solid-state phase transformation

[zelin zhang]

Dear Bernd,

I would like to ask for your help regarding a problem I encountered in MICRESS during pearlite transformation simulations.

In my simulations, the pearlite transformation behaves normally when stress coupling is not considered. The lamellar structure forms correctly and remains stable throughout the simulation.

However, after enabling stress coupling, I observe a different behavior:

In the early stage, the microstructure evolution is still reasonable, and the pearlite lamellae form as expected.
In the later stage, the cementite lamellae become unstable. They tend to break, merge, or become distorted, leading to abnormal pearlite morphology.

I have also tried to adjust the interface properties between ferrite and cementite. Specifically, I reduced the interface mobility and increased the interface energy between these two phases. However, these modifications did not resolve the issue, and the instability of cementite lamellae in the later stage still persists.

I would like to ask how this problem could be addressed, and whether there are recommended strategies to improve the stability of cementite lamellae under stress coupling.

Any advice or suggestions would be greatly appreciated.

Thank you very much for your help.

Best regards,
Zelin Zhang

[Bernd]

Dear Zelin Zhang,

This is difficult to say without having more details on the simulation and the type of instability. The molar volumes of the phases as well as the stress boundary conditions have the most strong impact on the mechanical coupling. However, other settings may also have an influence and trigger such a behaviour.

The next steps should be to find out how the instabilities play out exactly. Are they only connected to the elastic driving force (.dGsp-output)? Is this effect realistic with respect to its relative strength? Or is there also some impact of stress on the concentration distribution in the phases, which may indirectly lead to instabilities?

If you get stuck with this problem, you may send your input file (including all required input so that I can try to reproduce the described behaviour).

Bernd

[zelin zhang]

Dear Bernd,

I will send you the driving file via private message, and I would be very grateful if you could take a look and help me with this issue.

Thank you very much for your help.

Best regards,
Zelin Zhang

[Bernd]

Dear Zelin Zhang,

Thanks for sending your input file. But for running your input file I also need the .ges5-file (FeC_ALL.ges5)...

Bernd

[zelin zhang]

Dear Bernd,

Thank you very much for your help. I have sent you the required .ges5 file via private message, and I would greatly appreciate it if you could take a look.

Thank you again for your time and support.

Best regards,
Zelin Zhang

[Bernd]

Dear Zelin Zhang,

As you already suggested in your latest PM, the large differences in interfacial energies are the main source of your instabilities, although I still do not fully understand how this is connected to coupling with stress. Please note that strongly differing interfacial energies within a common triple/multiple junction can lead to instabilities, especially with older MICRESS versions like 7.0. As I have tested, having all energies within a factor of 2 definitively removes all instabilities!

Another problem of your setup is that you do not apply any updating of TQ data for the phase interactions. This leads not only to wrong results but also to negative compositions in large parts of the simulation domain and to error messages and problems at a later stage. It is absolutely necessary to update linearisation data in a regular interval, either globally for all phase interactions (in section "Database") or individually for each phase interaction (in section "Phase Interactions").

Further, in your approach, temperature is decreasing linearly, quickly leading to very low temperatures, where diffusion is getting extremely slow. After 80s, diffusivities have already dropped by 5 orders of magnitude, so that the grid spacing is not sufficient anymore to resolve diffusion fields, and any phase transformation is frozen. Pearlite formation should have finished already at this point. Later on, temperature gets even negative (newer MICRESS versions would stop automatically here).

Further suggestions:

- I would tend to use high interface mobilities (>1.E-4 cm**4/(Js)) for all interfaces between different phases, and hereby allow the "mob_corr" option to adjust the numerical mobility for diffusion-limited growth. This assures that pearlite formation can finish before the temperature gets too low. The interface mobility between same phases is a physical quantity. If grain ripening is required and must be included, you should use temperature dependent mobility input from file in order to avoid unrealistic grain coarsening at low temperatures.

- The definition of a minimum time-step (currently at 1.E-2 s) can help to improve performance, but is dangerous if the numerically required value is much below (see .TabT output). Especially when using my suggestion of using higher interface mobilities, it is necessary to reduce this value to at leat 1.E-3 s.

- Using an interface thickness of 5 cells is a waste of ressources, at least under these conditions. I would suggest to use 3 cells.

- Using default numerical parameters for the stress solver leads to an unnecessarily high computational effort. My suggestion is:

#
# Stress solver
# -------------
# convergence criteria for BiCGStab-solver (stress calculation)
# and (average) strain calculation? (real) (real)
5.0E-05 5.0E-06
# max. number of iterations for BiCGStab-solver
# and (average) strain iterations? (int) (int)
500 20
#

Berst wishes
Bernd