Yade

Effect of particle size on noise in the results

Asked by SO TA

Hi all,

I am new to DEM and I have very little experience with YADE and DEM. My question is more of a DEM nature rather than YADE. I hope someone would be able to provide some advise.

I am simulating MBD-DEM ( movement of a bucket into a stockpile) and I noticed that when I use large particle sizes (300-400mm) the results are noisy. I was advised to repeat the simulations by slightly shifting the bucket to right and left to evaluate the error and the error was huge (around 40% error compared to the simulations with initial location of the bucket).

I am not sure what could be the source of such error. If it is an issue with DEM then how someone possibly could simulate the large particles using DEM.

I appreciate any advise or resources in advance.

thanks
sota86

Question information

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

Hello,

> I am new to DEM and I have very little experience with YADE and DEM

welcome :-)

please read [1] and provide (much) more information, ideally also a MWE if possible.

> I am simulating ... a bucket ... a stockpile ...

What is "a bucket" and "a stockpile"?
What shape do they have?
What are they composed from (spheres, facets, ...)?
What is their mutual sizes and mutual sizes of the components?
Are the components regular or not?
... ?
... ?

> when I use large particle sizes

What is "large", large with respect to what?
300-400mm alone is useless information...

> the results are noisy

What is "the result"?
Positions, velocities, energies?
Numerical result or visual state?
... ?

> slightly shifting the bucket

what is "slightly"? with respect to what?

> the error was huge

error of what?

Cheers
Jan

[1] https://www.yade-dem.org/wiki/Howtoask

Revision history for this message
SO TA (sota86) said (last edit ): 		#2

Hi Jan,

Sorry for my not well-defined questions! I'll response to each section under your comments.

> I am new to DEM and I have very little experience with YADE and DEM

welcome :-)
Thanks

please read [1] and provide (much) more information, ideally also a MWE if possible.

> I am simulating ... a bucket ... a stockpile ...
please note that the simulation is not done using Yade. It is in Algoryx. As I mentioned in my first question, my question is of more DEM nature rather than a YADE one. I added image1 to provide you with an idea of what are bucket and stockpile.
  https://www.dropbox.com/sh/f0acr2dmqvfnsl0/AAC5yAqt0tFgrTdSycplH4bea?dl=0

What is "a bucket" and "a stockpile"? explained in the image (above link)
What shape do they have? explained in the image (above link)
What are they composed from (spheres, facets, ...)? the particles are spheres with uniform PSD. bucket is a rigid solid body.
What is their mutual sizes and mutual sizes of the components? the size of the bucket and stockpile is explained in the image (above link). four particle samples of 100mm, 200mm, 300mm and 400mm are simulated.
Are the components regular or not? not sure what does this mean?
... ?
... ?

> when I use large particle sizes

What is "large", large with respect to what?
300-400mm alone is useless information...
large with respect to the size and capacity of the excavator

> the results are noisy

What is "the result"?
Positions, velocities, energies?
Numerical result or visual state?
... ?
Numerical results including cycle time (the time takes the bucket to go through the granular material) and payload (the amount of the particles in the bucket at the end of the cycle)
> slightly shifting the bucket

what is "slightly"? with respect to what?
I moved the bucket in relation to the centreline of the stockpile (Image 2) and re-run the simulations and compare the results (cycle time and payload) with those from previous simulations. for the smaller particle sizes (100, 200, 300 mm) the difference was less than 3% while for the 400mm particles the difference between the results was more than 40%. I have been advised that using large particles in DEM results in the noise in the output and I do not quite understand why. If that is the case then how should we simulate phenomenon with large particle sizes?
> the error was huge

error of what?

Cheers
Jan

[1] https://www.yade-dem.org/wiki/Howtoask

Thanks
sota86

Revision history for this message
Best Jan Stránský (honzik) said : 		#3

> Are the components regular or not? not sure what does this mean?

how are the particles generated? As regural packing, e.g. hexagonal, or randomly? (if randomly, then how randomly?

> I have been advised that using large particles in DEM results in the noise in the output and I do not quite understand why.

Take an extreme. Particles 2 m size (roughly half the bucket width). Then (assuming random particle positions) sometimes one particle ends in bucket, sometimes two, in some cases no particle (I guess).

"Small" particles behave "like continuum", uniform matter, the differences between individual runs (shifting bucket) should not be significant.
"Large" particles, specifically their mutual positions, can influence the results (or its scatter) much more.

> If that is the case then how should we simulate phenomenon with large particle sizes?

not relying on one simulation, but make same "statistics" from several simulations.

Cheers
Jan

Revision history for this message
SO TA (sota86) said : 		#4

Thanks Jan Stránský, that solved my question.