How much sweep is too much sweep?

[Mihai]

I'm designing a swept flying 'wing for thermaling, so efficiency is of utmost importance. So far many things seem to improve with an increase in sweep - wing loading, size of fins, amount of wing twist etc. However, I understand that there is such a thing as too much sweep - after some points the efficiency of the wing starts to decrease. Where is this point, and why is the efficiency decreasing? Any way to quantify this decrease in efficiency? Will XFLR5 help?

Thanks,
Mihai

[nmasters]

Here’s a graph showing the induced drag increment due to sweep (http://www.rcgroups.com/forums/attachment.php?attachmentid=1666994) . Sweep forward is on the left. Notice that the top of the curve is not at 0 degrees sweep. You actually get an induced drag bonus up to 5 degrees sweep back. At 10 degrees the induced drag of a swept back wing is the same as a wing without sweep.

--Norm

[Brandano]

I think 90 degrees sweep is too much, YMMV. I suggest you also have a look at
http://www.continuo.com/marske/ and in particular at http://www.continuo.com/marske/butterfly1/flying%20wing%20theory.htm and

[Mihai]

Norm, how is the sweep defined in that graph? Is it the angle between the center line and the quarter chord span line?

That graph looks OK until about 25-30 degrees or so.
Thanks,
M.

[BMatthews]

It would be to the 1/4 chord line. It's all I've ever seen when discussing wing sweep in swept parallel chord wings or when looking at the effects of tapering the chord.

Choosing airfoils with inherently positive pitching moment at the tips and small negative pitching values for the root would go a long way to reducing the need for strong sweep angles as well as reducing how much washout you need. A lot of the MH airfoils are designed to minimize the negative pitching moments. And the MH series also contains some positive pitching moment airfoils.

[CloudyIFR]

Mihai,

Have you seen my spreadsheet that uses the late Dr. Walter Panknin's calculations for determining wing twist?

It's available here:

http://h1.ripway.com/cloudyifr/files.htm

Curtis
Montana

[Mihai]

Curtis, I've seen it and used it before, but for this wing, I had to implement it myself - the washout it not big, as I use a very small pitching moment airfoil, so washout savings are minimal, but I need the sweep for yaw dampening and reduce the need for tip ballast - it seems that the more sweep, the better in this areas.

[John235]

I have a feeling the graph dosent tell the whole story. As you probably know already, for thermal swept wings you generally need significant twist. Increasing the sweep means increasing the twist. Since the twist is only ideal for one AoA (alpha value) you will get increased drag for other speeds. My reading is that for a thermal wing, you have to trade-off the efficiency in a certain speed range when you increase the sweep.

[MarkusN]

...So far many things seem to improve with an increase in sweep - wing loading,...
If you consider that (http://www.rcgroups.com/forums/showthread.php?t=942105), wing loading improves only theoretically. As only the normal component of the speed contributes to lift, the effective wing loading remains the same. But wetted surface increases. So you actually pay a drag penalty.

[nmasters]

Norm, how is the sweep defined in that graph? Is it the angle between the center line and the quarter chord span line?

That graph is figure 2.20 from “Tailless Tale”. The formula with it appears to be using ¼ chord sweep. Figure 2.23 in the same book shows a similar graph of the CL increment.

That graph looks OK until about 25-30 degrees or so.
Thanks,
M.

He has worked examples for 30 degrees sweep that show a 15% increase in induced drag and a 16% decrease in Cl for a given AoA. So although the slope of the curve isn't very steep on either graph they both show an undesirable change in opposite directions ie the theoretical 3D L/D goes into the toilet. Fortunately sweep does not decrease Clmax, only makes the lift curve shallower and Cdi isn't a big problem at high speed. What this says to me is that there should be little effect on the high speed glide but minimum sink will be higher.

