Interface movement

solid-solid phase transformations, influence of stresses and strains
Post Reply
AZheng
Posts: 6
Joined: Thu Jun 14, 2018 2:47 pm
anti_bot: 333

Interface movement

Post by AZheng » Thu Jun 14, 2018 2:55 pm

Hello,

We are trying to simulate the phase transition from ferrite to austenite.
The grain input setup we use from_file option and the nucleation position is interface. At first there was only one horizontal straight line(as interface), everything was fine. But when we changed to nonhorizontal lines, e.g. bias, the lines became horizontal gradually as nucleation proceeded. In fact the lines need to be stable. Why it happened? Is this problem related to interface energy? And how to change the parameters?
Your help is highly appreciated!

Best regards,
AZheng
Attachments
bias_2s.PNG
bias_2s.PNG (5.59 KiB) Viewed 2236 times
bias_0s.PNG
bias_0s.PNG (6.91 KiB) Viewed 2236 times
REV_96_dri.txt
(16.55 KiB) Downloaded 123 times

Bernd
Posts: 1267
Joined: Mon Jun 23, 2008 9:29 pm

Re: Interface movement

Post by Bernd » Thu Jun 14, 2018 9:07 pm

Dear AZheng,

Welcome to the MICRESS Forum!

As far as I can see from the input file, you essentially have two problems:

1.) There is always an interaction of the grains in the simulation domain with the external boundary conditions. You have chosen isolating conditions which corresponds to the requirement that grain boundaries form right angles with the boundaries. Thus, what you observe is easy to understand: The curvature implied by the boundary conditions automatically will move your grain boundaries until all of them are perpendicular or parallel to the sides of the domain (or have vanished). The time scale for this movement only depends on interface mobility (1/1) and interface energy.
What you can try is to use gradient boundary conditions for phase-field instead of isolation. The idea of that would be to "continue" the grain boundary and avoid the strong forces. However, this method is only "first order" and thus probably will not completely stop the observed movement.
The problem essentially is quite old, and there is no general solution. Maybe, in your case, a specific solution could be to provide an additional boundary condition for phase-field which just sets the interface mobility at the boundary to 0...

2.) You use a extremely fine resolution which is essentially atomistic. The consequence of using a mean-field model at atomistic length-scales is exaggerated curvature which increases the problem discussed under 1.) and which also makes nucleation very difficult (see herefor discussion). A possible way to address the problem could be the use of a corrected (i.e. smaller) interface energy which could reduce all the mentioned problems.

Bernd

Post Reply