camber distribution

[gedaso]

(Posted this in the hand-launch forum but didn't get much of a response)

I once bought a book by Martin Simons called Model Aircraft Aerodynamics. In it, he discusses a method of designing efficient wings by:
- increasing camber towards the tip combined with
- a small amount of geometric washout to account for the difference in zero-lift angles of attack.

This gives a wing that has constant lift coefficient along the span (ie is efficient) yet doesn't suffer from tip stall.

Looking at most of Dr Mark Drela's recent designs, he uses a different approach - LESS camber towards the tips (thinner too) but still with some washout.

What gives? Does anyone know the theory behind the Drela designs ? Are there other (eg Reynold's number) issues at work that suggest the Drela approach is better than the Simons approach?

[rcjetpilot]

Signing up to read responses.

I posted a "like" question on this subject a while back, looking for a way to improve tip stalling. The answers I received here were more of the "latter" decreasing camber towards the root and with a mild increase in thickness, no washout.

I'm very interested to hear others on this since I'm about to have some ribs laser cut. My planform uses two 6-digit airfoils:
Panel span 20 inches.
Root @ 12.5pct thickness - 2:camber - is that 2 pct or 20pct:)
Tip @ 15pct thickness - 0 camber.

Bob.

(EDIT - added link)
http://www.rcgroups.com/forums/showthread.php?t=166363&highlight=camber+bob+thickness+span

[Sparky Paul]

Both Drela and Simons use washout. Which designer has the higher performing plane?
For all-out competition design, Mark is the guy.
For putting around the sky with the least complication, go with Martin.

[BMatthews]

Both methods work but in the Drela option you need to have the advantage of the Xfoil low reynolds number prediction software at your fingertips and know how to use it do design special airfoils to suit your needs.

There's a number of "rules of thumb" for delaying the stall in our models. First is to use a thicker airfoil, Up to a point this will delay the stall. Second is to add camber and adjust the incidence so the tip and root reach their Cl=0's at the same time. And finally moving the airfoil's high point forward by a small relative amount can also delay the stall at the expense of the maximum attainable lift.

[gedaso]

Both Drela and Simons use washout. Which designer has the higher performing plane?
For all-out competition design, Mark is the guy.
For putting around the sky with the least complication, go with Martin.

But WHY is the Drela approach better performing? By combining washout and decreased camber, you are effectively increasing the overall washout. This should increase drag at high-speed flight.

I have an incredible amount of respect for Mark Drela and his designs. I'm just trying to understand what his design goals were, and why he chose the options he did.

[Sparky Paul]

Dr. Drela is a competitor. His designs are intended to win against equally able designers and fliers.
I don't see that specificity in Simons, he's more interested in getting the information assembled on how to make practically anything fly reasonably well.
Competition finely hones the equipment to where frequently it is single-purpose, and does one thing well.
Looking at the "one thing" requirements then forces a design solution which must work well, for those requirements.
But you must start with the generalities of Simons, to get to the specififty of Drela.
...
Later on that same day:
With flight experience, stability begins to lose priority. The pilot participates more in flying the airplane, rather than guiding it. A high degree of stability is nice in a trainer, but not desired in a high performance airplane which be manuvered a lot... finding and staying in thermals, aerobatics..
Efficiency of flight becomes more important.. getting the most from the plane makes the plane more a tool for the flier's purposes, rather than an end in itself.
Novices always want a lot of stability, which they find isn't really needed once a lot of stick time has been accumulated.
The dreaded "tip stall" gets a lot of attention, yet that attention wanes with experience.. the plane is designed to not tip-stall, or the flier doesn't let it.
The latter has a better handle on flying by knowing all the characteristics of his model.

[SoarNeck]

I once bought a book by Martin Simons called Model Aircraft Aerodynamics. In it, he discusses a method of designing efficient wings by:
- increasing camber towards the tip combined with
- a small amount of geometric washout to account for the difference in zero-lift angles of attack.