Since you thermal at minimum sinking speed high sweep appears costly to thermallers. Since 'wings have to have more wing area than conventional planes of the same weight and speed anyway just make the span greater. The wetted area goes up a bit but wing area doesn't make as much form drag as the same amount of fuselage area :cool:

--Norm

[Mihai]

If you consider that (http://www.rcgroups.com/forums/showthread.php?t=942105), wing loading improves only theoretically. As only the normal component of the speed contributes to lift, the effective wing loading remains the same. But wetted surface increases. So you actually pay a drag penalty.

Sorry, I was too brief - what I really meant was that with a swept wing I'll need less ballast in the nose to balance it, so the overall plane weight reduces for the same wing area, so the wing loading reduces (only due to the reduction in ballast). At least this is how numbers come up for my wing (and it seems to be the case for others, as planks are notoriously tail heavy without ballast, while my Bee, for example, is nose heavy without ballast :eek: ).

M.

[Mihai]

That graph is figure 2.20 from “Tailless Tale”. The formula with it appears to be using ¼ chord sweep. Figure 2.23 in the same book shows a similar graph of the CL increment.



He has worked examples for 30 degrees sweep that show a 15% increase in induced drag and a 16% decrease in Cl for a given AoA. So although the slope of the curve isn't very steep on either graph they both show an undesirable change in opposite directions ie the theoretical 3D L/D goes into the toilet. Fortunately sweep does not decrease Clmax, only makes the lift curve shallower and Cdi isn't a big problem at high speed. What this says to me is that there should be little effect on the high speed glide but minimum sink will be higher.

Since you thermal at minimum sinking speed high sweep appears costly to thermallers. Since 'wings have to have more wing area than conventional planes of the same weight and speed anyway just make the span greater. The wetted area goes up a bit but wing area doesn't make as much form drag as the same amount of fuselage area :cool:

--Norm

After sleeping on it, I decided that 30 degrees is too much, now I was debating between 20 and 25 degrees. After you told me that both Cl and Cd suffer, I'll go back to 20 degrees (and a bigger chunk of lead in the nose, or a big battery :D ).

My span is limited by F3K rules to 1.5m, so this is all I have to work with,
but I'll make do :-). In all honesty at 1.5m it already doesn't fit in my trunk, so any bigger and I'll have to buy a new car.

M.

[Mihai]

Increasing the sweep means increasing the twist.

Not true - quite the opposite - increasing the sweep means decreasing the twist (assuming you have a negative pitching moment with high Cl). However, this doesn't apply to my case as I have an airfoil with very small negative pitching moment (may even be positive for some regimes, I need to check that), so my twist is very small to start with.

M.

[BMatthews]

The other way to decrease the washout with a lower angle sweep is moving the CG back closer to the dreaded neutral point. Keep in mind that the Panknin spreadsheet calls for you to fill in the CG.

And this is the real issue with swept flying wings. Unless you make a more open framed model that can be re-warped to adjust the washout angle then you're stuck with the CG that works with the angle other than a very minor amount of adjustment. Radially altering the CG results in larger than desireable trim inputs on the elevons and still leaves the airfoils working at less than ideal angles at some points on the wing.

It would be work making up a small all balsa test glider and playing with varying CG locations and washout angles to find a setup that results in guaranteed positive stability but only just barely. Higher speed diversions could be tolerated due to the ability with the RC verson to feed in elevon control. And of course the closer you get to the neutral point the more efficient the model is. The only real issue is retaining enough stability for the pilot to fly at longer ranges.

[John235]

Not true - quite the opposite - increasing the sweep means decreasing the twist (assuming you have a negative pitching moment with high Cl). However, this doesn't apply to my case as I have an airfoil with very small negative pitching moment (may even be positive for some regimes, I need to check that), so my twist is very small to start with.

M.

You are correct based on stability criteria, but there is more than one way to design twist for a swept wing. If your goal is to achieve an elliptical lift distribution, you will need more twist with increased sweep angle. I am not an expert on swept wing theory, but I think the best choice for sweep angle would consider both lift distribution and the pitch stability. It seems to me the link to the graph in the second post isn't very meaningful unless there is some more infomation about the twist distribution used.

