Thickness distribution on propeller analysis

[Dionysis Rouvas]

Hi everyone,

I am currently running unsteady rotating propeller analysis and I noticed something odd.

Since VSP uses VLM to model the surface it shouldn't matter what the thickness distribution is. Yet by running some cases with different thickness distributions I get different results. I have taken great care to make sure that everything else in the model is exactly the same (airfoil, chord, sweep and twist distribution).

Is it possible that I have missed something or does VSP take into account thickness in some way and incorporates it in the VLM? Does thickness implicitly affects the airfoil shape or camber? And if so what is the correct thickness I should go for. It seems that the thinner the blade the closer the results to the measurements I compare them against.

Thanks,

Dionysis

[Rob McDonald]

I would have to dig in further (which I don't have time for now), but what I think you're seeing is a change due to the implicit clustering of points on a thick surface....

The mesh points are arranged according to approximately equal arc-length by OpenVSP.  If the cluster parameters are set LE=1.0, TE=1.0, you should get equal spacing.

However, think about a cylinder -- equal spacing (in terms of arc length) will actually result in clustered spacing when those points are smashed down to the VLM thin surface.

This image is a default wing.  The airfoils have been changed to circles.  The point clustering has been set to uniform -- LE=1.0, TE=1.0.  Yet when viewed from above (or projected to a VLM surface), the points are clearly clustered at the LE and TE.

So, as you're varying your thickness (everything else held constant) you are actually also implicitly varying the clustering around the airfoil -- particularly at the LE, where the shape has lots of curvature.

You should be able to create two (otherwise identical) propellers.  One with thick foils, the other with thin.  Change the view surface mode to 'Camber Degen'.  Then adjust the LE/TE clustering on one of them until it most closely matches the mesh spacing of the other.  Hopefully you can get a fairly similar mesh between the two.

Then run both of those cases through VSPAERO.  I suspect their results will match more closely.

Let us know how it turns out.

This may mean that you want to perform a larger resolution and clustering study to come up with good values for your purposes.

Rob

 

--
You received this message because you are subscribed to the Google Groups "OpenVSP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openvsp+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openvsp/b74ffba6-f7cf-4caa-a04f-ae7e858f84bdn%40googlegroups.com.