Initial Microstructure
Initial Microstructure
Hi,
I am trying to use a real microstructure image to generate the initial microstructure in a periodic condition.
But the problem is how to get a periodic microstructure. Seems grains at the domain boundary need to be adjusted to connect to the other side. Is there any suggestion? Thank you.
I am trying to use a real microstructure image to generate the initial microstructure in a periodic condition.
But the problem is how to get a periodic microstructure. Seems grains at the domain boundary need to be adjusted to connect to the other side. Is there any suggestion? Thank you.
Re: Initial Microstructure
Dear zhubq,
Of course, using periodic boundary conditions with experimental microstructures is a fundamental problem, because experimental structures are never periodic. In the Grain-Growth examples on the MICRESS distribution CD, it can be seen what happens in that case: an extra grain boundary is formed all along the domain boundary. This looks very ugly in the beginning, but after some time of grain growth it is ok. I do not know any better way of treating the problem!
Bernd
Of course, using periodic boundary conditions with experimental microstructures is a fundamental problem, because experimental structures are never periodic. In the Grain-Growth examples on the MICRESS distribution CD, it can be seen what happens in that case: an extra grain boundary is formed all along the domain boundary. This looks very ugly in the beginning, but after some time of grain growth it is ok. I do not know any better way of treating the problem!
Bernd
Re: Initial Microstructure
Hi, Bernd.
Thank you for your explanation.
So MICRESS just regards the boudaries of the domain as the grain boudaries of those grains that touch the domain boudaries, right?
If we have to use real microstructure, which boudary condition do you suggest for recystallization and diffusion-type transformation, gradient or insulated? Because it may be OK for grain growth, but may not be applicable to those related to nucleation.
Ben
Thank you for your explanation.
So MICRESS just regards the boudaries of the domain as the grain boudaries of those grains that touch the domain boudaries, right?
If we have to use real microstructure, which boudary condition do you suggest for recystallization and diffusion-type transformation, gradient or insulated? Because it may be OK for grain growth, but may not be applicable to those related to nucleation.
Ben
Re: Initial Microstructure
This is correct!
If you don't want to live with the additional interfaces, gradient should be used because isolating boundary conditions as well as symmetric always lead to interfaces which are perpendicular to the border of the domain, which is bad for grain growth. But be carefull and use the gradient condition only for the phase-field and not for the concentration field! This would be not only quite unphysical but also lead to instabilities if interfaces between different phases touch the calculation domain boundary...
Bernd
If you don't want to live with the additional interfaces, gradient should be used because isolating boundary conditions as well as symmetric always lead to interfaces which are perpendicular to the border of the domain, which is bad for grain growth. But be carefull and use the gradient condition only for the phase-field and not for the concentration field! This would be not only quite unphysical but also lead to instabilities if interfaces between different phases touch the calculation domain boundary...
Bernd
Re: Initial Microstructure
Thank you!
So It is better to apply gradient condition to both phase fields and concentrations?
So It is better to apply gradient condition to both phase fields and concentrations?
Re: Initial Microstructure
No, you should not use the gradient condition for concentration in this case!
Bernd
Bernd
Re: Initial Microstructure
Oh, I misunderstood previously. Thank you. But is it OK to use the other boundary for concentration when gradient is applied to phase field?
Re: Initial Microstructure
Yes, it should. I never tried it personally, but it should!
Bernd
Bernd