Theoretically, they should work towards the same end. Airfoils with more camber tend to have lower zero lift angles of attack, so increasing camber towards the tip can effectively add washout. The problem with geometric washout is that it doesn't account for Re# effects, and so correcting low-speed issues with washout usually adds lots of drag at higher speeds.

I used the slight added camber trick on a model I just designed for F3B. It has zero geometric twist, but a tolerable tipstall margin.


This gives a wing that has constant lift coefficient along the span (ie is efficient) yet doesn't suffer from tip stall.


Sort of saying the same thing twice - the wing will stall at the location of the highest local lift coefficient. The tough part is figuring this stuff out, since spanwise flow is tricky to calculate quickly (spreadsheets and such help).


Looking at most of Dr Mark Drela's recent designs, he uses a different approach - LESS camber towards the tips (thinner too) but still with some washout.


He does that so that the wing is working effectively at all stations. Designing an airfoil for each location on the wing is an art, and there are a number of ways of doing this.

The Aegea was designed to have a huge speed range at low weight, and so it needed to have airfoils that responded well to camber. The root and tip have widely varying Re#, so it wasn't appropriate to just use the same airfoils (you can see this if you run the polars in XFoil/Profili2). His design choices reflect these goals.

Dr. D's design needs washout to balance the airfoil choice, in order to allow for proper stall behaviour. It's designed with that washout in mind (ie don't leave it out). It also means that the model isn't designed as a very high-speed aircraft, however...probably wouldn't make a good F3B competitor.

Thermals wonderfully, however, and it's amazingly responsive to camber (I flew an Aegea last week...what a treat!).


What gives? Does anyone know the theory behind the Drela designs ? Are there other (eg Reynold's number) issues at work that suggest the Drela approach is better than the Simons approach?

Re# has a big affect here. I had the same questions as you about two years ago when the Aegea came out, and Dr. Drela's design inspired me to figure these things out. The short answer is that the geometric washout that he calls for offsets a change in aerodynamic twist, and adds some tipstall margin as well (inner wingtip moving slower in a thermal turn).

I describe the effect in the thread in this forum called "Sailplane Design Spreadsheet". It's actually the tool that I developed to see these design parameters, and to work with them.

[SoarNeck]

But WHY is the Drela approach better performing? By combining washout and decreased camber, you are effectively increasing the overall washout. This should increase drag at high-speed flight.


It does...but the Aegea isn't designed to be efficient at extremely high speeds. Don't get me wrong, with reflex it comes back upwind like no 50 oz model should, but it wouldn't be as efficient at 100mph as a sailplane that was designed to fly that fast.

[markdrela]

It does...but the Aegea isn't designed to be efficient at extremely high speeds. Don't get me wrong, with reflex it comes back upwind like no 50 oz model should, but it wouldn't be as efficient at 100mph as a sailplane that was designed to fly that fast.

Correct. The Aegea wing is intended for F3J and TD, where you're not likely to glide much faster than 40 or 50 mph. It's overcambered for the 100 mph F3B Speed task, and the washout is also not exactly ideal for this case. On the other hand, you can still get reasonable 100 mph speed performance by over-reflexing the wing of something like -3 or even -4 degrees. Tom Kiesling says he gets noticably better zooms on his Supra (same airfoils) with this extra reflex. He also didn't do too badly flying it that way in F3B at the last Nats.

My approach is to design the airfoils first to match the local Re. This invariably requires less thickness and less camber towards the tip. Then I deal with the tip stall with a suitable taper/washout combination. Some iteration is usually needed.

[SoarNeck]

Theoretically, they should work towards the same end. Airfoils with more camber tend to have lower zero lift angles of attack, so increasing camber towards the tip can effectively add washout. The problem with geometric washout is that it doesn't account for Re# effects, and so correcting low-speed issues with washout usually adds lots of drag at higher speeds.


Oh, one thing I forgot to mention. One drawback to adding camber at the tip is that you might end up with an airfoil that won't take the same angle of attack range as the tip before stalling.

