Acceleration of aging simulation

dendritic solidification, eutectics, peritectics,....
Post Reply
G_azizi
Posts: 18
Joined: Wed Feb 17, 2021 7:59 pm
anti_bot: 333

Acceleration of aging simulation

Post by G_azizi » Mon Apr 18, 2022 4:44 pm

Hello,
I need to simulate aging of Pb-Sn solder alloy which is aged for 50 days.
The simulation time is very long even for 10s (almost 5 days). I tried to
accelerate simulations by increasing diffusion coefficient and interface
mobility but the outcome microstructure is awful because of instabilities.
Is there any technique/method to accelerate simulations? specially in solid
state?

Thanks

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

Re: Acceleration of aging simulation

Post by Bernd » Tue Apr 19, 2022 1:00 am

Hi,

First, it is necessary to find out, why the MICRESS simulation is so slow, and where all the time is spent. For that you should have a look at the tabular .TabP output which can give you this information. In which part of the code is most of the time spent?

Bernd

G_azizi
Posts: 18
Joined: Wed Feb 17, 2021 7:59 pm
anti_bot: 333

Re: Acceleration of aging simulation

Post by G_azizi » Mon May 02, 2022 7:35 pm

Thanks for your reply,

I checked and more than 60% of the time is spent for TQ time. based on this is there any way to accelerate simulation time?

Thanks

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

Re: Acceleration of aging simulation

Post by Bernd » Mon May 02, 2022 10:21 pm

TQ-time in most cases mainly is caused by the number of TQ-calls. This means that increasing the update interval or using more global update schemes should (e.g. "global" instead of "local") reduces TQ-time. Typically, properly selecting these parameters is a trade-off between exactness and performance.
If there are several phase interactions, it is possible to distinguish which phase interaction is causing most performance loss by referring to the .TabTQ output.

Sometimes high TQ-time values can also be caused by numerical problems (increasing the number of iterations per TQ-call), which can typically be recognized by large numbers of error messages in the screen output. In this case the reason of the instability needs to be found and removed in order to solve the problem.

Bernd

Post Reply