← Back to team overview

yade-users team mailing list archive

Re: Attributes of CohesiveFrictionalContactInteraction (bug ?)

 

Jerome Duriez a écrit :
> Hello everybody. I just noticed that the class 
> "CohesiveFrictionalContactInteraction" (in pkg/dem/Engine/EngineUnit/) 
> inherits from the "_ElasticContact_Interaction", whereas in the 
> registerAttributes (of CohesiveFrictionalContactInteraction) what is 
> called first is :
> _SimpleElastic_Interaction::registerAttributes();
>
> It seems to me that it is quite not logic...
> In fact some of the attributes which belong to 
> CohesiveFrictionalContactInteraction through ElasticContactInteraction 
> (but which are not in SimpleElasticInteraction) are however registered 
> (like tangensofFrictionAngle for exemple), but the work is not perfectly 
> done : for example "ks" is not registered.
>   
Thank you for the remark. I added "ks" in registered parameters. It 
should be enough to fix the problem, shouldn't it?
There is no real reason to consider this a bug. Registering 
ElasticContactIntercations would need extra disk space for variables 
like "equilibriumDistance", which are in fact useless in the current 
cohesive-frictional law.

> Fortunately it seems that this doesn't affect the computations because 
> in the CohesiveFrictionalRelationships we have (if the interaction is 
> "new" or not) :
>
> contactPhysics->ks = contactPhysics->initialKs;
>   
"Fortunately" is the good word... as there is really no reason to set

contactPhysics->ks = contactPhysics->initialKs

when contacts are not new.


> In fact this case occured also for the ElasticContactInteraction, and 
> was corrected in revision 1315 (14/04/08). But because such commits (in 
> which the list of registered attributes are modified) prevent users to 
> reload saves, once update is done, 
>   
We can now load simulations even when some parameters are not defined in 
the xml. The object will have default constructor values. At least this 
what I'm doing....
> (I take the opportunity to say that 
> this is a BIG disadvantage of Yade)
It would be... but it no problem really, you can load your good old 
xml's (without ks), even after I added ks as registerable parameters.
>  I didn't want to take the 
> responsability of commiting the corrections myself.
>
>   
In such case, perhaps you could talk about a possible commit on the 
mailing lists, then wait a day or two to see how people react, and then 
commit. One day you will have to anyway. :)

Bruno


-- 
 
_______________
Chareyre Bruno
Maitre de conference

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
________________

_______________________________________________
Yade-users mailing list
Yade-users@xxxxxxxxxxxxxxxx
https://lists.berlios.de/mailman/listinfo/yade-users



Follow ups

References