The tip will most times be working at an angle of attack that is effectively less than the root due to spanwise flow, but if you overdo the camber addition, you could get nasty high-speed stall characteristics (ie, the stab could cause the wing to stall by pushing the tip airfoil past its stall angle).

Something else to consider.

[rcjetpilot]

Novices always want a lot of stability, which they find isn't really needed once a lot of stick time has been accumulated.
The dreaded "tip stall" gets a lot of attention, yet that attention wanes with experience.. the plane is designed to not tip-stall, or the flier doesn't let it.
The latter has a better handle on flying by knowing all the characteristics of his model.

Not always true. Its a function of "need" for a given planform. Novices are not the only pilots who desire a "stable" planform. For some planes I want pure speed, thin wings, and requiring all my attention. Others are just for fun.

In other cases, a certain planform like highly swept wings with alot of tapper, its nice to modify the airfoil to "soften" the undesirable effects.

A novice == stability is not always true.

Bob

[gedaso]

Thanks SoarNeck and Mark for chiming in. I appreciate the knowledgeable comments.

Sort of saying the same thing twice - the wing will stall at the location of the highest local lift coefficient.
Not necessarily - stall would occur where the local lift coefficient exceeds the local stall lift coefficient. No??


The tough part is figuring this stuff out, since spanwise flow is tricky to calculate quickly (spreadsheets and such help).
I've written a vortex lattice program that interpolates profile drag from the Selig database to help me see these things. It also lets me see how close the local sections are to stall. I know its pretty tricky. Hence the questions :)


Oh, one thing I forgot to mention. One drawback to adding camber at the tip is that you might end up with an airfoil that won't take the same angle of attack range as the tip before stalling.
Perhaps you meant to say "as the ROOT before stalling"? This is what Martin Simons approach tries to avoid. He claims (and Profili/XFoil seems to support it) that the total available AoA (from zero-lift to stall) is larger with a more cambered aerofoil. This being the case, you shouldn't come across this problem. I think.


My approach is to design the airfoils first to match the local Re. This invariably requires less thickness and less camber towards the tip. Then I deal with the tip stall with a suitable taper/washout combination. Some iteration is usually needed.
If it isn't giving too much away, what are you trying to match to the local Re? Profile drag?

Thanks again to all. Enjoying the flow of knowledge..

[SoarNeck]

Not necessarily - stall would occur where the local lift coefficient exceeds the local stall lift coefficient. No??
True, but I’d wager that you’ll rarely see a situation where you have to investigate the local stall angle of each section of the wing to see if that dictates where the stall develops. Odds are that the transitional series used will have airfoils that are similar, and will hence have similar stall characteristics. If you’re designing a racing sailplane (F3B/F3F), it’s worth seeing whether you’re risking a high-speed snap by having the tip stall angle as the weak link, but remember that the tip lift coefficients are going to be quite a bit lower than those of the root.
I've written a vortex lattice program that interpolates profile drag from the Selig database to help me see these things. It also lets me see how close the local sections are to stall. I know its pretty tricky. Hence the questions :)
Good stuff, there aren’t many people who bother with this level of model design. It can be hard to find any help…I know that from experience :) How different is your program from something like LiftRoll (also vortex lattice)?
Perhaps you meant to say "as the ROOT before stalling"? This is what Martin Simons approach tries to avoid. He claims (and Profili/XFoil seems to support it) that the total available AoA (from zero-lift to stall) is larger with a more cambered aerofoil. This being the case, you shouldn't come across this problem. I think.
Yes, thanks…just typing too quickly is all. I would run the profiles through Xfoil before making too many sweeping generalizations, especially if you’re getting into the level of detail that you seem to be.

Actually, what made me dig up this thread again is that yesterday I went back and re-ran the numbers on the Supra to give myself a baseline for developing an F3J model (my F3B design is done, and under construction). I just wanted to show what I meant when I was talking about a sailplane that is designed for speed.

