Cellular Soldification. Low carbon steel

[salarniknafs]

Hi,

I am simulating the planar/cellular transition in Fe-C system. Attached is a image obtained from the laser scanning confocal microscopy showing the cellular delta-ferrite structure. The experiment showed that the perturbation wave start at the delta-ferrite GB's at the solid/liquid interface and eventually the whole interface develops into cells as shown in the microscope image.

In the simulations however, although I witness the planar/cellular transition starting at the delta-ferrite GBs, the cells initiated for the GBs grow more rapidly than the others. This is in contrast with the experimental observations. I use linearized phase diagram, calculated from ThermoCal. There is only one solute, 0.05(wt) %carbon and the rest of the data are defined as below. Could you point out the possible factors contributing in this error?

Regards,
Salar

# AnzX:
800
# AnzY:
1
# AnzZ:
1600
# Cell dimension (grid spacing in micrometers):
0.5



# Phase interaction data
# ======================
#
# Data for phase interaction 0 / 1:
# ---------------------------------
# Simulation of interaction between phase 0 and 1 ?
# Options: phase_interaction no_phase_interaction
phase_interaction
# Averaging length for driving force between phase LIQUID and 1 ? (real)[cells]
1000.000
# Maximal driving force allowed between phase LIQUID and 1 ? (real) [J/cm**3]
50.00000
# Type of surface energy definition between phases LIQUID and 1?
# Options: constant temp_dependent
constant
# Surface energy between phases LIQUID and 1? [J/cm**2]
202E-07
# Type of mobility definition between phases LIQUID and 1?
# Options: constant automatic temp_dependent sd_kth
constant
# Kinetic coefficient mu between phases LIQUID and 1? [cm**4/(Js)]
1.0000
# Which interaction model for interface between phase LIQUID and phase 1?
# Options: standard particle_pinning solute_drag
standard
# Is interaction isotropic?
# Options: isotropic anisotropic
anisotropic
# static anisotropy coefficient? (< 1.) (changed!!!)
0.2500000
# kinetic anisotropy coefficient? (< 1.)
0.2500000
#
# Data for phase interaction 1 / 1:
# ---------------------------------
# Simulation of interaction between phase 1 and 1 ?
# Options: phase_interaction no_phase_interaction
phase_interaction
# Type of surface energy definition between phases 1 and 1?
# Options: constant temp_dependent
constant
# Surface energy between phases 1 and 1? [J/cm**2]
468E-07
# Type of mobility definition between phases 1 and 1?
# Options: constant temp_dependent sd_kth
constant
# Kinetic coefficient mu between phases 1 and 1? [cm**4/(Js)]
1.0000
# Which interaction model for interface between phase 1 and phase 1?
# Options: standard particle_pinning solute_drag
standard
# Shall misorientation be considered?
# Optionen: misorientation no_misorientation
no_misorientation
#


# Temperature gradient in z-direction? [K/cm]
78
# Cooling rate? [K/s]
-0.08333

[Bernd]

Hi Salar,

I am not very familiar to the type of problems you are dealing with, and the information you give is quite scarce. But, nevertheless, I could imagine several factors which can influence the break up behaviour and explain differences which you observe between the experiments and the MICRESS results:

- The planar-cellular transition is a tpye of instability which I assume is reacting sensibly to all types of disturbations, as well in the experiment as in the simulation. So, many types of numerical parameters like the interface mobility, grid resolution and initial structure certainly will have a strong impact on the results, and a careful calibration is necessary.

- If MICRESS is used with a linearised phase diagram, in contrary to reality, practically no noise is present. Thus, breakup must start at the grain boundary, if you begin with a planar interface. I am not sure whether this also influences the lateral propagation of the transition.

- I do not know how exact is your knowledge the experimental conditions. The relation between the longitudinal growth velocity of a cell and the lateral propagation certainly depends on the undercooling of the interface in the moment of the transition. Thus, a too high undercooling in the simulation would explain the "wrong" behaviour. As well, wrong values of the interfacial energy between delta/liquid or the diffusivity of carbon in the melt could give such a deviation.

- The triple point angles at the contact point between the delta grains and the liquid may be very important. In reality they may be different from the 120° which are obtaind with older MICRESS versions. Newer versions allow the use of the "multi_obstacle" potential which makes the triple point angles dependent on the interfacial energies. (By the way, you seem to use an extremely old MICRESS version...)

- In the experiment, I guess, the grain boundaries are formed by delta grains which are growing together shortly before the planar-cellular transition. So, there may be a laterally unequal distribution of composition or front undercooling. If this is true, the whole process should be simulated, because the initial "flat" situation is unrealistic.

These are just some guesses without having more insight into the system and its characteristics!

Bernd