Can somebody please spare the time and explain to me (and other beginners here) what exactly is an airfoil family (or series)? What do they share in common, how are they different, and how are they to be used (which member of the family where and, more important, why)?

Thanks!
Mihai

Well, this picture should make a start at answering the question. I took the AG46ct and ran Type I polars for 10K through 500K at something resembling logarithmic intervals.

What should be obvious is the characteristics of the airfoil change considerably over that Re range. That, in a nutshell, is the issue. Things change too fast in the range we work with for any one airfoil to be close to optimal much of the time.

I'll probably provide a lot more on this subject, just not tonight. It will be piecemeal.

Gerald

So it's all about optimizing performance at different Re numbers due to tapper?
But normally on a wing you don't have nearly such a range - perhaps only an order of two between tip and root. On the other hand, the speed range of the plane is probably larger than that (from float to coming against the wind) - so it seems that there is more Re variation from speed than from wing tapper - what do you optimize for? Float?

M.

A good engineer should optimize for overall mission performance. Float is just one aspect of DLG contest performance. Other flight regimes to consider are launch, range (L/D at different speeds from slow to fast) straight line float, and thermal turn float (suprisingly different than straight line)...

Read through several of Gerald's recent threads about the Zone family of foils, the edge family, and the comparison of modern DLG wings...Your aero education will come along fast in those threads.

I did read Gerald's thread and I agree that overall performance is what is needed; in particular if I got it right for DLGs the two most important things are min-sink at float and good L/D at speed, perhaps launch follows from the second.

I was referring specifically to the Re optimization - it seems (from what I saw) that performance always gets better at higher Re, so if you get the foil efficient at low speed (= float in my book) it will also be efficient during launch and at high speed.

Please correct me if I'm wrong - as I stated in the thread start - I'm trying to learn.

Thanks,
Mihai

Mahai,

Unfortunately not that simple. A foil optimized for low Re is only a mediocre performer at higher Re, mainly due to Cl limitations. That'll be in my second installment when I get the chance.

Performance is not just about drag, but also about supported AOA range, and lift coefficient range, and a few other things.

Gerald

Oh... I see... then I'll just shut up and wait for the second installment .

Thanks for the explanation,
Mihai

Here are two sets of Type II polars, using AG455ct and AG47ct, at cambers of 0 and 5 degrees. The first data set shows what happens if both foils are used as a root at Re*sqrt(Cl)=52K. The second data set shows what happens if both foils are used as a tip at Re*sqrt(Cl)=25K. These are the design Re for these foils.

When compared as a root, you can see that the AG47 really is fairly low in drag, but unfortunately it is really picky of AOA. This shows on the L/D vs AOA plot on the lower right. Also you can see that the AG47 doesn't like higher AOA and cannot generate nearly the lift of the AG455.

When compared as tip, you can see that the AG455 just gives up too much performance. Again this really shows on the L/D vs AOA plot.

So, neither airfoil of this pair does the job of the other acceptably well. One could use the root as a tip, with a large performance penalty, but one couldn't get away with using the tip as a root. The plane would be too critical to fly.

Gerald

OK, I buy it (although the trade-offs are not at all crystal clear - the Cl/Cd of the tip airfoil at root looks awesome except for the narrow angle - it looks like for the Edge you preferred this behavior). A few questions:
- what is Xtrl on the x axis of Cl? I assume that it's some function of alpha, but not sure what.
- why is Re*sqrt(Cl) important, rather then just Re?
- why are they called a "family" rather than just a couple of airoils?
- how do you know which airfoil was optimized for which Re?
- how do you choose which airfoil you use where? If you look at Re*sqrt(Cl), now it looks like you optimize for some alpha~Cl and some speed~Re - how do you go about it?

Sorry, the list of questions is longer than I intended (and I'm not sure I'm done ).

Thanks,
M.

Mahai,

When one knows nothing, one doesn't even know what are the appropriate questions to ask. As one learns more, one has better questions. Your questions here are pretty good ones!

