Cr-Ni eutectic alloy simulation

[ralph]

Ah got it.
You switched off interaction of phase 1 and 2, but you try to give a phase diagram input for this.
Comment this part.

Ralph

[lihaoge]

Dear Ralph,

Thank you very much! I can get the lamellar growth now. Many thanks for you and Bernd's support!

lihaoge

[lihaoge]

Dear Bernd

The microstructure which I get from the .phas file seems like the lamellar. But there is something strange. I have defined 6 grains at the beginning as an initial lamellar structure but after 0.07s there are 3 of them seems like be swallowed up by the others. I increase the resolution and at this time, the lamellar seems like a "sine function".I have tried many times but it still seems strange. Do you have any ideas about this?
Many thanks for your support.

lihaoge

[Bernd]

Dear lihaoge,

This looks quite nice!

It is a well-known phenomenon that oscillatory modes appear in lamellar eutectic growth. In a binary eutectic, I think, it is a sign that you are outside of the stable range of spacings.

I would try to change 3 things:

a) Make the simulation domain broader. The small domain size (in x-direction) implies a strong restriction for selection of the spacing.

b) Play with the initial spacing and the initial undercooling. Perhaps, due to high initial undercooling, lamellae are vanishing at the very beginning and cannot be restored afterwards

c) Play with the nucleation condition. Perhaps, the formation of new lamellae is hindered by poor nucleation conditions.

Bernd

[lihaoge]

Dear Bernd

Many many thanks for your support for my problems. I have tried many times and still have some problems.
Like you said before, I am outside of the stable range of spacings and the lamellar seems like be swallowed up by the others. But when I made the simulation domain broader (in x-direction) like 500x1000, the oscillatory modes appeared. I have increased the initial spacing and it did not help. The eutectic temperature is around 1618K according to thermocalc so I defined the temperature at the bottom as 1600K. I have also tried to increase the cooling rate in order to get the lamellar but results show no change. Do you have some ideas which paramates that I should change? Thank you very much!

lihaoge

[Bernd]

Dear lihaoge,

You should not start with undercooling but directly at the eutectic temperature. If you start like in your case with almost 20K undercooling, the eutectic front has to advance around 1mm (according to your temperature gradient) before it can get to stationary growth (even without cooling rate). This is more than 10000 grid cells in your case!

In order to check the correct front temperature, you can use the optional "front_temp" keyword for the fraction table output (.TabF) which gives you the average temperature of the solid-liquid front.

Bernd

[lihaoge]

Dear Bernd

Thank you for your advice. I have started directly at the eutectic temperature this time and increase the grid resolution in order to be chosen such that at least 10-20 grid cells correspond to the expected lamellar spacing. But this time the results show strange. According to the _TabF:
# Simulation Temperature Fraction Fraction Fraction
# time [s] [K] Liquid Phase 1 Phase 2
0.00000 1616.00000 0.95531862 0.02302218 0.02165920
0.100000 1611.00000 0.88435646 0.03410492 0.08153862
0.200000 1606.00000 0.83008333 0.03664849 0.13326818
0.300000 1601.00000 0.77276865 0.03664698 0.19058438
0.400000 1596.00000 0.70649175 0.03665842 0.25684983
0.500000 1591.00000 0.62544923 0.03669240 0.33785837
0.600000 1585.76480 0.25098529 0.59915478 0.14985993
0.700000 1579.94400 0.23328320 0.55241119 0.21430561
0.800000 1573.90240 0.22673299 0.54945332 0.22381368
Phase 1(FCC) and phase 2(BCC) were always exist during the simulation, but according to the results, FCC vanished at 0.2s and appeared again at 0.51s. I put the results pictures to the attachments. And I increase the cooling rate from 50K/s to 100K/s and this time, FCC did not vanished but oscillating structures obtained at the beginning. In my opinion, high cooling rate can help me to stop the lamellae vanishing because interface has no time to vanish, low cooling rate lead to the smallest interface energy and interface vanish. Am I wrong?

lihaoge

[Bernd]

Hi lihaoge,

I did not get your argument with the high cooling rate, but the results now look quite reasonable. Nevertheless, you definitively need to check whether the interface is close to the eutectic temperature!

Bernd

[lihaoge]

Dear Bernd

I'm sorry to bother you again, I have a simple question that I can't understand. Is the grid spacing as same as mesh size? I want to increase the resolution so I want to decrease the mesh size. But I don't know how to get it.
Thank you very much！

lihaoge

[Bernd]

Dear lihaoge,

I am not very good in English neither, but typically "mesh size" means the same as grid spacing, namely the distance between two grid cell centers in the main grid directions.

In order to increase grid spacing, you go to the "Geometry" settings of the driving file (at the top) and reduce the "Cell dimension" accordingly. If you want the domain to keep the same overall size, you need to multiply the number of cells in x and z-direction by the same factor by which you divided the grid spacing.
In principle, that is all. However, in many cases you need to adapt numerical parameters which before have been optimized for the coarser grid.

Bernd