Initial and boundary conditions
Initial and boundary conditions
I am new in phase-field simulation of dendritic solidification. I tried to simulate isothermal dendritic solidification of binary solution, I have read some papers about such kind of simulation, but I am not sure what kind of initial and boundary conditions I should use for the phase-field function (phi) and the concentration (c). Can anybody give me some explanation? Thank you!
Re: Initial and boundary conditions
Dear shoney,
Welcome to the MICRESS forum!
The initial conditions for a simulation depend a lot on the simulation setup and the problem you want to treat. I you say isothermal, then it sounds like a free growing equiaxed dendrite, but it could also be a free growing columnar dendrite front. In general, it is wise to start with the smallest geometry possible in order not to lose too much time waiting for the results while you are still looking for proper conditions or making parameter optimisation.
A free growing dendrite typically starts from a defined position. The initial condition for the phase-field variable would be to set an initial grain into the lower left corner of a quadratic domain (say 100x100 grid cells). The concentration field is flat (apart from the grain in the corner, the temperature is constant without gradient.
There is no perfect equivalent in the examples directory of your installation, but typically one should start from an already existing driving file (otherwise one has to answer too many questions for a beginner...). I would propose to start with "AlCu_dri", which is also binary, starting form a grain in the lower left corner and producing a dendrite.
After running it successfully you start changing conditions according to your specifications. It is illustative and easier to find errors if you rerun the example after each change:
- make the domain quadratic (perhaps starting with 100 x 100)
- remove the 1D extension to the concentration field (no_1d_far_field in section flags, this implies also removing two input lines in the concentration input)
- turn the rotation angle of the grain to 45° (grain input)
- change the temperature gradient and cooling rate to 0 (boundary conditions)
- switch off the moving frame (no_moving_frame) and remove the related further inputs
- change all boundary conditions of phase-field and concentration field to isolated (iiii)
This should already provide you with a free growing solutal dendrite (neglecting latent heat and heat conduction). Probably you need to lower somewhat the inital temperature (also in boundary conditions) to have enough driving force.
To switch from AlCu to a specific system, you may
- adjust the diffusion data for your system
- change the thermodynamic data to your system
This is just a scratch, may be I have forgotten something. Of course, the "art" of phase field simulation includes a lot of numeric stuff like chosing a proper resolution, interface mobility etc.
Please ask if you have any doubts or if I misinterpreted your concern.
Bernd
Welcome to the MICRESS forum!
The initial conditions for a simulation depend a lot on the simulation setup and the problem you want to treat. I you say isothermal, then it sounds like a free growing equiaxed dendrite, but it could also be a free growing columnar dendrite front. In general, it is wise to start with the smallest geometry possible in order not to lose too much time waiting for the results while you are still looking for proper conditions or making parameter optimisation.
A free growing dendrite typically starts from a defined position. The initial condition for the phase-field variable would be to set an initial grain into the lower left corner of a quadratic domain (say 100x100 grid cells). The concentration field is flat (apart from the grain in the corner, the temperature is constant without gradient.
There is no perfect equivalent in the examples directory of your installation, but typically one should start from an already existing driving file (otherwise one has to answer too many questions for a beginner...). I would propose to start with "AlCu_dri", which is also binary, starting form a grain in the lower left corner and producing a dendrite.
After running it successfully you start changing conditions according to your specifications. It is illustative and easier to find errors if you rerun the example after each change:
- make the domain quadratic (perhaps starting with 100 x 100)
- remove the 1D extension to the concentration field (no_1d_far_field in section flags, this implies also removing two input lines in the concentration input)
- turn the rotation angle of the grain to 45° (grain input)
- change the temperature gradient and cooling rate to 0 (boundary conditions)
- switch off the moving frame (no_moving_frame) and remove the related further inputs
- change all boundary conditions of phase-field and concentration field to isolated (iiii)
This should already provide you with a free growing solutal dendrite (neglecting latent heat and heat conduction). Probably you need to lower somewhat the inital temperature (also in boundary conditions) to have enough driving force.
To switch from AlCu to a specific system, you may
- adjust the diffusion data for your system
- change the thermodynamic data to your system
This is just a scratch, may be I have forgotten something. Of course, the "art" of phase field simulation includes a lot of numeric stuff like chosing a proper resolution, interface mobility etc.
Please ask if you have any doubts or if I misinterpreted your concern.
Bernd
Re: Initial and boundary conditions
Thank you!
Another question
If the initial concentration is non-uniform, a function of space coordinates, is it okay?
Another question
If the initial concentration is non-uniform, a function of space coordinates, is it okay?
Bernd wrote:Dear shoney,
Welcome to the MICRESS forum!
The initial conditions for a simulation depend a lot on the simulation setup and the problem you want to treat. I you say isothermal, then it sounds like a free growing equiaxed dendrite, but it could also be a free growing columnar dendrite front. In general, it is wise to start with the smallest geometry possible in order not to lose too much time waiting for the results while you are still looking for proper conditions or making parameter optimisation.
A free growing dendrite typically starts from a defined position. The initial condition for the phase-field variable would be to set an initial grain into the lower left corner of a quadratic domain (say 100x100 grid cells). The concentration field is flat (apart from the grain in the corner, the temperature is constant without gradient.
There is no perfect equivalent in the examples directory of your installation, but typically one should start from an already existing driving file (otherwise one has to answer too many questions for a beginner...). I would propose to start with "AlCu_dri", which is also binary, starting form a grain in the lower left corner and producing a dendrite.
After running it successfully you start changing conditions according to your specifications. It is illustative and easier to find errors if you rerun the example after each change:
- make the domain quadratic (perhaps starting with 100 x 100)
- remove the 1D extension to the concentration field (no_1d_far_field in section flags, this implies also removing two input lines in the concentration input)
- turn the rotation angle of the grain to 45° (grain input)
- change the temperature gradient and cooling rate to 0 (boundary conditions)
- switch off the moving frame (no_moving_frame) and remove the related further inputs
- change all boundary conditions of phase-field and concentration field to isolated (iiii)
This should already provide you with a free growing solutal dendrite (neglecting latent heat and heat conduction). Probably you need to lower somewhat the inital temperature (also in boundary conditions) to have enough driving force.
To switch from AlCu to a specific system, you may
- adjust the diffusion data for your system
- change the thermodynamic data to your system
This is just a scratch, may be I have forgotten something. Of course, the "art" of phase field simulation includes a lot of numeric stuff like chosing a proper resolution, interface mobility etc.
Please ask if you have any doubts or if I misinterpreted your concern.
Bernd
Re: Initial and boundary conditions
Hi shoney,
if you want to start from a non-uniform melt, then you need to read in the initial concentration map from a file! Such an ASCII file you can create either from a former MICRESS result using DP_MICRESS, or from any concentration map which you can convert to ASCII or vtk format. More information about the input syntax you find here.
Bernd
if you want to start from a non-uniform melt, then you need to read in the initial concentration map from a file! Such an ASCII file you can create either from a former MICRESS result using DP_MICRESS, or from any concentration map which you can convert to ASCII or vtk format. More information about the input syntax you find here.
Bernd