Designing the ultimate DS plane

[bjaffee]

Have at it! I will sit back and listen quietly....

[Ollie]

Let me introduce myself. I am a retired engineer. I am 72 years old and beyond any capability of flying DS at high speeds. However, I am fascinated by DS and would like to participate vicariously through the design process, which I love.

The basic equation derived by Wurts and Drela states that the ultimate DS speed is some constant times the lift to drag ratio of the plane. The constant expresses the ratio of energy gained to energy lost in a circuit of the course. The greater the wind shear the greater the energy gained. The bigger the circle the more energy is lost. The trade is between the ability of the pilot to control the plane in a small circle versus the ability of the plane to maintain its shape and stay in one piece under high G forces.

So, the better the L/D the faster in given conditions and the smaller the constant can be for a given speed. I am supposing that the larger the circle for a given speed the easier it is to pilot the plane accurately. If the circle gets too small, pilot skill may become a limiting factor. The less the G forces the less strain on the structure and the bigger the circle the less the strain on the structure.

So, here is a straw design to comment on:

Span 10 feet
Aspect ratio 10
Root chord 16 inches
Design speed 250 MPH
Airfoil NACA23-210
Re at root and speed 3,120,000
Configuration, spanloader. By that I mean that there are multiple fuselages and tails distributed along the span. Specifically four fuselages, four vertical tails and four horizontal tails. The fuselage would consist of four graphite golf club shafts with four nose weights in streamlined bulbs in the nose. The battery, receiver and servos would be imbedded in the wing. The total stab area would be 10% of the wing area and the total vertical tail area 5% of the wing area. The horizontal tail volume coefficient would be 0.40.

Estimated weight 12.5 pounds. Estimated lift coefficient 0.32.
Estimated circle diameter at speed 200 feet.
Estimated transit time for one circuit 1.7 seconds.

I have to go now. I will put up more structural details and a weight budget later. Also, I will try to defend the configuration with detailed rationales.

[Ollie]

I selected the NACA 63-210 after realizing that the performance of full scale laminar flow airfoils is within reach when designing a large DS machine with a design speed of 250 MPH. The NACA 63-210 has a lift to profile drag ratio of about 80 at coefficients of lift from 0.3 to 0.8 at a reynolds number of three million. I looked at other airfoils in the NACA 63 family with lift to drag ratios as high as 120 but the high L/Do occured at high lift coefficients which would have resulted in unacceptably short course times, small circles and high G loads at design speed and coefficient of lift. The NACA 63-110 might be an even better choice for speeds above 250 MPH or even larger wing chords. The 10% thickness on a root chord of 16 inches provides plenty of room for radio gear and both chordwise and spanwise stiffness.

At a reynolds number of three million and above, increasing the airfoil thickness to 12 or 15% seems to have little effect on the drag coefficient in and just above the low drag bucket. So, thicker airfoils in the NACA 63 family could be considered.

Tomorrow I will discuss spar sizing and the rationale for a span loader configuration.

[Ade]

Do these foils work well enough at lower RNs to enable the model to get up to the "working" speeds?

Ade

[Craig Toutolmin]

Hi Ade,

The definition of "ultimate" was not made. From Ollie's discussion it is evident that top speed is his goal. 8-). To generate those speeds the wind on the front side will be in excess of 30 mph and most anything will fly enough to get up to speed.

I believe in type specific planes. If the goal is multi-tasking, a stout F3F ship has been offered as the best compromize.

[bjaffee]

Clarification then:

Ultimate as in no-compromises, possibly impractical, howling wind, fastest DS plane in the world. Not a sport DS plane.

Or if you want a more realistic/specific goal, a plane that can break 200mph with a reasonable saftey margin (so maybe an ultimate top speed of 240mph?...depending on conditions, of course).

[Ollie]

I don't have any hard windtunnel data on NACA laminar flow airfoils for the range between reynolds numbers of 300,000 and three million. However, the NACA 64A010 was tested by Selig back in the late 80's. It has an L/Do of 35.7 at a reynolds number of 300,000. Presumably the NACA 63-210, with 2% camber and a more forward thickness distribution would have an even higher L/Do at a reynolds number of 300,000.

It's not that laminar flow suddenly disappears below reynolds numbers of three million. Its that NASA didn't take data in that range or, at least, didn't publish it in Theory of Wing Sections by Abbott and Von Doenhoff. The minimum drag coefficien gradually increases from about 0.004 at a reynolds number of three million to about 0.0075 at a reynolds number of 300,000 as the boundry layer thickens.

By flying smaller diameter circles in the early stages, the energy gained to energy lost factor would be higher to compensate for the lower L/Do. I believe speed would be easily gained on every circuit.

The strategy I am advocating is to fly circuit times of the shortest duration that the pilot can handle from beginning to end while flying at the overall best L/D from beginning to end. This means the circle diameter increases with speed. This could not be a fun way to fly except for people who are trying to break records.

BTW, I estimate that the plane's over all L/D at 250 MPH and 200 ft diameter circle will be somewhere between 50 and 55, depending on the smoothness and perfection of contour of the airfoil (no insect splatters or rain drops allowed).

[Ade]

I was adding a little realism to equation. At the end of the day we are still going to be hand launching into slope lift. I didn't know how well the model would perform at the lower speeds.

Also how badly are the RG15 type sections performing at 150-250mph on a 2-2.5metre model?

Now lastly... Is a model small enough not to breakup at the speeds we are talking about using current technology big enough to extract enough energy from the DS?

Ade

[Craig Toutolmin]

A plane designed to reach 200 mph ground speed will need to designed for 230+mph airspeed. Design calcs are made in terms of airspeed and plane velocities are recorded in ground speed.

Another twist in the design process is that in higher wind conditions the circle is almost constantly changing shape. The conditions are also very often gusty. High wind and good separation are a hard combination to get at the same time.

[bjaffee]

Yeah, was just thinking that about ground speed., Ollie, what I wonder is, do you (or can you) take into account the large airpseed changes that occur when passing through the boundary layer. We're talking about a nearly instananeous change of possibly 30-50mph in air speed. Is it even possible to account for what happens on the wing in terms of lift/drag when that happens?

[Ollie]

Ade,

I'm glad you asked. I was going to save this for tomorrow morning but this discussion seems to be taking off.

The rationale for distributing the mass along the span similar to the lift distribution, is that it reduces the spanwise bending load to zero in the ideal case.

The problem is that nose weight is required to balance the aft mass of the wing, the tail boom(s) and tail. The solution is to compromize with multiple nose weights, fuselages and tails spread along the wing.
The reduction in bending load is proportional to the square of the number of load increments. I arbitrarily chose four increments which result in a 16 times reduction in spanwise bending moments compared to a single central fuselage with nose weight and tail.

The down side is that the spanwise mass distribution greatly increases the moment of inertia in roll and reduces the roll rate. I do not think that it is a serious problem for an all out DS machine which doesn’t need a high roll rate.

I used the NACA 63-210 maximum lift coefficient of 1.6 to calculate the maximum lift force that the wing could generate at 250 MPH. This is assuming that the wing were suddenly forced to high lift by a gust or inadvertant control input before it could slow down. The lift force cames to 3,000 pounds. Assuming that half the mass is concentrated in four nose weights, fuselages and tails, each fuselage assembly would impose an inertial force of 3000/4=750 pounds. Dividing the wing into eight spanwise segments of equal lift and eight moment producing bending loads acting on 7.5 inch moment arms, the peak bending moments are only 5,625 pound-inches. The minimum airfoil thickness is 0.9 inches and the spar depth could be as much as 0.8 inches, yielding a compression load in the upper spar cap of 7,030 pounds. Using a compression strength of 275,000 PSI yields a spar cap crossection of 0.026 square inches. A precured unidirectional carbon spar cap of 0.06 x 0.5 inches would be more than adequate. Twelve pound per cubic foot end grain balsa shearwebs the full width would easily take the maximum shear load of 375 pounds.

Because the normal bending loads will be associated with a five times smaller coefficient of lift, the actual bending with 200 foot diameter circles at 250 MPH will not be of much consequence.

As far as launching and landing is concerned, I estimate the maximum coefficient of lift at about 0.8. This results in a stall speed of 24 MPH. I don't think this will be a problem in 20 MPH wind even without flaps.

In the next installment I will discuss the rest of the structure and weight and balance considerations.

[Ollie]

Craig,
I don't think it is fesible to measure all the anomolies in the wind shear and gusts to have quantitative data to use. Even if you could get the data, I certainly couldn't come up with the equations to use the data.
Bjaffee,
I think a lot depends on the angle at which the plane's path grazes the shear boundry. In a dynamic situation, the L/D depends on the angle of atack change not the air speed change.

[Ade]

Originally posted by Ollie
Ade,

The down side is that the spanwise mass distribution greatly increases the moment of inertia in roll and reduces the roll rate. I do not think that it is a serious problem for an all out DS machine which doesn’t need a high roll rate.



Ahh... something that could be a problem is the yaw damping? I have seen some quite large yaw offsets in rough DS, once an oscillation gets going it could take a while to stop?

http://www.liftzone.com/~ade/vids/crooks_wiggle.avi

that shows quite a bad but not uncommon wiggle. The model lost a lot of speed from that and went quite "mushy" during it.

