Dear Deep,
if there are no confidentiality issues I would like to just paste the driving file I have used. Do you agree ? Otherwise I would send you a PN .
Bernd
---
original message from Bernd
Using the seed density model for heterogeneous nucleation
Re: Using the seed density model for heterogeneous nucleation
Sir,
can you just mail me that ,if possible, on my address.
---
original message from Deep
can you just mail me that ,if possible, on my address.
---
original message from Deep
Re: Using the seed density model for heterogeneous nucleation
file has been sent!
But be aware that random generators are machine-dependent! It may happen that instead getting one primary bcc grain you may get two ore none - if this happens then you should either set the primary grain manually using the grain input or increase the calculation domain or the seed density!
Bernd
---
original message from Bernd
But be aware that random generators are machine-dependent! It may happen that instead getting one primary bcc grain you may get two ore none - if this happens then you should either set the primary grain manually using the grain input or increase the calculation domain or the seed density!
Bernd
---
original message from Bernd
Re: Using the seed density model for heterogeneous nucleation
Sir,
I recently posted 1 question in the forum which is now nowhere.
I repeat my doubt once again,if possible please resolve it:
whether it is possible to simulate undercooled liquids, i.e., you cool the liquid so fast that nucleation and solidification do not have time to start.
Something like that:
http://www.esrf.eu/UsersAndScience/Publ ... /MAT2.html
if I am not wrong ,is it equivalent to using seed undercooling model with high undercooling?
Sir, please suggest some method by which the abovesaid conditions can be simulated, wherein the metal gets undercooled below its liquidus without solidifying and then solidifies due to localised fluctuations in the configuration space.
I would be grateful.
Thanking You,
Sincerely,
Deep
---
original message from Deep
I recently posted 1 question in the forum which is now nowhere.
I repeat my doubt once again,if possible please resolve it:
whether it is possible to simulate undercooled liquids, i.e., you cool the liquid so fast that nucleation and solidification do not have time to start.
Something like that:
http://www.esrf.eu/UsersAndScience/Publ ... /MAT2.html
if I am not wrong ,is it equivalent to using seed undercooling model with high undercooling?
Sir, please suggest some method by which the abovesaid conditions can be simulated, wherein the metal gets undercooled below its liquidus without solidifying and then solidifies due to localised fluctuations in the configuration space.
I would be grateful.
Thanking You,
Sincerely,
Deep
---
original message from Deep
Re: Using the seed density model for heterogeneous nucleation
Dear Deep,
I think you are right, simulating undercooled melts is very easy in the sense that you just do not nucleate or set a high critical undercooling in the seed undercooling model. But to be clear, we have no physical model implemented for nucleation under extremely high undercoolings. The seed density model works for heterogeneous nucleation only, the seed undercooling "model" consists of just choosing the correct value for the critical undercooling and is not physical at all...
If you find out a good model for that could think about implementing it in MICRESS!
By the way, I sent you an answer to your prior question, which also disappeared, probably due to the moving of the page to another server. I will ask our administrator whether we can get them back!
I will be on holidays next week, so please be patient if you have more questions...
Best wishes
Bernd
---
original message from Bernd
I think you are right, simulating undercooled melts is very easy in the sense that you just do not nucleate or set a high critical undercooling in the seed undercooling model. But to be clear, we have no physical model implemented for nucleation under extremely high undercoolings. The seed density model works for heterogeneous nucleation only, the seed undercooling "model" consists of just choosing the correct value for the critical undercooling and is not physical at all...
If you find out a good model for that could think about implementing it in MICRESS!
By the way, I sent you an answer to your prior question, which also disappeared, probably due to the moving of the page to another server. I will ask our administrator whether we can get them back!
I will be on holidays next week, so please be patient if you have more questions...
Best wishes
Bernd
---
original message from Bernd
Re: Using the seed density model for heterogeneous nucleation
Sir,
Thanks a lot for this reply...
