yade-users team mailing list archive

[chiara modenese <c.modenese@xxxxxxxxx>]

On 25 June 2010 19:26, Bruno Chareyre <bruno.chareyre@xxxxxxxxxxx> wrote:

> Hi,
>
> On the way back to Prague (yes, Vaclav lives in the countryside!), I
> simplified a little our last idea :
>
> bool incremental;
>
> void precompute (...,state1,state2,...)
> {
>   if (incremental) {us = incrementalUs; axialRotation=...}
>   else us = ... (totalUs);
> }
>
> The rotate function can be called with no penalty from the Law2s, as soon
> as the axial rotation (a vector3r) is precomputed with state1 and state2,
> and cached in the geometry class. For the other rotation perpendicular to
> contact plane, it only needs previous and current normals, no need to
> access  states. The rotate function would then use normals and precomputed
> axial rotation. It doesn't have to be virtual I think.
>
> This solution doesn't handle the relative velocity, which would not be
> fully optimized, probably not a big deal. Another solution would be to
> precompute relative vel. instead of us increment if incremental. It would be
> more general but it would need to update code in the existing Law2's.
>
Hi Bruno,
but what about simply using rotateAndGetShear, which is already available?
Splitting this function, as it is partially done, it is useful but in cases
in which some operations into the contact law are necessary (like damping).
You said that that function could be used to rotate any vector (in this case
the input would be a total shear displacement vector instead of shear force)
and it will return the increment shear displacement. No need to update code.
It would be only two lines code more (one line to call the function and save
the returned increment, the other to add the obained increment to the total
shear displ vector).


>
> Bruno
>
>
> On 25 June 2010 13:41, chiara modenese <c.modenese@xxxxxxxxx> wrote:
>
>> 2010/6/25 Václav Šmilauer <eudoxos@xxxxxxxx>
>>
>> Hi Chia, nice synchronicity, we 've been just discussing this about 5
>>> minutes ago with Bruno and Jan. Greetings to Oxford :-)
>>>
>> Hello guys! Hope you are having a great time all together :-) Could not we
>> organize a course/meeting among Yade users/dev? I have already mentioned to
>> Vaclav that this would be a nice idea. I hope it will happen at some point
>> in the future.
>>
>>>
>>>
>>> For the total shear displacement, we don't currently have that value
>>> stored anywhere. What you could do is to have elastic+plastic and
>>> elastic displacements and track both of them using ScGeom::rotate; the
>>> latter would be changed during plasticity activation, while the former
>>> one would not.
>>>
>> Thanks for suggestions, I will try it. Actually the elastic part, once
>> calling ScGeom::updateShear, is already stored in ScGeom shear variable.
>> Perhaps the easiest thing would be to create a new variable into the contact
>> law which stores what is returned by updateShear, and, as you suggest,
>> simply avoiding to pass it through the failure criterion.
>>
>> cheers, Chiara
>>
>>>
>>> _______________________________________________
>>> 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
>>>
>>
>>
>> _______________________________________________
>> 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
>>
>>
>
> _______________________________________________
> 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
>
>