EDIT: Apart from the issue of lift distribution, if you don't have enough twist on a swept wing, the stall behavior may have a problem due to pitching up when the tips approach the stall.

I apologise if I seem to be pushing this point, but I have previously read that a small amount of forward sweep had the lower induced drag. I wonder if the positive angles on the right hand side of the graph (http://www.rcgroups.com/forums/attachment.php?attachmentid=1666994) may be for sweep forward.

[Mihai]

You are correct based on stability criteria, but there is more than one way to design twist for a swept wing. If your goal is to achieve an elliptical lift distribution, you will need more twist with increased sweep angle. I am not an expert on swept wing theory, but I think the best choice for sweep angle would consider both lift distribution and the pitch stability. It seems to me the link to the graph in the second post isn't very meaningful unless there is some more infomation about the twist distribution used.


Hmmm... I didn't quite consider using the twist for lift distribution - I was thinking to use the twist for pitch stability, then use the taper for lift distribution - I'm shooting for a bit less load at the tips, so I have a bit of cushion for tight thermal turns.



I apologize if I seem to be pushing this point, but I have previously read that a small amount of forward sweep had the lower induced drag. I wonder if the positive angles on the right hand side of the graph (http://www.rcgroups.com/forums/attachment.php?attachmentid=1666994) may be for sweep forward.

Actually, I'd like to double check the information in this graph if possible - although overall it looks reasonable to me (I have no idea if slight swept forward or backward is a bit more efficient than exactly straight), I'd like to understand more where that is coming from - to me it seems that the shape of the graph should depend (a lot) on the aspect ratio of the wing, and possibly on the existence of winglets. Unfortunately a short Google search came up with nothing useful.

Best,
Mihai

[Mihai]

The other way to decrease the washout with a lower angle sweep is moving the CG back closer to the dreaded neutral point. Keep in mind that the Panknin spreadsheet calls for you to fill in the CG.


Oh, about that - in Panknin's formula (and in computing the ballast) I put the stability margin at 3% without any significant motivation for it - is that a reasonable number (I usually like to fly with an aft CG for thermals)?


It would be work making up a small all balsa test glider and playing with varying CG locations and washout angles to find a setup that results in guaranteed positive stability but only just barely. Higher speed diversions could be tolerated due to the ability with the RC verson to feed in elevon control. And of course the closer you get to the neutral point the more efficient the model is. The only real issue is retaining enough stability for the pilot to fly at longer ranges.

Nah, no balsa planes for me - I'd rather build several composites than do balsa. I may end up doing this, but I'd like to get as close as possible to the target in the first (few) step(s). I realize that theory will only get you as far if you don't spend major time on computing lift distributions when thermaling and the corresponding aileron deflection, and/or if you don't have the right simulators and the knowledge to use them (like Dr. Drela has).
Actually I was debating with myself if I should just stop designing and start building or if I first should go to XFLR5 to get (maybe) even closer to the target, and then start building. I'll have to learn how to use it first, but that is something I'll have to learn eventually anyway.

Thanks,
M.

[CloudyIFR]

Oh, about that - in Panknin's formula (and in computing the ballast) I put the stability margin at 3% without any significant motivation for it - is that a reasonable number (I usually like to fly with an aft CG for thermals)?



I've very new at this thermal flying wing business but when I build Tinamou, see here: http://h1.ripway.com/cloudyifr/files.htm

I found that when I had the CG at 3% static margin that when I slowed for landing she got very pitch sensitive with the high AoA. I've since moved it back to 6% and she's flying in all realms of flight very well.

Mihai, once you get things dialed in I'd be very interested in a quick writeup of what your final design is; area, wing loading, span, airfoil, twist, tiplets/fins, how you plan to launch etc.....

