Sorry for bothering you with somewhat simple things… but I could not be very sure from documentation whether a system of extended granular objects can be simulated with LAMMPS or not? Documentation says granular potentials work with spherical objects. Does it mean it does not work with non-spherical ones?
If by "extended" you mean an ellipsoid, then Axel is correct.
You would have to write a pair_style that works with pairs
of ellipsoids by detecting their overlap and contact point, etc.
That's a hard problem.
But you can combine groups of spherical particles into
rigid bodies via fix rigid and use the granular pair styles
with those extended objects.