Relations between time step, velocity and volume fraction in DES-Mixture model sims
 

Hi,

I am currently running a simulation of a dilute gas-liquid mixture inside a pipe with some internals. I am using the DES turbulence model (+RNG model) together with the mixture multiphase model (obtaining the interphase slip velocity through an algebraic equation, rather than solving transport equations for each phase).
To speed up the flow development in the early stages of the simulation, I adopted larger time steps and everything worked as expected. Just when I reduced the time step to increase accuracy I noticed that lowering the time step (I'm using implicit second order discretization) the dispersed phase velocity increases to values well above the expected magnitude. With an inlet velocity of around 1m/s I get characteristic velocities close to 7 m/s in the pipe. This also affects the gas-phase volume fraction that decreases (I guess to get a gas-flux matching the one imposed at the boundary conditions).

I've never noticed this behavior neither with single-phase DES nor with RANS Eulerian two-fluid model. Has someone experienced something like this before? Do you have any suggestions?
Thanks,

H.

P.s.
In single phase the DES simulation with the current mesh gives mean and fluctuating velocities that almost perfectly match the experimental observations. RANS TFM simulations with the same mesh gave realistic results, which were validated with experiments.
H.