reg high temperature and cooling rates solidification

[swetha]

Hello Sir,
I am new to micress and am trying to simulate a ternary alloy system single phase solidification initially with selective laser melting(slm) process parameters which involves very high temperature gradients and cooling rates around 35000k/cm and 10^4-10^5 k/s.I am not getting a stable simulation and i am getting error messages like serious error in relinearisation and error in interface 0/1.What parameters i have to take into concern for simulating such high process conditions.I have attached my driving file for that ternary system of which single phase(fcc_a1) solidification is the primary work with slm process conditions.What changes can be done to this driving file to get a stable simulation?Looking forward for your reply.

Regards,

[Bernd]

Dear swetha,

Welcome to the MICRESS forum.

There are some points in your input file which are obviously not correct. First, there is a fundamental issue with your time and length scales: If you intend to simulate SLM, I would expect a very short time scale and very fine dendrites. This is not reflected in the driving file (spatial resolution ("cell dimension", output times, intervals for updating diffusion data, thermodynamic data, etc.). The essential and first step must be to define the proper scales.

Apart from that, I found two other problems:

- You do not allow diffusion in the liquid phase (phase 0):

# How shall diffusion of component 1 in phase 0 be solved?
diagonal n
# How shall diffusion of component 1 in phase 1 be solved?
multi gg
# How shall diffusion of component 2 in phase 0 be solved?
diagonal n
# How shall diffusion of component 2 in phase 1 be solved?
multi gg

This would not allow dendrites to form and lead to numerical problems.

- You fix the concentrations at the top boundary condition to a composition which is different form your alloy:

[/i]# Boundary conditions for concentration field in each direction
# Options: i (insulation) s (symmetric) p (periodic/wrap-around) g (gradient) f (fixed)
# Sequence: W E (S N, if 3D) B T borders
sssf
# Fixed value for concentration field for component 1 in T-direction
3.0000
# Fixed value for concentration field for component 2 in T-direction
3.0000
# Unit-cell model symmetric with respect to the x/y diagonal plane?
# Options: unit_cell_symm no_unit_cell_symm[/i]

What is the reason?

Bernd

[swetha]

Dear Bernd,
Thanks for your reply.I will correct those and try to simulate.

Regards,
swetha.

[Bernd]

Dear swetha,

If you get growth with reasonable microstructures at less extreme cooling conditions, then you obviously managed to find proper length and time scales for simulation. You should check the .driv output like explained here.

The question what happens at much higher cooling rates and temperature gradients is an important but unsolved scientific question: Up to which cooling rates can we still assume local (quasi-)equilibrium? If we have too high rates and massive solute trapping as a consequence, equilibrium-based thermodynamic databases cannot be used any more...

Bernd

[swetha]

Dear Bernd,
Thanks for your reply.So,Is there any possibility/solution to simulate such high cooling rates and thermal gradients without those databases .You have any idea regarding this? Also at times there are warning statements like" warning of demixing of the component Cr in phase 0/1". what is the reason for this?

Regards,
Swetha.

[swetha]

Dear Bernd,
Also,with lower cooling conditions with which a simulation I got ,has no interface area at a particular location(very little part) at the bottom...what must be the reason for that ?

Micress has few modifications to be done regarding high temperature gradients and cooling rates like few parameters to be included (velocity dependant terms) which makes such process conditions simulate or is it already existing which i am unaware of?Can you please help me with this Looking forward for your reply.

Regards,
swetha.

[Bernd]

Dear swetha,

to avoid the "demixing" problem you should switch to diagonal extrapolation method for redistribution which you can do by using the keyword "interaction" in the "Phase diagram - input data" section without further parameters. This avoids the demixing artifact of the somewhat more exact multi-binary extrapolation method. The loss of exactness can be compensated by a (slightly) reduced relinearisation interval.
With respect to the high cooling rates, there are no databases existing for that . The first step should be to find out whether the local quasi-equilibrium approach still holds or not, and the only way to find out this, I guess, is to compare the results of simulations with experiments.
The problems you describe for lower cooling rates are probably due to not well-adjusted numerical parameters. You should post some pictures (.conc and .driv output).

Bernd

[Bernd]

Dear Swetha,

it makes little sense to discuss your images in the forum if you do not put them here...

Apart from that, it seems that the simulation results do not fit to the input file which you sent together by PM: The results are coarse (Deltax=1mum), while in the input file the resolution is much finer. In case of the results you show, resolution was definitively too low!

Bernd

[swetha]

Dear Bernd,
Thank you for your previous replies.Sure,I will post everything in forum here after.diagonal extrapolation method helped to rectify the error of demixing .Now,with increasing velocities I am simulating for the same ni-cr-mo system,where for 1cm/s velocity ,the screen shows the following error message

"Intermediate output for t = 0.30000 s
CPU-time: 1863 s
Current phase-field solver time step = 2.32E-06 s
Unable to read from file Results_v3/v3G.korn
STOP in routine outScroll"

Can you please say what does it mean?

Regards,
swetha.

[Bernd]

Dear Swetha,

when using "moving frame" and the option "out_moving_frame", a set of output files is created which sum up all the areas of the domain which have been moved out during the simulation run. These files have an additional "G" at the end of the file name (e.g. ResultG.conc1) and have to be opened for appending many times during the simulation run.
The error message you show says that MICRESS cannot open that file with specified name (v3G.korn). This can have various reasons, e.g. the file has been deleted during runtime, network connection to the storage device is lost, the file is used by another application, a second MICRESS instance is writing simultaneously to the same file, ...
You should check whether this file is in its correct location (the directory where the results are written), and whether it is readable.

Bernd