First, the Supra. Design weight of 48 oz, “thermal” condition evaluated at 3 degrees of flap and 15mph (RE# to match plan spec). The local lift distribution looks like the top chart below, according to my spreadsheet, and the wing is producing about 0.92 oz of drag. Cruise is at 50mph (same weight), and the wing is up to 5.41 oz of drag.

The plots:

[SoarNeck]

Next, my plots of my Frostbite wing. The thermal condition is plotted for a 90 oz sailplane flying at 20mph under 7 degrees of flap. Total wing drag is about 1.57 oz. This isn't a direct comparison because the model is heavier (much stiffer/stronger), and is carrying too much flap to thermal well (Dr. Drela would probably call this "crawl mode"), but I was curious how slow it would fly at 90 oz. Plus the airfoils on the Frostbite carry less camber to start with, as a result of being designed for speed.

The second plot is the cruise config, reflex of 1.0 degree across the wing. The Supra wing above uses 2 degrees of reflex BTW. Total wing drag is 4.93 oz, thereabouts, and the planform and transitional series was designed to (hopefully) avoid high-speed stalls. I was also required to avoid any geometric washout because of the building method I chose, but I would have tried to do that anyway to maintain clean high-speed behavior. The Supra above incorporates 0.5 degrees of washout in the mid panel, and can probably circle tighter as a result.

Can't wait until my wing molds are done! (only 200 hours of labour to go... :rolleyes:)

[Lazy]

As I understand Adam, you not play with different airfoil across span ?

[SoarNeck]

As I understand Adam, you not play with different airfoil across span ?

Yes I do, there are five airfoils across each span. I use a blend of the AG42d and 43d at the second-last chord station of the Supra, since the spreadsheet wants four panels, and the Supra only has three per halfspan.

[markdrela]

First, the Supra. Design weight of 48 oz, “thermal” condition evaluated at 3 degrees of flap and 15mph (RE# to match plan spec). The local lift distribution looks like the top chart below

Hmm. I get something quite different with AVL. See plots below.
Are you sure the airfoil and twist are set up correctly in your model? The twist angles are
0.0, 0.0, -0.5, -0.5 for the four defining stations.

[SoarNeck]

Hmm. I get something quite different with AVL. See plots below.
Are you sure the airfoil and twist are set up correctly in your model? The twist angles are
0.0, 0.0, -0.5, -0.5 for the four defining stations.

Hi Mark,

Yep, I was pretty sure that was how it was setup. I've got a fifth chord station stuck between your 3 & 4, but I just blend the airfoils together to get a "halfway" point.

I haven't had a change to work with AVL yet, I'm afraid, since the user's manual is a little large, but could you answer a quick question? When the graph shows "aileron 0.0, flap 3.0" for example, is that saying only the flaps are drooped, or are the ailerons drooped as well?

[Lazy]

Hi
Look Adam...Few examples.
First and second pictures show result of calculation, using a modified by you spread sheet.
Last picture - Nurflugel 2.17 ( remember ? )
Same airfoils, same design speed, weight and so on...
From root to 3/4 span I use RS001, transponde to MH42 on tip.
To big differences :)

[SoarNeck]

Sorry, I'd need to know more information on how you're using the sheet. How are you doing the airfoils across the middle of the wing on my sheet? I've never used the second program, so I really can't say why the differences are there. Can you specify changing airfoils in the second program?

I've used my own sheet to design a number of sailplanes, so it does work. If there's an issue somewhere in there, it's probably just a bug that somehow crept in. I'll take a look, at any rate.

[SoarNeck]

Any chance you can get some values on the Nurflugel graphs axis', BTW - it's tough to compare otherwise, since the graphs can appear different if the ranges are stretched.

[Lazy]

Adam, tomorrow I send you example.
Some changes need to make on design and we back to discuss.
Regards
Vladimir

[Lazy]

Sorry , Adam :)
I no say - spreadsheet wrong or not work. Absolutely NO :)
After consult with Petri and discuss about Eppler and XFolil ( kindest regards, professor :) ) code situation clear for now.
But look on you mailbox :p
Regards
Vladimir