New feature: Global linearisation

new developments, future functionalities, hints for further applications of MICRESS
Post Reply
Bernd
Posts: 1504
Joined: Mon Jun 23, 2008 9:29 pm

New feature: Global linearisation

Post by Bernd » Thu Feb 26, 2009 9:28 pm

Hi all,

from version 5.404, the use of global linearisation parameters with thermodynamic coupling is possible. This means one can use one set of linearisation parameters per interface (of grain pair k1/k2) instead of one set of linearisation parameter per interface cell. The feature can be activated per phase pair by the optional keyword "global/local" after the "database" keyword for each phase interaction in the phase diagram input data, that means that also mixed mode use is possible.
The main advantage of the global linearisation is an enhanced performance and stability. It should not be used if grains are large and cover a wide range of compositions or/and temperatures, because the averaging effect of a global linearisation could give less exact results in those cases (e.g. in the case of long dendrites during simulation of columnar growth). There are still not enough experiences to finally recommend the use in general, please tell us about your experiences!

Bernd

nokkikku
Posts: 25
Joined: Wed Nov 25, 2009 1:07 pm
anti_bot: 333

Re: New feature: Global linearisation

Post by nokkikku » Thu Aug 18, 2011 12:57 pm

Hi Bernd,

could you explain also about the option database_global/database_local in the 'concentration data'?

Thanks
nokkikku

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

Re: New feature: Global linearisation

Post by Bernd » Thu Aug 18, 2011 2:27 pm

Hi nokkikku,

this is not related to the new feature 'global linearisation' although the keywords are neary identical (just the underscore is different). The keywords 'database_global' and 'database_local' are used in the input of the diffusion data to retrive only the diagonal terms of the diffusion matrix. The difference between 'database_global' and 'database_local' is just whether the diffusion coefficients are taken according to the average composition (_global) or interpolated according to the local composition.
Details have been provided here by Philippe who programmed that.

If necessary I can give you further explanation!

Bernd

nokkikku
Posts: 25
Joined: Wed Nov 25, 2009 1:07 pm
anti_bot: 333

Re: New feature: Global linearisation

Post by nokkikku » Tue Aug 23, 2011 2:11 pm

good to know that. Thank you for now. I am sure I have to ask you more in the near future.

nokkikku
Posts: 25
Joined: Wed Nov 25, 2009 1:07 pm
anti_bot: 333

Re: New feature: Global linearisation

Post by nokkikku » Tue Aug 23, 2011 2:17 pm

Maybe near future can be now also.

If I activate the option 'phase_interaction' and 'redistribution_control' for the phase interaction 1/2 but I select 'normal' from all the Options: nple para paraTQ normal ATC, in the section 'Input of the phase diagram of phase 1 and phase 2',


Will it be EXACTLY the same as activate only 'phase_interaction' for the phase interaction 1/2 (which of course will deactivate all the options nple para paraTQ normal ATC)?

nokkikku

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

Re: New feature: Global linearisation

Post by Bernd » Tue Aug 23, 2011 5:16 pm

Hi nokkikku,

It will be exactly the same! And, I am sure, it should also numerically be EXACTLY the same...

Bernd

Post Reply