PF simulation of Eutectic solidification
Re: PF simulation of Eutectic solidification
Dear Omid,
the second option is better. The reference point I is just a mathematical reference, thus it needs not to be on the "real" part of the liquidus intersection, and extrapolation is fine.
Bernd
the second option is better. The reference point I is just a mathematical reference, thus it needs not to be on the "real" part of the liquidus intersection, and extrapolation is fine.
Bernd
Re: PF simulation of Eutectic solidification
Dear Bernd,
In some timesteps the minimum concentration of component 2 becomes a negative number, although the average is positive and and still in a wisely range, the minimum reaches to 6000 %
Is this normal or there is an obvious error which I don't know?
And another strange error! when almost 85% of the domain got solid, and the temperature reaches to 3 degrees less than melting point, it goes back and turns in to liquid! completely liquid! any advises? you can find the dri file as attached.
Omid.
In some timesteps the minimum concentration of component 2 becomes a negative number, although the average is positive and and still in a wisely range, the minimum reaches to 6000 %
Is this normal or there is an obvious error which I don't know?
And another strange error! when almost 85% of the domain got solid, and the temperature reaches to 3 degrees less than melting point, it goes back and turns in to liquid! completely liquid! any advises? you can find the dri file as attached.
Omid.
 Attachments

 4phases_100KCR_dri.txt
 (50.5 KiB) Downloaded 151 times
Re: PF simulation of Eutectic solidification
Dear Omid,
to me it appears obvious that both problems are due to an inconsistent phase diagram description. The difficult question is whether and how you can figure out what is wrong (or whether it is at all possible to construct a linear phase diagram for 4 phases which is consistent over all the temperature range...).
I would try to visualize the diagram by doing some kind of 3D plots with an appropriate software. I must admit that I have no experiences with that...
Bernd
to me it appears obvious that both problems are due to an inconsistent phase diagram description. The difficult question is whether and how you can figure out what is wrong (or whether it is at all possible to construct a linear phase diagram for 4 phases which is consistent over all the temperature range...).
I would try to visualize the diagram by doing some kind of 3D plots with an appropriate software. I must admit that I have no experiences with that...
Bernd
Re: PF simulation of Eutectic solidification
Dear Bernd
Many thanks for your efforts. Which software do you use for diagram visualization?
Omid.
Many thanks for your efforts. Which software do you use for diagram visualization?
Omid.
Re: PF simulation of Eutectic solidification
Dear Omid,
as I said, I have no experiences with that. But I imagine one could use Maple, MathLab or something similar...
Bernd
as I said, I have no experiences with that. But I imagine one could use Maple, MathLab or something similar...
Bernd
Re: PF simulation of Eutectic solidification
Dear Bernd,
I have turned off the Limits for phase 3 and it got better! only at one time step it gets minus but I still get strange values in concentration of content B.
Although the average is fine but in some points the maximum reaches to 50000 at.%!! how could it be possible? What is the routine of Micress for calculating the composition? May be it's due to the convergence of equations and so on! Do you have any idea? How can I switch off the minimum and maximum values or determine the limits for both?
kindly have a look at attached file.
Regards,
Omid.
I have turned off the Limits for phase 3 and it got better! only at one time step it gets minus but I still get strange values in concentration of content B.
Although the average is fine but in some points the maximum reaches to 50000 at.%!! how could it be possible? What is the routine of Micress for calculating the composition? May be it's due to the convergence of equations and so on! Do you have any idea? How can I switch off the minimum and maximum values or determine the limits for both?
kindly have a look at attached file.
Regards,
Omid.
 Attachments

