Flying wing: moveable CG stabilization

[deh6]

Is there merit in the following?

1) Take a straight (i.e. unswept) wing with a pod suspended below (similar to a hang glider) and an AoA sensor attached to the wing. The pod would house the batteries, radio, motor--all the heavy stuff. The suspension could be a flat piece of aluminum, steamlined of course. The suspension point on the wing would be at the NP and the wing would be free to pivot about that point; the angle with respect to the suspension arm would be controlled by a servo. The pod would also pivot about its CG and be controlled by its servo. A controller continuously positions the wing servo to match the angle of attack commanded from the ground, and it also positions the pod so that it is always at zero AoA for minimum drag.

No elevators, and no reflexed airfoil required. Positioning of the pod relative to the wing adjusts the CG to produce the AoA commanded. Ailerons control roll, and split drag flaps at the wing tips for yaw.

2) Using the concept of 1) above, add a boom and fixed elevator to the pod (and while we are at it, add a vertical fin for yaw stabilization). The pod/boom becomes the AoA sensor and elminates the AoA sensor on the wing as well as servo on the pod. Shorten the suspension rod. It now *looks* like a conventional configuration, but is a flying wing with CG for pitch control. Or, one might view it as a conventional configuration where the CG is continuously adjusted so the elevator produces no positive or negative force.

The idea comes from pursuing the following thread:
http://www.rcgroups.com/forums/showthread.php?threadid=146767&goto=newpost

coupled with Martin's diagram showing three types of stability for flying wings:
http://www.mh-aerotools.de/airfoils/index.htm

Has someone tried this before (and failed, since if it was a big improvement we would all be using it!)?

Several problems come to mind. One, the pod's drag needs to be lower than the wing so that in a dive the force causes a pitch up moment. Another, is that the servo in the wing must generate enough torque to maintain the CG position, which, depending on the airfoil, may shift considerably over the operation range. For 2), the slipstream might obviate using the pod/boom for AoA sensing; it'd be OK for a glider, however. Getting the dynamics of the control loop sorted out might take some experimenting (busted planes), but that part looks doable.

Regards,

Don

[BMatthews]

A few things off the top of my head.

The wing DOES need reflex or a positive pitching moment. There's nothing automatic in your scheme to replace that. IF that AoA sensor and a few more sensors were connected to a computer that was connected to a fast servo perhaps the model would have a psuedo stability but not otherwise. That material of Steve Morris' and the others in that thread applies in that case. If you persue the active stabilization then perhaps this could work.

The center of drag will be below the wing. No way to avoid that with a pod there and not much above the wing. So the drag center will tend to promote a dive. So no assistance to your pitch stability I'm afraid.

And finally I have some concerns about how fast the CG changing will affect the plane. Also the fact that the CG will move as the attitude of the model changes. You may get this to work in level flight but if you command a quick pitch change the effect won't be linear across the range of pitch change. Something that movable surfaces don't suffer from.

And since you're going to use movable controls for the other functions with clamshell surfaces at the tips for yaw control I don't see where there's going to be any advantagious reduction in drag worth mentioning.

On the other hand if your goal is to replicate and apply Steve's work with trying to utilize more efficient airfoils for flying planks then it'll be interesting but I think there's many a problem to solve on your road to sucess.

[deh6]

BMatthews,

Thanks for your thoughts. It certainly helps my thinking to have people raise issues.

Martin H's (url above) says that for when the CG is far below the wing the airfoil does not need to be reflexed for stability, though he mentions that airfoils with negative pitching moments are typically limited to "moderate."

Concerning coming out of a dive--The CG will be well below the wing (heavy pod compared to the wing), but the center of drag will be above the CG assuming the pod has lower drag than the wing (which should be the case). If the pitching moment of the airfoil at zero lift is less than pitching moment of the center-of-drag|CG, then it should right itself.

The whole idea is to use active control, i.e. a control loop (which nowadays includes a computer). Steve's work focused on a fixed CG; elevons control the angle of attack, whereas the idea here is to move the CG for control.

