Initial Concentration

technical aspects of .dri file generation (e.g. debug mode ) etc...
R.Hess
Posts: 22
Joined: Fri Aug 12, 2016 2:14 pm
anti_bot: 333

Initial Concentration

Post by R.Hess » Tue Dec 06, 2016 2:38 pm

Hello,
i want to simulate the following phase transitions:

Ferrite/Pearlite -> Austenite (heating)
Austenite -> Martensite (fast cooling)

It is well known, that there is no way to represent Pearlite as a separate Phase in Micress. Because of that, it seems to be very difficult to set a FP strucuture (as a mixture of Ferrite and cementite) as initial strucuture.

I changed some settings in the GammaAlphaPearlite_TQ example (initial composition, cooling rate) and my idea was, to set the last "step" of this example as initial for my first simulation. But how can i do that?

I tried it with the "restart" -function. But for that, i need some further informations:
1) what does the "shiftX , shiftY .." mean?
2) when i try to to it that way, what about the initial concentratin input and grain input, etc? is everything replaced by the inputs of the restart_file?
3) what about the diffusion input?... what i want to know is, is it possible, to change the diffusion input data in the "second" simulation? (when in the "first" simulation, the diff. input was coupled to TQ, is it possible to set it in the "second" simulation manually?

Are there any other ways? I think, it is not possible to use those ASCII-Files, because of the carbon concentration, or am i wrong?

It would be great, if someone has some helpful ideas :)

Furthermore, is it possible to start a "noTQ simulation", from a initial strucuture of a "TQ simulation" ?

Best regard

Raphael

R.Hess
Posts: 22
Joined: Fri Aug 12, 2016 2:14 pm
anti_bot: 333

Re: Initial Concentration

Post by R.Hess » Tue Dec 06, 2016 2:46 pm

Additional Information:

when i try to run the simulation with this restart option, following error occurs:
Unable to read from File .....
Reading nucleation data



what does that mean? Or why is it possible to read all informations except the nucleation data?

Bernd
Posts: 1504
Joined: Mon Jun 23, 2008 9:29 pm

Re: Initial Concentration

Post by Bernd » Wed Dec 07, 2016 5:18 pm

Dear Rafael,

I think most of your questions have already been answered in this thread. Therefore I will answer in short form:

1) The "shiftX , shiftY .." is asked only when you use the restart option "structure_only" (what you should do by the way). This most flexible restart option allows you to read only parts or multiples of the simulation domain of the first simulation as initial microstructure. In this case, "shiftX , shiftY .." permits to shift the domain relative to the origin at coordinates [0,0,0] (lover left corner). The 3 shift values (plus a rotation parameter) are optional, you can just omit them if you want to continue with the identical domain.
2.) When using restart, in general all those data which are used from the restart file will replace information which may have been specified in the input file. But the amount of data which is read from the restart file varies depending on the restart option (see here). In case of the initial concentrations, the original information is always lost. However, these values are additionally used as starting values in case of TQ convergence problems. Therefore, reasonable values should always be given.
3.) Changing diffusion data like from TQ to manual is possible with options "reset_time" and "structure_only", but not for normal restart without optional keyword. The reason is the timer data for updating diffusion coefficients from TQ which are contained in the restart file. The same holds for nucleation data (see error message you posted).

Reading initial structures and concentrations from ASCII files is possible if they stem from a prior simulation. If they are from experiments, consistency problems may arise due to measurement errors or missing data. Generally, due to information loss by only considering sharp geometries, restart is preferable to reading from ASCII files.

no_TQ simulations can be restarted from TQ-simulations and vice versa, if the option "structure_only" is used.

Bernd

R.Hess
Posts: 22
Joined: Fri Aug 12, 2016 2:14 pm
anti_bot: 333

Re: Initial Concentration

Post by R.Hess » Mon Dec 19, 2016 12:50 pm

Hello,
thank you for your reply.

I have a further question. When i run the simulation with the "structure_only" option, micress gives me the following errors:

Image



what can i do against this problem ? Do you need more information about the input files ?

Best Regards
Raphael

Bernd
Posts: 1504
Joined: Mon Jun 23, 2008 9:29 pm

Re: Initial Concentration

Post by Bernd » Wed Dec 21, 2016 3:43 pm

This definitively should not happen!

Can you please send me the input file(s) and the .rest file by PM or to my e-mail account (b.boettger@micress.de)? I would like to check what happens (probably, I will not be able to do that before January...).

