VSPAero Errors in Panel Method Analysis of a BWB Aircraft with Propeller

[Runxi Chen]

Hi all,
I'm analyzing a blended wing body (BWB) aircraft with a propeller using the panel method.
Individually, the aircraft and propeller run fine. When combined (inserted the propeller in the BWB file, both under the vehicle tab), I encounter the following:
1. Panel Method + Rotating Blade → VSPAero shows error codes (see attachment 1).
2. VLM Method → Runs without errors.
3. Panel Method + Actuator Disk → Runs but only simulates 0 RPM (blades stationary) despite setting 6500 RPM. This is confirmed by the trailing wakes in the viewer (attachment 2).

I'd also like to confirm:
1. A 2020 post mentioned control surfaces not working in the panel method. Do they work now?

Thanks for your help!

Best regards,
Runxi Chen

[Runxi Chen]

Here is the full model

[Rob McDonald]

Control surfaces on thick geometry should work in 3.45.X.  However, there seems to be a bug on some platforms that I'm working through.

When using the actuator disk, you need to set the propeller to 'Disk', not 'Both'.  If you've done this correctly and are still getting both showing up, then you've run into a bug and should move to 3.45.X where it is fixed.

Rob

[Runxi Chen]

Hi Rob,

Thanks for the update! 

I tried and was able to get the BWB (thick model) + Propeller (VLM) model running. While the control surface is added, it still gives error codes. (I am on Windows+Python 3.9 platform) Do you know which platforms work?

Also, for the 3.45.1 viewer, the pressure range might not be displayed correctly. The entire BWB, except the winglet, is displayed as high pressure (all pink), and the model also looked a lot darker than the previous version. Did I set something wrong in the viewer?

I know that I might still need to fix the tessellation to make the model behave better, but these are the problems I currently have.

Thanks for your help!

Best Regards,

Runxi Chen

[Rob McDonald]

Control surfaces currently work on anything but Windows.  I have a fix for this that will go out in the next release.

I'm not sure what changed, but the vspviewer isn't behaving as well as it used to.  Fortunately, all this can be worked around easily...

In the bottom right corner of the viewer, there is an option to reverse the normal vector direction on various surfaces.  When a surface shows up very dark to black, it means you need to flip the normal vectors.

Similarly under one of the menus, there is an option to adjust the range of the contour scale.  You'll want to re-adjust that.

Often, with rotating propellers, you can either choose a reasonable scale for the prop -- or the non-rotating geometry -- but not really both.  The velocity of each is so different that they throw the respective Cp scales way off.

Rob

[Runxi Chen]

Hi Rob,

Thanks for your instructions! Now the viewer works.

Just one additional quick question:

Was the solver rewritten or modified in this update? The results seem a bit different between the two versions, and the new solver seems to take more iterations (the old solver takes ~120 iterations, and the new solver takes 300+).

Here is a comparison between the two versions running the same case:

1. The high-pressure area around the trailing edge of the BWB at its upper surface is gone in the new solver. Overall, the pressure becomes lower on both the upper and lower surfaces in the new solver.

2. The new solver does not really like the winglet mesh, as the CL of the winglet went really high (~180, not the case in the old solver).

Old version:

New version:

Thank you so much for your help!

Best Regards,

Runxi Chen

[Rob McDonald]

What is the difference between a re-write and modifications?

There were many and extensive modifications all around.  There were enough changes that you should throw out your old experience with solver parameters (things like number of iterations, number of points in the wake, etc) and start experimentation again.

Something is certainly wrong with your winglets in both cases.  I would start by fixing those and then working to figure out whatever is next.  Did you build it with zero thickness airfoils or something?

You also don't need nearly that much spanwise resolution on the center body and the transition.

Rob

[Runxi Chen]

Hi Rob,

I just tried different things and still have some problems with the winglets. Is there a standard practice for how the winglets should be handled?

I did not use zero-thickness airfoils; the airfoil for winglets is NACA 0012. OpenVSP might consider the airfoil as zero thickness (the top view in the viewer does not show any thickness for these winglets, as shown below). The winglets are 90 degrees from horizontal.

Currently, the winglets give extremely high/low CL values, which are not reasonable. For the end cap of these winglets, I know that a flat cap is probably the worst practice. However, it seems to give a lower CL in both new and old versions (New version: flat: ~600, round ~100000, old version: flat: 2, round ~300). Increasing tessellations of the winglets helps, but CL is still out of the reasonable range for the new version.

My next steps are reducing tessellations for the body and the transition section. There are a lot of tessellations initially because there used to be control surfaces in that location. Since those surfaces are for trimming the aircraft, I just embedded them in the airfoil.

Thank you so much for your help!

Best Regards,

Runxi Chen

[Rob McDonald]

There is an option to rotate airfoils to match dihedral.  I suspect you need to turn that option on.

You may also want to blend the winglet with an intermediate section.  Check out the videos on the Ground School and past Workshops on Blended Wings / Advanced Wing Modeling.

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 visit https://groups.google.com/d/msgid/openvsp/833ee440-ef23-4c3b-88ad-5fc191d102b4n%40googlegroups.com.