As you are on holidays next week,let me put just 1 question regarding the solid state transformation of aluminum bronze,as next week is the second last week for me...
Sir, as prescribed by you, I had used the seed density model from the bulk for the new grains in the gamma phase to emerge from the bulk of alpha phase present.
However, finally I haven't obtained any gamma phase ,the .Tabf output, on the other hand shows that there is third phase present.Moreover in the .phas file,the parameter varies from only -1 to 2. So it gives me an impression that there is no scope for the third phase to appear.
Also as seen in the examples, only 2 phases maximum is considered. So, is it possible in Micress to do such kind of simulation. Is yes, then can you suggest some ways to incorporate the third phase in my results.
I can also send the input file , if you need.
Thanking You,
Sincerely,
Deep
---
original message from Deep
Thanks a lot for this reply...
As you are on holidays next week,let me put just 1 question regarding the solid state transformation of aluminum bronze,as next week is the second last week for me...
Sir, as prescribed by you, I had used the seed density model from the bulk for the new grains in the gamma phase to emerge from the bulk of alpha phase present.
However, finally I haven't obtained any gamma phase ,the .Tabf output, on the other hand shows that there is third phase present.Moreover in the .phas file,the parameter varies from only -1 to 2. So it gives me an impression that there is no scope for the third phase to appear.
Also as seen in the examples, only 2 phases maximum is considered. So, is it possible in Micress to do such kind of simulation. Is yes, then can you suggest some ways to incorporate the third phase in my results.
I can also send the input file , if you need.
Thanking You,
Sincerely,
Deep
---
original message from Deep
Re: Using the seed density model for heterogeneous nucleatio
Dear Deep,
To be honest, I do not really understand what you mean...
In a previous answer to your question about how to nucleate the gamma phase, I sent a phase diagram which showed the solidification sequence:
LIQUID-->LIQUID+BCC-->LIQUID+FCC+BCC-->BCC+FCC-->FCC-->gamma+FCC
I hope you read this answer because it is one of the messages which disappeared in the course of switching the forum server. As I am no not at ACCESS, I am not able to send it again .
If you agree on this sequence then you need to consider three different solid phases. You have to put it into the .GES5 file using Thermo-Calc and to define three solid phases in MICRESS.
I guess you already did that, because you say you have seen the third phase (=gamma) in the .TabF output, i.e. you have seen a fraction >0 for phase 3 in MICRESS!
If you did not see this phase in the .phas output, I guess it is just because no individual grain of phase 3 ever got big enough to show up in this output - only grid points with more than 80% (as far as I remember) of a phase are shown with the phase number value, otherwise -1 indicating interphase is displayed.
Anyway you should see the nucleation output on screen (or in the text file where you redirected the output to).
If it is like that (i.e. you got nucleation, but the grains of phase 3 did not grow very big), then it could be because the mobility of the 0/3 interface is too low.
On the other side, in the last part of your question, you give the impression that you did even not consider a third phase to appear
...
If this is the case then the steps would be:
1.) Create a new .GES file which containes at least the four phases which you need for your simulation. Unfortunately I have actually no Thermo-Calc available, and can tell you the phase names only from memory: LIQUID:L, BCC_B2, FCC_A1 and GAMMA_??.
2.) Add gamma as phase number 3 to MICRESS. You will have to add a lot of phase interactions: 0/3, 1/3, 2/3, 3/3. Only for 2/3 you specify "phase_interaction", for the others "no_phase_interaction", as these phase pairs should never appear. Furthermore you have to add diffusion data, molar volumes and so on... The screen output will guide you through the input until you have everything consistent.
3.) Choose the nucleation interval for nucleation of phase 3 (reference phase is FCC) such that MICRESS starts checking for nucleation not at too high temperatures. The TQ initialisation of the phase interaction could fail if you check nucleation at temperatures where the phase is not existing at all!
