[sammm]

Hi everyone,

I am currently working on a CFD study of an axial fan installed inside a straight pipe about 20 cm long (the fan diameter is roughly 10 cm).
The main goal is to compare the suction performance of different fans. The fan rotation is modeled using an MRF approach. I am using k-omega SST model.

I initially set up the case by embedding the pipe and the fan inside an external ambient domain. One important detail is that the inlet is not the pipe inlet surface directly in front of the fan, but a small inlet “spot” placed upstream of the system. This spot is the only region where the velocity is different from zero; the rest of the external domain represents still air.

After some initial tests, I realized that the external domain was too small, because the air exiting the pipe was being recirculated back toward the inlet region. I then significantly enlarged the ambient domain, which reduced this effect, but I am still unable to achieve full convergence. The residuals decrease initially and then reach a plateau where they stabilize without further reduction.

My question is the following:
what type of boundary conditions would you recommend for the outer boundaries of the ambient domain, in order to allow flow to be generated only when required by the internal suction of the fan?

I have tried several boundary condition setups found in online tutorials and YouTube videos, but I am not confident that they are physically correct for this problem.

Any suggestions or best practices for this kind of “open ambient” fan simulation would be greatly appreciated.

P.S.
I am also trying to study the system without the external ambient, with inlet and outlet sticked to the tube.

Thanks in advance.

[tomf]

Hi Sammm,

My suggestion would be to combine a totalPressure boundary condition for the pressure with pressureInletOutletVelocity for velocity. inletOutlet for k and omega, calculated for nut.

Both k and omega should have low values (~0.01) to have something non-zero.

Best of luck,
Tom

[sammm]

Originally Posted by tomf

Hi Sammm,

My suggestion would be to combine a totalPressure boundary condition for the pressure with pressureInletOutletVelocity for velocity. inletOutlet for k and omega, calculated for nut.

Both k and omega should have low values (~0.01) to have something non-zero.

Best of luck,
Tom

Hi Tom, it works! I have to fix something about the mesh but the residuals are very good now, thank you!