Potential Flow
 

Hi all, I am working on an academic problem of potential flow through venturimeter type geometry. For defining potential flow i need to specify viscosity=0 and no slip boundary condition.

I created a modified fluid with all properties of water except viscosity=10^-10 (to specify viscosity close to zero) and i used free slip boundary condition at walls and k-omega turbulence model.

Upto 30 iterations the solver was working fine with a convergence rate close to 0.98. But after that the convergence rate is close to 1(most of the time above 1) and now circulating flow is developed (indicated by the "wall placed" in solver ) and after 77 iterations 42% area is covered with walls to prevent inflow.

If the fluid would have not been potential flow, a recirculation zone is expected in the venturimeter type (contracting-expanding) geometry. But, potential flow should not show any recirculation.

Why i am getting recirculation, even though i have defined the fluid close to potential flow? And why the recirculation started showing after 30 iterations.

Please let me know.

Regards, Swapnil

Re: Potential Flow
 

Hi Swapnil,

If you are trying to model potential flow you should not use a turbulence model, that is run a laminar flow simulation with almost zero (eg 10^-10) viscosity. Also use slip wall boundaries.

Glenn

Re: Potential Flow
 

Hi Glenn,

since the viscosity is very low (10^-10), the Reynolds no is of the order of 10^13 and if i use the laminar flow model it gave me error that Re number is out of range. That's why i used turbulence model. Then the solver didn't give the Re no. error.

Re: Potential Flow
 

Ignore the error message. It is just a warning that considering your Re number you are in the turbulent region. If you are not going to solve the boundary layer then do use laminar.

Re: Potential Flow
 

Bart, what would happen if you just turned the turbulence equations off using the expert paramerters ? Bob

Re: Potential Flow
 

Don't know actually. Propably the same as laminair. But it's easy to test. Just try it both ways and let me know the difference. You know, you can also post this on the CFX forum at the CFX site. It's just started and works great (for me)