Simulation of solidification of Gamma prime alloy during SLM process

dendritic solidification, eutectics, peritectics,....
CharMIC
Posts: 213
Joined: Mon Aug 21, 2017 5:01 pm
anti_bot: 333

Re: Simulation of solidification of Gamma prime alloy during SLM process

Post by CharMIC » Mon Jul 29, 2019 1:49 pm

Hi Bernd,

In my simulations I mainly get error to wards the end of the simulation where small amount of Liquid (>0.02%) is present. So I did some single phase equilibrium calculations with thermo-calc. considering the liquid composition. What I see is depend up on the temperature, major constituents of γ′ phase change. Is this is expected. Is this the reason behind errors?

Bernd
Posts: 1221
Joined: Mon Jun 23, 2008 9:29 pm

Re: Simulation of solidification of Gamma prime alloy during SLM process

Post by Bernd » Mon Jul 29, 2019 4:55 pm

Hi Chamara,

I am not sure whether I fully understand your question, but I believe what you describe is very typical. Towards the end of a multi-phase solidification simulation, when the rest liquid amount is quite small (<~0.1-0.01%), solidification is not proceeding further like expected but more or less retarded. If you do a Thermo-Calc equilibrium calculation for the rest liquid, a large part of the liquid is predicted to be solid, with most of it consisting of fcc or gamma prime. Reasons for this behaviour are:

- not all required phases are present in every rest liquid inclusion like assumed by Thermo-Calc equilibrium. An extremely dense nucleation would be required to achieve that...
- not all rest liquid regions have the same composition. Thus, there may be a specific enrichment of an element which is not reflected by the average liquid composition but which would require additional phases to be solid.
- numerical issues can arise from the non-equilibrium situation in individual liquid regions. In the same way, extreme composition values can arise due to numerical issues...
- in 2D simulation rest liquid solidification is even more difficult due to the lower specific surface (compared to 3D).

Depending on the given situation, I prefer the following strategies to cope with the problem:

- If I am interested only in solidification and the resulting microstructure (phase fractions, concentration distributions, etc.), the simulation can be stopped once the rest liquid fraction is below a relevant value (e.g. 0.1%).

- If the solidified microstructure is to be used as input for further simulations (e.g. heat treatment), we must get rid of the rest liquid. A convenient method is to use the "add_to_grain" feature which allows us to switch the rest liquid to fcc or gamma prime phase by nucleation (see here).

- If the final stage of solidification is really important (e.g. for getting the correct solidus temperature), it is absolutely necessary to have sufficiently frequent and realistic nucleation checks for all potentially relevant phases. The compositions of individual separated rest liquid regions should be checked with Thermo-Calc in order to be sure that all necessary phases are really provided by nucleation. Certainly, 2D simulation would lead to different results, so that 3D simulation should be preferred. Grain boundary wetting can be important, which can be achieved by selection of suitable interfacial energies and selection of the "multi-obstacle" flag and sufficient fine spatial resolution. I personally believe that interface adsorption may also be important, what cannot be accounted for with MICRESS by now. All in all, the correct prediction of solidus temperatures with MICRESS is still a challenge...

Bernd

CharMIC
Posts: 213
Joined: Mon Aug 21, 2017 5:01 pm
anti_bot: 333

Re: Simulation of solidification of Gamma prime alloy during SLM process

Post by CharMIC » Sat Dec 21, 2019 4:54 am

Dear Bernd,

Now I am doing the same simulation using TTNI8/MOBNI1. This is because TCNI8 did not give good equilibrium description of the alloy.

in TTNI8 i have
1. FCC_A1 (gamma),
2. FCC_A1#2(MC)
3. GAMMA_PRIME.

during the simulation I get error=3 and error=1611. they come during relinearisation. There after I get other errors also. I have defined the following also based on our discussion.

2 1-9
limits 2 9
30
100
diagonal
no_more_stoichio

Could you help me to find what is the root cause for the errors? I have attached .scr file. I have edited it to see mainly the error messages.