Ade

PS we all appear to be online at the moment.... why not take this to Chat? Follow the link near the top of the page.

[Ollie]

Ade,

I hadn't considered yaw damping until you mentioned it.

Yaw damping is proportional to the square of the tail moment arm and directly to the vertical tail area. I don't think it would be a problem to increase the vertical tail to 10% of the wing area and increase the tail moment arm by 13 or 14 inches. That would tripple the yaw damping. I don't know if it would be enough.

It would increase the tail boom stiffness requirement. Instead of golf club shafts, it would probably be necessary to go to tail booms of an inch in diameter tapering to a half inch with a wall thickness of around 1/16 inch. Now we are talking custom made tail booms.

[Craig Toutolmin]

Ollie,

I understand that it is not something that needs to be completely calculated. It just need to be taken into account for max air speed and aero load calculations. The destructive gusts occur at the bottom turn.

The most interesting thought to me, and one that is incorporated into my current project is the selection of an airfoil based on the flying style (circuit size and shape) of the pilot.

[Ollie]

Craig wrote:

"I understand that it is not something that needs to be completely calculated. It just need to be taken into account for max air speed and aero load calculations."

I think I took into account the maximum lift that the wing could produce at 250 MPH when I did the spar design. The maximum lift was 3000 pounds at a lift coefficient of 1.6 and an airspeed of 250 MPH. There is no way the wing could produce more lift than that under those conditions unless flaps came into play.

The advantage of the spanloader configuration is that the mass of the wing and the mass of the four fuselages, nose weights and tails, also, to a large extent, float on the lift. This, in turn, reduces the spar strength and stiffness requirements by a factor of 16 (compared to a conventional configuration) even under the worst possible conditions. The incredibly small spar is for the worst possible case. The spar cap crossection requirement is reduced from .416 square inches to 0.026 square inches by the spanloader configuration!

The spanloader configuration also reduces the torsional load requirements on the wing and the bending load requirements on the tail booms by dividing and distributing the loads. It combines the structural advantages of a small model with the aerodynamic advantages of a large model.

To my way of thinking, all these advantages of the spanloader configuration overcome its disadvantages of high roll and yaw moments of inertia which, hopefully, can be delt with.

[Craig Toutolmin]

I'm definately up for the spanloader. Is this going to be P-38 ish? I'm having hard time visualising the elevator configuration.


The NACA 63-210 sure is a sexy looking foil.

[Ade]

That is a good point actually. what is the tail configuration?

If thought out the tail could be used to reinforce the Fuz strength. In yaw at least, If you see what I mean.

Ade

[Ollie]

For the purpose of placing the four fuselages, the wing would be divided in to four panels which produced equal lift. If the wing was untwisted and of one airfoil, then the tip panels would have somewhat longer span than the inner panels. The fuselage associated with each panel would be located at the mean aerodynamic chord of each panel. The inner pair of fuselages would be spaced a bit closer then the outer pairs. In any case the spacing would be a bit less than 30 inches. This is too large a spacing for either one or two horizontal tails to span because it would result in too high an aspect ratio, too thin an airfoil and too narrow a chord for either one or two horizontal tails. I think the only practical alternative would be four horizontal tails, one associated with each fuselage. Starting with a total horizontal tail area of 144 square inches and dividing it into four equal areas gives an individual horizontal tail area of only 36 square inches. Chords between 3 and 3.5 inches and spans between 10.28 and 12 inches would seem about right.

If we increase the total vertical tail area to 10% of the wing area to increase yaw damping, then the vertical tails might have similar dimensions.

The average chord of the wing is 12 inches and the MAC will be a bit more. To get a generous tail volume coefficient and better yaw damping, the tail boom lengths would be about 60 inches. The nose lengths would be about 24 inches to keep the nose weights within the weight budget.

There is no doubt that the configuration would look goofey but, we are not designing for style but for a very specialized function.

If the configuration were three fuselages, the obvious name of the plane would be Trident. Should we name the four fuselage configuration Dinner Fork? Any one that got hit with the thing would be Forked for sure.

The kinetic energy at 250 MPH would be about 53 thousand foot-pounds, which is no joking matter.

[Ade]

Hmm.... What about an inverted Vtail between the 2 inner fuzes? I bit of extra area can't be worse than all those tip voticies sp?from those seperate tails?

Ade

[Craig Toutolmin]

Would three fuses be used for four panels? Or will there be a fuse at each tip? Controlling the torsional loads on the tails will be a challenge. At what point does the parasitic drag from the fuselages reduce the advantage over building a traditional design with a huge spar(s)?

[Ollie]

Ade,

You keep making me rethink my initial approach. That's good!

The angles between the inverted V-panels would have to be about 90 degrees for the substantial yaw damping we are looking for. If we mounted a single V- tail between the two inner booms, we would eliminate the two outer tail boom but retain all four nose booms to carry the nose weight distributed along the span. This would put considerable torsional load on the wing panels between the inner fuselages and the outer nose booms.

The single, central, inverted V-tail would have a projected span of 30 inches or so, a developed span of 42 inches and an average chord of about 6.8 inches. The proportions sound acceptable. There will be interference drag at the 90 degree intersection but, probably not serious.

An advantage would be that the booms could be evenly spaced and the nose weights and fuselage assembly weights adjusted to match the lift distribution. This arrangement would concentrate a little more of the mass toward the center, reducing the roll and yaw moments of inertia some. I dread having to estimate the moment of inertia of the airfoil so that I can see how much the wing panels will twist under a 240 G maneuver. Since the mass would be more concentrated, the lift distribution could be adjusted a little to shift the lift distribution to the center. The penalty would not be large since a slightly less efficient lift distribution would be applied to the already low induced drag. Another small disadvantage would be an increased tendency to tip stall at low speed requiring more stall margin while launching and landing.

I guess the most irksome aspect of a central V-tail in combination with distributed, multiple nose weights is that it greatly complicates the design process both aerodynamically and structurally. I think I will let someone else tackle the specifics.

BTW, the tip vortex (induced ) drag is zero when the stabs are not loaded and increases as the square of the coefficient of lift of the stab. The load on the stabs depend on the pitching moment of the wing and the CG placement. The aft CG placement depends on the neutral point. It is possible to make the tail big enough to be able to place the CG far enough aft to reduce the tails' vortex drag at the expense of increased parasitic drag. I haven't looked into where the balance of conflicting objectives falls but there is a satisfactory answer without considering if four stabs are too many.

[Ollie]

Craig,

There would be four wing panels of aproximately 30 inches span. The fuselages would be located near the center of each panel. Post number 16, first paragraph deals with a method for exact sizing of the panel spans and placement of the fuselages spanwise.

The parasitic drag summed across all four skinny fuselages will only be 3 or 4% of the total drag.

Please be a little more specific about what you mean by torsional loads on the tails.

[Craig Toutolmin]

I'm picturing it now. A fuse will be in the center of each panel. I'm a little slow sometimes.

There are significant twisting loads on the tail surfaces. Some planes are speed limited from this force. My old Pike Extreme is an example. When Gavin Baskin owned the plane he had the V twisting back and forth so much that each side hit vertical. This was at only 130 mph. In one case the V-tail was torn from a plane. The failure was with the two blind nuts. The threaded sections broke away from the flanges. The flanges were still intact in the fuse and the threaded sections were attached to the bolts of the v-tail. Loosening of the V-tail is a common maintenance problem. Many choose to permanently bond it to the fuse.

A common in-air failure of T-tails is that the vertical fin sub-spar delaminates from the skin due to twisting forces. The standard fix is to use a basswood subspar that is both glued and screwed.

[Ollie]

Craig,

There are no serious torsional loads on a tail boom with a cross tail configuration where about half of the vertical tail is above and below the tail boom. The torsional loads on a V- tail boom are a very good reason for avoiding V-tails. I suggest that two equal sized stabs be mounted on either side of each of the four fins about 1.5 inches above the tail boom so that the pushrod from the wing mounted servo has a straight shot through the tail boom to the bottom of the stab bell crank. If the stab is carefully pivoted at the aerodynamic center there will be very little load on the elevator servos and their couplings. The pivot bar should be a generous sized carbon rod of about 3/16 inch diameter to take the bending loads. It should be supported by double ball bearings in the fin, one on either side of the bell crank.

A beefed up version of Dr. Drela's V-mount for a one piece stab mounted ahead of the fin is another good way to go. Such a V-mount has drag but it is offset by the saving in interference drag and gap drag at the root of a conventionally mounted stab.

The vertical tail extending below the boom is vulnerable to damage on landing. I think a soccer goal type catch net for landing is a good idea for many sites with rough terrain. This could be bought in common and used and maintained by all the fliers at a site.

[hkrussell]

Hi Ollie,

Thanks for contributing to the subject. You've already raised some ideas that I'm sure are totally unique. (at least they are new to me)..

I have some issues with you thoughts and I offer them in a constructive way (hopefully). Please don't take this as a personal or professional attack, because it's not intended as that.

As I understand the idea, you are proposing a design that has very little actual "load" on it. Lift and drag of course, but the other forces will be neutralized/cancelled out by design.

