2 Liquid phases during dissolution

multicomponent diffusion,...,...
Post Reply
bfaivre
Posts: 2
Joined: Thu Mar 23, 2023 12:55 pm
anti_bot: 333

2 Liquid phases during dissolution

Post by bfaivre » Thu Mar 23, 2023 1:37 pm

Hi everyone ! Hope you are doing great.

I am currently working on the addition of FSM alloy in iron cast (especially the dissolution). In my first simulation, I put a grain of MG2SI in a Iron cast liquid (T=1200C), during the dissolution, a Mg liquid phase should appear as the Mg has a bad solubility in cast iron. But Micress seems to not take into account this other liquid phase and only discern one liquid phase.
I also implement mobility data concerning the diffusion in the liquid phase.

Can someone help me on this problematic and check if my driver file looks correct ?

Thanks in advance,

Baptiste
Attachments
TC2023a_1phase.GES5
(144.29 KiB) Downloaded 108 times
MG2SI dissolution.txt
(16.88 KiB) Downloaded 118 times

Bernd
Posts: 1504
Joined: Mon Jun 23, 2008 9:29 pm

Re: 2 Liquid phases during dissolution

Post by Bernd » Thu Mar 23, 2023 5:04 pm

Dear Baptiste,

Welcome to the MICRESS Forum!

If a phase which is described in a Calphad database has different stability ranges in composition space with miscibility gaps in between, then this is treated as "composition sets" in Thermo-Calc. Essentially, those sets are treated as if they were different phases with identical Gibbs energy description, but different start values for numerical iteration (default major composition). While Thermo-Calc does, MICRESS does not create such composition sets automatically. Instead, this has to be done via Thermo-Calc when creating the .GES5-file (using command "amend_phase_description").

Then, in MICRESS, the new composition set would appear in the phase-list of the "database" as LIQUID#2, and you can add it as extra phase in MICRESS. Please be sure that in the .GES5-file, the default major composition is chosen such that the new phase is assumed to have a high Mg content, while the original one (LIQUID) has a low one.

Further, in MICRESS you need to actively nucleate the new phase at the interface between LIQUID and MG2SI. Then, this phase should a second liquid "particle". Please don't expect any "fluid" behavior from the liquid phases, which is beyond the scope of the current MICRESS version.

You can have a look here to be aware of further potential issues coming from the existence of composition sets.

Best wishes
Bernd

bfaivre
Posts: 2
Joined: Thu Mar 23, 2023 12:55 pm
anti_bot: 333

Re: 2 Liquid phases during dissolution

Post by bfaivre » Thu Mar 30, 2023 9:41 am

Dear Bernd,

I managed to create a second liquid phase with a high Mg content, thank you.
However, my simulation generated a lot of interface errors and the system seemed out of balance.

I decided to take the problem differently to start: 2 liquid phases (iron cast and MG) and a MG2SI phase. In order not to generate the nucleation of liquid Mg but rather a magnification of phase.

At the beginning of my simulation, I had a lot of errors like: "Fast moving interface(s) relinearized LIQUID/MG2SI_C1 (11)
Fast moving interface(s) relinearized MG2SI_C1/LIQUID#2 (23)
" or "trying hard phases 0 2 level: 3 zp= 14827 error= 30200". Reading some of your answers, I used correction functions like "mob_corr" or interaction .

The mobility of the interfaces then seems slower to accompany the diffusion of the elements. However, after a certain calculation time, these errors reappear and concentrations turn negative. The simulation also takes more time..

Does Micress take into account the maximum sobulity of an element? In addition, interfaces are created in the middle of a phase (see photo). Do you have any ideas to solve this problem?

I send you my main file and the GES5 in PM so that you can see more clearly. I use the TCFE12 and MOBFE7 database.

Thanks in advance,

Baptiste
Attachments
interface 2.png
interface 2.png (81.15 KiB) Viewed 11650 times
interface 1.png
interface 1.png (21.47 KiB) Viewed 11650 times

Bernd
Posts: 1504
Joined: Mon Jun 23, 2008 9:29 pm

Re: 2 Liquid phases during dissolution

Post by Bernd » Fri Mar 31, 2023 5:10 pm

Dear Baptiste,

The problem you pose definitively is not an easy one. It took me some time to figure out what are the main problems here. Given that this is not a usual case, the usual advice may not work here. Technically speaking, the problems are the strongly "demixing" interaction of the two liquids, the extreme composition differences, and the extreme out-of-equilibrium initial situation.

I found that maybe the most fundamental error is to try to assume diffusion-limited conditions in this case. Although this is almost in all cases a good starting point, here it produces problems, because the "mob_corr" model requires at least some stationary diffusion fluxes and interface movement. Thus, it is necessary to use "normal" or "atc" without "mob_corr" for the redistribution options (or remove completely the "redistribution_control" switch). As a consequence, the interface mobilities now matter and have to be chosen by hand.

To make initial conditions a bit less extreme, I propose the following:
1.) Don't use 0 a initial composition for phases with solubility range (i.e. 0 and 2). Chose something small like 0.01 wt% instead in order to reduce the danger of running into negative values, e.g. due to cross diffusion.
2.) If you only specify the initial composition of phase 0 and chose "equilibrium", then the initial situation will be at least diffusionally equilibrated (quasi-equilibrium)
3.) Growth of phase 2 is expected to happen only at places where it has contact to both other phases (the lower left corner). The process will be much smoother if you set the Mg2Si into the center, so that it remains in contact to both liquids even when it shrinks.
4.) An interface thickness of 6 cells is much too large and additionally destabilizes processes with large gradients. 3-4 cells are sufficient.

Let's see how far you get with these hints. By the way, the messages with "fast moving interfaces" is not an error. If you a lot of errors still, please watch out for negative compositions. They are generally a problem for all types of thermodynamic calculations and may occur e.g. as a consequence of cross-terms of diffusion or too far thermodynamic extrapolation (too little updating).

Bernd

Post Reply