← Back to team overview

yade-users team mailing list archive

Re: Wall stress and problem with small density


Kien Dang, Mr a e'crit :
> Hi,
> I am working file generator modelling a odometer test. It is very similar to Triaxial test (in fact, I just modify the triaxial test a littble bit). I got several probs:
> 1. Wall stresses: personally, I think they are supposed to be increased with iteration; However, they were decreasing. Let's say I want to apply 500 of sigma_iso to the wall top, and then, at the begining, the wall is around 10,000 and then it increases with the iterations and when the wall stress ~ 500, the unbalance force is also small. What I think is the wall stress should increase from 0 to 500??? (does it make sense?)
I don't understand precisely what you mean. What "should" be is : if the 
flag "wall_XXX_activated" is true, the stress on boundary XXX should 
tend to sigma_iso (so it will increase/decrease depending on the value 
of sigma_iso). If not, you just found a bug (or perhaps you are 
interpretating data the wrong way).

Remember to commit your OedometerTest to SVN (or let us commit it for 
you) when it will work. :-)
> 2. Density: when the density is small (like 1 or 2), the balls disapear (or explose), however, when I tried with other file generator, there is no prob with the small value of density. Did I miss any treatments for this?
It sounds like a problem with the timestep : lower density needs smaller 
timesteps. Dt should be updated automaticaly in the TriaxialTest but 
there are still few problems, often due to a difficulty setting the 
timestep at iteration 0. The consequence is you start the simulation 
with Dt higher than a critical value and the intergation scheme 
diverges. Try setting Dt manually (if you divide the density by x, 
divide Dt by sqrt(x)).
If other exemples ar ok with lower density, it means that Dt could have 
been higher with the high density. Note that the procedure to compute 
the timestep in the TriaxialTest is not the same as in other preprocessors.

> 3. Is there any damping for velocity existed? I remembered that some one posted his paper involving to some damping in YADE, is it velocity damping and can I have that code?
It is force damping. See CundallNonViscousForceDamping.cpp and 
CundallNonViscousMomentumDamping.cpp (or browse mailing list archives?).


Chareyre Bruno
Maitre de conference

Institut National Polytechnique de Grenoble
Laboratoire 3S (Soils Solids Structures) - bureau E145
BP 53 - 38041, Grenoble cedex 9 - France
Te'l : 33 4 56 52 86 21
Fax : 33 4 76 82 70 00

Yade-users mailing list