Issues:
1) multiple fuses:
will they be connected at the tail end?
If not - you will have multiple tail sections going in multiple different directions because DS'ing is hard on parts and they tend to bend/deform/etc.

If so - the connected parts will tend to become unconnected as the multiple fuselages bend/flex/etc differently throughout the DS circuit.

connected or not - if you have multiple stabs, there will be a unique angle of decalage for each one because parts tend to bend differently through the DS circuit. Bad for controlled high speed flight.


2) minimalist spar constuction - My last spar was 1/2" end-grain balsa with .120 carbon fiber uni cap strips, so I am having a hard time with the .06 thick C/F you stated would be sufficient at speed.

How would that sort of spar handle negative 'G' loads? How would it handle simultaneous negative 'G' loads on one tip panel and positive 'G' loads on the opposite panel?

My above mentioned spar was built into a 100" flat wing, but if I ran a line that was too high and hit a bad spot in the circuit, the wing would bend anhedral (looked like a seagull's wing) about 6" or so if measured at mid-panel. Craig saw this at Parker one day. Pretty strong wing because that didn't kill the plane, but would the same thing happen with your wing?

Especially if one wing panel goes negative and the other goes positive with all that weight spread across the panels.

3) construction problems -

all those little wing panels means multiple control surfaces on the wing. Does that mean 6 servo wings? A pretty nice idea because it would reduce surface loads and eliminate fultter, but tough to get our radio's to do that.


I really like the discussion, getting nitty gritty with a fresh mind is good.

Ollie, thanks for your time.

[Ollie]

"Issues:
1) multiple fuses:
will they be connected at the tail end?
If not - you will have multiple tail sections going in multiple different directions because DS'ing is hard on parts and they tend to bend/deform/etc.

"If so - the connected parts will tend to become unconnected as the multiple fuselages bend/flex/etc differently throughout the DS circuit.

"connected or not - if you have multiple stabs, there will be a unique angle of decalage for each one because parts tend to bend differently through the DS circuit. Bad for controlled high speed flight. "

Answer: The center lines of the outer fuselages are about 90 inches apart. The turbulence which causes deflections would have to have substantially different velocities over a space of 90 inches to produce the effects of deflecting the tail booms in different directions. If that were the case, I think single fuselage designs would become uncontrollable in such turbulence. The inertial loads causing deflections will be the same for all four fuselages. In this case, the solution is the same. You just have to make the structure stiff enough that the deflections are small enough. This is definitely a problem but, it is definitely solvable. Under inertial loads the deflections will be in unison.

"2) minimalist spar constuction - My last spar was 1/2" end-grain balsa with .120 carbon fiber uni cap strips, so I am having a hard time with the .06 thick C/F you stated would be sufficient at speed.

"How would that sort of spar handle negative 'G' loads? How would it handle simultaneous negative 'G' loads on one tip panel and positive 'G' loads on the opposite panel?

"My above mentioned spar was built into a 100" flat wing, but if I ran a line that was too high and hit a bad spot in the circuit, the wing would bend anhedral (looked like a seagull's wing) about 6" or so if measured at mid-panel. Craig saw this at Parker one day. Pretty strong wing because that didn't kill the plane, but would the same thing happen with your wing?

"Especially if one wing panel goes negative and the other goes positive with all that weight spread across the panels."

Answer: If the mass distribution, and, therefore the inertial load distribution, exactly matched the lift distribution along the span there would be no spanwise bending load on the structure and no spar would be needed. To determine the bending load on a wing requires that the distance between an increment of lift and an increment of load be multiplied by the load to get the bending moment. When the distance is zero so is the bending moment.
With four fuselages, each individual load is four times smaller and with four wing panels associated individually with each fuselage the distances between lifts and loads are four times shorter. Therefore, the maximum bending moments along the span are 4x4=16 times smaller. If we had five panels and five fuselages the bending moments would be 25 times smaller!

The only reason that we use fuselages at all is to put the necessary nose weights out ahead of the wing. If we used zero or positive pitching moment airfoils we wouldn't need horizontal tails, much of a tail boom or as much nose weight. In such a case, we could come much closer to the ideal spanloader.

"3) construction problems -

all those little wing panels means multiple control surfaces on the wing. Does that mean 6 servo wings? A pretty nice idea because it would reduce surface loads and eliminate fultter, but tough to get our radio's to do that."

You can hook all four aileron servos to one channel with three Y-harnesses.
The same thing could be done for each of the other channels. So many digital servos will be quite expensive and the battery drain will be large but these arenot so much technical as economic issues. You could even use two receivers and associated batteries and eliminate two thirds of the Y-harnesses.

[Craig Toutolmin]

Using appropriate sections, a foamie prototype would be the next step. They suffer from aero elastic problems at lower speeds. Feedback on the design would come quickly and cheaply.

[Ollie]

I suggest using two layers of glass reinforced strapping tape at plus and minus 45 degrees to the spanwise direction on an EPP foam wing. See if you can get away with no spar. That should prove the spanloader structural advantage.

Teardrop shaped fishing sinkers would make good nose weights. Fiberglass or aluminum, 32 inch long arrow shafts would make good fuselages. A five foot span and 6 inch constant chord would be about right. Target weight would be about 25 ounces. Aileron and elevator control with four micro servos and a small battery should almost fit in the wing. It might be necessary to go to a 12% thick airfoil to imbed all the radio gear. The fuselages could be strapped directly to the wing with glass reinforced packing tape, spaced 15 inches apart. If the ailerons are at all effective at landing speed then the roll moment of inertia issue would be settled.

I suspect that the weakness that will show up is the need for more torsional stiffness in the wing between the fuselages, or maybe excessive spanwise bending in the wing depending on how fast the 1/2 scale model will go.

[hkrussell]

Sorry Ollie,
I think the plane will need a spar strong enough to soak up the abuse from turbulence. I understand the spanloader concept and I think it has merit under ideal conditions, but there are lots of turbulence and uneven air that we fly through.

In addition, as the plane is coming up the vertical line to the top of the circle, the wings see the air at different angles of attack and at different times during the circle. (High wing gets air first and higher angle of attack).

Minimizing the spar because there is 16 times less bending force on the wing is the same thinking that gets "moldie" flyers in trouble. They have fast race planes that are built with minimum structural strength, they aren't built for the stresses of DS and they generally come apart at slow speeds.

I understand your point though that the spanloader would be designed to minimize the loads, so the structure wouldn't need to be as strong.

[Ollie]

I think it's time to update the design and discuss structures and safety factors.

NACA 66-206, Cdo=0.0035 at Cl=0.2 for Re=3x10^6 (new airfoil)
Span 120 inches
Root chord 16 inches
Tip chord 8 inches
Four vertical tails of 36 square inches each
Four horizontal tails of 36 square inches each
Tail moment arm 60 inches
Wing weight 76 ounces
Radio weight 25 ounces
Total fuselages weight 20 ounces
Total V&H tail weights 16 ounces
Nose weights total 80 ounces (20 ounces per nose)
Gross weight 14.5 pounds
Wing loading 23.2 ounces per square foot
Maximum possible lift coefficient 1.0 at stall
Stall speed 24 MPH (not a problem launching into a 20 or 30 MPH wind)
The circle diameter at 250 MPH and Cl=0.2 is 263 feet and the circuit time is 2.25 seconds making it easier to fly than the first design.
The L/D is a respectable 33 even at the lower lift coefficient.
At 250 MPH and maximum lift coefficient, G=310 and at normal lift coefficient G=32.
The wing spar is designed for the maximum possible lift at 250 MPH.
Normal operation is at a five times lower lift coefficient so the safety factor is five for the spar.
The wing skin consists of four layers of 2.9 ounce per square yard unicarbon in laminating epoxy top and bottom over a Rohacell 71 core. The core has a compression strength of 219 PSI. The carbon is layed on at plus and minus 45 degrees for maximum torsional rigidity. This gives an additional bending strength safety factor of 2.5 for normal operation. Between the spar and the stressed skin bending strengths the total safety factor is 7.5 for normal operation and 1.5 for the worst possible case.
The tail surface structures consist of two layers of 2.9 ounce unicarbon at plus and minus 45 degrees, in laminating epoxy over Rohacell 71 cores.

That's about $400 in unicarbon, $50 in epoxy, $300 in Rohacell and $640 in digital servos. Probably over $1500 in materials alone. Then there is the cost of machining the cores or the countless hours to hand sand them to shape accurately. Rohacell isn't safe to cut with a hot wire. If the cost and labor could be divided among several people it wouldn't be so bad. All the design calculations and assumptions should be checked independently before starting on construction.

[Ollie]

Following up on the last post:

If a curing oven is available for the wing and tail panels, then considerable money and weight can be saved in the wing skins and the skin strength doubled by the use of 150 gram per square meter prepreg carbon. The cost would be less than $100. About 20 ounces would be saved. The reduction in weight of the back half of the wing structure would save a bit of nose weight.

The strength increase would improve the safety factor at a coefficient of lift of 0.2 and 250 MPH to 10 and the worst case safety factor to 2.

The tails could be skinned with two layers, top and bottom, of 50 gram per square meter prepreg carbon for an additional saving in tail weight and nose weight. This would also increase the strength of all the tail surfaces.