4.) Once the phase nucleates properly (and the initial phase diagram linearisation which you find in the .log output seems reasonable) you have to find a reasonable value for the mobility:
see here: http://board.micress.de/viewtopic.php?f=17&t=17
I hope this helps!
I am on holidays next week, but maybe I can l lock in from somewhere and see further questions...
Best regards
Bernd
---
original message from Bernd
To be honest, I do not really understand what you mean...
In a previous answer to your question about how to nucleate the gamma phase, I sent a phase diagram which showed the solidification sequence:
LIQUID-->LIQUID+BCC-->LIQUID+FCC+BCC-->BCC+FCC-->FCC-->gamma+FCC
I hope you read this answer because it is one of the messages which disappeared in the course of switching the forum server. As I am no not at ACCESS, I am not able to send it again .
If you agree on this sequence then you need to consider three different solid phases. You have to put it into the .GES5 file using Thermo-Calc and to define three solid phases in MICRESS.
I guess you already did that, because you say you have seen the third phase (=gamma) in the .TabF output, i.e. you have seen a fraction >0 for phase 3 in MICRESS!
If you did not see this phase in the .phas output, I guess it is just because no individual grain of phase 3 ever got big enough to show up in this output - only grid points with more than 80% (as far as I remember) of a phase are shown with the phase number value, otherwise -1 indicating interphase is displayed.
Anyway you should see the nucleation output on screen (or in the text file where you redirected the output to).
If it is like that (i.e. you got nucleation, but the grains of phase 3 did not grow very big), then it could be because the mobility of the 0/3 interface is too low.
On the other side, in the last part of your question, you give the impression that you did even not consider a third phase to appear
...If this is the case then the steps would be:
1.) Create a new .GES file which containes at least the four phases which you need for your simulation. Unfortunately I have actually no Thermo-Calc available, and can tell you the phase names only from memory: LIQUID:L, BCC_B2, FCC_A1 and GAMMA_??.
2.) Add gamma as phase number 3 to MICRESS. You will have to add a lot of phase interactions: 0/3, 1/3, 2/3, 3/3. Only for 2/3 you specify "phase_interaction", for the others "no_phase_interaction", as these phase pairs should never appear. Furthermore you have to add diffusion data, molar volumes and so on... The screen output will guide you through the input until you have everything consistent.
3.) Choose the nucleation interval for nucleation of phase 3 (reference phase is FCC) such that MICRESS starts checking for nucleation not at too high temperatures. The TQ initialisation of the phase interaction could fail if you check nucleation at temperatures where the phase is not existing at all!
4.) Once the phase nucleates properly (and the initial phase diagram linearisation which you find in the .log output seems reasonable) you have to find a reasonable value for the mobility:
see here: http://board.micress.de/viewtopic.php?f=17&t=17
I hope this helps!
I am on holidays next week, but maybe I can l lock in from somewhere and see further questions...
Best regards
Bernd
---
original message from Bernd
Last edited by jan on Mon Jun 23, 2008 9:58 am, edited 2 times in total.
Re: Using the seed density model for heterogeneous nucleation
Sir,
Thanks a lot for the answer.
I considered the 3rd phase in my input file and also included it my thermo calc file( the phase is precisely ALCU_GAMMA_D )
yes, the nucleation is taking place but the grains are not growing for the third phase.
In the previous answer ,it was advised to keep the mobility of 0/3 interface to be high, but again you have said only to consider the interaction between 2/3, ,.as that is the only interaction happening physically.
The last part of my question was related to working of MIcress, It was asked that if i consider 3 phases in the input(as I have done), will micress be able to display all 3 ?
more recently, my simulations have started giving this error:
Old start values kept
CALFUN:Error 1611 301
Error 1611 in TQ routine!
----------------------------
time: = 8.671865155278343
So i am confused at this stage.
Please suggest me how to modify the input file so as to get the final result.
