escaping particles during compression

Asked by Wojciech Sobieski on 2019-12-05

Hi Everyone,

I prepared an script to make granular beds with the particle distribution from a cumulative curve ( One case works correctly (the data from file p06_w1_15.txt), but if I change the distribution function (file synlab_01.txt), some particles escapes from the space (they pass through the walls) during simulation. In the Yade interface all seems to be correct (see the screenshot: yade-interface.png), however if I try to see data saved in VTK files, I see only few particles there. I tried to change some settings (e.g. goals values or damping coefficient), but nothing helps. Where can be the problem?

In general, what is the strategy to obtain beds with the porosity to small as possible?

Files are here:

Regards from Poland,


Question information

English Edit question
Yade Edit question
No assignee Edit question
Last query:
Last reply:
Robert Caulk (rcaulk) said : #1


External links are discouraged for various reasons including: our goal of maintaining a persistent searchable knowledge base, embedded code encourages true minimal working examples, and of course the increased convenience for the volunteers providing assistance. Thanks for considering the simple requests found in our forum guidelines [1].

That said, based on your description it sounds like you may need to increase the stiffness of your walls (or decrease your timestep, but I assume you are using an auto-timestepper).




Launchpad Janitor (janitor) said : #2

This question was expired because it remained in the 'Needs information' state without activity for the last 15 days.

sorry for external links, but I wanted to add a screenshot.

I resolved the problem by decreasing the Young modulus and the density of the spheres material. However, I do not understand one thing. In DEM a lot of discussions is related to model contacts and physical phenomena. But, if I try to create a virtual bed representing a real sample with relatively small porosity (but the same like in the measurements - in others words I know, that such a bed is possible in real world), than it is very difficult with the use of real material properties. And if I decrease the Young modulus and/or the material density, than my model becomes unrealistic. In this context: why I should care about proper contact models at all? With wrong material data, these models do not will work correctly. I am wrong? Is there another way to obtain beds with small porosity, but with the use of realistic properties of used materials? I found in the TriaxialStressController a parameter called thickness (I thought that maybe ma walls are too thin). However, if I try to increase its value (now I have 0), than I have an error during simulations. Maybe I should do something with time step. Could somebody give mi some tips?

Regards from Poland,


Launchpad Janitor (janitor) said : #4

This question was expired because it remained in the 'Open' state without activity for the last 15 days.

Jérôme Duriez (jduriez) said : #5

Just a minor comment regarding TriaxialStressCcontroller.thickness: this one is just meant to measure correct dimensions for the packing which is enclosed within surrounding boxes/walls, for the purpose of stress and strain measurements.

(the engine first starts with differences of position -- along each axis -- of the center of mass of these surrounding bodies, but it then needs to subtract the thickness to have the correct "inside" length)

Obviously TriaxialStressCcontroller.thickness has to match the actual thickness of that bodies in your simulation, but this is usually achieved automatically.

All in all, should not be a big source of major problems anyway.

Robert Caulk (rcaulk) said : #6

User indicates the question is solved.

Robert Caulk (rcaulk) said : #7

> . And if I decrease the Young modulus and/or the material density, than my model becomes unrealistic.

This is why I suggested increasing the stiffness of the walls instead.