[Craig Toutolmin]

You are recommending a 60% laminar 6% thick foil design to operate at cl.2?
Will you need camber for cl .5 in the turns?

[Ollie]

Craig,

I'm designing on the basis that the plane be flown at a nearly constant angle of bank and a nearly constant coefficient of lift of 0.2. The only deviation would be to compensate for drift due to the wind. At the beginning of the flight the "circles" would be small enough for the time to complete a circuit to be about 2.25 seconds. As the speed increased the "circles" would get bigger while maintaining a lap time of about 2.25 seconds. I feel that this would be a good compromise between ease of accurately piloting the plane and maximizing the L/D for low energy loss per lap.

I, of course, having never flown DS, need your guidance in this. Do you think maintaing 2.25 second laps is something you could do while maintaining a nearly constant angle of bank? The reason I am advocating this flight plan is that I think it will keep the plane flying nearer peak performance over the course of many laps.

The reason I switched from the NACA63-210 to the NACA 66-206 is that it improves the Cdo at Cl=0.2 from Cdo=0.0048 to Cdo=0.003 for a 60% reduction in profile drag. The reason I shifted to to a lower lift coefficient is it makes the circumference of the "circle" bigger for easier piloting and for lower G forces.

If you would like to specify a flight path by shape, minimum lap time, etc. that you think would be more practical, I could design to your flight path specifications. The higher the lift coefficient, the smaller the circle and the shorter the lap time the higher the G forces for a given top speed. On the other hand, the higher the design lift coefficient, the better the L/D can be when picking an airfoil. It seems to me that we have a case of trying to find the best balance between conflicting design alternatives.

[Craig Toutolmin]

Maintaining a constant bank angle should not be a problem if the plane is flying fast enough. 2.25 seconds seems like a lot of time. I think this is good for controllable high speed DS. I can check some flying tapes and time the circles.

My main concern with the 6 series foil was with the performance at lower reynolds numbers. I don't have the coodinates for this foil to see what x-foil predicts.

How significant is air density? Cape Blanco is at sea level and cold, my local hills are around 1000' and cool to warm, and Parker is around 3800' from freezing to warm.

[Ollie]

For the airfoil coordinates, see:
http://www.pdas.com/sections6.htm#66-206

An altitude increase of 3,800 feet will increase the airspeed for a given Cl about 12%. In addition, a temperature increase of 15 degrees C will increase the the airspeed for a given Cl another 17% at Parker Mountain. Air density affects L/D hardly at all over the ranges of interest. All my calculations so far have been for a standard atmosphere (15 degrees C near sea level). See:
http://wahiduddin.net/calc/calc_da.htm
Lift and drag are directly proportional to density. For a given lift or drag coefficient, speed and wing area the force is directly proportional to air density.

[hkrussell]

I like the idea of using the speed of the circle to measure the average lift coefficient. I find it tough to measure the radius of the circle. It would be a great way of maintaining a constant Cl. for the foils.

Thinking about the video of Paul Naton's run at Kiona Butte with the Speed Runner (?) on the video, it seems he was using BIG circles. I'd guess those were about 2 sec. or so. SD 6062 foil right?

Ken

[Ollie]

One more datum for the design-speed circle: At 250 MPH, 263 foot diameter lap, lift coefficient 0.2, lap time 2.25 seconds and, G=32, the angle of bank is arccosine 1/32 = 88.21 degrees. This will undoubtedly require opposite aileron to maintain the angle of bank. If the bank angle and lap time is maintained constant and the circle size is allowed to grow as the speed, the coefficient of lift, and G's will also remain constant.

Every time the plane gets kicked in the tailfeathers the pilot makes the circle larger in proportion to the kick. A pair of lap timers each timing alternate laps give the pilot feed back on adjusting the circle size according to the lap time. If the time is less than 2.25 seconds increase circle size and vis-versa. Alternatively a metronome set to 2.25 seconds could provide the pilot feed back on lap time deviation and therefore, circle size adjustment needed. Helpers with sighting devices to check bank angle could provide additional feed back to the pilot. I think it would take a lot of practise and team work to make this work without distracting the pilot too much. With sufficient performance margin, small deviations from the ideal flight path would be of little consequence except to reduce the speed gain per lap a bit.

As pilot and team ability improve, it might be possible to use an airfoil with a higher working lift coefficient that gives shorter lap times with higher lift to drag ratios and better energy gained to energy lost ratio too.

BTW, if a radar gun is aimed tangent to the flight path and perpendicular to the wind direction, the gun will measure airspeed too because under those conditions, airspeed equals ground speed.

[hkrussell]

Does it have any effect on the calculations if the actual bank angle of the plane.

A DS circle is about 30-60 degrees off vertical (depends on the shape of the hill and the conditions). So where you have an angle of bank at 90 degrees, from practical experience I would expect the angle to be more like 20 degrees of bank on the bottom, 90 degrees on the verticals, and 130 degrees at the top.

Ken

[Ollie]

Ken,

As I understand it, the "circle" has to be inclined prependicular to the wind and also inclined in the direction of the wind so that the lowest part of the circle is somewhere between something like 30 and 60 degrees to the down wind direction. The angle of bank is determined by the component of the lift vector that opposes the weight of the model and the component of the lift vector that provides the centrifugal force. When the plane is flying at 32 G's acceleration toward the center of the circle the angle of bank relative to the plane of the inclined circle is arccosine of W/(WxG)= 1/32. Arccosine of 1/32= 88.21 degrees relative to the plane of the circle. This is only an approximation. Relative to the horizon the bank angle around a lap will vary from 88.21 degrees by plus or minus the tilt angle of the circle relative to the horizon. The approximation is a fairly good one because the centrifugal force is so much larger than the force that supports the weight. I should have made it clear that the angle of bank was relative to the plane of the circle. Because we naturally assume the reference of bank angle is to the horizon, it will take a lot of pilot practise to reorient himself to the plane of the circle when judging the angle of bank. To overcome the approximation I will need to know the tilt angle of the circle for various sites and wind conditions.

There are two tangents to the inclined circle that are horizontal. If the helpers position themselves so that they are sighting along these tangents they can judge the angle of bank relative to the horizon (the natural way). These bank angle observing helpers will be of great help to the pilot in reorienting his bank angle reference during practise flights.

I apologise for this very serious omission on my part.

This is getting fairly complicated. This complication may be one of the reasons why the world speed record hasn't been broken by means of DS yet.

[Ollie]

When the circle is tilted 60 degrees the vertical vector opposing the weight goes through an angular change of plus and minus 60 degrees in direction relative to the plane of the circle and the centrifugal force vector in the plane of the circle. The lift force is the vector sum of the other two vectors. The change in direction of the vector opposing weight only results in a 5.4 % change in the magnitude of the lift vector.

In the case of a circle tilted 30 degrees the lift force only changes 3.125% in magnitude over the course of the circle.

These percentage changes in lift, coefficient of lift and other parameters of the flight plan do not affect the flight plan appreciably but they do have a major affect on the pilots perception of the bank angle changes during a lap, at least until the pilot trains himself to perceive the proper bank angle as nearly constant relative to the plane of the circle.

[Craig Toutolmin]

DSing has a loooooong way to go in order to set the speed record.

I did some big round circles today in between turbulent conditions. While 2.25 seconds seemed like a lot of time, it turns out to be very nice.... just a little challenging, definately not booring. The plane is a moderate weight 82" plane going 135-145 mph. Circle bank angle was about 20 deg.

[Humbro]

Why is it that the composite planes all seem to be of conventional stabilizer and fin configurations when the EPP's mostly seem to be a flying wing planform.

To me it kind of makes sense to have a composite flying as a DS machine, there is no tail boom to flex, the elevons are so big that elevator response is good, there is less drag, cheaper to build etc.

Certainly the EPP wings designed for DS all seem to go pretty well… imagine what a well build composite one without all that twist, flex and flapping tape would be like. I’m sure it has been done its just you rarely hear of any.

Luke

[hkrussell]

Humbro,

Flying wing DS planes have been tried and although they look impressive, apparently they aren't as fast, nor do they handle as well. I've thought about building one, but there don't seem to be enough advantages to a flying wing.

Craig,

"Circle bank angle" is the angle from a vertical line?? At Shell right?

Ken

[Humbro]

Thanks hkrussell but I was hopeing for a bit more in sight.

From my understanding the only serious thing that my limit a flying wing is the lower tail moment (which can be easily fixed) and that the foils are a bit less efficient dur to the reflex.

As fas as an efficient planform they should be no less efficient than a conventional plane.

Luke

[bjaffee]

I had heard that the NCFM guys were working on a large (8ft?) composite/molded Bluto for a shot at the DS record. Haven't heard anything about it for several months though. Not sure what became of it, but it would have been interesting.

[Humbro]

If that’s the case then they must have faith that a wing can push the record, and they have the experience with wings as well as conventional.

Luke

[Ollie]

One way or another almost everything about DS from design to flying style comes back to the dominant forces. The centrifugal force and the lift force opposing it are many, many times bigger than any other forces.

The centrifugal force is produced by accelerating the model's mass toward the center of the "circle." The centrifugal force is distributed across the model as the mass is distributed. The spanloader approach to configuration attempts to match the lift distribution to the mass distribution. To the extent that this is achieved, the mass "floats" on the lift and structural stresses and strains are reduced or eliminated.

