[deprecated list] Peak results

[Anton Gladky]

Has anybody the problem with "peak results"?
I mean, when you record the data from anything, for example force affecting
on facet or sphere,
sometimes you get "pick results" which are twice or more larger than normal
result.
It happens just for a short period, like a noise.

Thanks in advance
______________________________
[ENG] Best Regards
[GER] Mit freundlichen GrÌßen
[RUS] С МаОлучшОЌО пПжелаМОяЌО
[UKR] З МайкращОЌО пПбажаММяЌО

Anton Gladkyy

[Václav Šmilauer]

Post by Anton Gladky
Has anybody the problem with "peak results"?
I mean, when you record the data from anything, for example force
affecting on facet or sphere,
sometimes you get "pick results" which are twice or more larger than
normal result.
It happens just for a short period, like a noise.

Eh, probably everybody has had that problem.

Taking force on 1 particle makes no sense, you have to average it
somehow. DEM doesn't solve for equilibrium, therefore you can either let
the simulation stabilize somehow (dissipate all the kinetic energy), or
average somehow what you see over larger area (or both). You can compute
forces on a specimen cross-section or some meaningfully defined set of
particles; if you have a facet that touches many spheres, that could be
a good candidate as well, as long as the sphere move independently.

You can try taking force from 1 single particle and smooth it using
moving average or similar techniques (they are available in the numpy
package), but that is just "graph aesthetics".

e.g. examples/concrete/uniaxial sums forces (in UniaxialStrainer) on all
particles that are considered supports and lower and higher supports are
then averaged. There are still some osciallations (as there are waves
going back and forth in the specimen), but that is not so marked unless
you go to very high loading rates. The same example also plots
cross-section forces (i.e. forces in interactions that cross that plane)
and you can see that they differ.

Vaclav



_______________________________________________
Mailing list: https://launchpad.net/~yade-users
Post to : yade-users-oU9gvf+***@public.gmane.org
Unsubscribe : https://launchpad.net/~yade-users
More help : https://help.launchpad.net/ListHelp

[Anton Gladky]

Ok, thank you, will try to "struggle to it"

http://yade.wikia.com/wiki/FAQ#I_have_a_problem_with_.22peak_results.22._What_is_wrong.3F


______________________________
[ENG] Best Regards
[GER] Mit freundlichen GrÌßen
[RUS] С МаОлучшОЌО пПжелаМОяЌО
[UKR] З МайкращОЌО пПбажаММяЌО

Anton Gladkyy

Post by Václav Šmilauer

Post by Anton Gladky
Has anybody the problem with "peak results"?
I mean, when you record the data from anything, for example force
affecting on facet or sphere,
sometimes you get "pick results" which are twice or more larger than
normal result.
It happens just for a short period, like a noise.

Eh, probably everybody has had that problem.
Taking force on 1 particle makes no sense, you have to average it
somehow. DEM doesn't solve for equilibrium, therefore you can either let
the simulation stabilize somehow (dissipate all the kinetic energy), or
average somehow what you see over larger area (or both). You can compute
forces on a specimen cross-section or some meaningfully defined set of
particles; if you have a facet that touches many spheres, that could be
a good candidate as well, as long as the sphere move independently.
You can try taking force from 1 single particle and smooth it using
moving average or similar techniques (they are available in the numpy
package), but that is just "graph aesthetics".
e.g. examples/concrete/uniaxial sums forces (in UniaxialStrainer) on all
particles that are considered supports and lower and higher supports are
then averaged. There are still some osciallations (as there are waves
going back and forth in the specimen), but that is not so marked unless
you go to very high loading rates. The same example also plots
cross-section forces (i.e. forces in interactions that cross that plane)
and you can see that they differ.
Vaclav
_______________________________________________
Mailing list: https://launchpad.net/~yade-users<https://launchpad.net/%7Eyade-users>
Unsubscribe : https://launchpad.net/~yade-users<https://launchpad.net/%7Eyade-users>
More help : https://help.launchpad.net/ListHelp