Yade

Validation of Cundall Strack contact law?

Asked by Hasan

Hello all,

I am trying to validate some contact laws that used in Yade before doing anything else . I started with a very simple example of sphere dropping and calculate the displacement at different speed , after that I began to compare the analytic solution with Yade results. But there was some deviation in the analytic solution.

I don't know if I am using different contact law or will be there some error percentage between analytic and numerical solution.

- Could some one please direct me to the papers that contain the exact contact law for Cundall Strack (ElasticContactLaw) that used in Yade , and alos for Hertz Mindlin law if that's possible.

- How much will probably be the percentage of error for a very simple test like this one?

Thanks,

Hasan

Question information

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

Hello Hasan,

> I am trying to validate some contact laws that used in Yade before doing
anything else . I started with a very simple example of sphere dropping
 and calculate the displacement at different speed , after that I begun
to compare the analytic solution with Yade results. But there was some
deviation in the analytic solution.

Would it be possible to provide us with the yade script and also
script/formulas you use as the analytical solution? Maybe we could find the
problem directly.

The Cundall Strack law in normal direction is the simplest possible, force
= stiffness * displacement. So your problem might be caused by wrong values
(perhaps stiffness?) In analytical solution or too large time step in yade
solution.

> - How much will probably be the percentage of error for a very simple
test like this one?

Depends on your time step, theoretically could be any value, practically in
stable regime should be "very small".

cheers
Jan

Revision history for this message
Bruno Chareyre (bruno-chareyre) said : 		#2

> Could some one please direct me to the papers that contain the exact
contact law for Cundall Strack (ElasticContactLaw) that used in Yade

The best reference is given in the class documentation: [Pfc3dManual30]
There are subtle differences but they are only in the shear component,
and unpublished yet.

> How much will probably be the percentage of error for a very simple test like this one?

For small enough timesteps, the error should be quickly less than 10e-6.

Bruno

Revision history for this message
Hasan (hassanmsahli) said : 		#3

Thanks cheers and Bruno, that really gave me very good ideas to what to look at.

Revision history for this message
Luis Barbosa (luis-pires-b) said : 		#4

Hi Hasan,
I'm working with this validation, did you do validation to Law2_ScGeom6D_CohFrictPhys_CohesionMoment ?
Thanks

Revision history for this message
Bruno Chareyre (bruno-chareyre) said : 		#5

There is a validation in [1], but it is only for the elastic part of the
law.
The other things (e.g. rolling friction) have been (and are being) used
many times by serious users. They would have notice very bad bugs, but I
would not call this a true validation.

Bruno

[1] http://www.sciencedirect.com/science/article/pii/S0925857413001936

Revision history for this message
Bruno Chareyre (bruno-chareyre) said : 		#6

The above message (#5) is for Law2_ScGeom6D_CohFrictPhys_CohesionMoment.
For ElasticContactLaw I can't cite a published paper, but if you find a bug in it you will have my bravo.

Revision history for this message
Luis Barbosa (luis-pires-b) said : 		#7

Hi Bruno,
My question about validation is just to learn how the model works and not say if is correct or not.
But thanks for the paper.