[ddreelan7]

Hi All,

I am an OpenFOAM novice and am working on a Cellular Automata model for alloy solidification.

When a grain is nucleated at a cell, I want its value (switch i.e. true or false) to change, indicating that it is now solid. As far as I have been able to find out, there is no volField which uses a bool value, or even an integer value i.e. no volSwitchField / volLabelField or equivalent.

I am assigning it a random crystal orientation, which I am recording as a volVectorField, whose 3 components are the 3 Euler angles, so if I could just modify the volVectorField class to also record if the cell is solid or liquid that would be perfect.

When a cell is liquid, it has no orientation angle, so a possible work around would be to set the angles to an invalid value, e.g. greater than 360 degrees, or NaN.

However, as the CA algorithm will be constantly checking the state (solid or liquid) of neighbouring cells, a switch value would be more natural and probably more efficient.

For now I am using a volScalarField and setting its value to 0 or 1, and when checking I am allowing some tolerance.

Any help would be greatly appreciated, as I am very new to this and digging through source code to find answers is difficult at my current level.

Best Regards,
Danny

[DoQuocVu]

Hi Danny,


Why don't you try using a boolList or PackedBoolList instead?

[tom_flint2012]

Hi Danny,


How far did you get in implementing the cellular automata method in openfoam?



I am more familiar with phase field methods, but I'm also thinking of implementing CA solidification model in my solver. I have a 'simple' multi-component description of diffusion for liquids, with N phases (n-1)^2 diffusion coefficient matrix.



If you have something working it would be great to see it in action.


All the best,


Tom

[tom_flint2012]

Hi Danny,


Not sure if you're still working on this. Bu I just finished my cellular automata implementation. I used dynamic lists to store al the data for the growing octahedron in each cell. Then when a particular vector of that octahedron entered a new cell I used this point as the new origin for that cells octahedron - copying over the three euler angles to that cell etc. It works very well. I had to use the lists in order to parallelise.


There was some memory distribution fighting to be done but I'm really happy with it. Let me know if you do want a hand with your implementation.


My work is multiphase so I dont think I'll be able to simulate the solute rejection at the growing interface easily. I think I would need a phase field model for the moving boundary concentration condition. I can do it for a binary alloy but struggling with my N-component mixture.


Do you know if anyone has done any work in multicomponent solute partitioning in OpenFOAM?



I think I'll just grow the dendrites by curvature and local undercooling. I'd be interested to hear your thoughts on this.


All the best,


Tom