 10) by definition of a stabilization value as second optional parameter in the same input line.
If you are unsure about the physically correct values, and you want try out the effects of changing their values, I would strongly advice to use a smaller subdomain for testing, given that simulation of the whole domain is too slow to progress by trial and error.

2.) The question whether you get a flat or finger-like advance of austenite into the pearlite is expected to be the result of an interplay of interface and kinetic contributions. So, it should strongly depend on interfacial energies and their differences, leading to more or less wetting of cementite by austenite, as well as on the relation of interface (interface mobility) and diffusion (diffusion coefficients) kinetics.
While diffusion data come from the mobility database and can be accepted to be known and probably correct, mobility data is typically unknown. My advice is to assume diffusion limited phase transformation kinetics as long as there is no clear indication against. This implies using a (sufficiently) high value of the interface mobility together with thin interface corrections (mob_corr and perhaps also anti-trapping correction atc). If you have trusted sources for physically correct interface mobility values, you still should use thin-interface correction (mob_corr) to correct them for artefacts, which at the same time makes sure that kinetics can never be faster than diffusion limited.
Of course, thin-interface correction applies only to interfaces between different phases, while same-phase interactions remain uncorrected. For those, you need to make assumptions on the physical mobility which could be based on grain growth experiments.

3.) This behaviour appears realistic to me, as the amount of carbon which the austenite is able to take up, is not sufficient to dissolve the cementite lamellae. Obviously, cementite can only be completely removed if austenite grows on both sides.

4.) You should have a look on the .TabP output file to see where the simulation time is going to. I can imagine that very high interface mobility of the α/α-interface leads to very small phase-field time steps, leading to large computation times for the interface-related program parts (list time, PF-time) in the .TabP). You can get more details on the time-stepping selection from the .TabT output file.
Strongly increased simulation times could also be a sign of numerical instabilities or a completely altered behaviour. You should avoid "flying blind" by using frequent tab_log outputs (which do not hurt because they are tabular only) and, if necessary, extra main outputs (during testing).

Hopefully, these general advice is helpful. Perhaps I could give you more concrete advice if you attach or send the complete input file (.in).

Best wishes

Posts: 1221
Joined: Mon Jun 23, 2008 9:29 pm

Re: Pearlite dissolution

Post by Bernd » Mon Mar 15, 2021 12:39 pm

Hi Vitoria,

based on the .in-file you sent me by PM, I have two additional comments:

1.) Currently you do not do any updates of diffusion data. In view of the heating rate of 200K/s I would recommend updating at least all 0.1 seconds.

2.) You use a "maximum number of new nuclei" for both seed types. I strongly recommend to use this option for testing only, because it has no physical background and can lead to very strange effects!


Post Reply