← Back to team overview

yade-users team mailing list archive

Re: Cell mass

 



    since as you say it is a mass density, that would not work. I
    mean, I have tried different cell dimensions (I guess that is what
    you mean with size of the period) and have employed different size
distributions. In all the cases, the same problem was shown.

I meant, find a value suitable for you, and then change cell dimensions, everything else equal. Do instabilities appear in this case?


    This mass value influences very much the stability, and one has
    only to guess about it (I mean, it is a random number, depending
    also on the micromechanical parameters chosen). Besides this, it
    does not have a proper physical meaning for the purpose of the
    simulation. That is why I do not understand its role.

It has a physical meaning and a clear role : it defines an inertia linking time derivative of strain rate (strain acceleration) and a stress. I recognize however there is a dimensional mismatch at the moment, the reason is unclear to me (there should be a squared length somewhere). It is integrated in an explicit scheme and coupled with elastic response of the sample, so there is no surprise if it is unstable in some cases, independently of dimensional consistency.

    I think we could write the algorithm in a simpler way, if you have
    any ideas how to do it differently do let me know.

I have no idea. Another way is non-dynamic control of stress in the period. If it is dynamic, it needs an inertia. Let you suggest better definition of inertia if you have some, numerical experiments could give indications (I never got instabilities in dynamic cells, so I don't really know what triggers them).

Bruno

Follow ups

References