Dear MICRESSers, I have two questions about Al-Si eutectic simulation:
1) TQ-coupling is always give the following error messages in spite of whatever I input for the interaction parameters.
STOP in SolveCCalFunTQ 18
interface between phases 1 2
not converged, error= 1 in interface 1 2
try harder! Error = 1
not converged, error= 1 in interface 2 0
try harder! Error = 1
not converged, error= 1 in interface 2 0
try harder! Error = 1
But when I use the same parameters with linear phase diagram instead of TQ coupling, the calculation is finished smoothly. Can you tell me how to solve this "not converged error" in TQ coupling?
2). When the solidification is finished, I compare the composition data stored in these two files. One is named "AlSi_eutectic_conc1_00000800.txt", and the other is named "AlSi_eutectic_TabC.txt". As you can see from these attached files. the highest composition of Si in entire 2D field shown in file one is only about 40%, but Si in phase 2 shown in file two is 100%. How to understand this difference between these two files?
Best regards
Two questions about Al-Si eutectic simulation
Re: Two questions about Al-Si eutectic simulation
Dear Kong,
welcome to the MICRESS forum! I will try to answer your questions:
1.) If you are using TQ-coupling, MICRESS is transferring concentration data from the interfaces to the Thermo-Calc libraries in order to obtain the local driving forces and information about the redistribution behaviour. Numerical problems can arise if those data are unreasonable (e.g. negative), which can be due to poorly chosen numerical parameters (like interface mobility). If you do the same with a linearised phase diagram description, you will not see any complaint even if the concentrations are negative!. In so far, TQ-coupled simualtions are much more sensitive to numerical instabilities, but one should always try to achieve numerically sound conditions, also with linear phase diagram descriptions!
A good candidate for checking numerical stability would be the .driv output (look here).
In complex simulations, it may sometimes be unavoidable to get some of those messages, so few of them may be still ok...
2.) Unfortunately, I cannot see your attachments, but I can guess what you got: If the size of the silcon particles is less than the interface thickness, one will obtain only "diffuse" particles with correspondingly lower Si (mixture) content in the .conc1 output. If you graphically like to see the Si phase composition, use the the .cPha1 output with
out_conc_phase 2
in the output specification, which means that in the interface regions, the phase composition of phase 2 (if this is the Silicon phase in your example) is written instead of the mixture composition. This output should be directly comparable to the .TabC.
The reason for "diffuse" growth of secondary phases may be either a too low spatial resolution or, again, poor numerical parameters.
Bernd
welcome to the MICRESS forum! I will try to answer your questions:
1.) If you are using TQ-coupling, MICRESS is transferring concentration data from the interfaces to the Thermo-Calc libraries in order to obtain the local driving forces and information about the redistribution behaviour. Numerical problems can arise if those data are unreasonable (e.g. negative), which can be due to poorly chosen numerical parameters (like interface mobility). If you do the same with a linearised phase diagram description, you will not see any complaint even if the concentrations are negative!. In so far, TQ-coupled simualtions are much more sensitive to numerical instabilities, but one should always try to achieve numerically sound conditions, also with linear phase diagram descriptions!
A good candidate for checking numerical stability would be the .driv output (look here).
In complex simulations, it may sometimes be unavoidable to get some of those messages, so few of them may be still ok...
2.) Unfortunately, I cannot see your attachments, but I can guess what you got: If the size of the silcon particles is less than the interface thickness, one will obtain only "diffuse" particles with correspondingly lower Si (mixture) content in the .conc1 output. If you graphically like to see the Si phase composition, use the the .cPha1 output with
out_conc_phase 2
in the output specification, which means that in the interface regions, the phase composition of phase 2 (if this is the Silicon phase in your example) is written instead of the mixture composition. This output should be directly comparable to the .TabC.
The reason for "diffuse" growth of secondary phases may be either a too low spatial resolution or, again, poor numerical parameters.
Bernd