As far as speed of pitch change is concerned, it should be fast. It is merely a case of rotating the wing with respect to the suspension arm, i.e it will be as fast a the usual elevator movement. When the NP of the wing and the pivot point coincide the servo should have no load other than the moment of rotational inertia of the wing which is small in that axis.

The dynamic aspect will take some tinkering with the math. It's been *many* years since I was in-the-swing of doing this type math analysis, so it's slow going.

Don

[vintage1]

It might fly, but looping could be 'interesting' :D

Think about it...

[BMatthews]

A small free flight model would be a good first step to see if this is a viable plan. If it can fly free flight then adding control via radio should be simple. If it won't fly as a free flight without stability augmentation through active electronics then it's not much good as a hobby item.

A couple of other observations. Depending on the required distance from the pod to the wing landings may be "interesting" Obviously there's a certain amount of vertical separation for this scheme to work. It seems to me that the usual contest type spot landings are out of the question since with the high CG location relative to the pod nose the model will just somersault on any landing other than a very smooth arrival.

In thinking about this I came up with this sketch. Since you need a strut anyway why not incorporate a conventional fin and rudder. I showed this with the "elevator" servo in the fin and the rest of the gear in the pod but I forgot about aileron servos needed in the wing. And since this will be a fairly critical balance of pod strut length to wing pitching moment damping I don't see using regular flaps for glide path control. Or if you do use something it would need to be very carefully designed such that any pitching from the airflow disturbance is offset by a suitable change in the drag center to keep the system in balance. I'm thinking that a clamshell split rudder may be a suitable option since it's drag center would be very close to the CG axis.

And finally there's launching. A normal hook on the pod belly is pretty much out of the question. The towline forces are too far away from the CG to allow any form of effective control during a tow The bridle hooks on the wing are similarly too far from the CG. For other flying wings I've seen pivoting hooks that have their pivot located a tiny bit ahead of the true CG up inside the fuselage or at the wing pivot location Since the CG of this arrangment is out in space between the wing and pod some form of curved track inside the pod with it's virtual pivot just ahead of the CG would be needed. Then as the pod moved the pull would shift about the CG and the model would respond accordingly. This track would need to be very free in operation. Certainly the hook car would need to be equipped with a ball bearing arrangement so it can shift small amounts very easily.

All in all an intruiging experiment but I don't see it being on the podium of the the world FAI meets any time soon.

Here's my take on the rear strut fin idea.....

[BMatthews]

2) Using the concept of 1) above, add a boom and fixed elevator to the pod (and while we are at it, add a vertical fin for yaw stabilization). The pod/boom becomes the AoA sensor and elminates the AoA sensor on the wing as well as servo on the pod. Shorten the suspension rod. It now *looks* like a conventional configuration, but is a flying wing with CG for pitch control. Or, one might view it as a conventional configuration where the CG is continuously adjusted so the elevator produces no positive or negative force.


I missed this the first time around what with thinking about the massive pendulum effect idea.

To my mind adding all this "stuff" is just shooting yourself in the foot. The goal here, as I understood it, was to achieve lower drag and higher efficiency for a flying wing. If you add all this "stuff" you may as well just make a regular sailplane. Certainly all the wetted surface area and drag is there. In conventional practice we achieve a near to 0 tail load through using the dive testing to set the CG very close to the neutral point anyway.

Here's something else to ponder. If you choose to fly some aerobatics either by choice or in response to turbulent upsets a variable CG will have no control authourity at some point near the vertical positions like in a vertical dive or the nose up portion of a loop At that point you must rely on that "undesireable" pitching moment of the section rather than control surface forces. At least with your hang glider pod like arrangement you would have the high drag center of the wing to assist with restoring level flight.

And if you were to try to fly inverted with your "conventional" planform or if the speed over the top of a loop was to fall off to where the model was flying at 0 or "negative" G (as in inverted) then your "up" weight shift is nuetralized or reverses its effect and the model will try to pitch the wrong way.