The plank wing configuration eliminates the mass and drag of the horizontal stabilizer which are both small advantages. On the down side, the lower moment of inertia of the "plank" in pitch makes it more responsive to gusts and control inputs. The low moment of inertia in pitch can be increased by putting smaller nose weights on longer nose moments. It should be possible to find a low drag airfoil with a slightly positive pitching moment coefficient that will perform as well as negative pitching moment laminar flow airfoils. The things to look for are airfoils that were developed for the tips of helicopter rotor blades which go fast and have positive pitching moment coefficients.

The reason it makes sense to refer the angle of bank to the plane of the "circle" is because the dominant forces lie in the plane of the "circle." If the dominant forces are to be accurately controlled by the pilot, then the pilot's orientation has to be the plane of the "circle." If the pilot were in the plane he would have the seat of his pants to help orient him to the "circle." the pilot isn't in the plane with the G force reorienting him from the normal horizontal reference. Pilot reorientation is a matter of lots of DS flying while trying to concentrate on the plane of the "circle" rather than the horizon as an orientation reference. This is a difficult exercise but it can be done.

[VinceS]

great thread guys...

although i dont have the techincal background im not finding it hard to follow


I have never DS'ed but i plan to as soon as i make that extra bit of money.


Ollie - you have loads of knowledge. We are priveledged to have you in this discussion

[Ollie]

I was able to down load an NACA report on windtunnel tests of some cambered and reflexed airfoils originally intended for helicopter rotor blades. See:
http://naca.larc.nasa.gov/reports/1946/naca-rb-l5k02/
The airfoils were tested at reynolds numbers of1.8 and 2.6 million. The NACA 8-H-12 airfoil looked promising with a pitching moment coefficient of plus 0.005, so, I designed a plank style, spanloader flying wing around it.

Span120 inches
Root chord 12 inches (R=2.34 million at 250 MPH)
Tip chord 8 inches (R= 1.57 million at 250 MPH)
Design Cl= 0.25
Stall Cl =1.3
Maximum possible lift at 250 MPH = 1733#
Design lift at Cl=0.25 is 333#
Lap diameter +252 feet
Lap time at 250MPH =2.16 seconds per lap.
Tail moment arms 60 inches
Nose moment arms 30 inches
L/D=35.7 (with 10% Vert. Tail area)
L/D=36.8 (with 5% Vert Tail area)
Estimated weight < 10 pounds
Estimated wing loading 19.2 ounces per square foot

Because the design lift coefficient is at the extreme low end of the low drag bucket of the 8-H-12 airfoil, lap times in excess of 2.16 seconds will seriously degrade the L/D because of sharp Cdo increase. The wind tunnel test at modestly high reynols numbers allow chord reduction and aspect ratio increase with confidence.

The advantages are that the down load on the tail boon is eliminated with the elimination of the horiziontal tail, making graphite golf club shafts for the tail booms an economical practicality. The nose weights are reduced by eliminating the need to balance the horizontal tail weight. Four costly digital servos are eliminated. Nose weights are further reduced by extending the nose moment arms. Aspect ratio is increased to compensate for the higher design lift coefficient. Wing area is reduced, reducing the cost of materials for the wing. The moment of inertia in pitch will be similar to conventional, tailed designs which should improve the pitch response to gusts compared to other plank configurations.

[Ade]

I have been thinking about this for a while... As somebody else mentioned the G forces from turning aren't the only forces to worry about. My relatively bendy Acacia F3F has been seen on many occasions with a rather nice S bend in the wing as she approaches and passes through the boundary. This doesn't appear to be to do with the fuz at all, I assume this is to do with different angles of attack as the model passes through the boundary.

In some ways the span loader system could actually make this worse?

I used to have a combat foamy that was built light, It didn't have a spar, if you taped ballast on the CG then a resonance would get setup and it would just flutter. However if you split the ballast into 2 lumps and put it on the CG about 1/2 way out on the wing it performed much better.

Now that is with the extra mass on something near the centre of lift. The problem I see is that you are going to have lumps in front of the centre of lift for each panel, As the angle of attack changes across different parts of the panel this is going to cause some acceleration in some direction or other compared to the rest of the wing. This could setup stresses similar to my foamy with the weight in the centre.

Ok... what about this... Keep the nose weight in the normal place but also place evenly out along the wing on chord wise centre of lift for the model.

Although the total weight of the model would go up. The loading on the spar would be more even across the span of the model rather that being concentrated on one section.

Its amazing what you think of when you have had a couple of beers. I am really not sure of the validity of the above... feel free to shoot me down in flames!

Ade (going to bed)

[Ollie]

Ade,

You can build the spar as strong as you see fit. It won't do a lot of good to over build the spar but it won't be much of a penalty for attaining 250 MPH. There will just be a few ounces of over built spar just going along for the ride. You could distribute the balance weight along the leading edge to good effect. The balance weight would have to be three to five times heavier because it would be on a shorter moment arm. It would help the fore and aft balance if you mounted the radio gear up inside the D-tube too. You could put some of that lead weight into an almost solid carbon leading edge, D-tube and spar back to about 20% of the wing chord if you really wanted excessive strength and stiffness and were willing to foot the bill for the carbon. The main draw back would be that the wing loading and stall speed would be so high that you would have to bungee launch. It might also be a bit hard on the soccer type catch net for landing. You would end up with a safety factor of 10 or more against the worst possible case.

[Craig Toutolmin]

Is here a happy medium between all of the balance weight in the LE and all of the balance weight in the fuse?

[Ade]

Well... the trade off is gonna be in the torsional stiffness needed I think, The further foward the nose weight the more leverage it has so the more torsional stiffness will be needed.

Ade

[Craig Toutolmin]

Unless it cancels out the torque from the weight aft of the aero dynamic center.

[Ollie]

The 8-H-12 airfoil has its aerodynamic center (AC) a bit farther aft than one might expect. The NACA report gives the AC location at 27.8 % of the chord rather than the usually assumed 25%. The center of gravity of the nose weight, fuselage, vertical tail sub assembly is positioned to balance the aft half of the wing weight, the subassembly CG will be ahead of the AC but not by any great distance, only enough to establish the desired pitch stability. The overall CG will probably be one to five percent of the wing chord ahead of the wing's aerodynamic center after the plane is trimmed out. The resultant torque by inertial forces on the wing will be quite managable. Even this managable torque can be reduced by distributing the wing mass more forward so that the mass of the aft half of the wing is much less than the front half of the wing. How much mass is concentrated in the forward quarter of the wing should be desided by balancing the confliction objectives of low landing and launch speeds associated with low wing loading versus structural considerations. Another consideration is the moment of inertia in pitch. Keeping the moment of inertia in pitch similar to a conventional configuration so that the plane will respond less to turbulence, requires that considerable mass in nose balancing weight be placed on a long nose boom. Concentrating the balancing weight in an over built leading edge structure instead of on long nose booms will reduce the pitch moment of inertia making the plane more responsive to gusts. The same thing can be said about the moment of inertia in roll. The greater the moment of inertia in roll, the less the roll response to gusts. I expect the handling in roll to be even groovier than a conventional configuration and at least as groovey in pitch.

Planes that were designed for F3B and F3F have low moments of inertia about the roll axis so that they can signal lift by responding to vertical gusts and can roll from level to highly banked as quickly as possible. This is contrary to the design objectives of an all out DS machine, both aerodynamically and structurally.

[Ollie]

For those that may still be having difficulty with the spanloader concept. All planes have zero bending moments at their wing tips. Picture four planes connected wing tip to wing tip. At the connection the bending moments are zero. The individual wings only have a span of about 30 inches and an aspect ratio of 2.5 to 3 making them able to carry their bending loads with a modest sized wing spar. At the same time, the tip vortices are canceled where the wing tips connect. This leaves only two tips seperated by ten feet for a reasonably low induced drag. The wing panels that connect together are designed to carry great torsional loads in case any stray forces due to turbulence try to twist the wings. The vertical tails will have their area split above and below the tail booms to cancel any torsional loads on the tail booms. The nose weights and tail weights will balance each other, cancelling most of the torsional loads on the wings. The proposed wing structure will be several times stiffer and stronger in torsion than any F3B or F3F designs. The proposed wings will withstand torsional or roll producing loads due to gusts that would make F3F or F3B planes uncontrollable. The proposed wing spar will withstand the worst possible case bending loads with a safety factor of two.The moments of inertia in roll and yaw will be comparable to a conventional configuration or more, giving the spanloader groovy handling qualities especially in roll.

[Ollie]

I've been thinking about a DS plane's response to gusts. Gusts involve a gradient in relative wind speed and direction. Here is what Dr. Drela had to say about entering a thermal (one type of gust).

"What does make the glider change pitch attitude is flying through a lift GRADIENT. This is because the lift gradient makes the wing and tail see effectively different AoA's, which is equivalent to a decalage change or an elevator input. The glider effectively gets up-elevator when flying into lift, and down-elevator when flying out of lift. The closer the glider is to neutral stability (aft CG), the stronger its lift-gradient pitch response will be. See the CG article at www.polecataero.com for a little cartoon of this."

