yade-dev team mailing list archive
-
yade-dev team
-
Mailing list archive
-
Message #02587
Re: PeriTriaxController
Václav Šmilauer a écrit :
I am little lost in the H notation (generic linear transformation,
right?), but I think the point is very clear. Is it "better" to change
positions or velocity?
I'll reply to that in details after discussion here in Grenoble.
if we have imposed stress along x axis : d^2 X/dt^2 = sigma/m;
To me, it is just not clear how you relate stress to bulk mass (m), but
I will think about it. It is true that it would be nice to do without
stiffness, since computing that sucks (and PeriTriaxController does it
in order to eliminate the loop over all interactions every time; so it
just takes last stiffness value and updates the stress estimate based on
sctrain increment. clearly it is overestimated (I didn't go in detail),
and it reminds me of convergence implications of secant/tangent
stiffness matrix in FEM... see attachment :-) ).
Note that underestimating stiffness, on the other side, is not really a
problem. For instance, if estimate = 0.3 * current, the error decreases
like 0.7^n, which is good enough to reach the goal in a few iterations.
So, if you have a doubt on stiffness, you can just divide it by 10 and
see the result.
But yes, an inertia would be simpler. Gael is just using the inertia of
one grain, which is not exactly "realistic", but the point here is to
not to be realistic, just to keep a constant stress in the packing.
Bruno
v
------------------------------------------------------------------------
_______________________________________________
Mailing list: https://launchpad.net/~yade-dev
Post to : yade-dev@xxxxxxxxxxxxxxxxxxx
Unsubscribe : https://launchpad.net/~yade-dev
More help : https://help.launchpad.net/ListHelp
--
_______________
Chareyre Bruno
Maître de Conférences
Grenoble INP
Laboratoire 3SR - bureau E145
BP 53 - 38041, Grenoble cedex 9 - France
Tél : 33 4 56 52 86 21
Fax : 33 4 76 82 70 43
________________
References
