Triggering Dendrite Arms

[bjoern]

Dear Bernd,

starting with the P-Peak example, I am trying to simulate spheroidal graphite iron with primary austenitic dendrites as they appear in experiments. However, I obtain no dendrite arms at all. I decreased interface energy and increased mobility in reasonable range. I varied also anisotropy factors up to .9 without success. Decreasing dG averaging didn't help either. Do you have an idea what parameters could help to make dendrite arms appear

Thank you
Björn

[Bernd]

Hi Björn,

Welcome to the MICRESS forum!

Could it be that dendrite arms are "destroyed" by the graphite particles which have a strong attraction to the dendrite?

Apart from that, missing dendrite arms can be due to

- a low interface mobility (if the curvature contribution to the driving force, as seen in the .driv output, is small in contrast to the kinetic contribution),
- the physical setup of the problem, i.e. the combination of the cooling condition, the interfacial energy, the domain size and the diffusion coefficients which does not allow dendrite arms to form,
- a missing anisotropy of the interface,
- or anything else which could destroy the correct interface profile.

What you should look at first is the .driv output, and then whether possible there was a typo in the interface energy liquid/fcc...

[bjoern]

Dear Bernd,

I wanted to give short feedback on my work with regard to DAS. It seems to be the case that the physical-chemical setup of the problem is such, that dendrite arms hardly grow, that is, the growth kinetic of the FCC phase is low in that setup. However, in reality we can see dendrite arms grown due to perturbation of the interface, not due to graphite nodules. The first problem was, I had to reduce the time stepping to 1e-4. I could reproduce dendrite arms assuming some unrealistic parameters: Interface energy 1e-5, delta stiffness 0.9, rotating the dendrite by 15°.

Since I was unhappy with the setup and in addition it didn't work properly in terms of fragmentation, I tried out harmonic expansions. After looking in Janin's theses, I found that Hoyt, Asta, Karma 2001 deduced coefficients for Ni eps1:0.02269 eps2:-.01168. Based on these coefficients I played around and found that the double value of these coefficients gives satisfactory results for the case of SGI. Therefore, I could switch back to a realistic interface energy and no rotation of the dendrite. Right now, I'm happy, I hope you too

Thanks for helping
Björn

[Bernd]

Hi Björn,

if you are happy, that also makes me happy

Apart from what you did to solve the problem, you could also consider to increase the domain size if it does not correspond to the correct grain size (or 1/4 of it). This would also change side branch behaviour, especially when latent heat is taken into account!

Bernd