I think this sort of analysis can be applied to gusts. It helps explain why planes designed for thermal duration (F3B and F3J), where signaling lift is a virtue, are at a disadvantage for DS.

The closer together the wings and tail, the smaller longitudinal gust gradients will be and the smaller the response to gusts. On the other hand, the moment of inertia goes up as the square of the distance between masses. Damping goes up as the square of the tail moment arm length. Tail area can be reduced in proportion to the tail moment arm length. Low response to gust favors high moments of inertia, small tail areas and long tail and nose moment arms. In the case of a plank type tailless configuration with high moment of inertia in pitch, you not only have reduced the distance between the two angles of attack to less than the wing chord, but you get all the other advantages as well.

If the graident of a gust is spanwise, then the wing tips will experience different angles of attack and the plane will tend to roll. Here again low roll rate associated with high moment of inertia in roll are an advantage in reducing gust response sa in a spanloader configuration.

[Craig Toutolmin]

Would the spanloader fuselages decrease in size or weight as they are placed along the span from root to tip to match the lift distribution of the wing?

I think I understand the spanloader concept. I need to figure out some of the details and more importantly finish my current project before starting another.

Another concern is that I think two types of planes might be necessary. One for big smooth circles and another for hard turning ovals.

Flew a new spot that Tim B turned me on to the other day. Big smooth circles worked ok (around 125), but a deep oval with very hard bottom and top turns bumped the speed into the low 150's (154 max) when the conditions allowed. The oval is a similar size to what you'll find a Parker (onshore, right side) and Blanco (Dave's line with the Wiz). It is about 75 yards deep with about a 40-50' radius turn.

Is there a camber changing foil that can adapt well to both conditions or will this be too much of a performance compromise?

[Ollie]

I would be concerned about the additional pilot work load when using camber momentarily on a turn that takes only a fraction of a second.

I think a simpler solution would be to select an airfoil that has a low drag bucket which covers the necessary changes in coefficient of lift. Such an airfoil will be a compromize that lowers the average L/D roughly 10% over a whole lap. This doesn't sound to me like it would be much of a limitation except in marginally low wind conditions. The NACA 66-209 would be a good choice for an oval flight path with an L/Do of about 60 from a Cl=0.2 all the way up to Cl=0.5.

The NACA 8-H-12 wouldn't be such a good choice because the drag rises sharply for coefficients of lift less than 0.25. I haven't been able to find a cambered, reflexed aitfoil that would do for a flying plank and an oval flight path. The best i have found so far is the NACA 66-015 which has a Cdo=0.0046 from Cl=0.2 to Cl=0.3.

There are at least two ways of designing a spanloader. You can devide the span equally among the panels and match the fuselage subassembly weights to the lift of the panel or you can make the fuselage subassembly weights equal and adjust the panel spans for equal panel lift. There is not a lot to choose between the two methods. In either case the wing lift distribution is the basis for adjusting the weights and panel lifts so that they match.

[VinceS]

I am not blessed with the knowledge ollie has, but I can think logicaly.
I understand the spanloader concept, but with such emphasis placed on the spanload concept, I cant help wondering if a wing would indeed be a better solution.

6 foot with only slightly swept wings, if at all. (the less the better)
high wing loading at the CG. Slightly thinner version of the airfoils we typicaly see in wings, large elevons, span loader concept built into the wing.

It just seems to make sence! (Unfortunatly I have nothing better to go on)

If my instincts are correct, you could turn much tighter circles while loosing less energy.

If indeed I am correct, how would you design such a wing ollie? (and others)

Think outside the box =D

I would however love to see your spanloader flying.. it would be a sight!

[Ollie]

Vince,

A swept back flying wing presents a couple of hurdles. One is that it is stealthy. It is hard to see edge on and seeing is a necessity at the speeds of interest. Another is that the lift distribution typically (ala the Horton brothers designs) has a bell shaped curve for stability. This lift distribution produces a lot more speed limiting induced drag even at low lift coefficients than an elliptical lift distribution. That is, the wing tips lift in the opposite direction from the rest of the wing so that the inertial forces on the wing tips add to, rather than subtract from, the aerodynamic lift. This force arrangement produces very, very large twisting forces on the wing structure. Irv Culver has proposed a twist distribution for swept back wings that has a more elliptical lift distribution overcoming many of the problems associated with the Horton lift distribution but I don't know if the Culver lift and twist distribution has ever been tried. See:
http://www.b2streamlines.com/Culver.html
The visibility can be improved with generous vertical tails at the wing tips ala the Swiss SB-13 but that just adds mass where there is no lift to support it. Another problem with a twisted wing, whether the twist is Horton type or Culver type, is that it limits the coefficient of lift range for low, over all, drag. When a swept back spar bends it introduces additional twist into a swept back wing. Because of this effect swept back wing spars have to be designed for great stiffness not just for strength.

The Swept wing problems can be mitigated by limiting sweep back angle. When you do that you need vertical tail area for lateral stability and end up moving the vertical tails inboard laterally. Once that process begins, optimization leads to the plank type flying wing spanloader which solves most of the problems.

[Craig Toutolmin]

Back to the Spanloader.

If the spanloader is to have dihedral, how would the tail surfaces be oriented? Perpendicular and parallel to the local wing spanward plane?

[Ollie]

Make all the horizontal tails oriented in the horizontal plane, parallel to a line between the wing tips and not in the plane of any wing panel with dihedral. Make the vertical tails perpendicular to the horizontal tails.

[Pismo]

Oh man, what happened. I was enjoying this thread so much and then it just stopped. Did you guys go and start building this thing? :D

Thanks for sharing all the knowledge! If there is anyone with some drawing talent out there I would love to see some concept drawings.

Thanks again & let's hear where it's at!

Jim

[Ollie]

I'll draw some conceptual sketches, photograph the sketches and try to post the results later.

[Ollie]

Pismo,
I drew up a pencil sketch of my concept of a spanloader, both with and without horizontal tail. I took a picture of the sketch with my digital camera set on the lowest resolution and tried to send it as an attachment to a post in this forum. The file was 123 K and about 3 K too big for this forum to accept.

If anyone is interested in this sketch, I will be glad to send it to them in a private E-mail. All you have to do is E-mail your request to ocwilson@gls3c.com.

To bad I don't have a CAD program to sketdh in.

[bjaffee]

Ollie,

Can you shrink it down and post it here?

[Ollie]

The contrast is poor and the notes are illegible. I'll redo it in ink and post a better sketch.

[Ollie]

This should be more legible and less distorted.

[bjaffee]

Wow, that really is radical.

One question (forgive me if this has already been covered), but does this actually distribute the total wieght of a single fuse over the wing, or do the 4 fuses actually add up to the wing carrying a greater amount of total weight?

I realize the spanloader concept would probably make an additional amount of weight versus a single fuselage be inconsequential, but I'm just curious.

[Ollie]

A single empty fuselage for a conventional ten foot span F3B type model with stiffened tail boom, will weigh about about 12 ounces. There is another 10 ounces of radio gear in the F3B model fuselage. The golf club fuselages will weigh about 5 ounces each for a total of 20 ounces. That's defecit of two ounces for the fuselage function of the DS model. The nose weight for the F3B reinforced fuselage will be about four ounces versus about 60 ounces of nose weight for the DS model. That's a premium of 56 ounces for the DS model. The net of all this is that the DS model fuselages and gear will weigh about 54 ounces more than the F3B type model. When I calculated the wing spar I took the additional 54 ounces of inertial load into account and still ended up with spar caps of 0.06 x 0.5 carbon for the maximum coefficient of lift at 250 MPH.

[Ollie]

Graphite golf club shafts 46 inches long (driver) are often available free of charge from golf club rebuilding operations. Such shafts come in a wide variety of weights, bending stiffness and torsional stiffness. The shaft forward of the joint would only use about 32 inches of the original 46 inch long shaft. If the free shafts from club rebuilders were missing a couple of inches of length where the club head was cut away, then the shafts' lengths would be about 44 and 34 inches. Most shafts will take a half inch diameter carbon joiner bar with a slightly loose fit and could easily be epoxied in place for a solid joint. To preserve the integrity of the shaft it might be best to run the control pushrods on the outside of the shafts using teflon tubing with 0.07 inch diameter or larger pultruded carbon pushrods. Teflon pushrod housings could be attached to the outside of the shafts by wrapping the asembly with light cloth in epoxy. This would avoid stress concentrations where the pushrod penetrated the wall of the shaft.

If shafts with high torsional stiffness were used it might be possible to move all the vertical tail area above the shaft and avoid the landing vulnerability of the large subrudder shown in the sketches. In that case, a vertical tail of lower aspect ratio would be an advantage in limiting the torque on the shaft with rudder deflection. It might also be possible to use an all moving vertical tail pivoted at the 25% chord line to limit the load on the rudder servo.

The possible variations on the spanloader theme are virtually endless.

One of the things I neglected to show in the sketch was the higher aspect ratio proposed for the NACA 8-H-12 airfoiled wing. In that case the root chord would be reduced from 16 to 13 inches and the tip chord reduced from 8 to 7 inches.

