Aeroelasticity... counteracting blade twist 1

[PaulPounds]

Here's something topical for those following some of the work being done in aeroelsticity.

The rotors I've been working on for my robot helicopter are tremendously thin and seem to be twisting something awful. I've put together a computer simulation of the blade-twist/rotor speed system and it seems to confirm my hunch that the blades are twisting into stall under load. We can't easily change the materials we're using (two-ply carbon fibre prepreg) or the shape of the blades themselves.

Can anyone suggest a good method for reducing blade twist out near the tip?

I've considered trim tabs to provide a downwards moment at the rotor tips, but those might be difficult to attach to the blades in a way that won't add a heap of drag or go flying off.

I've considered 'pre-decrementing' the twist of the rotors by affixing them at a negative angle at the hub so that they will twist into position under load. According to my simulation, the loading of the blades would be dominated by the aerodynamic forces, which will simply twist them into stall anyway (probably breaking my rotors while it's at it).

I've even go so far as to consider some variation of chinese weights to offset the twist...

I'm open to any practical suggestion!

-Paul

 

[Intermesher]

Paul

A few quick unqualified thoughts;

Split the tip weight in two and relocate them in the 'corners' of the leading edge and the trailing edge. The tennis racquet effect from centrifugal force will attempt to put the blade tip into a flat pitch.

Vary the mass of the two tip weights between the leading and trailing edge, to change the chordwise center of gravity.

Relocate the axis of pitch on the blade so that its percentage of chord differs at the tip from that of the root.

Other comments appreciated.

 

[PaulPounds]

I've thought about tip-weights, but the problem is that it's difficult and dangerous to put weights on the tips (they are extremely thin). Furthermore, I want to keep the Lock number down, as this does good things to its passive stability.

I can't add tip sweep at this point, as that would require redesigning the blade moulds. It was a chicken and egg problem - we didn't now how much twist we'd have until we made the moulds but we couldn't really make the moulds properly until we knew how much twist we'd have...

-Paul

 

[PaulPounds]

Actually... my calculations for tip weights look promising. Maybe it's not so impractical after all.

Can anyone direct me to a good source on calculating tip-weights?

-Paul

 

[GraviMan]

Just an extreme suggestion, but it is a model...

Put a rotor ring around the outside of the blade tips. Get the compliance between this and the tips right (damping if needed), and the tips should behave exactly as you want. May even have aerodynamic benefits, such as reducing power settling from the vortex ring state. Will also avoid all the headaches of aero elastic flexing and coning.

Just a suggestion, but I'm curious about the thoughts of others on this...

Mart

 

[Intermesher]

Paul,

This may be of some help;

http://www.unicopter.com/0735.html

Dave

 

[tbuelna]

PaulPounds:

This article should interest you:

http://www.boeing.com/news/releases/2004/q2/nr_040518t.html

Unfortunately, it is likely a more expensive solution than you are looking for.

Terry

 

[GraviMan]

Out of interest Paul,

What simulation software do you use to simulate aeroelasticity? I am familiar with Nastran Super Element formulation - exporting the results of modal anaylsis for other explicit codes to use. In my case it was for use in ADAMS to simulate ground vehicle dynamics, but I would imagine that CFD would use superelements in a similar way.

Is there actually an explicit code that enables structural interaction with aerodynamic models? I am familiar with the good ol' Navier-Stokes equations, but never got the opportunity to try out CFD. I'm trying out response dynamics in Master Series, and wonder if the technique is adaptable...

Mart

 

[PaulPounds]

Uh... wellll, nothing quite so fancy as that!

My 'simulation' is more of an emulation. I use an iterative algorithm coded in matlab to calculate the moments, lift and drag on each radial station of the rotor blade. It then uses these values to figure out the resultant rotational velocity of the rotor based on the electric motor current draw.

After a couple of iterations it settles down to an equilibrium point that seems to reflect what I'm seeing in experiment. It's really dodgy and needs heaps more testing, but it gives me a taste of how the behaviour changes when I vary things like stiffness and angle.

The simulation is no good for blade stall conditions or forward flight conditions.

If you can recommend a good (free, or inexpensive) aeromechanical simulator, I'd be very interested.

One thing I am having trouble in is finding accurate polar data for stalled wings. My blades are very thin and so I can probably model them as flat plates at and after stall, but I've never seen good data for it.

Anyone know where such a thing might be found?

-Paul

 

[GraviMan]

"Uh... wellll, nothing quite so fancy as that!"

Hehehe, you should see how I am virtually testing my truck suspension - takes 4 hours to run. Must admit I miss the old cost no object days (for design procedures) in the auto industry. To some extent it sounds like you are solving the problem explicitly (i.e. iteratively).

"If you can recommend a good (free, or inexpensive) aeromechanical simulator, I'd be very interested."

Me too! CFD is still relatively new, so I'm not aware of any packages combining CFD with FE. My best guess would be Ansys, since they market themselves on interface type problems. Might be an idea giving them a prod to see if they will give you a demo version. I'm tempted by this myself for another problem I am working on.

I would really enjoy trying out a few of my own helicopter ideas in my "spare" time. ;-)

Mart

 

[PaulPounds]

Yeah - I have this crazy dream of free engineering tools for non-commercial use, like RC aircraft and scale modelling, but I don't think I'll ever see it.

Still, the thought of being able to design an airfoil, then a rotor, then the whole craft on my PC is a seductive thought.

For now, I'll just crank out stuff in Matlab.

-Paul

 

[GraviMan]

It's a shame most software manufacturers don't allow a reduced cost licences for private individuals to "play" with the software. Gone are the days when you need a supercomputer, and most companies only look at what packages you already know, not learning potential.

I'm convinced individual licencing would benefit all - including software writers. Maybe we are the enlightened few...

Are you going to make your Matlab simulation public domain? Intermeshers amazing site might be the ideal place for this sort of thing to be centralised. I appreciate that much sweat and tears will likely have gone into it, so maybe a contribution system.

Mart

 

[PaulPounds]

Ah, I'd do that in a flash except for two reasons.

1. It's buggy as hell and I'm not even sure the results it gives are meaningful

2. It's probably the IP of the uni at which I work... they have this thing about 'anything you produce blah blah blah is ours, plus your immortal soul'.

Mind you, I'll discuss it with my supervisor and see what he thinks.

-Paul

 

[Vilayti]

One of the problems of using torsionally soft rotor blades occurs at high collective pitch, if the blade is highly twisted. In addition to Propeller moment, there is a tension term acting on the blade which is normally small in comparison to the propeller moment. At low collective and high twist, these moments act in the same direction. However at high collective, a highly twisted blade will suffer nose up tension moments causing stall. I wonder if your blades are suffering this. Is the pitch hinge inboard of flap and lag?

Have your tried using blade sweep? Sweep will change two things; namely the C.G. distribution and thus the propeller moment acts in a different direction and generate some pitch-flap coupling that can be stabilising as far as the blade twisting is concerned.

 

[PaulPounds]

Our blades are fixed-pitch, so I'm not sure they'll have that problem.

Blade sweep isn't something I've considered, but it sounds intruiging. I'll have a look into it - thanks!

-Paul

 

[GraviMan]

Must admit, I have frequently wondered why torsionally soft helicopter blades don't have an "aileron", or tab, on then tip trailing edge. Since tip AOA is more critical than root AOA the tab would directly control cyclic input, assuming actuation mech was stiff enough to avoid introducing more aeroflexure. Swept back blades would go even further down this route (assuming higher rotor rpm is the way to go).

Keep hub control and you can go towards active blade twist, assuming you have already rejected bi-flow airofoils...

Mart