PS I'm sure you know but I see this posted in error quite often but your wing loading will not be like you figure for a tailed airplane, as the outer portion of the wing is producing the down force that's counter acting the negative pitching moment of your lifting airfoil, how much, well I guestimate by reducing the overall wing loading by 30% and call it "effective" wing loading.

Curtis
Montana

[CloudyIFR]

Mihai,

The flying wing spreadsheet only allows for a single tapered planform, that's all that Dr. Panknin provided. I have put together a spreadsheet that will allow you to use a three panel wing. I haven't posted it for download as the formulas for determing the taper ratio as far as I can tell are accurate, but I'm not fully certain how it affects the Panknin formula for twist.

If you'd like to mess around with it and report your findings send me an email and I'll send it to you. It might take me a few days as I have to do some minor cleanup on the file.

Curtis

[Mihai]

I've very new at this thermal flying wing business but when I build Tinamou, see here: http://h1.ripway.com/cloudyifr/files.htm

I found that when I had the CG at 3% static margin that when I slowed for landing she got very pitch sensitive with the high AoA. I've since moved it back to 6% and she's flying in all realms of flight very well.


Oh, I'm sure this will increase the required twist.


Mihai, once you get things dialed in I'd be very interested in a quick writeup of what your final design is; area, wing loading, span, airfoil, twist, tiplets/fins, how you plan to launch etc.....


I'll be sure to do that! It will take a bit though as I don't have all the things I need, but I'm close (I'm missing the vacuum bagging setup, but I have the pump :-).


PS I'm sure you know but I see this posted in error quite often but your wing loading will not be like you figure for a tailed airplane, as the outer portion of the wing is producing the down force that's counter acting the negative pitching moment of your lifting airfoil, how much, well I guestimate by reducing the overall wing loading by 30% and call it "effective" wing loading.

Curtis
Montana

This is an interesting observation - it didn't quite occur to me, but this is true - even with a small twist the outboard panels will do less work than the center, to the center will have to pick it up, so higher wing loading - darn! I was doing so well! :). This is probably at the core of the observation that "Flying wings at the same wing loading fly faster than conventional planes" - because the "effective" wingloading is higher although the "mathematical" (by the definition) wingloading may be the same.

Thanks for the offer for the multi-panel spreadsheet, but I'm resolved to have the first wing use a single panel (and likely the same airfoil at tip and root, but I'm not as resolved about this :D.

Best,
M.

[CloudyIFR]

....... but I'm resolved to have the first wing use a single panel (and likely the same airfoil at tip and root, but I'm not as resolved about this :D.



Hmm, well I think you can get away with this if you're using a very low pitching moment airfoil, i.e. low Cmo. Such as the PW series or airfoils or even some AG airfoils....

The reason you twist the wing is that the outer portion of the wing is there to counteract the negative or nose down pitching moment of your lifting airfoil. A conventional airplane does the same thing but with the tail.

So what I did was use a fully symmetrical airfoil that has a zero pitching moment in my outer tips and added washout or twist to counteract the negative pitching moment of my lifting airfoil

Then I split my trailing edge surfaces so that the outer elevons were my roll and pitch control and the inboards were flaps.

You're a few months ahead of my plans as I'd like to try a DLG flying wing, however, I'm still not past the idea that the flying wing with the reduced "effective" wing area and the restricted span that effeciency can be gained in the Cl or launch height.

It's very interesting to watch what you're doing!

Curtis

[John235]

Mihai, you asked about XFLR5.. Personally I think its a great program. I can't comment about how accurate the results are, but I think its possible to learn a lot about the effects of the various parameters in wing design. My advice would be to start entering the data for an existing design. For example why not compare the two designs presented in this document. (http://www.glide.net.au/on-the-wing2/101-Two-HLGs.pdf)

XFLR5 won't do anything unless you start the analysis by doing full batch analysis to generate the polars for the airfoils over a full range of reynolds numbers (ie 20,000 to 1,000,000) for alpha values of 0.25 degree increments. This can take some time if your PC is not the very latest technology, but it is important not to take any shortcuts with this step. You should be able to find the airfoil co-ordinates in .dat files required for input at various places on the web. The airfoil analysis is done by selecting the "Xfoil direct analysis" from the application menu of XFLR5. The polars calculated are then stored in the .wpa project file together with the configuration data for all the wing shapes and everything else that you will enter as part of the analysis. Its a good idea to make a backup copy of the .wpa file every week or so in case the worst happens.

