Initial Condition setting for recrystallization modeling

[Xin-Bo]

Dear MOOSE developers,

I encountered a problem when modeling recrystallization. 

As we know, when recrystallization begins, there are many small grains newly formed on the boundaries of the deformed grains. However, it is found that the current ICs for polycrystalline cannot meet the requirement, since these ICs cannot control the grain size and distribution of the new formed small grains.

My current idea is to modify the "PolycrystalVoronoiVoidIC" class, as it can distribute the voids on the grain boundaries. By setting the variable that this IC acts on to another grain "gr", we may convert the voids to the ordinary small grains. However, since an extra "gr" is introduced, other objects, such as the relative Kernels, also have to be modified.

Can anyone give me some advices on this problem?

Thanks very much!

Xin-Bo

[Cody Permann]

I've collaborated with one developer and we've done quite a bit of work on recrystallization. I'm a little confused by your statement here. It seems like you are trying to address a problem that occurs _during_ the simulation. However, you are wanting to handle this by changing the behavior of one of the Initial Conditions. I'm not sure if I'm capturing your concern correctly or not, please clarify.

I've revamped and created a system for Polycrystal initial conditions and updated the normal Voronoi condition along with the EBSD and Hex ICs to use the "new" system. I guess I haven't worked with PolycrystalVoronoiVoidIC much but I know I didn't change that class around at all. It may need some additional treatment. Anyway, if you can help me understand more about how you'd like this IC to behave, I may be able to advise you on how to proceed.

Cody 

--
You received this message because you are subscribed to the Google Groups "moose-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to moose-users...@googlegroups.com.
Visit this group at https://groups.google.com/group/moose-users.
To view this discussion on the web visit https://groups.google.com/d/msgid/moose-users/535fab45-5f64-4b6b-955d-2b7ff112048e%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Xin-Bo]

Thanks for your reply, cody. 

I may give an example by explaining the problem clearly. 

In the "DeformedGrainMaterial" Class, we can only set the number of the deformed grains, see here. This means that we cannot control the size and distribution of the deformed grains (as well as the recrystallized grains). However, if we want to simulate recrystallization, we need to give these informations in the initial conditions, for example, the recrystallized grains may distribute on the grain boundaries of the deformed grains. In the MOOSE current test file, the deformed grains and the recrystallized grains experienced the same voronoi tessellation. So, my attempt is **to find an IC that can distribute the recrystallized grains on the boundaries of the original grains**.

From your reply, as far as I understand, you added the small recrystallized grains __during__ the simulation and not __before__ the simulation. My question is how to add new grains in this process, since it means the new OPs are also needed to add __during__ this process?

I hope my statement is clear enough to understand. 

Looking forward to your reply!

在 2017年11月23日星期四 UTC+8上午12:22:12，Cody Permann写道：

[Cody Permann]

On Wed, Nov 22, 2017 at 6:34 PM Xin-Bo <qix...@gmail.com> wrote:

Thanks for your reply, cody. 

I may give an example by explaining the problem clearly. 

In the "DeformedGrainMaterial" Class, we can only set the number of the deformed grains, see here. This means that we cannot control the size and distribution of the deformed grains (as well as the recrystallized grains). However, if we want to simulate recrystallization, we need to give these informations in the initial conditions, for example, the recrystallized grains may distribute on the grain boundaries of the deformed grains. In the MOOSE current test file, the deformed grains and the recrystallized grains experienced the same voronoi tessellation. So, my attempt is **to find an IC that can distribute the recrystallized grains on the boundaries of the original grains**.

From your reply, as far as I understand, you added the small recrystallized grains __during__ the simulation and not __before__ the simulation.

That's correct.

 

My question is how to add new grains in this process, since it means the new OPs are also needed to add __during__ this process?

No, not quite. The point of the GrainTracker is to efficiently distribute a small fixed number of order parameters among a much large number of grains. In our recrystallization model, we have a fixed number of OPs for representing grains over a certain threshold and a few more "reserved" OPs where new grains can be nucleated. The GrainTracker is designed to migrate newly nucleated grains from the reserved OPs to the stable OPs keeping those reserved parameters open for new grains. In this manner we can add new grains all throughout the simulation.

We really need to publish that result... Are you interested in continuing this work? I'll get in touch with my collaborator and see where we are at on this.

Cody

 

I hope my statement is clear enough to understand. 

Looking forward to your reply!

在 2017年11月23日星期四 UTC+8上午12:22:12，Cody Permann写道：

I've collaborated with one developer and we've done quite a bit of work on recrystallization. I'm a little confused by your statement here. It seems like you are trying to address a problem that occurs _during_ the simulation. However, you are wanting to handle this by changing the behavior of one of the Initial Conditions. I'm not sure if I'm capturing your concern correctly or not, please clarify.

I've revamped and created a system for Polycrystal initial conditions and updated the normal Voronoi condition along with the EBSD and Hex ICs to use the "new" system. I guess I haven't worked with PolycrystalVoronoiVoidIC much but I know I didn't change that class around at all. It may need some additional treatment. Anyway, if you can help me understand more about how you'd like this IC to behave, I may be able to advise you on how to proceed.

Cody 

On Tue, Nov 21, 2017 at 11:08 PM Xin-Bo <qix...@gmail.com> wrote:

Dear MOOSE developers,

I encountered a problem when modeling recrystallization. 

As we know, when recrystallization begins, there are many small grains newly formed on the boundaries of the deformed grains. However, it is found that the current ICs for polycrystalline cannot meet the requirement, since these ICs cannot control the grain size and distribution of the new formed small grains.

My current idea is to modify the "PolycrystalVoronoiVoidIC" class, as it can distribute the voids on the grain boundaries. By setting the variable that this IC acts on to another grain "gr", we may convert the voids to the ordinary small grains. However, since an extra "gr" is introduced, other objects, such as the relative Kernels, also have to be modified.

Can anyone give me some advices on this problem?

Thanks very much!

Xin-Bo

--
You received this message because you are subscribed to the Google Groups "moose-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to moose-users...@googlegroups.com.

Visit this group at https://groups.google.com/group/moose-users.
To view this discussion on the web visit https://groups.google.com/d/msgid/moose-users/535fab45-5f64-4b6b-955d-2b7ff112048e%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "moose-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to moose-users...@googlegroups.com.
Visit this group at https://groups.google.com/group/moose-users.

To view this discussion on the web visit https://groups.google.com/d/msgid/moose-users/1320867d-31af-42fd-a372-29f3ae366474%40googlegroups.com.

[Xin-Bo]

Thank you very much, Cody.

These days I am keeping thinking on this topic, and have tried an idea to tackle this problem. 

The detailed input file is here.

The main idea is to use the PolycrystalVoronoiVoidIC.

Firstly, set the total OP number is 13, see here.

Secondly, use PolycrystalVoronoiIC to generate the Voronoi tessellation of the deformed grains, in this process, at most 12 OPs are used to represent these grains, see here (in this input, 8 OPs are used).

Then, use the variable named "gr12" to represent the recrystallized grains, which are distributed at the boundaries of the previous deformed grains, see here. 

I think this kind of setting may represent the recystallization correctly, although it seems a bit ugly since it cannot set the the paramter "op_num" in the [Global Parameters] block and have to put it at several places.

Do you have some advices on this idea?

Looking forward to your reply!

在 2017年11月28日星期二 UTC+8上午6:46:51，Cody Permann写道：