On 25 March 2010 11:57, Anton Gladky <[email protected]> wrote: > Hi, Chiara! > > Recently I have got some problems with damping in shear directions, but I > don't know the reason. > > When my RockPM had only normal forces, it worked normally. When I added > shear forces, bodies started to "hang" in the air. If I reduce dampingCoef, > it become better, but too low dampingCoef brings other problems. > > I use only NewtonIntegrator for damping purposes, so I have to check it > there. >
Hi Anton, the local damping (alias the one defined used in NewtonIntegrator law) is working properly in the shear direction as well as in the normal. This time I have not properly checked with a closed solution (it is a little bit more complicated than with viscous damping) but numerically I get the expected behavior (so I see an energy dissipation and a tendency to an equilibrium position). I do not know about your specific problem, maybe you can try to use the viscous damp and see what happens (but wait till Sergei will have updated his class). cheers, Chiara ______________________________ > > Anton Gladkyy > > > 2010/3/25 chiara modenese <[email protected]> > >> Hi Sergei, >> >> I think that the global damping (the one at the contact level) as it is >> now implemented in Yade (class ViscoelastiPM) is wrong in the shear >> direction. >> >> At the moment we do the following (I only refer to the shear direction): >> >> First we rotate Fs_tot(old); >> Then: >> deltaFelastic=ks*deltaUs; >> Fvisc=cs*deltaVrel_n; >> Fs_tot(new)=deltaFelastic+Fvisc+Fs_tot(old); >> >> Then we check Mohr-Coulomb on Fs_tot(new); >> >> The wrong thing (I suppose) is that we store Fs_tot including the viscous >> component and then we go for the next step. Instead we should only store the >> elastic part and then add the viscous part if we pass the Mohr-Coulomb >> criterion (Bruno was right in pointing this out). Otherwise the final effect >> is that we are not dissipating energy but only changing the amplitude and >> the frequency of the oscillation. I did a comparison between the analytical >> solution, Yade code and what I coded for the shear direction (I took a >> simple example to do that). I attach the comparison. >> If you think in the normal direction we do exactly the same. We work out >> the normal elastic force as: >> >> Fn_tot_elastic=kn*Un_tot; >> Fvisc=cn*deltaVrel_n; >> Fn_tot=Fn_tot_elastic-Fvisc; (minus or plus depending on how we work out >> the relative velocity) >> >> Next step we get a new Fn_tot_elastic that does not include the hystory of >> the viscous force, and then we simply add the incremental current viscous >> force. >> >> This is a total formulation but we could use the incremental one also for >> the normal part (as in Bruno's notes). So you see that in the normal >> direction there is no history of the viscous force. And this is correct, in >> fact Un_tot (as well as Us_tot) includes the damping effect since it is a >> result of the motion. >> >> I wrote a new class that adjusts the implementation of the damping in the >> shear direction as explained above. Should I commit it? Or would you prefer >> to modify your existing one (ViscoelasticPM)? If you agree with me, of >> course. >> Any comments would be appreciated. >> >> Cheers, Chiara >> >> >> _______________________________________________ >> Mailing list: >> https://launchpad.net/~yade-users<https://launchpad.net/%7Eyade-users> >> Post to : [email protected] >> 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 Post to : [email protected] Unsubscribe : https://launchpad.net/~yade-users More help : https://help.launchpad.net/ListHelp
