yade-dev team mailing list archive

[Bruno Chareyre <bruno.chareyre@xxxxxxxxxxx>]

> (As said, I never used that code; what I was driving at was to get
> comparison of critical timestep given by PWaveTimeStep and
> GlobalStiffnessTimeStepper.)
>
>
>   
This is another interesting discussion I wish we could have when you 
were here...
Actually, the relation between the two is relatively weak.
In short, they scale the same way, but the factor between them depends 
on (always the same set) porosity, contact anisotropy, PSD,...
8 contacts per grain or 4 contacts per grain really means different 
critical timesteps for the explicit scheme. It is accounted for 
explicitely in GlobalStiffness since it is a sum on all contacts (you 
will find a factor sqrt(2) betwween the dt's), in PWave it is just ignored.

The good point of global stiffness is it is dynamic (changing with 
packing state), so it can really be as close as possible as the exact 
current critical value. Even if GS gives dt only 10% higher than PWave 
(it is in fact more based on what you put as default factor in 
dt=0.5*PWave), it is worth using, since GS doesn't take more than 10% in 
cpu time. The weak point is GS gives nothing when there is no contact, 
thats why I use both in scripts now. PW gives an initial value, and GS 
adjusts it dynamicaly once contacts are created.

Bruno


>   


--
_______________
Bruno Chareyre
Associate Professor
Grenoble INP
Lab. 3SR
BP 53 - 38041, Grenoble cedex 9 - France
Tél : 33 4 56 52 86 21
Fax : 33 4 76 82 70 43
________________