Thanking You,
Sincerely,
Deep
---
original message from Deep
Thanks a lot for the answer.
I considered the 3rd phase in my input file and also included it my thermo calc file( the phase is precisely ALCU_GAMMA_D )
yes, the nucleation is taking place but the grains are not growing for the third phase.
In the previous answer ,it was advised to keep the mobility of 0/3 interface to be high, but again you have said only to consider the interaction between 2/3, ,.as that is the only interaction happening physically.
The last part of my question was related to working of MIcress, It was asked that if i consider 3 phases in the input(as I have done), will micress be able to display all 3 ?
more recently, my simulations have started giving this error:
Old start values kept
CALFUN:Error 1611 301
Error 1611 in TQ routine!
----------------------------
time: = 8.671865155278343
So i am confused at this stage.
Please suggest me how to modify the input file so as to get the final result.
Thanking You,
Sincerely,
Deep
---
original message from Deep
Re: Using the seed density model for heterogeneous nucleation
Dear Deep,
you are right, my previous posting was a bit confusing
! I meant "2/3" interface instead of 0/3...
I had a look on your example "input1". In principle it seems to be just a problem of numerical parameters. Try the following:
- increase the maximum time step value from 1.E-5 to 1.E-3. This is still stable with respect to the temperature coupling, but speeds up the simulation tremendously
- reduce the 1/2 mobility by a factor of 10. This is still enough!
- reduce the 1/3 mobility also at least by a factor of 10.
- increase the seed density for seed type three. I would expect a higher density for phase 3 particles.
- You are checking for nucleation of phase 3 a bit too late. Increase a bit the upper limit of the checking temperature interval
- diffusion in phase 1 is slow, so that the 1/3 transformation cannot be really fast. On the other side the constant heat extraction which you assume is unrealistic, because typically the heat flux decreases if you get close to room temperature. You could try to reduce the heat extraction rate with time using several connecting points in the boundary condition input.
With some these changes I have run your example, and even with the low resolution of 2.5 microns I could see phase 3 as well in the .TabF output as in the .conc1 image. I have to admit that I did not use exactly the same MICRESS version, but this should not change too much!
If you have still problems, please send me also the .log output!
Good luck!
Bernd
---
original message from Bernd
you are right, my previous posting was a bit confusing
! I meant "2/3" interface instead of 0/3...I had a look on your example "input1". In principle it seems to be just a problem of numerical parameters. Try the following:
- increase the maximum time step value from 1.E-5 to 1.E-3. This is still stable with respect to the temperature coupling, but speeds up the simulation tremendously
- reduce the 1/2 mobility by a factor of 10. This is still enough!
- reduce the 1/3 mobility also at least by a factor of 10.
- increase the seed density for seed type three. I would expect a higher density for phase 3 particles.
- You are checking for nucleation of phase 3 a bit too late. Increase a bit the upper limit of the checking temperature interval
- diffusion in phase 1 is slow, so that the 1/3 transformation cannot be really fast. On the other side the constant heat extraction which you assume is unrealistic, because typically the heat flux decreases if you get close to room temperature. You could try to reduce the heat extraction rate with time using several connecting points in the boundary condition input.
With some these changes I have run your example, and even with the low resolution of 2.5 microns I could see phase 3 as well in the .TabF output as in the .conc1 image. I have to admit that I did not use exactly the same MICRESS version, but this should not change too much!
If you have still problems, please send me also the .log output!
Good luck!
Bernd
---
original message from Bernd
Re: Using the seed density model for heterogeneous nucleation
A small addition:
Do you really think that diffusion in phase 3 is that fast as you specify (2.E-6 cm2/s)? This also can intoduce numerical issues which are not necessary!
Bernd
---
original message from Bernd
Do you really think that diffusion in phase 3 is that fast as you specify (2.E-6 cm2/s)? This also can intoduce numerical issues which are not necessary!
Bernd
---
original message from Bernd