I also suggest you read the excellent tutorial (http://www.rcsoaringdigest.com/pdfs/RCSD-2008/RCSD-2008-02.pdf) that was published in RC soaring digest in Feb 2008.

[HerkS]

Mihai - twist and stability

Twist and/or reflexed airfoils don't actually confer stability.

Stability comes from having the CG ahead of the Neutral point.

Twist or reflex establish trim at a specific flight speed (design point) with the elevons at neutral. For example a high speed slope ship would have low twist or minimal reflex to trim out at high speed, whereas a thermal model would have more twist or reflex to trim out at slow speed. The CG being ahead of the NP creates a nose down moment. That has to be counteracted by aerodynamic forces established by twist, reflex or elevon trim. The larger the distance between the CG and the NP, the greater the compensation needed (at the design speed.)

[Mihai]

Curtis, I've been sitting on this idea for a while, but now I just got around to put the numbers in a program to optimize the planform and see what kind of wing loadings I can get - they're good even when considering the loss efficiency due to twist.

John, I'll postpone starting to cut foam to learn the ropes of the XFLR5 - I saw that tutorial and it looked like an awesome program. However, I am not very sure what I'll learn after I model my wing in there - maybe the lift distribution after I deflect the ailerons and try to optimize for that (i.e., to avoid tip stall), but I'm not even sure how much I need to deflect the ailerons for a "tight" turning ratio.

Herk, you need both having the wing with an overall positive pitching moment and the CG in front of the NP for pitch stability - other conditions for other types of stability. Honestly I'm not sure at what speed to optimize it because I want it to launch high (at high speed, with small drag), and they fly slow and efficient. However, it seems to me that, given the shape of the drag bucket, if I optimize for slow flight, the high speed will not have a high drag, so that's what I'll shoot for.

M.

[HerkS]

Mihai > high speed, with small drag

Every design is a compromise. You have choices. Design for the high speed of the launch - low twist - low reflex, design for thermal flying lots of twist or reflex, or design to something in-between - like good L/D for penetration and lift search.

Fortunately you can design to one point, but use long span elevons to adjust the effective reflex in flight for the condition you want. If it was me, I'd probably build it biased toward something between cruise and launch phase and retrim the elevator function of the elevons to get the reflex I need for varying phases of the flight.

[nmasters]

You also have options for landing flaps.

[Mihai]

If it flies the way I want it, I'll give it landing flaps - the AUW will probably not go up more than 2-5 grams as I can use the flap servo as ballast in the nose. First test will be without flaps.

I put the wing in XFLR5 and it looks sweet :). I also played with airfoil analysis - not quite sure how to compare different airfoils (in the 2D Xfoil simulation results), but I'm getting there. The first victory was in fixing the airfoil as Xfoil kept quitting after not converging in 100 iterations. Found the bug in the airfoil and fixed it, now it cranks up numbers.

Best,
Mihai

[Mihai]

Crap, XFLR5 cannot handle twin fins at the tip of the wings - it lets me define them, shows them nicely to wet my appetite, but then fails (due to a singular matrix somewhere in its calculation) when I actually want to compute the polars. At least I was able to figure out that the lift distribution seems "safe" for all relevant angles with only a small amount of twist.

M.