BR
Chamara
Attachments
Support_MICRESS.scr
(107.4 KiB) Downloaded 117 times

Bernd
Posts: 1221
Joined: Mon Jun 23, 2008 9:29 pm

Re: Simulation of solidification of Gamma prime alloy during SLM process

Post by Bernd » Sun Dec 22, 2019 7:52 pm

Hi Chamara,

From your screen output is seems obvious that the problems come from the MC carbides in your simulation. There is the "error=30201" which is relatively frequent and which means that the quasi-equilibrium results often end with identical compositions for fcc (phase 1) and MC (phase 2). It is difficult to say whether this is the initial problem or just a later consequence. What you should do is to check the phase composition outputs (.c*pha*) for early deviations of the compositions of the MC phase (or possibly also the fcc phase). You can also try to trace the places using the location information ("zp=****") in the screen output and DP_MICRESS.
If you e.g. were able to identify such early deviations, then you could use "limits" or similar option for this element...
In any case, it is important to find out where the problems exactly come from.

I recently added some comments to the specific problems with MC carbides here.

Bernd

CharMIC
Posts: 213
Joined: Mon Aug 21, 2017 5:01 pm
anti_bot: 333

Re: Simulation of solidification of Gamma prime alloy during SLM process

Post by CharMIC » Mon Jan 06, 2020 9:09 am

Dear Bernd,

Thanks for the reply. what cause to get the Error=3 and Error =1. Is this specific for problems in .GES file?

Bernd
Posts: 1221
Joined: Mon Jun 23, 2008 9:29 pm

Re: Simulation of solidification of Gamma prime alloy during SLM process

Post by Bernd » Mon Jan 06, 2020 10:52 am

Hi Chamara,

No, these are just two common exit errors when recognizing that the Newton-Raphson algorithm for obtaining quasi-equilibrium is not converging...

Bernd

CharMIC
Posts: 213
Joined: Mon Aug 21, 2017 5:01 pm
anti_bot: 333

Re: Simulation of solidification of Gamma prime alloy during SLM process

Post by CharMIC » Tue Jan 07, 2020 9:03 am

Hi Bernd,

When I look at the composition (in at%) in MC and FCC_A1, I Hf and C composition fluctuating a lot. especially I get negative compositions in FCC_A1. I tried first defining limits (min 0, max 50) for C composition in both MC and fcc phase. But I see still C composition goes to negative values in certain locations in fcc phase.

Bernd
Posts: 1221
Joined: Mon Jun 23, 2008 9:29 pm

Re: Simulation of solidification of Gamma prime alloy during SLM process

Post by Bernd » Tue Jan 07, 2020 12:22 pm

Dear Chamara,

It does not make any effect to limit carbon in MC between 0 and 50%, because these limits are already given by the sub-lattice structure of FCC_L12 phase. What you may want to prevent is that MC switches to FCC when x(C) gets too low, say below 35%. Thus, you should try to limit C between 35 and 50%.

Bernd

CharMIC
Posts: 213
Joined: Mon Aug 21, 2017 5:01 pm
anti_bot: 333

Re: Simulation of solidification of Gamma prime alloy during SLM process

Post by CharMIC » Tue Jan 07, 2020 4:19 pm

I tried limiting C in MC between 30 and 50 at%. Still I see the composition fluctuate above 50%. I get error =10209 indicating that limit criteria has been violated (if I am correct). also I get error = 30201. Do I need to limit some other elements also?

I see negative C nd Hf compositions in some locations during some time steps in Liquid as well.

Bernd
Posts: 1221
Joined: Mon Jun 23, 2008 9:29 pm

Re: Simulation of solidification of Gamma prime alloy during SLM process

Post by Bernd » Tue Jan 07, 2020 6:55 pm

Hi Chamara,

Yes, error 10209 means that in phase 2 component 9 has crossed a limit during iteration of quasi-equilibrium. And 30201 means that during this process both phases 1 and 2 got identical composition. If component 9 is C in your case, it could be the same thing. However, FCC could also switch to MC...

When you say you see composition fluctuations, do yo mean the phase compositions of MC, or the mixture composition? Are there also fluctuations of other carbide forming elements like e.g. Ti or Ta?

Bernd

Post Reply