[Ollie]

The proposed NACA 8-H-12 wing has 20% less area than the NACA 66-206 wing so, each of the four vertical tail areas can be reduced from 36 square inches to 28.8 square inches. Using a low aspect ratio vertical tail with an average chord of about 4 inches and a height of 7.2 inches will will reduce the torque on the tail boom to managable levels allowing it to be mounted entirely above the tail boom. This makes an all moving vertical tail practical which in turn reduces the load on the rudder servos to near zero with an aerodynamically balanced vertical tail. With the elimination of the mass of the horizontal tail and a 20% reduction in the mass of the vertical tail the nose weights can be reduced. All this weight reduction in the point loads of the four fuselage sub assemblies reduces the bending loads on the wing spar too. The fuselage sub assembly weight reductions would increase the absolute, worst possible case safety factor to more than 2 and the safety factor for normal operation at 250 MPH to more than 10.

Like Don Stackhouse says," everything affects everything else."

[Ollie]

bjaffee wrote:

"One question (forgive me if this has already been covered), but does this actually distribute the total wieght of a single fuse over the wing, or do the 4 fuses actually add up to the wing carrying a greater amount of total weight?"

I don't think I answered this question adequately in an earlier post. The spanloader I sketched has four stiffer, stronger, heavier tail booms than a reinforced F3B fuselage. The back half of the wing is also stiffer, stronger and heavier. The tail surfaces are stiffer, stronger and heavier too. Off setting this somewhat is the removal of the mass of radio gear from the fuselage to the wing. Even so, much heavier balancing weights are required in the noses. The net effect is heavier for the four fuselage sub assemblies. Yet the spanloader is so efficient structurally that smaller spar caps can be used with the "ultimate dynamic soaring (UDS) machine."

The typical F3B machine can take 50 to 60 maneuvers. The UDS can take 300G maneuvers. The pilot can't guide any model accurately through a sustained 100 G maneuver because things are happening too fast for human reflexes to respond.

[Ollie]

I compared the maximum bending moment of the NYX F3F wing (asuming half the gross weight is in the nose weight, radio, fuselage and tail) with the maximum bending moment of the spanloader UDS wing. Even though the span of the NYX is about 8 inches shorter, the NYX weighs only 40% of the UDS, the NYX has only 60% of the wing area, the UDS has about 20% of the maximum wing bending moment of the NYX when both planes are at the same speed and coefficient of lift. This means that for the same size spar, the UDS can go about 2.24 times (square root of 5) as fast before folding the wings.

[keithsmith]

I just read this thread end to end for the first time...what a monster! Thanks for getting it started, Brett.

I have two points, and I'll make it quick.

Ollie, you said you hadn't DS'd. I've probably had no more than 30 sessions, so I'm still a beginner, but I can speak to two concerns you've brought up regarding pilot workload...

1) You were concerned about the pilot having to abandon the horizon as a frame of reference. If I'm understanding you, then from my limited experience, I can say that isn't a problem at all. When I fly circles, I pretty much ignore the horizon. It's all about the plane. I usually picture a disc. The two variables I play with are a) the inclination of the disc and b) testing racetrack vs circle patterns for the current conditions.

I figure if it's coming naturally for me, and I'm new, then a hardcore pilot with experience sure isn't going to have the horizon problem.

Alright, so far the post is going slower, and considerably more boring than I expected. I can juggle, but that barely helps in real life, let alone here...moving along...

2) Maintaining constant lap times is going to take a moderate amount of concentration, but isn't a show stopper. If conditions are good, and consistent, I've found myself settling into a cadence where the top and bottom 'whumps' happened at a pretty consistent pace. As the speed increased, the circles got bigger, but the laptimes felt about the same.

So, in summary...keeping the plane on a plane isn't so hard. Maintaining a constant laptime (once the pilot is comfortable and conditions remain unchanged) isn't so hard.

That's it for now, I've actually bored myself on this one.

Keith

PS: My kingdom for a radar gun the other day...I'll start a different thread for that one, though. This one is already has a worthy focus!

PPS: Ollie, your knowledge of aerodynamics is inspiring. I hope you can try DSing some time, it doesn't have to be heart stoppingly fast all the time!

[Ollie]

Craig and Keith,

It is reassuring to know that my proposals are not pushing the pilot work load too much.

Both of you have mentioned that something takes place at the bottom of the circle. Craig says that's where wings fail. Keith says that's where the model gets another "whump." I'd like to understsand what is taking place at the bottom of the circle. I have been assuming that it is no different than flying through the rest of the air below the shear interface.

[bjaffee]

You can get another whump at the bottom if the rotor (or perhaps some other mechanism?) is causing the air down there to actually blow back up the backside of the hill.

[keithsmith]

Ollie and Brett,

My understanding is that the bottom whump happens for the same reason the top one does, a sudden gain in airspeed (if you enter the dead air flying in a tailwind).

I've seen ppl begining their turn before they're in the dead air, so once they DO hit the dead air, they don't have the tail wind they once had, hence less of a whump.

If your exit line is directly into the head wind, and your entry line is done with the tailwind, the whumps should be equally audible, save for the fact that the pilot is usually further away from the lower one.

The two reasons that the 'lower' whump is usually perceived as quieter is because a) it's further away from the pilot, b) ppl mess up the entry line more often than the exit line from my observations (myself included if I get lazy).

The presence of a rotor isn't required for the lower whump, but will make it more apparent.

Once you get moving, the lower whump is hard to miss, and it's a god-send...it takes all the guess work out of when to turn :) I first experienced it with my Hades, and let out a bit woohoo after the session. Thankfully nobody was around.

It's amazing how quickly the whump fades away as you get further away. I've hit the layer with the plane 20 feet away from me, and it's much much quieter than when the plane enters 5-7 feet away. I don't know enough about acoustics, but it may be an exponential drop off.

Stand at the bottom turn sometime, you'll hear the whump loud and clear, just as though you were standing at the top. I have some footage of Reed at the Secret Spot (nor cal coast) at the bottom turn, and the sound is there on tape.

Keith

[bjaffee]

Keith,

Yeah, that's why I mention some "other mechanism."
because, I'm not entirely convinced what I've observed is nessecarily from the rotor. At the hill I do most of my DS'ing on, the downhill "whump" isn't always apparent, either in sound or and increase in speed. When it does occur, though, it occurs at a point much further down the hill then the uphill whump occurs on the other side.

I guess this could be a rotor. Or, it could be that the irregular shape of our hill just happens to pull the boundary layer down further on the downhill portion of the circle, making the re-entrance into still air occur further down than the exit from still air on the uphill side.

[keithsmith]

Brett, I think you called it just perfectly...separation must be further down the hill at that spot.

At the new site I'm flying at, the layer is just a shade below eye level, and the speed is manic (by my standards)

On the first day I flew there, it was blowing 25-30mph (estimated), when I started DSing, I was just afraid to turn hard because it was going so fast in so little time...I wasn't even warm yet! :)

Keith

[Ollie]

If I understand correctly, there is a shear interface at the top of the circuit and also at the bottom, at least for some sites and for some conditions. How dependable is this second shear interface at known DS sites? Are there DS sites or wind conditions that produce one shear boundary but not a second shear boundary?

[keithsmith]

Check out the picture on this site...I don't care much for the representation of the path, but the boundary depiction isn't bad...
http://members.tripod.com/douglasturner/id27.htm

There's only one boundary.

Visualize the most familiar looking part of the circuit, the bottom turn, with the plane coming back toward the pilot. The plane is in dead air. At some height above the ridge, just as it's about to pass the pilot, it crosses the boundary, from being under it, to being over it....whump #1.

The 180 degree turn is performed entirely above the boundary, so now we're in a tailwind, hauling more ass than when we started.

At some point on the downwind line, as the plane descends, it crosses that SAME boundary again. whump #2.

A 180 degree turn is performed in the dead air, and we're back to where we started, with the original airspeed + (2*windspeed) - speed lost along the way.

That's my understanding...and when I've been diligent with my lines and timing, I've gotten my best speeds.

Brett's point, I think, is that the height of the boundary diminishes at his hill at you go further back.

At some hills, the separation is best closest to the hill, and diminishes further back. Without a sickeningly difficult line, the pilot doesn't get a chance to make a clean entry to get the second whump.

Sorry if I ended up hijacking this thread...the original topic was fascinating, I didn't mean to change it. Ollie, if you'd like to start another thread, please go ahead.

Slightly more on topic, http://www.x-plane.com is a flight sim for the PC and Mac that comes with a pretty amazing aircraft designer. The aircraft could be mocked up and tested to some extent. The sim generates all sorts of stats that might be useful. It's capable of handling up to 8 wings and 8 fuses on a single ship if memory serves. It also comes with an airfoil designer. It can be downloaded for free. The flight sim stops running after 6 mins if you don't pay for it, but the foil and plane designer run forever.

