[viking]

Dear CFX-Users,

i have a problem simulating a single fluid case with variable density. i used an expression in CEL for modelling the density. the density decreases for low pressure values and should represent the increasing bubble friction (desorption of non condensable gases) in areas of pressure below the atmospheric pressure.

in my simulation model there is couette and poseuille flow through a gap (rotor-stator). the boundary conditions are total pressure at the inlet and opening entrainment at the outlet.

i allready have done some other simulations with these boundaries and the total energy model but different fluid propperties as variable densities on multiphase flow. they did all work. for the single fluid simulation the solver writes an error "divide by zero" and even an initialisation from a constant density simulation does not solve the problem for total energy model. i also tried some oter timesteps.

furthermore the obove described CEL density expression workes fine with an thermal energy model or isotherm as well.

can anyone tell me, why i get the divide by zero error?
the error does not occure whithin my CEL expressions, in this case the solver would tell me the problematic CEL.

kind regards
yours viking

[ghorrocks]

When you put weird kinks and bends in the density curve that always makes convergence difficult. Your problem is simply difficult numerics (assuming there is no gross error in your equations).

One hint is to make sure that the density is continuous against pressure (I assume you have density as a function of pressure), and preferably keep the first derivative continuous as well.

[ghorrocks]

And also consider using the cavitation model for this as well. The physics are quite similar so you may be able to adapt the cavitation model to this. That model has stuff built in to handle the convergence difficulties.