Dear Oh Jeong Min,
Welcome to the MICRESS Forum.
Before coming to your problem, I would like stress that, especially as a beginner, you should try to start with a more simplified approach, and not put too many complications at time. Are you sure that all elements which you included are really relevant for your simulation goal? The elements Cu, S and P e.g. have only a very small content, but are segregating strongly. Having them included without caring for the specific precipitates they may imply can be very problematic and lead to numerical issues, even if they are not relevant for your research question. In any case, having many elements makes the simulation slower!
It would be much more easy for me to give advices if you would explain which goals you follow with this simulation approach. I also cannot reproduce the errors because you did not attach the required .ges5-file. But what I can tell you is that error 15 indicates problems in the equilibrium calculation when phase diagram slopes for a phase interaction are calculated (see
here). It probably occurs when nucleation of phase 2 is checked, and the 1/2-interface is initialized.
Phase 2 in your case represents TIC-carbides, which are treated in the TCNI12 database as a composition set of the FCC_L12-phase. This type of treatment makes the phase numerically very difficult, often leading to specific problems (see
here), not only for MICRESS but also for Thermo-Calc and Dictra. To be on the safe side, you should define all elements of this phase manually as stoichiometric by including the line
stoichiometric 2 1-11
before the "no_more_stoichio" keyword (or "end_of_options") in section "Numerical Parameters/Concentration Solver".
Further comments:
1.) You use Fe an matrix component in MICRESS, although your database is NI-based. This is unusuall and may cause problems. You should rather use Ni as matrix element in MICRESS unless there is a specific reason not to do so.
2.) Once the calculation of the initial equilibrium is successful, you should look up the linearisation data in the .log-file and check the composition of the phases. With composition sets it can easily happen that compositions are switching, so that e.g. you get two identical fcc-compositions instead of fcc and TiC.
3.) The length and time scales of your simulation setup seem to be not yet adapted to your application case. At present, the shield distance of seed_type 2 is 1 µm, which is identical to the grid spacing. This means no shield is active. Also, you use "global" updating of linearisation data with a distance criterion of 1 µm. This again is much too small to make any sense. I would recommend to use "global" without distance criterion in a first step, which is numerically most safe, and which is justified because you do not have a temperature gradient. Alternatively, you could use "global fragment", so that each TIC-precipitate gets its own set of linearisation parameters.
4.) You have defined "IONIC_LIQ", which numerically is very "ugly", as phase 0. What is the idea behind? However, if you never plan to use phase 0 in your simulation setup, it does'nt matter...
Best wishes
Bernd