Hi,

I've been working on a problem with a fan for a while and can only come up with a numerical solution within 40% of the experimental results. I've done grid independence testing, looked at the sensitivity of the residuals, changed the turbulence models and tried different boundary conditions, but still can't get a better agreement than 40% which just isn't good enough.

I've been speaking with one of my lecturers today and he's suggested that the problem may be in the airflow entry point into the fan. Currently I have the set up of the fan in an open circular duct (pipe) using a mass flow on the outlet and an opening in the shape of a half circle on the front (see http://www.dezignit.com/domain.jpg). I have set the boundary condition as an opening with reference pressure 1 atm, and relative pressure on the surface as 0Pa. I'm using opening pressure and direction and flow normal to boundary for the opening. The suggestion is that due to the shape of the inlet not all the air flow will flow with constant velocity through the fan. The air entering from the side of the duct will turn in with a higher velocity, so there may be a false representation of the actual mass flow entering into the fan compared with the real situation of the fan in the experiment.

I would like to try putting a variance of pressure across the opening and wondered if anyone else has tried this or has any other suggestions. I definitely have a discrepancy between the mass flow in the numerical experiment and in the actual experiment. The numerical model is showing the fan to be incredibly efficient, but this can't be a real situation even allowing for losses which would be present in the 'real world' situation.

Any suggestions anyone can give would be great on how to put a variance of pressure, or whether this would even be the correct approach. Thanks in advance.

Jenny

Hi Jenny,

If you have set a specific mass flow outlet condition, and the only way for air to enter is through the opening boundary condition, then as long as your solution is converged and imbalances are small, the mass flows should be identical. Are you saying that some air circumvents the fan through a tip clearance, and this is smalller/larger than it should be?

It also wasn't clear to me from the picture how you are modelling your fan - is it with a rotating domain (with frozen rotor interfaces), or with momentum sources?

Could you elaborate further?

regards,

Johnson

Hi Johnson,

Thank you for your reply. I'll try and elaborate a little better on the physical set up of the system. I've added some text to the diagram at http://www.dezignit.com/domain.jpg to clarify the parts of the geometry.

Basically I have used the frozen rotor approach with interfaces between the spherical "inlet" and the small rotating cylinder containing the fan, and between the rotating cylinder and the stationary outlet duct.

The problem I'm having is that when I use the mass flow measurements from the actual experiment they are giving me a pressure reading much lower than the experimental value. They even become negative as the volume flow rate increases through the duct, so since I am using gauge pressure somehow there seems to be too much mass flowing through the system than it is capable of handling if I use the experimental value. We were trying to work out how this would happen.

In my model you can see that the inlet has a flat plane of zero pressure across the front of the rotating cylinder containing the fan. What we suspect might be happening is due to the fact some air must essentially turn to get into the duct it needs to speed up to get in, so in reality the pressure is not constant across this plane as we are forcing it to be. So basically it looks to the system like a denser amount of air is coming in at these areas. If this makes sense?

It was just a way we thought might explain the increased mass flow. I wanted to test the theory by putting a variation of pressure across this plane, so it isn't just a relative pressure of 0Pa.

I hope that makes sense. Thanks in advance for any light you can throw on this.

Best wishes, Jenny

Hello Jenny.

I feel the way you have considered open atmosphere is creating problem. Try following things..

1) Instead a hemisphere have a complete sphere (minus pipe) to consider atmosphere AND

2) Increase size of this sphere (This will take care of pressure variation as you reach near to fan)

3) Define opening boundary on this whole sphere

I hope this should help.

Best luck,

Chirag