MICRESS Forum

Hi, Bernd!

Recently, I am doing 3-D pearlite growth. Since the Carbon concentration is so different in ferrite, austenite and cementite, I can distinguish each phase in conc.mcr. However, I find a mismatch with the phas.mcr output for the same timestep output. It seems that the conc.mcr output is close to the real case, while the phas.mcr seems to be impossible. With simulation time goes by, the mismatch increases.

However, according to the basic theory of phase field, there should be no mismatch. I set the interface thickness to be 4 cells, is it because the thickness is too small that the phas.mcr wrongly mark the phases? Which output should I believe?

If you can leave me your mail address, I can send you my result.

Best wishes!

Yang Zenan

Dear Yang,

If you do not wish to share your results by uploading them here e.g. in form of pictures, you could also send them to me as a Personal Message (PM). But, anyway, I believe I can give you a helpful answer without having seen them:
The .phas output is not always the best choice when examining the progress of phase transformations, because it gives only the correct phase number when the highest fraction of the locally coexisting grains exceeds 0.7! Otherwise, a value of -1 is shown which indicates the presence of an interface, even in case it is an interface between grains of the same phase!
To follow up the development of the local phase fractions, you should write out and examine the .frac<n> outputs which you can activate by e.g.

out_fraction 0 1 2

which would write out the fractions of phase 0, 1 and 2 as graphical outputs (.frac0, .frac1, .frac2). Using "out_fraction" without extra parameter activates only one output (.frac) of the fraction solid which is not helpful in case of solid-solid transformations.
The .intf output also gives you insight into the local interface structure:
-1: dual interface
-3: triple junction
-6: quadruple junction
-10: quintuple junction
etc.

Best regards

Bernd

Dear Yang,

As got clear from the information in your PM, there is no problem with the .conc or .phas output but with the numerical stability of the ferrite/cementite interface!
What you should first do is having a look on the .driv output which shows you first whether there are numerical problems and when they start. Then you should try getting the interface stable by trying to change numerical parameters, starting from low mobility values. Thermo-Calc related error messages are always a good hint that numerics are outside of the stable range.
If you do not succeed in getting the simulation stable, please attach the input file (including other relevant input data like the .ges file and tabular input files if applicable) to this thread, and I will try to run the simulation here and find out what is wrong. You can also send me the data by PM if you feel they are confidential!

Best wishes

Bernd