 r.png (44.57 KiB) Viewed 2808 times
Re: PF simulation of Eutectic solidification
Dear Omid,
Redistribution of elements between different phases can easily lead to extreme compositions in a phase because mass conservation for each element is necessary. Imagine for example a stoichiometric phase which is growing too fast. If the total composition of the grid cell is different from the stoichiometric composition of this phase, the other phase is forced to compensate for that  in case of a very small fraction of this other phase, it must adopt extreme (including negative) compositions to achieve conservation. By diffusion, these extreme phase compositions can be transported to other grid cells and get visible as extreme total composition. Even without stoichiometric phases, redistribution with linear phase diagrams easily cross the physical limits (composition between 0 and 1) when you go very high or low with temperature.
In normal cases, such a process which creates extreme or wrong compositions should be avoided by the high driving force which is created and which should work against it. But this requires thermodynamic consistency, i.e. extreme compositions always should produce correct driving forces which drive the system back to reasonable compositions.
A lack of thermodynamic consistency, in general, can have different origins.
Linearized phase diagrams may be inconsistent by design, in general or only for certain composition or temperature regions where consistency is lost. When using TQ coupling, inconsistency may be due to extrapolating too far from the reference quasiequilibrium. A third reason may be a very high kinetic undercooling which brings the system into regions where the phase diagram description is not consistent any more.
The use of "Limits" may be helpful in some cases but must be used very carefully. User limits, apart from controlling quasiequilibria in case of TQ coupling, create a local penalty driving force when local compositions reach the defined limit composition.
Bernd
Redistribution of elements between different phases can easily lead to extreme compositions in a phase because mass conservation for each element is necessary. Imagine for example a stoichiometric phase which is growing too fast. If the total composition of the grid cell is different from the stoichiometric composition of this phase, the other phase is forced to compensate for that  in case of a very small fraction of this other phase, it must adopt extreme (including negative) compositions to achieve conservation. By diffusion, these extreme phase compositions can be transported to other grid cells and get visible as extreme total composition. Even without stoichiometric phases, redistribution with linear phase diagrams easily cross the physical limits (composition between 0 and 1) when you go very high or low with temperature.
In normal cases, such a process which creates extreme or wrong compositions should be avoided by the high driving force which is created and which should work against it. But this requires thermodynamic consistency, i.e. extreme compositions always should produce correct driving forces which drive the system back to reasonable compositions.
A lack of thermodynamic consistency, in general, can have different origins.
Linearized phase diagrams may be inconsistent by design, in general or only for certain composition or temperature regions where consistency is lost. When using TQ coupling, inconsistency may be due to extrapolating too far from the reference quasiequilibrium. A third reason may be a very high kinetic undercooling which brings the system into regions where the phase diagram description is not consistent any more.
The use of "Limits" may be helpful in some cases but must be used very carefully. User limits, apart from controlling quasiequilibria in case of TQ coupling, create a local penalty driving force when local compositions reach the defined limit composition.
Bernd
Re: PF simulation of Eutectic solidification
Dear Bernd,
I am trying to start the same simulation mentioned above with solid state. I got two concerns: firstly how can I define one of those solid phases as super saturated phase with specific composition? please note that it is defined as stoichiometric phase. And secondly I get an error like below:
Routine init calls routine initTwidth
# tWidth_max( 1 : 2 ) = 1.5200000E04 s
# tWidth_max( 1 : 3 ) = 1.5200000E04 s
# tWidth_max( 2 : 3 ) = 1.5200000E05 s
# tWidth_max( 3 : 3 ) = ************* s
tWidth value too large.
With the chosen phasefield parameters,
tWidth must be less than 1.521521521521522E009 s
STOP in routine initTwidth.
How can I solve this problem?
Regards,
Omid.
I am trying to start the same simulation mentioned above with solid state. I got two concerns: firstly how can I define one of those solid phases as super saturated phase with specific composition? please note that it is defined as stoichiometric phase. And secondly I get an error like below:
Routine init calls routine initTwidth
# tWidth_max( 1 : 2 ) = 1.5200000E04 s
# tWidth_max( 1 : 3 ) = 1.5200000E04 s
# tWidth_max( 2 : 3 ) = 1.5200000E05 s
# tWidth_max( 3 : 3 ) = ************* s
tWidth value too large.
With the chosen phasefield parameters,
tWidth must be less than 1.521521521521522E009 s
STOP in routine initTwidth.
How can I solve this problem?
Regards,
Omid.
Re: PF simulation of Eutectic solidification
Dear Omid,
When you switch from ternary your eutectic system to a solidstate system, you need a new definition of the (linearized) phase diagram. Note that the assumption that all solid phased have a stoichiometric composition was an approximation which was valid only for the case of ternary eutectic reaction. Once you want to incorporate solidsolid interactions you need the corresponding phase diagram information (or a thermodynamic database) and need to figure out which compositions are really stoichiometric and which not.
With respect to the error you showed above, the problem is the negative value of the minimum time step. Probably you made a wrong input in form of a negative input value which cannot be negative, e.g. interface mobility, anisotropy coefficients, factors in misorientation model, etc. I am not sure which of these cases is checked already at input time and which could produce such an error
Bernd
When you switch from ternary your eutectic system to a solidstate system, you need a new definition of the (linearized) phase diagram. Note that the assumption that all solid phased have a stoichiometric composition was an approximation which was valid only for the case of ternary eutectic reaction. Once you want to incorporate solidsolid interactions you need the corresponding phase diagram information (or a thermodynamic database) and need to figure out which compositions are really stoichiometric and which not.
With respect to the error you showed above, the problem is the negative value of the minimum time step. Probably you made a wrong input in form of a negative input value which cannot be negative, e.g. interface mobility, anisotropy coefficients, factors in misorientation model, etc. I am not sure which of these cases is checked already at input time and which could produce such an error
Bernd
Re: PF simulation of Eutectic solidification
Dear Bernd,
Regarding the phase diagram, I have already made two isotermal cross section out of the phase diagram and have defined the new phase diagram but both the sections are above the eutectic point. I have attached the dri file ans could you please have look on it for checking the obvious problems and give me some advices regarding the diffusion definitions? I mean considering the point that at first we have all domain occupied with the super saturated Mo phase and the other phases would nucleate during the heat treatment, the diffusion part should be more focused specially at the boundaries and just simple diff key does not satisfy of course!
Thanks agian for your helps,
Regards,
Omid.
Regarding the phase diagram, I have already made two isotermal cross section out of the phase diagram and have defined the new phase diagram but both the sections are above the eutectic point. I have attached the dri file ans could you please have look on it for checking the obvious problems and give me some advices regarding the diffusion definitions? I mean considering the point that at first we have all domain occupied with the super saturated Mo phase and the other phases would nucleate during the heat treatment, the diffusion part should be more focused specially at the boundaries and just simple diff key does not satisfy of course!
Thanks agian for your helps,
Regards,
Omid.
 Attachments

 solidstate_dri.txt
 (35.66 KiB) Downloaded 168 times