Regards,
Keith

[Ollie]

I think I am finally starting to understand the stresses that the plane goes through when it traverses the shear boundary. I was assuming that the plane went through the shear boundary at a low enough grazing angle that the wind gradient wasn't all that abrupt. You experienced fliers were feeding me the clues but I wasn't putting them together in my thinking. The "whump" sound that Keith mentioned, is the result of a pulse of down wash behind the wing during the windward crossing of the shear boundary and the opposite when recrossing at the lee end of the circle. The failure of wings at the lower turn that Craig mentioned is probably due to a high, momentary negative /G load. The plane passes through the shear boundary in something like a fiftieth to perhaps a hundredth of a second. If the shear boundary is sharply defined, the wing will experience up to a nine or ten degree change in angle of attack in that very short time. The wing and tail are in different parts of the gradient and will experience different angles of attack than their relationsip through decalage. This can highly stress the horizontal tail and bend the tail boom. It happens so fast that the plane doesn't change pitch atitude much because of its rotational inertia. This may also explain why short coupled plank configurations seem to be speed limited. In this case the moment of inertia in pitch is considerable less and the plank changes pitch attitude more when traversing the shear boundary. The plank's change in pitch atitude has a breaking effect due to the momentary increase in drag associated with the pitch change. As hkrussel pointed out, the wing tips experience different angles of attack when passing through the gradient of the shear boundary. Up until now all my thinking has been based on a kind of dynamic equilibrium and I wasn't taking into account the transients at the shear boundary. I'll have to read up on transient lift phenomenon and do some more thinking.

So far, I'm beginning to question the advisability of a thin airfoil with a narrow low drag bucket. It may be that thicker airfoils with wider low drag buckets will experience less drag during the transient change in angle of attack. I think I must go back and reengineer the tail boom stiffness and horizontal tail bending strength. I think I must take a closer look at the torsional loads on the wing caused by four seperate horizontal tails experiencing different angles of attack because of their locations in the shear gradient. My notion that high moment of inertia in pitch and roll is an advantage holds up for now. Also, I think the spar design holds up.

[bjaffee]

Originally posted by Ollie
I think I must take a closer look at the torsional loads on the wing caused by four seperate horizontal tails experiencing different angles of attack because of their locations in the shear gradient.

I was wondering about this as well. It isn't too uncommon to see a plane suddenly roll in the direction of the hill quite violently if it's turned to early in the boundary layer at the top of the circle. Our assumption has always been that this is caused by the gradient of the layer, with the higher/outside wing seeing a higher windspeed then the lower/inside wing. I imagine this could put quite a bit of twisting load on the opposite sides of the span loader.

[hkrussell]

I was wondering if the wings failing at the bottom of the turns was because the plane is carring more actual 'weight' on the bottom than on the top, hence it fails there first.

I'm thinging it carries more weight because the plane is essentially in a zero/negative 'G' situation on the top of the turn vs. a heavy positive "G" on the bottom turn.

Whumping.

I would agree that Whumping is caused by going through the boundry layer a second time. Under DS conditions the plane will whump on the bottom turn also. Might not hear it if the wind is too strong, but it's there.

I've also heard a plane make a 'ripping' sound on the bottom turn and was told it was because the plane couldn't make it under the boundry layer and was flying through the boundry layer.

Ollie,
Sorry if I wasn't clear in my explanation of the torsional loads on a DS plane going through the boundary layer. I was just ass/u/ming...

Keith,
I remember you...

Ken Russell
Shell Ridge Sissy Boy...

[bjaffee]

Originally posted by hkrussell

I'm thinging it carries more weight because the plane is essentially in a zero/negative 'G' situation on the top of the turn vs. a heavy positive "G" on the bottom turn.


Hrrmmm...I tend to think the plane is in a pretty serious positive g situation on both the bottom and top turn. I suppose there might be 1 or more g's on the bottom vs the top, so the plane might be pulling 19g's instead of 17 or 18g's. I guess that could make a difference though.


I've also heard a plane make a 'ripping' sound on the bottom turn and was told it was because the plane couldn't make it under the boundry layer and was flying through the boundry layer.


I've heard this sound as well (just today, in fact) but always assumed that this was the rotor everyone always talks about, because it usually occurs quite far down the hill. I've always associated it with improving conditions, since whenever i hear it start happening, I know I'll be going much faster in the space of a lap or so.

[Ollie]

Here is what Robert T. Jones says about gusts in Wing Theory, Chapter 4, pages 61 and 62.

"A thin symmetrical airfoil was weighted and balanced so that its center of gravity was exactly at the 25% chord point. The airfoil was dropped from the high ceiling of the laboratory through the open horizontal jet of a wind tunnel. Evidently moving through the jet without rotation."

Perhaps this is why we don't see an abrupt change in pitch attitude when a plank wing hits the shear boundary at high speed.

" Starting the airfoil from rest with infinite acceleration would result in an infinite force if the fluid were incompressible. In a real compressible fluid, such a start will generate sound waves from the upper and lower sides of the airfoil - - -."

Perhaps this is analagous to what causes the "whump."

[Reed]

I don't mean to self-aggrandize, but I am something of an expert re: the aforementioned "whump" phenomena. Clearly, when DSing, the "whump" is caused by the airframe impacting the planet's surface. The various audio levels described by Keith are simply a matter of groundspeed; e.g.: 80»0 mph groundspeed expresses lower db than 120»0 mph. Other important "whump" factors are AUW and percentages of carbon, S-glass, kevlar, ply and balsa used in the airframe's structure. 5.7 oz carbon on a bias with an endgrain balsa, carbon-capped spar expresses very high frequency "whump" (aka "whimp"). Balsa built-up fuselages with balsa sheeted, white foam wing structures tend to express a lower frequency "whump", although it should be noted that these structures provide a very satisfying additional tone often described as a "whump-crunch."

There are so many other factors, of course, including recent rainfall and local vegetation, but this should suffice as a rudimentary introduction to the "whump" phenomena.

[Ollie]

Reed,

When the four fuselages of the spanloader are properly tuned, they can play a low frequency, somewhat atonal, chord known as the whimp, whamp, whump and thump on earth impact. When impacting flesh, the chord is usually drowned out by a blood curdling scream unless, of course, it happens to be a head shot.

[Reed]

mmmmmmmm, headshots.

[hkrussell]

Ah yes Reed, I understand now!!
my planes go Thump or Thud. Recent rainfall and vegetation doesn't seem to matter, except it limits the rebound and contains the debris field.

Ken

[Reed]

Sorry for the detour, I just wanted to contribute something to the conversation.

Actually, I'm interested in something Tom Seitz was talking about the other day. He suspects that when the plane unloads coming out of a hard bottom turn there may be a sort of shudder in the wing like if you bent a ruler and let it go suddenly. Does the description make sense? In this case, there would be a lot of very sudden positive and negative bending stresses along the wing. Not sure how this relates to the spanloader but it could be a reason that so many wings fail at that point in the circle as spars break loose and skin delams. Might also have some impact on control at that point, even if the wing doesn't fail structurally.

Thoughts?

[Ollie]

Many F3F and F3B wing spars are biased for positive G's with more carbon in the top spar cap than the bottom spar cap to take advantage of the carbons greater strength in tension than compression. When transiting the shear boundary at the bottom they are momentarily subject to negative G's I think. This might be one possible cause. I think it would be well to keep the spar caps the same size and allow an adequate safety factor even for worst cases.

[Reed]

hkrussell said the same thing. Sounds like with DS we might consider building with symmetrical spars rather than the usual bias toward positive Gs. Craig, if you're listening, did you consider this when building your Mofo? If so (and I assume so), did you make any adjustments in spar design?

[Craig Toutolmin]

I designed/built for a smoother flight pattern. No turn banging. Caressing is a more accurate term for what I have in mind.

I have resigned to listening to the design process of the span loader in this thread. In this thread design parameters based on a traditional airframe have been shunned. I have been told that the cubic loading basis for scaling up the Circle Jerk to the Mofo is "dead wrong". I have committed to this traditional airframe design with "enhancements" and intend to fly it to its potential over the comming months.

I'm open to the span loader concept and infact have already applied the concept to the traditional airframe. I think the current state of the design would last about three laps in moderate air before the flying fork looked more like a high amplitude tuning fork. When the concept gets refined perhaps we'll have a "building the ultimate DS plane" thread. A progression of smaller prototypes can be built and compared to current designs. You'll know before 100" if the concept has promise.

[Ollie]

Foam stiffness measurements by Mark Drela and Phil Barnes indicate that Spyder foam, Hi Load 60 and Hi load 100 have about the same or better stiffness than Rohacell 71. These polystyrene foams are less dense, less expensive and can be hot wired. That makes them better in every way than Rohacell as wing cores for the spanloader.

[Reed]

Anyone still working on this stuff?

The Mofo has proved itself over 200mph without reaching its potential. I hope I'm there when Craig gets a chance to wring it out sometime in the fall.

Spanloader? Is that still happening?

I've finally started getting serious about my little DS monster. My planform keeps the spar straight along the quarter cord which, without completely understanding why, I assume is a good thing. The layup will include both uni-di CF and biased CF as well as biased FG. 74" wingspan for reasons of transport, strength, agility and building ease/speed. This will be my first bagged wing so wish me luck.

Trying to figure out an ideal tail moment - any ideas?

[Ollie]

The original motivation for the spanloader was to avoid structural limitations that may eventually crop up as model size increases in search of a higher L/D. It appears from Craig's MOFO and its outstanding performance that the structural limit is still out there somewhere and there is not yet sufficient motivation to seriously continue the spanloader development for now. The risk of running into serious problems with a radical design isn't waranted as long as progress is being made with a conventional approach in my opinion.