[John235]

Mihai, It seems you have made a good start so far. I think the most useful thing you can do with XFLR5 in the early design stage is to check the pitch stability. You can follow the method described in the RCSD tutorial to evaluate the pitch stability for various CoG locations. Try to find a combination of airfoil usage, sweep, twist, and CG location, that gives you good positive pitch stability. You can look at the Lift/drag polars for the entire wing by choosing "wing design" for the application menu, then "polars" on the view menu. You may want to try changes in the design to see how the polars are affected. You can also choose "operating point" on the view menu to see the lift distribution across the wingspan. Each of these operations is described in the RCSD tutorial.

[John235]

Crap, XFLR5 cannot handle twin fins at the tip of the wings - it lets me define them, shows them nicely to wet my appetite, but then fails (due to a singular matrix somewhere in its calculation) when I actually want to compute the polars. At least I was able to figure out that the lift distribution seems "safe" for all relevant angles with only a small amount of twist.

M.Instead of defining them as fins, try setting the winglets as wing panels with a dihedral angle of around 80 degrees.

[Mihai]

XFLR5 is way cool (the wing moment being one such thing), and from what I saw on its website matches reality pretty close (at least for the conventional planform they tested). I fixed the fins problem - it seems that if I add a second panel to the fins it crashes if I use twin fins, but with only one panel it works fine. Some strange results toward the tips (big spikes in local lift), I'll have to see if that's an artifact of the fin location or real, but overall it looks cool. I have more playing to do.

The 80 degree panel winglets trick doesn't quite work in my case as I want to have the fins both above and below the wing (i.e., the bottom of the winglet is not at the bottom of the wing).

Best,
M.

[kral27]

Thanks for posting this information and perhaps this is the key for the next step.

[kral27]

Here’s a graph showing the induced drag increment due to sweep (http://www.rcgroups.com/forums/attachment.php?attachmentid=1666994) . Sweep forward is on the left. Notice that the top of the curve is not at 0 degrees sweep. You actually get an induced drag bonus up to 5 degrees sweep back. At 10 degrees the induced drag of a swept back wing is the same as a wing without sweep.

--Norm
Ignore my last post as it makes little sense without the reference... now take the last post and combine it to this post and it makes sense...

[Texas Buzzard]

I'm designing a swept flying 'wing for thermaling, so efficiency is of utmost importance. So far many things seem to improve with an increase in sweep - wing loading, size of fins, amount of wing twist etc. However, I understand that there is such a thing as too much sweep - after some points the efficiency of the wing starts to decrease. Where is this point, and why is the efficiency decreasing? Any way to quantify this decrease in efficiency? Will XFLR5 help?

Thanks,
Mihai

Some may want to disagree with me, that's alright. It is said that he is designing a swept wing for thermaling. May I say that if swept wings were promising in thermaking the sailplane manufactuers would be making them.

I appreciate the experimental mind but the swept wing isn't very effecient when compared with conventional ships. Longitudinal stability requirements mandate that s ignificant area of the wing to have negative incidence so you loose area when you go swept wing.

If you could build a "magic" plane with high strength and low weight you may have some fun with it. Sorry but to me to make a swept wing glider into an effecient thermal machine is not possible. :)

[Mihai]

Some may want to disagree with me, that's alright. It is said that he is designing a swept wing for thermaling. May I say that if swept wings were promising in thermaking the sailplane manufactuers would be making them.

I appreciate the experimental mind but the swept wing isn't very effecient when compared with conventional ships. Longitudinal stability requirements mandate that s ignificant area of the wing to have negative incidence so you loose area when you go swept wing.


Unless you use a profile with very small (or even positive) pitching moment - then you don't have to worry about stability - like a plank. Then you can have the entire wing at positive incidence (with very small twist).


If you could build a "magic" plane with high strength and low weight you may have some fun with it. Sorry but to me to make a swept wing glider into an effecient thermal machine is not possible. :)

I'm so new at this plane design thing (and it's not my day-job), so I can be awfully wrong -however, I had a lot of fun designing it - I'm stuck building it though as I'm very short on free time (I know everybody is). However, I'm sure that I'll eventually build it (surely before the summer), and I'll certainly report on the massive failure or small success, whatever may come first.

Best,
Mihai