cavitation
 

Hi, Does anybody use the cavitation model from Star-CD? Because my work goes on a hydraulics valve with 2 pressure boudaries (one for the inlet with a pressure about 500 bar, the second for the outlet with a pressure of 1 bar), but as I do a steady calculation: first I don't have the convergence; second I have crazy values ( velocities about 2000 m/s!!! and negative absolute pressure!) So I tried to turn on the cavitation model and followed the manual's instructions but the calculations don't seem to start! Could anybody help me? Thanks

Maxime

Re: cavitation
 

The last time I used the Cavitation model you had to run Transiently (not Steady State). You can use the steady state solution as a starting point for the time dependent solution. However, you need a decent steady solution first.

The negative absolute pressure is a good indicator that you will see some cavitation.

The 2000 m/s is not a good thing. Check your cell quality (warpage, internal angles, etc..)

Re: cavitation
 

Negative densities and high velocities are quite usual in high pressure flows in hydraulic systems. You can check you values by a simple estimation using the Bernoulli-quation. Unusual high velocities can also occure due to coarse or inadequate grid structures near sharp edges.

The cavitation model can only be used in transient mode. Be sure to use the latest version; earlier version had some bugs. Sometimes restarts from single phase solution are not converging because StarCD does not make a real restart, but only uses some initial values. It could be necessary to start with a lower pressure difference and then use a ramp to reach the high pressure inlet condition. The default value for the cavitation model to set in is 80 time steps which is usually too low. If the model does not set in, you have probably messed up your load step definitions. You also should use the cavprop user-function and check the values for vapour pressure and velocity of sound. Subcyling has to be turned on, a maximal value of 10 cycling steps shoud be enough. The BTF model is quite unstable so you have to live with the barotropic model. Surface tension values have to be set in rcon 187. The model sets its own time step ! The actual time step can be calculated by using the postdat user-function.

Good luck.

Michael