Designing my own wing - RC Groups
Thread Tools
Aug 21, 2017, 04:40 PM
Registered User
Discussion

Designing my own wing


Hello all,

As I will have some more spare time in the next few weeks I've decided to learn how to design my own flying wing. I've been designing autopilot systems in the last 8 years or so, and my personal objective has always been to use them to break my own endurance records.

So I want to build a simple, not too large (+- 1.5m) , lightweight flying wing to improve my endurance records. After viewing the inspirational speeches by Al Bowers I suppose there is no other option left in my mind than building my very own Bell-distribution Prandtl-D alike wing.

Because I could not find any ready to use designs/plans, I guess I will need to make my own.
My main source of information is the paper by Al Bowers and the design he made for Koen's dragon wing.

I am writing all my findings down as a personal logbook, and hopefully you guys will learn from the mistakes I will make. Also, I hope some of you will point out my mistakes and guide me in the good direction :-)

Thanks for reading!
Sign up now
to remove ads between posts
Aug 21, 2017, 04:57 PM
Registered User

Airfoils


The wing Al Bowers designed for the dragonwing had its own airfoils. Because this wing is a large one (12 meters or so), I doubt the airfoils will still be valid at my scale (1.5 meters).

I know X-Foil is widely used to investigate airfoils, but its complicated to use. XFLR5 should be an easier to use alternative. I watched this tutorial to learn how to use it.

This is what the root airfoil looks like
Name: RootAirfoilDragonwing.png
Views: 33
Size: 11.6 KB
Description: Root airfoil from the full-sized dragonwing design

Assuming I will fly around 10m/s, my reynolds numbers should be between 50 000 and 250 000 (I used this tool).

Putting all this data into XFRL5 gives me these graphs:
Name: RootAirfoilDragonwingXflr5.jpg
Views: 26
Size: 316.2 KB
Description: First analysis of the root airfoil in XFLR5
I am by no mean an expert like this, but it seems obvious that at the lower reynolds range, this airfoil just doesn't seem to work well. I also made some similations at bigger reynolds numbers and the performance degradation it even more obious then.

So I'm afraid I will need to start looking for a new root airfoil. My current assumption is that the zero lift angle (C_L,0) should be zero, and the pitching moment (Cm) should be low as well. Any suggestions are welcome! :-)
Aug 21, 2017, 10:52 PM
internet gadfly
nmasters's Avatar
See how this compares. Change the file extension to .DAT and run <refine globally> in the direct design window to increase the panel density.
Aug 21, 2017, 11:22 PM
Registered User
tspeer's Avatar
Take a look at Drela & Youngren's AVL (http://web.mit.edu/drela/Public/web/avl/) and XFOIL (http://web.mit.edu/drela/Public/web/xfoil/) for designing your planform and sections. The AVL examples include input files of Drela's gliders (http://www.charlesriverrc.org/articles.htm). His Supergee II would be the perfect baseline for comparison with your design. You could aim for similar spiral stability, performance, weight, etc., to be competitive with a conventional configuration.
Aug 21, 2017, 11:31 PM
Registered User
tspeer's Avatar
You may want to take a look at NACA TN 2249, "The spanwise distribution of lift for minimum induced drag of wings having a given lift and a given bending moment", http://hdl.handle.net/2060/19760012005. For a Horton style swept wing, the pitching moment and the bending moment go hand in hand, because as you move the loading inboard to reduce the bending moment, it also moves forward and creates positive pitching moment. The bell-shaped lift distribution Bowers advocates is simply the byproduct of a linearly tapered downwash distribution that is negative at the tips.

With AVL, you can play with planform, twist and sweep to achieve the downwash distribution you want to get the best compromise between induced drag and stability. If you start with flat plate sections, you can do most of the planform and twist design work independent of the section shapes. Then add the real sections with their zero lift lines aligned with the flat plate chords to complete the 3D shape. That will maintain the spanwise lift distribution as you iterate on the section shapes. The section design then becomes a tradeoff between profile drag and Cm0.

