Yade

optimum number of balls

Asked by Yuri Bezmenov

Hello community,
I am simulating Triaxial Test. Currently, I'm calibrating my contact properties. Can you suggest me some method to optimize the number of balls, otherwise it takes me very long to run a single simulation. Currently, I am using around 40000-50000 balls with PSD.

Question information

Language:
English Edit question
Status:
Solved
For:
Yade Edit question
Assignee:
No assignee Edit question
Solved by:
Yuri Bezmenov
Solved:
Last query:
Last reply:
Revision history for this message
Jan Stránský (honzik) said : 		#1

Hello,

this is really very subjective, optimum value may vary with respect to what optimum you define.
How realistic it should be (=the more balls the better)?
How fast it should be computed (=the less balls the better)?
Even these two most basic optimum criteria are mutually contradicting and it depends on how you define weights for them..

Cheers
Jan

Revision history for this message
Yuri Bezmenov (yuribezmenov) said : 		#2

Hii Jan,
>Even these two most basic optimum criteria are mutually contradicting and it depends on how you define weights for them.
 Exactly, that's the problem. I want it to represent the most possible real scenario but fell short of computational resources.

I'm trying to compare the results from 2d and 3d tests, it that works I'll update here.

Thanks for your comment.

Revision history for this message
Yuri Bezmenov (yuribezmenov) said : 		#3

hi folks,
I thought a lot but couldn't find any way to optimize the number of balls.
I am going with, checking the most critical parameter of my problem with the number of balls as we do in FEM. that's the only way I could figure out.

Thanks Jan again.