Error when linearisation!

dendritic solidification, eutectics, peritectics,....
Bernd
Posts: 1179
Joined: Mon Jun 23, 2008 9:29 pm

Re: Error when linearisation!

Post by Bernd » Fri Dec 07, 2018 12:10 pm

Dear Chamara,

If you list the phases in Thermo-Calc while creating the .ges5 file, you will not see the composition sets because they are regarded as the same phase. However, once you open the .ges5 file in MICRESS you will see them. I am not sure how you can see them directly in Thermo-Calc, because there composition sets are handled dynamically.
If the expected composition set is really not in the .ges5 file, you use option 3) of my explanation you cited above to create a new one.
Thus, the trick is to first create a .ges5 without modification/addition of the composition sets. Then check in MICRESS, and modify the script for creating the ges-file either by adding ("composition_sets") or modifying ("mayor_constituents"), and rebuild it.

Bernd

CharMIC
Posts: 211
Joined: Mon Aug 21, 2017 5:01 pm
anti_bot: 333

Re: Error when linearisation!

Post by CharMIC » Wed Jan 16, 2019 9:26 am

Hi Bernd,

I get the following error when I do a heat treatment simulation. I have made a .ges file has you have suggested and treated phase 2 (Laves) as stoichiometric.

What I could not understand is that I only starting to get this error after 11600 s in to the HT simulation. I do the HT at 1033K. So there is negligible dissolution kinetics for Laves. Apart from Laves, I have delta and gamma'' in the simulation microstructure.

BR
Chamara
Attachments
Error.PNG
Error.PNG (44.14 KiB) Viewed 2543 times

Bernd
Posts: 1179
Joined: Mon Jun 23, 2008 9:29 pm

Re: Error when linearisation!

Post by Bernd » Wed Jan 16, 2019 12:10 pm

Hi Chamara,

This type of error is not very specific. What you can see from the messages is that errors probably occur during complete relinearisation, and that they occur only in 2 locations (zp=8155 and zp=7650).

I would advice you to check out these two places with DP_MICRESS in order to see what could be wrong there. Be aware that if you use global relinearisation schemes, these two points are just representing one place in the whole interface or fragment for which the relinearisation is calculated.

If there is no obvious problem, I would check the phase compositions (*.c*pha*) output which per default is in atom percent (up to Version 6.4). Typical problems are negative concentrations, or local compositions which indicate a switch to a "wrong" composition set (e.g. LAVES).

Another place to look at is the .TabLin output. In case of global linearisation, you can directly see the thermodynamic linearisation data which are calculated in the 2 locations if you figure out the corresponding grain numbers first (.korn). The linearisation data can also point to the source of the problem (extreme slopes, switch of composition set, etc.).

It is important to find out what goes wrong before you can know how to prevent it!

If you can't find the origin, and the problem is only local and does not hurt, you can also just ignore it - in complex systems with a big simulation domain (many phases and precipitates) only two such (non-catastrophic) errors per complete relinearisation would be still fine...

Bernd

CharMIC
Posts: 211
Joined: Mon Aug 21, 2017 5:01 pm
anti_bot: 333

Re: Error when linearisation!

Post by CharMIC » Wed May 22, 2019 11:22 am

Hi Bernd,

I am getting similar error in a simulation where I simulate Alloy 247

During the solidification simulation I allow to nucleate MC (FCC_L12#3) phase starting from 1570K. As soon as the first nucleation was set, I get the following error message.

Thermo-Calc error 1611 MICRESS error 24 phases 2/ 2.

After this error I get "trying hard phase.." errors.

here phase 2 is FCC_L12#3. when I created he .GES file I adjusted the major constitutes as your suggestion in this discussion thread.
Amend_phases.PNG
Amend_phases.PNG (118.91 KiB) Viewed 2428 times

I get correct initial quasi equilibrium for the FCC_L12#3. high HF,TA and C.
Initial quasi equlibrium.PNG
Initial quasi equlibrium.PNG (25.77 KiB) Viewed 2428 times
During the nucleation, the initial linearisation parameter seems to be OK. eventhough I am not sure if Ti is high.
Initial linearisation parameters.PNG
Initial linearisation parameters.PNG (84.62 KiB) Viewed 2428 times
I used stoichio condition for the FCC_L12#3 for the all the elements.

