Hi, I would like to model a propeller through a momentum source model. I am thinking about adding some momentum into the model in a cylinder shaped subdomain with the same radius as the impeller. I would like to reproduce the swirling effect by adding momentum in the tangential direction, but also momentum in the axial and radial directions and turbulence quantities should be applied in the subdomain. Is it possible to make an expression in cel and how should I define it? My plan is to apply the cel expression to the subdoamin.

I should probably say that I have already made a detailed sliding mesh computation of the problem, but since I want to make some longer transient computations where the propeller is only part of a larger geometry, I would like to see if I could calculate the effect of the impeller by a less computationally expensive method. So, if anyone have experience with modelling propellers with momentum source models in CFX-5.7.1, I would be thankful for some advice.

Regards Jesper

In 5.7/5.7.1 you don't need CEL expressions to enter radial/tangential/axial momentum components. You can simply setup your momentum source as cylindrical components instead.

Neale

Hi Neal, thank you. I had actually tried that, but I did not think it was working as expected. I had however overlooked the coordinate axis definition. After I defined the right coordinate axis it works just fine.

Best regards Jesper

Hi Jesper,

You can do this through momentum sources, but if you're going to include a subdomain anyway, why not just include 1 pitch of the impeller with a very coarse mesh? GGI it with a stage interface and you should get a better representation than your sub-domain will offer. All you need is ~20k nodes.

The stage interface will not introduce any transient effects, but neither will your momentum sources.

Regards, Robin

Well,

20k nodes on a propeller?? Have you ever done one?

I have seen simulations that take 200k and have errors on the order of 20%.

20k might work on a high solidity centrifugal passage because it is just a spinning duct...not likely on an open propeller!

But why not do it and prove me wrong!

Bak_Flow

And do you expect to have lower error with a momentum source??? I'm just suggesting that it would be a better approximation. If you want accuracy, include the rotor with a fine enough mesh.

-Robin

Hi,

no offense intended, but if you know the global change in angular, axial and radial (small except at low advance coef opperation) momentum, say from an experiment or a detailed model, which is what I assume the user has. Then you get the exact effect in the global sense although the distribution will be wrong...unless you know that too for appropriate S'''_u_i distributions.

If one does not know the above...then your suggestion may be useful. I am just a bit cautious on propellers as they are very hard to model accuratly.

Regards,

Bak_Flow