I've had a lot of difficulties modelling an experimental setup with a fan and have posted questions to this group before. After developing quite a complex numerical model I am now getting results which are not too bad, when compared with the experimental values, but I have a couple of questions which I hope someone who has fan modelling experience may be able to help me with.

I am using the SST turbulence model as it is very important for my modelling to be able to see where separation occurs on the blades. I have been using Y+ values of 1 and 2 to calculate the first prism height of the boundary layer, which I understand is neceesary for SST, along with having at least 10 nodes within the boundary layer.

To see just where the automatic wall function is being employed I have made contours of yplus on the blades and I can't get these below 2 unless I use a y+ of 0.5, which means I'm predominantly using the auto wall function in the modelling. The residuals never go much below 10E-3, but I am monitoring the torque and this flattens off quite well, which I am taking to be a converged solution.

I have also experimented with different time steps (currently I'm using 1/rotational speed), but this makes little difference to the convergence. I am getting quite big residuals for momentum in certain parts of the domain, when I alter the domain size.

I am a bit lost as to how to proceed, since I seem to be dealing with quite a few variables.

My question is to anyone who has used SST for fan modelling and wants to get a good picture of the flow over the blades, what type of convergence do you get and what yplus are you using.

Thanks in advance for any suggestions.

Jenny

Hi,

What is the blade Reynolds number? Are using the turbulence transition model? Do the blades have a gross separation or are they near the design operating point and are largely attached flows?

Glenn Horrocks