So what is the reason here. What am I doing wrong.

BR
Chamara

Bernd
Posts: 1179
Joined: Mon Jun 23, 2008 9:29 pm

Re: Error when linearisation!

Post by Bernd » Wed May 22, 2019 3:20 pm

Hi Chamara,

It is hard to say what is the reason for the problems in your case without having more information.

The first picture you show is how you created the .ges5 file. This looks quite well, assuming that the two composition sets FCC_L12#2-3 already existed before. I am somewhat astonished that Al has been offered as default major composition in both cases. It could indicate some unusual behaviour and a possible mistake, but I do not really know about the Thermo-Calc internals...
Anyway, in the second picture, the default major compositions (it is not the initial quasi-equilibrium!) show that you obviously got what you expected.
And the initial equilibrium in the third picture also looks quite good. Typically, at that high temperature, you get lower Ti contents, which is ok.

Thus, after successful initial equilibrium, you get error 24, which I know occurs while trying to calculate the molar volume of FCC_L12#3. This only makes sense if we assume that the composition of the initial seed (other than the initial equilibrium) is wrong. What I would start looking for is

- Which the initial composition of phase 2? You can check either .TabC or .c*pha2, you may need to request more frequent outputs.
- Do you nucleate at an interface or in the bulk, i.e. do you have other initial equilibria like FCC_L12#2/FCC_L12#3? Perhaps this one failed...
- Which kind of error happens after error 24, which error number occurs? How are the new linearisation parameters looking like (.TabLin)? This can give us hints what is going wrong.
- It could be that you just got a stupidly wrong molar volume, which causes irrational driving forces, and thus the subsequent errors. You could try to input the molar volume of phase 2 manually and then see whether the problem vanishes.
- Re-check the values for interface mobility, diffusion coefficients and interface energy in order to see whether there is something unrealistic there..
- the observed behaviour would be logic if you would try to set a big seed of phase 2 (r>Δx) instead of a small one!
- redo the .ges5 file just to be sure there was no error...

Hopefully, one of these checks gives more hints for the problem!

Bernd

CharMIC
Posts: 211
Joined: Mon Aug 21, 2017 5:01 pm
anti_bot: 333

Re: Error when linearisation!

Post by CharMIC » Wed May 22, 2019 5:08 pm

Thanks Bernd.

found the problem for error. it was big seed of phase 2 (r>Δx) instead of a small one!

However now I am run into a different problem. I now get this error.
Error.PNG
Error.PNG (43.68 KiB) Viewed 2423 times
when I check .TabC out put, once the FCC_L12#3 nucleation (at interface) happen, minimum composition value of some elements in the LIQUID phase and FCC_L12 phase become negative. in FCC_L12#3 all the minimum compositions are positive.

Bernd
Posts: 1179
Joined: Mon Jun 23, 2008 9:29 pm

Re: Error when linearisation!

Post by Bernd » Wed May 22, 2019 6:06 pm

Hi Chamara,

Did you set the initial radius for the seeds of phase 2 to zero or to something bigger (0<r<Δx)? To set it bigger is allowed in principle, but when nucleating a stoichiometric phase this will initially lead to negative phase compositions in the liquid if the fraction value φ of phase 2 (which corresponds to the 3D radius r) is so high that there is not enough carbon, hafnium or tantalum inside the grid cell to form the carbide. However, this should only be an initial problem which is corrected by diffusion afterwards...

Bernd

CharMIC
Posts: 211
Joined: Mon Aug 21, 2017 5:01 pm
anti_bot: 333

Re: Error when linearisation!

Post by CharMIC » Wed May 29, 2019 9:00 am

Hi Bernd,

How can we locate error position "zp" in DP_MICRESS?

ralph
Posts: 113
Joined: Wed Apr 27, 2011 4:42 pm
anti_bot: 333

Re: Error when linearisation!

Post by ralph » Wed May 29, 2019 10:58 am

As far as I know, 'zp' is a cell id.
Virtual EDX -> Pick points/lines with coodinates -> cell

Best,
Ralph

Bernd
Posts: 1179
Joined: Mon Jun 23, 2008 9:29 pm

Re: Error when linearisation!

Post by Bernd » Wed May 29, 2019 12:52 pm

You can see the coordinates and zp in DP_MICRESS already by moving your mouse pointer over the result image...

Bernd

Post Reply