Bernd
Posts: 1504
Joined: Mon Jun 23, 2008 9:29 pm

Re: Initial Concentration

Post by Bernd » Fri Dec 30, 2016 5:50 pm

Hi Raphael,

I tried to reproduce this error, but I could not. Which exactly was the MICRESS version which you have used? You find that in the header of the .log output.

Perhaps, the error is linked with the fact that you use the "restart structure_only" in a very unusual way: You specify grain number 40 as target for the .rest file, but you have only grains 1-8 in the initial structure. Thus, restart is doing nothing...
I am not sure whether I understand your intention: If you just want to use the content of the .rest file as initial structure you should set the number of grains in the grain input section to 0 (i.e. only grain number 0 is there as default "background"), and then specify this grain number 0 in the restart options for replacement (the reason why you need to specify this grain number and optionally shifts and rotation is that the option "structure_only" is more powerful and also allows replacing only part of the domain with the microstructure from the .rest file).

Wish you all the best for New Year

Bernd

R.Hess
Posts: 22
Joined: Fri Aug 12, 2016 2:14 pm
anti_bot: 333

Re: Initial Concentration

Post by R.Hess » Mon Jan 02, 2017 9:36 am

Hello and a happy new year,
i use the Micress version 6.200(windows).

I changed the things you told me, but i got the same error-message as before.

I tried to run the same process, but with a thermocalc coupled simulation. the error message i got here, is :
unable to read from file [...]
STOP in routine initRestart
Reading nucleation data
It doesn´t matter, if i simulate the process TQ_coupled or without a database. the most important thing is, that it works ;)
But i thought, that a simulation with a linearised PD would be preferable, because of the high heating rate.

I hope we can find a solution

Best regards
Raphael

Bernd
Posts: 1504
Joined: Mon Jun 23, 2008 9:29 pm

Re: Initial Concentration

Post by Bernd » Mon Jan 02, 2017 6:44 pm

Dear Raphael,

that is weird because for me it runs without this error message when using Version 6.200 under Windows and the restart file which you sent me...
Also, you should not get the error you showed in your latest post when using "restart structure_only", because with this option, no nucleation data are being read at all!

I propose you switch to the new MICRESS version 6.300. This hopefully will make things easier and potenially prevent us from "hunting old bugs"

Bernd

R.Hess
Posts: 22
Joined: Fri Aug 12, 2016 2:14 pm
anti_bot: 333

Re: Initial Concentration

Post by R.Hess » Thu Apr 27, 2017 1:39 pm

Dear Bernd,
since my simulation of austenitisation works (by the way: your approach was very helpful - thank you !) i want to simulate martensitic formation.
But i got another problem. Because of the complicated circumstances, my explanation could be confusing. so i will try to explain as simple as possible(I will call the austenite-formation-simulation "sim1" and martensitic "sim2").
Sim2 should (re)start from the austenitisation-model.

I think the most common way (with less problems) to restart a simulation is:
...
restart structure_only
1
0
FileName_rest
...
thus, the concentration and grain input data are overwritten by the restarted file. Right?

But, the problem now is:
if i do it that way, i can observe, that after initialisation, the average concentration of Carbon in sim2 is too high! it says, "avg.conc of Component 1: 1.8% ."that is not right. the average concentration of sim.1 is 0.4. This is the value i need for sim.2, too(Ms depends highly on carbon content).
Confusing for me are the following two things.

1.) the avg. conc of comp2 is read and assumed right. (0.9%). How is it possible, that on the one hand i got the right conc. and on the other hand, for another component, i got a wrong concentration?


2.) if i add a database the sim.2 , the avg. conc of com1 is read and assumed right, but runtime explodes because of updating diffusion data. to my knowledge, during martensitic formation occurs no diffusion of carbon (or every other alloying element). only the strucutural rearrangement of the unitcells cause the phase transformation. therefore, i wanted to prohibit diffusion of carbon (no_diff in every line). Furhtermore, i need to use my "own" phase diagram (linearized). Summarized, i don´t really need a database for this simulation.
The simulation with TQ_coupled runs and the result looks quite good - but simulation time is high and i allowed diffusion (what is not right).

I hope you can unterstand the problem and can tell my, what could be wrong.

Best regards

Raphael

R.Hess
Posts: 22
Joined: Fri Aug 12, 2016 2:14 pm
anti_bot: 333

Re: Initial Concentration

Post by R.Hess » Thu Apr 27, 2017 2:04 pm

the second "problem" seems to be solved. i allowed diffusion of comp2 and prohibited diffusion of comp.1
But, nevertheless, what is the reason behind it, that i need a database to receive correct concentration data?

Post Reply