I don't think that this problem has a simple "mixing" answer. Without more details about impeller type and location, I can only generalize about the problem and possible solutions. The mixer size and general description sounds appropriate for PVC. However, PVC reactions are typically carried out with overhead condensation of monomer for heat removal. The return and rapid mixing of the condensed monomer is critical for good polymer bead formation. The high porosity and odd shaped particles sound as if the monomer is remaining on the liquid surface and not being quickly incorporated back into the batch. The usual method for testing mixer performance is with a water batch and either adjusting the liquid level for the baffle effect to create a strong, but not too deep vortex on the surface. Vortex characteristics are difficult to predict and must be tested in the specific reactor. This problem is less about the reactor design and more about the operation of the reactor. Formulation can only do so much to overcome mixing problems.