"Cl/Cd of the tip airfoil at root looks awesome except for the narrow angle - it looks like for the Edge you preferred this behavior"

Actually, the Edge behavior is VERY different from this. The Zone has this sort of behavior just not this good/bad. Maximize L/D and the pilot had better be good enough to fly it (or fly a forward CG to make up for the difference). It is a good news, bad news situation. Yes, the L/D will be high but that is mostly because the drag is quite low. The lift is also low so one would have to build a light plane. One could do so. At that point, if flown correctly, then yes the L/D would be very good. I've been saying for quite a while that most DLG wings are too thick. That L/D is showing something about what could be done with thinner wings. But one can go too thin or not have the right shape, and that is the case here. Look at the lower right graph which is L/D vs AOA. See how sharp the L/D peak is. If one flies at a slightly high or slightly low AOA, one looses a lot of performance. Contrast that with the L/D peak of the real root airfoil. It is MUCH more forgiving of errors in flying.

Now one might think one could get away with it anyway, by using a forward CG (which isn't the minus most seem to groupthink). And one could, in calm air. But as soon as the air becomes active, then it becomes physically impossible to tightly control the AOA of the wing. So average performance starts to suffer. Lets look at the float camber case. The performance advantage of the AG47 as root compared to the AG455 is small to start off with. Notice how it only out-performs the AG455 from about 2 to 2.5 degrees AOA (at float camber). It won't take much turbulence before the pilot can't maintain that level of accuracy. Notice how much the AG455 out-performs the AG47 as root outside that range, particularly when on the slow (higher AOA) side? In a thermal I doubt one can maintain the required AOA range to have better performance with the AG47. Thermals are turbulent.

At speed mode, the AG47 has a little more performance advantage, and holds it over a 2 degree AOA range. The L/D across the peak of that range is relatively flat. Personally I consider that extremely flyable. I don't really see a downside here.

So here is a crude rule of thumb. Thinner airfoils as root on a DLG can give stellar performance in speed mode. That's pretty much a no-brainer but this one is an example. But thinner airfoils are more difficult to make forgiving in performance at higher camber settings. We also see that here. If we never had to thermal our DLGs then that wouldn't matter. But it would be a shame to be the one who launches highest, gets to the thermal first, higher than anyone else, and then gets to watch everyone else sky out in that thermal while the one falls out.

Ideally an airfoil gives not just performance when flown correctly, but performance when flown incorrectly. A good part of the reason is that under active air conditions, it is pretty much impossible to fly correctly, regardless of the skill of the pilot.

"what is Xtrl on the x axis of Cl? I assume that it's some function of alpha, but not sure what"

That's a tricky one to guess. When I started work on the Zone series I didn't know what that graph was, so I ignored it. Generally you can ignore it when not designing airfoils but are using existing ones. But it is important in design.

The graph shows the upper and lower surface transition point location, normalized (0 -> leading edge, 1 -> trailing edge). One could take a lift coefficient, 0.8 for example, and check on the AG47ct float camber graph (bright green), and see that the transition point is around 26% back from the leading edge. Whereas for the AG455 at that same Cl, would have its transition point at about 72% back. So the AG47 is on the hairy edge and the AG455 is doing fine at Cl=0.8.

"why are they called a "family" rather than just a couple of airoils?"

Mainly because they were designed from the start to work well together. One could take random airfoils, say some decent root airfoil like the HM-51, and pair it with a decent tip airfoil, like the AG47, and make a wing. But chances are the wing would behave rather strangely. These two were not designed to play well with each other.

Playing well - there are a few things which go into this. Please don't consider this an exhaustive list!

1) AOA compatibility. In a typical wing design, we unload the tip compared to the root, to give high resistance to tip stall. So if the root is working at Cl=0.8, floating along, then we might have the tip working at Cl=0.5. These are made-up numbers and not off any specific wing so don't use them as guidance! Now at Cl=0.8, the root airfoil requires an AOA of, say, 2.5 degrees. Let's say the wing has a degree of washout. Then the tip AOA would be 1.5 degrees... but it is not that simple because of some aero effects. The AOA the tip sees will be different than 1.5 degrees because the air flow is 3D and it is a full wing not just an airfoil section. So the tip really sees an AOA of X. Does the tip generate a Cl of 0.5 at an AOA of X efficiently? Or is is way off its L/D peak?

