yade-users team mailing list archive

[luc scholtes <lscholtes63@xxxxxxxxx>]

Unfortunately, the correction does not affect my results. Indeed, as I said,
the 3 strain rates were equal, and this is what the equations below should
provide if I understand well. Maybe I wasn't clear enough. I attached here
the evolutions of the 3 strains during the same test. You can see the delay
between the 3 strains to reach the desired stress level. I told you about
"relaxation" effects because of the transition between compression and
dilation when the axial stress (see E2) level is reached. I tried by
decreasing the maxVel and by decresing the time step, but I still have the
same delay...

In fact, I don't really know if my problem (say, less deformation in the
axial direction than in the radial ones to reach the same stress level) is
really a problem... I would expect to have identical deformations (and I
insist on deformation and not displacement!) in each directions to produce
the same increase in stress, assuming that the sample is isotropic and
homogeneous, but it seems that I was wrong...

Many thanks for your help

  Luc

ps: I found another tiny error in TriaxialStressController. It was in the
computation of the wall thickness (see rev 2068). It affected the case where
the thickness parameter was not specified.



2010/3/9 Luc Sibille <luc.sibille@xxxxxxxxxxxxxx>

> Hi,
>
>  Your are right Bruno, I made an error by introducing max_vel1, 2 and 3,
> sorry. Your correction should solve the problem.
>
>  Luc
>
> Bruno Chareyre a écrit :
>
>  Hi Luc Scholtès (and Luc Sibille, read that too)
>>
>> I think the problem had been introduced with these lines in TTController
>>  :
>> -               max_vel1=3 * width /(depth+width+depth)*max_vel;
>> -               max_vel2=3 * height /(depth+width+depth)*max_vel;
>> -               max_vel3 =3 * depth /(depth+width+depth)*max_vel;
>>
>> Now (r2066) replaced by :
>>
>> +               max_vel1=3 * width /(height+width+depth)*max_vel;
>> +               max_vel2=3 * height /(height+width+depth)*max_vel;
>> +               max_vel3 =3 * depth /(height+width+depth)*max_vel;
>>
>> I didn't check the result yet. Tell me if that's better.
>>
>> Bruno
>>
>>
>>
>>
>>
>> luc scholtes a écrit :
>>
>>> Hi all,
>>>
>>> (Bruno, you probably have an idea about this)
>>>
>>> A remark for those who are interested in doing triaxial loadings on
>>> samples with height/width ratio different from 1. Here is the test I did:
>>>
>>> 1 - take 2 samples, one with a ratio equal to 1 (a cube typically) and
>>> another with a ratio equal to 2.
>>> 2 - give them exactly the same micro properties (it appears to me that it
>>> is more interesting with a high E(Yade Young Modulus)/confiningPressure
>>> ratio)
>>> 3 - from an initial dense state (say that you have already created a
>>> dense sample with the compaction procedure of the triaxial state), apply
>>> them an hydrostatic loading in order to reach a higher confining pressure,
>>> with a sufficiently small wall velocity to avoid inertial effects (setting
>>> the maxVel parameters small enough)
>>> 4 - plot the evolution of the 3 principal stresses as a function of 1 one
>>> the 3 principal strains (the corresponding strain rates are equal according
>>> to the servo-controlled controlller) for both of the 2 samples.
>>>
>>> -> see the attached figure for the results
>>>
>>> As you can see, in the case of the cube, the 3 principal stresses evolve
>>> in an identical way with the deformation, which is a good news, but,
>>> unfortunately, for the sample with a ratio heigh/width = 2, we can see that
>>> it is not the case at all. The stresses reach the desired confining
>>> pressure, but there is a kind of a delay between the axial stress (S2) and
>>> the radial ones (S1,S3) that occurs during the loading: the axial stress
>>> increase more rapidly than the 2 others... This leads to less deformation in
>>> the axial direction than in the radial ones to reach the desired confining
>>> pressure.
>>>
>>> I tried with several velocities and with very low damping values
>>>  (wallDamping=0, damping(NewtonIntegrator)=0.05), and I always get the same
>>> result. The sample is supposed to be isotropic and homogeneous in regards to
>>> contact forces and orientations distribution.
>>>
>>> One of my experienced colleague though about relaxation effects that
>>> should occur in the sample, creating this delay between the axial and the
>>> radial responses of the system.
>>>
>>> If you have any advice....
>>>
>>> Cheers
>>>
>>>  Luc
>>>
>>>
>>>
>>>
>>> ------------------------------------------------------------------------
>>>
>>> ------------------------------------------------------------------------
>>>
>>> _______________________________________________
>>> Mailing list: https://launchpad.net/~yade-users<https://launchpad.net/%7Eyade-users>
>>> Post to     : yade-users@xxxxxxxxxxxxxxxxxxx
>>> Unsubscribe : https://launchpad.net/~yade-users<https://launchpad.net/%7Eyade-users>
>>> More help   : https://help.launchpad.net/ListHelp
>>>
>>>
>>
>>
>>
> --
> Luc Sibille
>
> Université de Nantes - Laboratoire GeM UMR CNRS
>
> IUT de Saint Nazaire
> 58, rue Michel-Ange - BP 420
> 44606 Saint-Nazaire Cedex, France
>
> Tel: +33 (0)2 40 17 81 78
>
>
> _______________________________________________
> Mailing list: https://launchpad.net/~yade-users<https://launchpad.net/%7Eyade-users>
> Post to     : yade-users@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~yade-users<https://launchpad.net/%7Eyade-users>
> More help   : https://help.launchpad.net/ListHelp
>

Attachment: Hydrostatique_EpsVsIterations.eps
Description: PostScript document