Please keep the mail list in the email replies.
So instead of “spheroid” you really mean ellipsoid.
If you want to the neighbor cutoff to be different
in different directions, you should figure
out if that model makes sense for Peridynamic,
I.e. what happens if the sample rotates.
Maybe Rezwan or Mike Parks (Ccd) can comment
on your model and what you want to do.
Possibly it would be modeled by using a different
lattice spacing in different dimensions for your initial
particles. With a spherical
cutoff, that might give the equivalent effect you
are asking for.
If not, the neighbor list in Peridyanmics is only
used once to create the initial bond list. So it
wouldn’t be hard to add an option to the PD commands
to use an ellipsoidal distance metric for determining
neighbors.
Steve
Wan,
I agree with Steve. In addition, you need to check if the ellipsoidal neighborhood region affects the PD formulation. I am guessing you want to introduce anisotropy. If so, you can do that by choosing effective influence function too. Off course, the deviatoric component of the force state needs to be redefined based on your anisotropic influence function. See Proposition:15.1 in Stewart Silling’s paper “Peridynamic States and Constitutive Modeling”. The typical isotropic PD formulation depends on spherical influence function. For anisotropic case the expression in proposition: 15.1 will be much complex.
I hope this helps.
–Rezwan
Please keep the mail list in the email replies.
So instead of “spheroid” you really mean ellipsoid.
If you want to the neighbor cutoff to be different
in different directions, you should figure
out if that model makes sense for Peridynamic,
I.e. what happens if the sample rotates.
Maybe Rezwan or Mike Parks (Ccd) can comment
on your model and what you want to do.
Possibly it would be modeled by using a different
lattice spacing in different dimensions for your initial
particles. With a spherical
cutoff, that might give the equivalent effect you
are asking for.
If not, the neighbor list in Peridyanmics is only
used once to create the initial bond list. So it
wouldn’t be hard to add an option to the PD commands
to use an ellipsoidal distance metric for determining
neighbors.
Steve
The existing models within the peridynamics pair styles in LAMMPS are all isotropic. If you are trying to introduce a simple anisotropic model, you can follow this suggestions in this manuscript (http://digitalcommons.unl.edu/cgi/viewcontent.cgi?article=1083&context=mechengfacpub) where the authors alter the bond strength based upon the direction of the bond in the reference configuration. It would be possible to modify the PMB pair style to do this without a lot of work. You still use spherical neighborhoods in this approach, so no need to change that.
Best,