Hello Bernd,
Thank you for the suggestions. This will help.
Could you please explain the seed_density model in nucleation?
My simulation domain is small and very fine grids but if I used seed_density instead of seed_undercooling then I can see only 3-4 seeds are appearing. But, seed_undercooling there are many nucleation sites. Hence, I would like to know seed_density model in details so that I can use appropriately.
What is meaning of seed_density output:
Seed type: 1 (4: 1/2)
Regards
Parimal
CMSX4
Re: CMSX4
Dear Parimal,
The seed-density model has been specifically designed for heterogeneous nucleation. Starting from a given seed-density distribution (consisting of explicit classes of virtual seeding particles with given average radius), explicit randomly distributed nucleation sites with discrete value of critical undercooling (calculated from seed radius and transformation entropy) are created at initialisation time. During runtime, only these places are checked for possible nucleation, whereby the analytical curvature model is used as default for the "small grain" stage.
The advantage of this model (compared to the "seed_undercooling" model) is that the model has a strong physical background (although seed density distributions are mostly unknown and need to be "guessed") and the seeds appear in random places. The number of potential places which are randomly selected depends on the seed particle density (particles per cm3) and is written to the .log file. When during runtime a seed is nucleated, the number of the seed class and the the number of seeds which already nucleated and which are in total available for this seed radius class is written to the screen and .log output.
If the seed_undercooling model is used for nucleation in the bulk or in a region, the number of checked points depends on the nucleation distance or shield distance (if no nucleation distance is given). Due to the algorithm of scanning the bulk or region, partially regular pattern may evolve. If the checking distance and the checking time interval are chosen too small, a considerable performance loss may be experienced.
The seed density model also has been discussed in other places:
About seed_density model
Seed density model for 2D simulations
Bernd
The seed-density model has been specifically designed for heterogeneous nucleation. Starting from a given seed-density distribution (consisting of explicit classes of virtual seeding particles with given average radius), explicit randomly distributed nucleation sites with discrete value of critical undercooling (calculated from seed radius and transformation entropy) are created at initialisation time. During runtime, only these places are checked for possible nucleation, whereby the analytical curvature model is used as default for the "small grain" stage.
The advantage of this model (compared to the "seed_undercooling" model) is that the model has a strong physical background (although seed density distributions are mostly unknown and need to be "guessed") and the seeds appear in random places. The number of potential places which are randomly selected depends on the seed particle density (particles per cm3) and is written to the .log file. When during runtime a seed is nucleated, the number of the seed class and the the number of seeds which already nucleated and which are in total available for this seed radius class is written to the screen and .log output.
If the seed_undercooling model is used for nucleation in the bulk or in a region, the number of checked points depends on the nucleation distance or shield distance (if no nucleation distance is given). Due to the algorithm of scanning the bulk or region, partially regular pattern may evolve. If the checking distance and the checking time interval are chosen too small, a considerable performance loss may be experienced.
The seed density model also has been discussed in other places:
About seed_density model
Seed density model for 2D simulations
Bernd
Re: CMSX4
Bernd wrote:Dear Parimal,
unfortunately I am not able to reproduce your problem:
I copy and paste the content of the .TCM which you posted into a Thermo-Calc 2016b console which automatically creates the CMSX4.ges5 file. I take the original CMSX4_dri.txt example from MICRESS Version 6.3 and adjust the linkage between phases:
# The database contains 3 phases:
# 1: LIQUID
# 2: FCC_L12
# 3: FCC_L12#2
# Specify relation between phase indices Micress -> TC!
# The matrix phase has in MICRESS the index 0
# Thermo-Calc index of the (MICRESS) phase 0?
1
# Thermo-Calc index of the (MICRESS) phase 1?
2
# Thermo-Calc index of the (MICRESS) phase 2?
3
# 0 -> LIQUID
# 1 -> FCC_L12
# 2 -> FCC_L12#2
#
and it runs perfectly...
Could it be you use a very old Thermo-Calc Version?
Bernd
Hej Bernd
I am new to MICRESS and I am also trying the CMSX4 example with TCNI8+MOBNI4 data bases. I only change the data bases and the phases (as you have suggested in this post) form the original example.
My simulation run without any errors. But when I looked at the results, the simulation with TCNI8+MOBNI4 shows faster growth of Gama phase compare to the original example results. I have attached a figure that shows this at 2 sec after solidification.
My question is:
What could be the reason for this. Is it because of the change of the data base. I want to clarify this before proceeding further changes.
Many thanks in advance!!
Best regards.
Chamara
- Attachments
-
- Comparison CMSX4 example with and TC data base
- CMSX4 example with TC.PNG (407.86 KiB) Viewed 5067 times
Re: CMSX4
Dear Chamara,
Welcome to the MICRESS forum!
You should not tell that you are new to MICRESS - you already joined a MICRESS basic training course
What you did (and what is correct) is just the technical stuff which is necessary to switch the database. What remains are the differences which are due to the different models and model parameters of the two databases.
The most pronounced difference (at least as it appears in the MICRESS simulation at the beginning) is the difference in Liquidus temperature for CMSX-4 of about 6-7K. For thermodynamic people this may be little, however it shifts the initial undercooling at a given temperature. Thus, to get similar results, you need to increase the initial temperature such that you get the same initial undercooling or driving force (you can check in the initial linearisation parameters in the .log file, given that you equally shift the temperature for initial equilibrium).
After having done that, the dendrites will look pretty similar. Of course, there will be differences later e.g. with the amount of gamma prime or with the segregation direction of some elements.
Bernd
Welcome to the MICRESS forum!
You should not tell that you are new to MICRESS - you already joined a MICRESS basic training course
What you did (and what is correct) is just the technical stuff which is necessary to switch the database. What remains are the differences which are due to the different models and model parameters of the two databases.
The most pronounced difference (at least as it appears in the MICRESS simulation at the beginning) is the difference in Liquidus temperature for CMSX-4 of about 6-7K. For thermodynamic people this may be little, however it shifts the initial undercooling at a given temperature. Thus, to get similar results, you need to increase the initial temperature such that you get the same initial undercooling or driving force (you can check in the initial linearisation parameters in the .log file, given that you equally shift the temperature for initial equilibrium).
After having done that, the dendrites will look pretty similar. Of course, there will be differences later e.g. with the amount of gamma prime or with the segregation direction of some elements.
Bernd