Propeller performance curve with VSPaero

[Sam Smith]

Hello,

I reproduced the geometry of an APC propeller in OpenVSP (v3.51.3), based on the geometry file available on APC website, and tried to compute the performance at different J for two different RPMs.

At high advance ratio the thrust or power curves are not so far from the APC data, but I'm a bit surprised by the trends at low advance ratio, and the linearity of the thrust curves... I used the Unsteady Rotating blade option, but the Pseudo Steady gave me similar results.
Do you have an explanation? Do you have a template file that works better ?


Thank you in advance for any hints 

Best regards

Sam S.

[Rob McDonald]

At low advance, the airfoils start stalling.  Are the Reynolds number, Mach, and clmax settings into VSPAERO appropriate?

Rob

[Zach Hazen]

It might be good to compare with wind tunnel results from UIUC https://m-selig.ae.illinois.edu/props/propDB.html

That way you have two predictions and one test: APC proprietary, VSPaero, and UIUC test.  APC data tends to be only "so-so" against wind tunnel data ie: this 10x5 courtesy of propfolio

[Sam Smith]

Thank you for your hint.
I use the Propfolio website myself, it’s very handy, and I know that some APC propellers have been tested in wind tunnel at UIUC, but unfortunately not the one I want to study with OpenVSP.

[Brandon Litherland]

Apologies.  I mistakenly rejected a reply that should have been approved.

Here is the post:

Thank you for your response.

Mach is correct, but it doesn't seem to affect the result when using “Unsteady Rotating,” does it?

As for the Reynolds number, I defined it based on the speed at 75% of the wingspan (forward speed + rotational speed)—is that correct?

For CLmax, do you mean I should enable the stall model and define CLMax2D and Clo2D?

Best regards

Sam

[Rob McDonald]

You should find the most similar propeller to yours that is in both the UIUC database and the APC data set and use both of those to validate your VSPAERO analysis.  Then you will have more confidence in your analysis applied to the propeller you are most interested in.

Rob

[Rob McDonald]

No, ReCref is the Reynolds number with reference length Cref and the reference velocity -- which is either Vinf or Vref if you are using Vref.

If you are using Vref, then it is whatever you set it to be.  Usually, we only use Vref when Vinf is zero (static conditions).

I meant the CLMax2D.

Rob

[Sam Smith]

Thank you for all your suggestions.

I ran a quick comparison for the APC 10x10 propeller, for which UIUC data is also available.

VSP overestimates the thrust compared to the wind tunnel data, but it matches the APC data fairly well.

On the other hand, VSP seems to predict power better at low forward speeds than APC data.

I think modeling the hubless propeller could explain the observed discrepancy in thrust (with less of an impact on torque and power).

Do you have any recommendations for how to better account for this? 

Best regards

Sam

[Rob McDonald]

I think this match is amazingly good and I would be thrilled to have it.

When plotted vs. J, we expect a fixed geometry prop to collapse onto a single curve.  The UIUC data does not do this.

If you read the reports that accompany the data, they explain it with Reynolds number differences.  I'm not sure I believe that -- I have also seen propellers at this scale with significant flex at high thrust.  That flex may cause bend-twist coupling -- effectively changing the pitch of the prop.  This is something that the APC and VSPAERO models will not capture.  I think the APC data will capture the Reynolds number variation, and VSPAERO will somewhat too -- but you have to run a different case for each Re and it can make sweeping J difficult.

My intuition tells me that the Re variation should affect Cp more than Ct (it is more of a drag effect than a lift effect - particularly away from stall).  And you see that in the Propfolio results (for an APC 10x10):

The Cp spread (in the un-stalled region) of the two results seems to be of similar magnitude -- i.e. both could be explained by Reynolds number.

However, the APC data has very little spread in Ct - while the UIUC data has considerable spread.  These do not seem likely to have the same explanation.

I always thought someone should set up one of these small props in a tunnel with reflective tape on the back surface along with a pulsed laser and the ability to trace where the laser reflection hits the tunnel wall as a way of measuring deflection and twist of the blade under load....

Rob