Issues with bigParticle and processor boundaries

Hello everyone,

while modelling a periodic vertical pipe flow with the cfdemSolverPiso I am running into issues with the processor boundaries of my decomposed model. The CFD and the DEM components of my set up are running individually without any problems.

The problems occur if I run the coupled simulation. The indicators of the problems are the following:
- the voidfraction field shows irregularities => the particle presence is cut off by processor boundaries (see figure)
- after crossing boundaries some particles reverse their moving direction. Others just go on as expected

I assume that this happens because of an error of the void fraction model. Currently I am using bigParticle even though I read that IB simulation would be better here. However, by reason of the limited computing power I have currently access on I am bound to unresolved...
I found this paper: http://www.cfd.com.au/cfd_conf12/PDFs/068HAG.pdf that implies for me that a cross boundary algorithm has been developed for the IB model. Is this implemented for the bigParticle algo too? If yes, why does it not work for my case?

My second question: Am I right that the reversing particle motion is a result of the missing field information? Is it possible that an error occurs if the gradPforce is calculated? In my understanding at one side of the particle no field of pressure is known for this processor that in turn causes a force towards the boundary. Or is this point of view too much 'discrete'?

I'm looking forward to any answers.
Thanks for any help!

Matthias