XFOIL's Type 2 polars will vary the Reynolds number with the lift coefficient, as would be the case
in the steady-state when the loading is held constant (lift = weight) and the speed is varied. This is the best way to evaluate the sections for performance.
Aug 22, 2017, 07:01 AM
Registered User
EdSoars's Avatar
Are you committed to the prandtl platform? If I wanted to increase endurance, and didn't need a wide speed range or high speeds, I would look no further than the RES Dart, elsewhere in this forum. Constant - chord, single - airfoil, chevron wings can be VERY efficient and spirally stable.
Oct 08, 2017, 12:03 PM
Registered User

More knowledge needed, project changed


Thanks all for your advice. I tried to model the prandtl non-lineair twist into xflr5, but the model didn't converge...

I decided that I needed more knowledge on wing design so I decided to start off with something more "simple". Basically I would like to design a wing that has a huge payload volume and is still as small as possible. Application could be attaching it to a weather balloon on making it fly home automatically using a autopilot system. So main requirements are:
- Thick airfoil ("payload area") in the center
- Small span because we don't need a big glide ratio
- Very stable, it should more or less glide and stabilize itself when dropped from the balloon.

My current design has 3 parts:
- a 18% thick airfoil with a chord of 35cm (Al bowers airfoil from dragonwing). This center section has a 10cm span.
- a small piece in between that makes the 18% airfoil change to a HS522, 22cm chord, 5cm span.
- a small wing part. HS522, 10cm span, 20° dihedral for stability, 20° sweep, -3° twist

According to xflr5 it should be "very" stable.
However, when flying it, I needed to add a considerable amount of twist to make it flyable (luckily not a big deal when assembled using double-sided tape). Even the reflexed trailing edge needed to be turned up a bit. Isn't this very strange for a wing designed with flying wing airfoils? I assumed it would need zero sweep or additional reflex. Anybody know why?

Something that needs improving is the pitch-self stabilization (does this have a better name)? When I toss the model high in the air, it will descend at a more than 50° downward angle and it won't recover from this... I would like the wing to return automatically to a zero-pitch angle. Do I need to adapt my design for this?

Also, it needs quite a bit of speed, but I guess that's due to the low aspect ratio...

Any advice on my questions, or general remarks are highly appreciated!

Tom


Edit: maybe the additional needed twist as shown in the photo isn't that big. About 5mm dropped LE at a chord of 255mm would mean only 1.3 degrees...
Last edited by dentompie; Oct 08, 2017 at 12:30 PM.
Oct 09, 2017, 08:49 AM
Registered User
Quote:
Originally Posted by dentompie
I decided that I needed more knowledge on wing design so I decided to start off with something more "simple". Basically I would like to design a wing that has a huge payload volume and is still as small as possible. Application could be attaching it to a weather balloon on making it fly home automatically using a autopilot system.
How far is it supposed to glide? What altitude is it to be released from?

Your wing has a very low aspect ratio which will contribute to a low glide ratio.

Additionally, it's essentially a plank configuration, which will make it have pitch issues.

-Dave
Oct 09, 2017, 10:10 AM
Registered User
miniphase's Avatar
Assuming that your cg is dialled in ok, I suggest that working at such a small scale isn't going to give you great results. I'm guessing that you'll not have flow attachment over the lifting body element of the model causing it to plummet in the way you describe.

Maybe a more productive approach would be to start with a simpler (but larger) delta type platform, get that flying and them incrementally increase the centre 'pod' thickness.

Don't give up!
Oct 09, 2017, 11:06 AM
Registered User
Low aspect ratio should be able to tolerate slower speed under normal circumstances.
So that leaves me wondering about your wing loading. Is it very high?
And yes, that very thick center section will encourage separation at the low Reynolds number you may be flying at.
Oct 09, 2017, 12:08 PM
Registered User
Don Stackhouse's Avatar
Not just the center section, the airfoils on the wing panels are too thick as well. At your Reynolds numbers ("Re"), there is likely to be large areas of separation on the aft portions of the top, bottom or both, at all angles of attack.

