question about Gamma Alpha Example

[Mehnoush]

Hi Bernd,

I am trying to work on Micress Example Gamma Alpha.
I dont completly understand some of the points in this *.dri file.
First is how you find the surface energy and kinetic coefficient between phases?
and more importantly about the temperature.
the

temperature at which the initial equilibrium will be calculated

and the

initial temperature at bottom

.


Best regards,
Mehnoush

[Bernd]

Dear Mehnoush,

The interfacial energy is a property of interfaces which in general is difficult to find. But in case of gamma-alpha transition it should be possible to find some values in literature.
If this is not the case, or if you are not interested in exact values at the current stage, you should at least use value which have a correct order of magnitude (1.E-5 J/cm₂). Be aware that this value directly influences the length scale of the microstructure pattern which will form, and also has a direct influence on the nucleation undercooling.
When using the "multi_obstacle" flag (taking triple point terms into account for obeying Joung's law), wetting angles will depend on the relative size of interfacial energy values of the different interfaces which form a triple junction. In case of gamma-alpha transformation, the relation of sigma(alphaalpha), sigma(gammaalpha) and sigma(gammagamma) affect the shape of the precipitates.

The "temperature at which the initial equilibrium will be calculated" is relevant only if there are several phases present at the beginning. Then, changing this temperature can help finding a good initial equilibrium in difficult cases. Otherwise, it should be equal to the "initial temperature at bottom".

The "initial temperature at bottom" is just the temperature at which your simulation starts. "bottom" means the lower boundary of the simulation domain. This further distinction is only meaningful if there is a temperature gradient along the z-direction (bottom to top).

Bernd

[Mehnoush]

Thanks alot for your response, the other question that i have is, that what kind of model you have taken for interface mobility is the mix methode, or diffusional method. there are different approaches in the literature for example by steinbach or Zwaag and etc. it is a little confusing.

Best regars,
Mehnoush

[Bernd]

Dear Mehnoush,

In MICRESS, there is no default model for interface mobility. That means that the user can (and has to) decide what is the physics behind the system he or she is simulating and which interface mobility he or she wants to apply. While solidification is relatively simple, the complexity of the physics of solid-solid phase transformation leads to very different strategies the user may follow:
- assume the interface mobility to be a physical interface property and chose it according to literature or by fitting to experiments
- take it as substitute for not included physical phenomena like mechanical stress or redistribution kinetics (nple/para) and calibrate by fitting to experiments
- assume it to be diffusion limited and calibrate with high resolution simulations (see here).
- assume it to be diffusion limited and use automatic mobility correction for finite interface thickness (mob_corr). Please note that this function is still under test (ask for latest version)

All the above strategies make sense under certain circumstances, and depending on how much physics you include in the simulation: Do you include
- substitutional elements
- redistribution kinetics of substitutional elements (nple, para),
- stress?

Unfortunately, this decision is not easy and depend on the goals which you follow with your simulations...

Bernd

[R.Hess]

Dear Bernd,
I´m working on the same project as mehnoush.
I have another question regarding the GammaAlpha example.

At the beginning , I started with some simple modifications of the examples to learn how to work with micress.
So I modified the initial compositions of C and Mn of the GammaAlpha-example.
In the examples the compositions was set to 0,1wt-% of C and 1,5 wt-% of Mn. Now I set 0,4 wt-% of C and 0,9 wt-% of Mn (like 42CrMo4) as initial compositions to look what will happen and what will be the difference.
Furthermore, I set the temperature at the bottom, and the temperature for initial composition to 800K.

Now, In the *phase_output, the initial phase is ,obviously, Phase 1(gamma). Then Phase 2 (alpha) begins to grow in the bulk and at the interfaces. Up to now is everything clear, but from now on, I don´t understand the following result:
After several time steps the alpha-phase used all the gamma phase up, so that only the alpha phase is left. Some timesteps later, the gamma phase begins to grow at the interfaces / grain boundaries. At the end of the simulation (at t=300) I can see a mixture of big alpha grains und smaller gamma grains.


Is this a realistic result? If yes, is it because of the equilibrium of the two phases at this given compostions and Temperatures? If not , what could be my mistake?

Best regards
Raphael

[Bernd]

Dear Raphael,

the way you describe it, the process does not seem logic. If the initial gamma phase is completely overgrown by alpha, i.e. no gamma phase is left, there is no mechanism to bring back gamma phase (in the original Gamma_Alpha example, nucleation is defined only for alpha phase). Perhaps you should re-analyse what exactly is happening. The .phas output gives no clear answer on that. Please check the .frac1 and .frac2 outputs to see whether the gamma phase is really vanishing.
What I could imagine is that due to insufficient or missing calibration of numerical parameters, the interface between the two phases gets unstable and spreads over large regions, leaving behind a phase mixture which is difficult to see in the .phas output. Apart from the .frac, the .intf gives you a good impression of sanity of the interfaces. If this is true, the diffuse gamma phase could aggregate with time and reappear in the .phas output.
By changing composition and temperature, the driving force for the gamma-alpha transformation can change drastically, leading to different conditions. The resulting microstructure would be finer, requiring higher spatial resolution.

Bernd