2) Camber change compatibility. We use variable camber wings. So we have to have AOA compatibility over the range of camber we intend to employ. As a rule of thumb that generally implies having the percentage chord devoted to the flaperons be the same root to tip, or pretty close to the same. But that is not a sufficient guarantee. Anyway one could take foils that are AOA compatible at some particular camber setting, and by juggling the washout and flaperon chords, make them fairly compatible at all desired camber settings.

3) Transition point compatibility. Where there is a transition point, there is going to be a little pressure change and flow velocity change. If the transition point has an alignment substantially off perpendicular to the local air flow, then it is possible that the angle of the pressure gradient will result in an increase in cross-flow along the wing. That would just show up as extra drag. Also when the transition point is angled to the flow, I really expect there to be some influence between adjacent stations and the resulting flow characteristics should deviate farther from design. I can't model that so I prefer to eliminate it as an issue. That way I have a better idea what the wing is doing.

"why is Re*sqrt(Cl) important, rather then just Re"

This one is pretty simple really. In steady-state flight, the wings will always be generating lift equal to the weight of the plane. But lift is a function of the square of the speed and Re is linear in speed. With Type I polars, where Re is held constant, when one flies at a different speed one must look up the results on a different trace since the Re has changed. With a Type II polar where Re*sqrt(Cl) is held constant, one is looking at a different location on the same trace. It makes it much easier to see what is going on with the flight behavior.

Type II polars are not appropriate for all situations. For instance you can't use them to effectively analyze the climb portion of a launch, or the behavior of tails. But they are generally more useful than Type I at least for most of what I do with wings.

"how do you know which airfoil was optimized for which Re"

Generally, you don't. In some special cases you do. For the Zone series, the label tells you the answer. Zone-52 was optimized for Re*sqrt(Cl)=52K. For the AG455,46,47 series, you can infer the answer from the Supergee-II plans which use the foils at 52K,40K,25K respectively. That's marked on the plans. In the end the Edge series will be labeled like the Zone series, starting at 60K.

But the bigger question would be "does it matter". The characteristics of a foil are going to change as Re is changes. That is pretty obvious in the Re range we play with for DLGs. Things change fast in this range. What matters is that an airfoil you choose has the characteristics you want at the Re range you intend.

To some extent you can eyeball an airfoil and get a good idea. Thin hingeline, thickness peak well forward, thin overall -> low Re. Fat foil, thickness peak closer to the middle of the foil, possibly high camber -> high Re. This is for cases where mach flow is ignorable.

"how do you choose which airfoil you use where? If you look at Re*sqrt(Cl), now it looks like you optimize for some alpha~Cl and some speed~Re - how do you go about it?"

Yes. That is a good question!

Gerald

Edit - I was going to not answer your last question as it would take a book, but I'll give you a start. Experience has shown that certain speed ranges are more useful than others when flying a DLG. These are what we consider the typical float, cruise, and speed settings. Given certain assumptions about plane design, this effectively translates to certain lift coefficients. Given a candidate set of airfoils, this translates into certain camber settings. Anyway one designs a wing for the candidate airfoils, and tests its performance over the camber settings desired to give the speed range desired. Then you try again to see if you can do better... All the while checking the performance and design against your requirements. Once requirements are met then you have a design that could work. How much time you invest after that to make it better is up to you. Personally I spend far longer making it better than in just meeting requirements.

Wow!!!

Thank you so much for the very detailed answers - I read them four times spaced out to sink it in. I learned a lot out of these answers, thank you for taking the time!

First, I apologize about the comment on Edge's narrow peak performance - I really meant Zone - I didn't see any Edge graphs (did you publish any?), while I looked quite close at the Zone graphs.

Second - it took me a while to figure out the Re*sqrt(Cl), but now it makes sense (strangely enough at first it made less sense than other things in your answer).

Finally, I think that the most important thing I learned is that it's extremely difficult to design a high performance plane. I thought that it's hard and time, consuming, but now it seems that this is an understatement.