This also means that your trailing edges are flying in a bubble of separated flow, which is why your reflex is not as effective as you expected.

If you are not getting positive static pitch stability (the plane will not pull out of a dive by itself), then your C/G is too far aft. The separated flow could be a factor in that. If the aft portions of your lifting body fuselage and your wings are all basically "along for the ride" due to separated flow, then the forward portions of the plane are doing all the work. That makes the plane appear to the air molecules as if the Aerodynamic Centers ("AC") of the wings and the lifting body fuselage are further forward than the basic geometry suggests, which makes the C/G appear to be further aft, in comparison to the effective local AC's, which makes the plane behave like it's tail heavy, and therefore statically unstable.

At your Re's, 18% thick is about three times too thick! For your size and airspeed, it should be more like 6% thick. Making airfoils too thick for planes that operate at low Re's not only increases drag, it also REDUCES lift, as well as reducing dCl/d-alpha, the lift curve slope (the slope of the plot of lift coefficient vs angle of attack), which also alters the stability.
Oct 09, 2017, 12:20 PM
Registered User
Don Stackhouse's Avatar
One more bit of reference info - at sea level, "standard day" (about 59 deg F) conditions,

Re = 778 * speed (MPH) * chord (inches)

As altitude and/or temperature go up, the Re goes down, although not by a huge amount in the places we typically operate.

Whenever you get below Re = 150K or so, things get tricky and quirky. Required thickness and camber both need to go down, and the high point needs to move forward to around 25%. At Re's below 100K this is even more of an issue, and things REALLY get weird below about 60K.
Oct 09, 2017, 04:22 PM
You know nothing....
Stuart A's Avatar
Question from the back of the class
Is the amount of dihedral on the model excessive/unnecessary for a swept configuration?
Oct 10, 2017, 07:02 AM
Registered User
Thanks all for your replies!

@Dave: Assuming the balloon would go 30km altitude and drift around 100km, the glide ratio should be between 1:5 and 1:10

@Nuteman
Oct 10, 2017, 07:14 AM
Registered User
Thanks all for your replies!

@Dave: Assuming the balloon would go 30km altitude and drift around 100km, the glide ratio should be between 1:5 and 1:10

@Nuteman: wingloading is low because it is just a throw-model for the moment

@Don and others: indeed separation might be the issue here. Also my glide tests are very inconsistent. The 18% thick airfoil might work for the application (flying back 15m/s, thus working at a higher Re), but is very hard to test and probably not predictable enough for all flight modes.
However I do not agree with your 6% thickness: I have flown a 20cm (4inch) plane using the ESA40 airfoil which is about 10% thick. It flew great!

@Stuart: it would be great to hear what others have to say about that. According to XFLR5 the dihedral should be a lot more stable and it no longer requires a horizontal stabilizer. Of course I will plane the elevons on the center section, otherwise I would have a massive adverse yaw.

Maybe a blended wing-body design at small reynolds numbers it not really workable? Maybe it is a better idea to build a neutral fuselage and just attach a wing to it. Any ideas?

Either way, I will start with creating a 12% thickness center section and see how it goes. Anything smaller is not usable as payload area...


Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
Help! designing my own impeller for water jet engine olcho Dock Talk 14 Oct 23, 2017 08:19 AM
Question How to build flightcontroller (circuit design + PCB) of my own to support raceflight? farazahmad759 Beginner Multirotor Drones 9 Aug 10, 2016 07:26 AM
Discussion "THE GO DRONE Aero" My own design looks awesome! MTNDrew Foamies (Scratchbuilt) 0 Feb 23, 2016 09:49 AM
Discussion My own design indoor plane rcnutt Scratchbuilt Indoor and Micro Models 11 Nov 28, 2015 08:43 AM