See - http://cfdem.dcs-computing.com/?q=node/920 . I didn't know if my comment would be more pertinent for a user forum or not.
I don't know why, but when using the "hybrid granular molecular" atom type, it remarks that "All masses are not set" even though the density and diameter of each particle are defined (read in from file using read_data). If I specify the mass of the atom type, it works fine, but this is not an ideal solution since the masses of the particles given in the input file (via diameter and density) are not uniform. This should be a pretty simple calculation, and it works when the atom style is "granular alone"