I am using Eulerian multiphase model coupled with DDPM to simulate solids in the Knelson concentrator and I have managed to run the simulation on my laptop (at least 300 iterations) and the residuals indicate that the simulation is stable over this duration, however, when I try to submit this simulation setup on the HPC, I get a divegence error just after 1sec.

And in the output it says "Divergence detected in AMG solver: pressure correction: Divergence detected in AMG solver: kDivergence detected in AMG solver: epsilonDivergence detected in AMG solver: vof-1Divergence detected in AMG solver".

could it be normal that a simulation that is running for more than 300 iterations and is stable over that duration simply fails on the HPC? or is there something else that I must be aware of ? Kindly assist, I have being troubleshooting this problem for more than a week now and I need 3-phase results soonest.

Some of the settings that I have tried out include: Decreasing time step size from 0.001 to 0.0001 and using an adaptive time-stepping method, the option not to track the particles with fluid flow time step, using a Rosin-Rammler diameter distribuiton for gold and gangue particles to reduce the number of DPM equation that must be solved, and also the settins in "advanced solution control" under multigrid i.e changing from flexible to W and F-cycle and increasing the termination restriction from 0.7 to 2,5 and 10 but I havent had any success!!

Additional infomation: the model I have is of a Knelson unit that separates heavy particles (gold) from light particles (gangue) in a centrifugal field and against fluidization water inlet from the pores in the concentrate rings. The heavy particles must remain in the concentrate bed formed in the concentrate capture rings. The Lighter particles are expelled from the bed and are carried in the upward-rotating liquid film into the tailings outlet.

Therefore, am using DDPM has this has the ability to set the wall conditions to “trap” so that particles that reach there are trapped and also because the solids concentration is more than 10-12 % v/v recommended for DPM (25 % v/v in my case).

ANSYS Fluent manual says that the DDPM formulation takes into account the discrete phase volume fraction and can handle solid mass loading approaching the particle packing limit, which for spherical particles translates to a volume fraction of 0.74. Hence, since my model is handles solids around 25% , I thought this would be a right way to approach the discrete phase modelling in my model.

Thank you in advance.