Courant number divergence

Hi everybody, I'm running a simulation of particles inside fluid and I'm having problems with the Curant numbre (C-n) value.
When I create the setup for the CFD simulation I adjusted the OpenFOAM timestep to keep the C-n less than 1, I understand that the Courant number It's afected by the mesh size, the value o U and finally the timestep. I checked the CFD simulation and everything it's ok.

The problem starts when I want to add the particles and run the coupling, after a couple of simulation steps the program gives me an error and suddenty stops with C-n values way over 1.
My question is: If the C-n only depends on the parameter that I listed above why diverge when I'm doing the coupling?

I've tried everything, reduce the CFD timestep, reduce the coupling interval, reduce the number of particles, etc... but I have the same problem over and over again.

PD: I succeed doing the coupling in laminar regime, I have this problem when I'm modelling turbulent flow (RES with k-epsilon model, pisoFoam solver).

If someone have an idea of where could be the problem it would be great!

Thanks,
Mauro