Hello,
I´m using MICRESS 7.1 and I have a problem while simulating dendritic growth wiht the bridgman approach.
I want to simulate dendritic grwoth of an Ni-Base superalloy (CM186LC). My Problem is, that during the solidification, a thick interface and holes in the dendrites are forming.
I Have tested different options (grid size, kinetic coefficient and interface stabilisation).
But the Problem still is there.
If anyone has any ideas for solving the problem, i would be very pleased.
With kind regards
Sean
Formation of a thick interface during dendritic growth
Formation of a thick interface during dendritic growth
- Attachments
-
- Unbenannt.PNG (255.52 KiB) Viewed 32748 times
-
- CM186LC_Längsschliff.dri
- (25.27 KiB) Downloaded 189 times
Re: Formation of a thick interface during dendritic growth
Dear Sean,
At the first glimpse it looks as if grid resolution would be too low, but only at the first glimpse. Given that you use mob_corr, low resolution would rather result in slow brick-like growth with straight interfaces. Furthermore, you say that you already have checked that...
I cannot test your .dri-file because I don't have the required .GES5 file. But everything looks quite fine there, except the value of 2.0 for the interface thickness in cells, which is not at all reasonable (you need at least 2.5 cells!). Probably you have tried that value in order to overcome your problem, so I don't think that this could be the point.
Thus, it must be something nasty which I cannot see directly. Could it be e.g. that reading diffusion data from database did not work like you expected? I also can't see any segregation in the .conc1 output. Simulating without diffusion in the liquid could easily explain the destroyed interfaces. You should check the .diff and/or the .TabD text outputs.
Further, if that also was not the problem, you should have a look on the distribution of all elements. Perhaps there is a numerical issue (which can easily happen when using the full diffusion matrix with local extrapolation in the solid phase). Check whether there are extreme or negative values anywhere. If you have a problem with switching of fcc to the ordered composition set (instead of nucleation of γ'-phase) you could cure that by using the "disordered" keyword in the numerical parameters section. Did you already have nucleation of γ'-phase? Diffusion of this phase could also be a source of problems.
And did you get any meaningful error messages?
Bernd
At the first glimpse it looks as if grid resolution would be too low, but only at the first glimpse. Given that you use mob_corr, low resolution would rather result in slow brick-like growth with straight interfaces. Furthermore, you say that you already have checked that...
I cannot test your .dri-file because I don't have the required .GES5 file. But everything looks quite fine there, except the value of 2.0 for the interface thickness in cells, which is not at all reasonable (you need at least 2.5 cells!). Probably you have tried that value in order to overcome your problem, so I don't think that this could be the point.
Thus, it must be something nasty which I cannot see directly. Could it be e.g. that reading diffusion data from database did not work like you expected? I also can't see any segregation in the .conc1 output. Simulating without diffusion in the liquid could easily explain the destroyed interfaces. You should check the .diff and/or the .TabD text outputs.
Further, if that also was not the problem, you should have a look on the distribution of all elements. Perhaps there is a numerical issue (which can easily happen when using the full diffusion matrix with local extrapolation in the solid phase). Check whether there are extreme or negative values anywhere. If you have a problem with switching of fcc to the ordered composition set (instead of nucleation of γ'-phase) you could cure that by using the "disordered" keyword in the numerical parameters section. Did you already have nucleation of γ'-phase? Diffusion of this phase could also be a source of problems.
And did you get any meaningful error messages?
Bernd