Actuator disk representation question.

[Cody]

Hello,  

I have a question about VSPAERO's implementation of a simple propeller analysis. How does VSPAERO take into account propellers represented as actuator disks when propeller representation is disabled ('Actuator Disk' is unchecked under propeller representation in the Advanced tab of VSPAERO,. but the disks are still included in the geometry of the model being ran)?

I had assumed it just neglected them, but I tried running an analysis with them included in the geometry but disabled under propeller representation, and ended up with somewhat different results. 

Thanks for the help! 

[Brandon Litherland]

I'm afraid that I don't quite understand the question.  We will need a bit more information about your particular use case and what you are trying to analyze first.

For example, are you running a propeller in isolation or with lifting surfaces?  Are you using Panel or VLM mode?  If a Prop is set in disk mode and included in the VSPAERO set, say Shown, unless the Actuator Disk button under Advanced is checked those components will be ignored by the solver in VLM or Panel mode.

Are you using Prop components or the heritage Disk component?

- Brandon

[Rob McDonald]

If you have a disk in your model, but you aren't using the actuator disk model, it may model the disk as a solid blockage.

If you want it left out, you should create a 'Set' that does not have the disk and tell the VSPAERO GUI to use that Set.

Rob

[Cody]

Thanks for the responses, guys. Brandon, I am in VLM mode running VTOL propellers modelled as actuator disks in tandem with lifting surfaces (I know that this may cause some inaccuracies in the modeling, but I am not sure how much exactly or if it is enough to render results not useful). I am trying to analyze the effect of powered VTOL propellers on the stability characteristics, drag, etc. of the aircraft during horizontal flight. However, I was noticing that even when actuator disk was unchecked in VSPAERO, if the geometry was in the analyzed set (shown), it provides different results than an analysis with the geometry not included. 

Rob, thanks for the answer and I will keep that in mind, thank you. I have a follow up question, then, if you don't mind: Is there an acceptable method for approximating drag across unpowered VTOL propellers during horizontal flight (ie, rotor tips facing streamwise and not rotating), since unpowered actuator disks would be modeled inaccurately as a solid disk? I have tried modeling this using blade mode instead of actuator disks, but that makes VSPAERO much more complex and was leading to errors for me. I tried modeling the props as fuselages with similar proportions and got what appeared to be reasonable results, but I am not sure if this simplification is able to work accurately with its implementation in VSPAERO. 

Thanks so much for the help. 

Cody 

[Rob McDonald]

Not really with a potential flow code.

The truth is, the flow on a stopped rotor -- edgewise into the flow like a surfboard is pretty ugly.

My best recommendation is to use other tools -- handbook methods or CFD to build an estimate of the drag on your stopped rotor.  Then add that as an increment to your drag buildup that includes both viscous and inviscid contributions.

Rob

[Cody]

I see, thank you. You say that that the problem is particularly strong for a stopped rotor, is it doable to model (if approximately) a moving rotor (actuator disk) edgewise into flow, or is CFD necessary here too? 

Thanks! 

Cody

[Rob McDonald]

It depends on the advance ratio and other aspects of the situation.

If the blade is rotating relatively fast compared to the forward velocity, then the airfoils will experience appropriate flow at moderate angles for a full revolution.  The flow will be attached and well behaved, the kutta condition will apply to the trailing edge as expected.

As you slow the rotor and/or increase forward speed, the retreating blade will start to see a very different flow condition.  The local angle of attack will significantly increase -- resulting in locally stalled flow.  Eventually, the local flow will actually be fully reversed.  There will be areas of separated flow on the blade for part of its rotation.  These phenomena will not be very well predicted by a potential flow method.

A tool that has a long history of being tailored for use on rotating wings -- like CHARM -- will have models and adjustments that will allow it to do better for longer in these situations.

However, VSPAERO's models have not been pushed hard into modeling retreating blade stall etc.

A stopped blade is a similar problem -- but potentially worse.  With a potential flow model, the Kutta condition is going to be applied to the trailing edges of both stopped rotors.  However, you have to consider whether that really matches the physics of the flow separation lines for that situation.  Probably not.

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/3c06b740-b789-4dca-9a90-084c817b47c8n%40googlegroups.com.

[Cody]

I see, it looks like I will need to consult some CFD software, then. Thanks a ton for all the help!

[Rob McDonald]

CFD is not an easy answer to this problem either.  You're looking to model an extremely complex flow.

You really need to think about what you're trying to accomplish.  What kind of fidelity do you need?  Will you have any experimental results to validate your computations?  How many points do you need to consider?

None of this is easy.  CFD might not be the right answer.

Rob

To view this discussion on the web visit https://groups.google.com/d/msgid/openvsp/2436c630-37ab-4f07-9652-4eb156289694n%40googlegroups.com.

[Brandon Litherland]

I'll second what Rob said.  High-fidelity CFD that "could" capture the separated, complex flows in these regions is VERY expensive to run even for clusters.  And you really only run that when you've got a good, detailed idea of what the geometry looks like.  Heck, even a pitch oscillation 2D airfoil is hard for CFD to get "right".  All CFD is, fundamentally, is using clever math (often with lots of corrections) and a whole lot of points.  If there aren't enough points, if the assumptions of a particular method don't apply to your case, if you don't let it run long enough, if the geometry isn't quite right, etc. can all give you incorrect, or even physically absurd, results.

In your case, at least for the power-off drag buildup, even modeling the fixed blades as fuselage components may be a step too far.  You may have more reasonable and physically representative results from simply adding some flat plate drag to the model.  Model the hubs as cylinders or ellipsoids and the blades as plates, check Raymer or another reference for appropriate form factors (the Parasite Drag documentation has some of these references) and see if the results are sufficiently conservative.  I've learned to just accept that what eventually gets built will most often perform worse than expected or at best break even because of all the little things we miss in the early designs.  This is also why historical regressions for transport aircraft are so useful.  They are all trained to fit real airplanes.  Advanced concepts tend to fall outside of these regressions so we just have to do the best we can with tried-and-true fundamental methods.  If you intentionally overcompensate for uncertainty in your conceptual models, the worst that can happen is you prove yourself right over time.  Often, however, you can take steps to gradually remove that uncertainty and improve the performance.  In that case, you'll only ever be pleasantly surprised.