All in all I think your hang glider format has some interesting possibilities but a "normal" layout using CG shift is just not worth the trouble. All that just to avoid ONE control surface and a small amount of induced drag from a very light tail loading? Because that is what it's come down to with this format.

I think I understand that you're trying to set up the model so that the airfoil can do it's job without the added drag of the tail load but with proper fine tuning (setting the CG using the dive test) it's possible to come very close to the optimum without resorting to onboard stability augmentation. Unless you can show some numbers to indicate that there's a significant drag reduction through avoiding any tail loading then I don't see that there would be any advantages to this variable CG over a conventional design.

And even if there is then I would suggest that a computer controlled CG and stability augmentation loop that activates a variable weight combined with a conventional elevator would be more effective. The CG would be set to minimize or eliminate the tail loading and the elevator commands would be mixed into the system which would respond with a mix of CG shift and elevator together with the CG shift being a minor player in any control commands to avoid the loss of effect due to any attitude related problems.

In proof reading this last paragraph it reminded me that I do believe there is someone that did just this. A variable CG with onboard stability augmentation on a conventional model. In fact I seem to recall that he flew it at a Nats. There was a pic in one of the magazines and a paragraph in the write up on it. That was about 10 years ago as I recall.

Anyway, I'd like to thank you for an interesting thread. Sure makes one think don't it. :D

[deh6]

BMatthews,

Your strut/fin diagram looks interesting. It has a nice graceful look.

My latest thinking is that it may not work! At least in the original configuration. Picture a wing with a pendulum hanging down from the exact NP (assume a symmetrical airfoil to keep things simple). Assume a "perfect servo" always holds the wing at the commanded AoA. Everything is in balanced as flies along. Now a small disturbance moves the pendulum forward. The servo dutifully keeps the wing at the commanded AoA with the result that in no force on the pendulum. The servo allows it to swing free unretarded, so there is no damping. Consequently it appears that it will continue to swing whilst the contraption glides along.

In detail, it more complicated. The wing will actually speed-up/slow-down, as well as have the load slightly change, due to centrifugal force as the pendulum swings. It may be possible to exploit this with the controller compensation to produce a damping, but at the moment it is not clear.

There must be some good stability analyses of hang gliders. If the basic configuration is subject to undamped oscillation, then pilot must be providing the compensation to damp it out, much the same as the pilot damps out the phugoid oscillation in conventional plane.

As you point out landing is problematic-high grass might be required. I toyed with some thoughts about how one might increase the AoA as it came in and when it was about stall/touchdown, pitch the wing nearly vertical. The pod would catch in the grass and then the wing flop down flat-side-first, which (might) be gentle, versus a forward tumble where the wing would hit the ground leading-edge-first. I also pondered if it could be made to "spin" like those toy propellers on a stick, a maple tree seed type of descent. The Tufflight (flying wings) www site talks about one model that can spin down to a landing-"Roto-landing". http://www.tufflight.com

Regarding inverted flight such as falling out at the top of a loop, (as vintage1 mentioned, the loop is indeed "interesting"), it looks to me like the control system would tilt the wing to max positive angle (with respect to the "strut") and assuming some forward motion (i.e. not a tail(less) slide) it would complete the loop in a somersault manner, not unlike a conventional plane that stalls in the inverted position. Sustained inverted flight is certainly not possible. I have never seen a parasail type of parachute jumper do a true loop, though at airshows I've seen them make what looks like a loop tilted 45 or more degrees-rather breathtaking to watch.

Concerning the moveable CG on a conventional design, my thinking (currently ;) is that the ground command would be an AoA (Gary Warner in the thread ref'd above mentions doing this). The error between this and AoA sensor on the wing would control the elevator. A second control loop attempts to keep the elevator at zero by moving the CG. The drag from the elevator would occur during changes in trim while the CG gets adjusted (which could be slow, e.g. a lead-screw scheme). Now, if the CG can be moved rapidly, then the elevator has no load and is therefore redundant. The "pendulum" idea offers a way of implementing fast CG change, but as I pointed out at the beginning there may be problems.

I need to tinker with some models.

Don