I was planning to undertake a similar exercise for a flying wing DLG, but I'm really discouraged now: it seems to be a daunting task for a conventional DLG - for a flying wing it's probably even harder, as you need to watch the momentum of the wing at the same time as you optimize camber (and you have separate flaps and elevators of variable size mixed in the equation).

I think I got all the answers regarding the airfoil family - it all makes sense now. Of course, I got new questions in the process (like why the heck is the transition point going back and then falling off at small Cl - is it getting laminar? - I thought that a back transition point is bad, but it seems that in some conditions that is not the case), but I don't want to abuse your patience anymore (and this question probably doesn't belong in this thread anyway).

Thank you very much for your help!
Mihai

Mihai,

You are welcome!

You can find everything published about the Edge series, so far, in the Edge thread. IIRC there is at least one polar including a number of Edge root candidates and the Zone-52 in comparison.

https://www.rcgroups.com/forums/atta...8&d=1257925902

Hmmm, I must not have published some of the stuff I had. I recall data that isn't in the thread. I couldn't find the picture I was looking for. Oh well.

One approach to designing a high performance plane is to copy an existing one, and then make the modifications you desire. One can then design a little, build a little, test a little, rinse and repeat. That way, it is not all that hard to do a high performance design. And one gets to learn as one goes. What gets hard is when working from a little closer to ground zero as I've been doing with my designs, or when trying to push the state of the art a little farther. Trying to be in front in any endeavor is always hard and time consuming or someone else would have already done it. Actually everyone else has already done it; you're just trying to do it better!

On flying wing DLGs, I know many seem to want to make them, but I'd recommend against the attempt. Flying wings are at a considerable disadvantage to start with compared to conventional arrangements, and then when it comes to the throw they are at an incredible disadvantage. DLGs don't really have any margin of performance to lose. But if you want to, go for it! Just don't expect to ever be able to equal the performance of, say, a Supergee-II. Sorry.

A rearwards transition point can be a good thing to my way of thinking, overall. It is a compromise of course. It depends on the bubble. Anyway I'd like to see 100% laminar attached flow at speed. Do-able on the tails, but not yet at a wing root. At float it matters less but laminar attached is less draggy than turbulent attached. What one wants to avoid is laminar detached, particularly where it stays detached right on past the trailing edge of the wing. Generally IMHO in our Re range we're going to get a bubble in the float camber range. Flow will detach, and then re-attach a bit later, on the top surface. Flow may (probably) circulate within the bubble. Now, from the point of view of the overall air mass, the airfoil as acquired a lump on top that is in the shape of the bubble. This makes the airfoil behave as if it were thicker and had higher camber. Therefore it can generate more lift. Done properly, it can manage this more lift without much more drag, so the L/D becomes higher. Now the transition point isn't really showing where the bubble is located, at least not directly. It is probably somewhere around the middle of the bubble. Quick link:

http://www.mh-aerotools.de/airfoils/bubbles.htm

If you look at the L vs D polar of an airfoil (Type II), and you see a region where the lift is increasing but the drag is constant or decreasing, then you are most likely seeing evidence of a significant bubble.

A lot of current low Re wing airfoil design, IMHO, is in deliberately creating and controlling the bubble to widen the performance envilope. One wants an airfoil thin and clean and close to 100% laminar for best launch and speed performance, but then one wants to slow it down and generate a lot of lift. Dropping the flaps, changing the Re, and controlling the bubble let us do that.

Gerald

Thanks for the pointers - it makes sense now (or at least I think so for now ).

Originally Posted by G_T

On flying wing DLGs, I know many seem to want to make them, but I'd recommend against the attempt.

It's a bit late: https://www.rcgroups.com/forums/show....php?t=1162415 .

The thing is that I realize that it will probably not be an SGII or a Taboo, but I do like working at new things - and for the conventional DLG I think that there is very little to be improved - it's probably already working over 90% efficiency (I may be wrong on this), while for a flying wing, there is a wide open space there - I have no idea what the current efficiency for a flying wing is, but I'd assume it's low.

Thanks again,
Mihai