Quantcast

Suitable airfoil large thickness, high RE

HomeBuiltAirplanes.com

Help Support HomeBuiltAirplanes.com:

berridos

Well-Known Member
Joined
Oct 10, 2009
Messages
1,024
Location
madrid
In order to reflex an airfoil I wonder if it is ortodox to preserve the points from the camberline up to the station from which i want to bend upwards the camberline and from that station on, propose for example 2 points (guess more points are required) moved upward and fit a polynomial with least squares in order to obtain a whole new camberline.
Would such a approach be reasonable?
With this new camberline i would reconstruct the airfoil preserving the distances from the camberline at each station.
 

cluttonfred

Well-Known Member
HBA Supporter
Joined
Feb 13, 2010
Messages
7,834
Location
Fort Walton Beach, Florida, USA
I think either way will work and it won't make much difference in practice. Here are a couple of examples using the Airfoiltools.com comparison tool graphics:

NACA 23112 seems to have the reflex distributed along the camber line compared to 23012, note how the camber lines intersect at the leading and trailing edges. It’s just that 23112 camber line (red) follows more of an S-curve, slightly higher towards the front but slightly lower towards the back.
naca 23112 23012.png
On the other hand, the reflexed Clark YH seems to be identical to Clark Y except for the upturned trailing edge. It's harder to see in the image but imagine rotating the Clark Y (blue) around the trailing edge so the flat bottoms of the two airfoils overlap.
clark y yh.png
So it looks like both approaches have been used.
 
Last edited:

berridos

Well-Known Member
Joined
Oct 10, 2009
Messages
1,024
Location
madrid
In the first approach you suggest (with matching LE and TE points) i would define an intersection point (somewhere behind the thickest section station) and move the points before up and the points behind down, but according to which function?
Thank you very much for the hints.
 

Norman

Well-Known Member
HBA Supporter
Joined
Nov 28, 2003
Messages
3,047
Location
Grand Junction, Colorado
The 23012 is reflexed?
The 230 mean line is not reflexed but the 231 is. Moving the point of maximum camber forward reduces the pitching moment but can't eliminate it so you add a little reflex to get that last 0.02. Extreme forward camber also causes a sharp stall so there's a limit to how far forward you can move it, the NACA 5 digit airfoils seem to have hit that limit. The biggest influence on Cm is the angle of the mean line at the trailing edge. Some books say that the mean line must cross the chord line at 75%c for Cm to be greater than zero but it's really the angle at the TE that determines Cm. In the case of the NACA 230 family there's not much to trim out so there's not much reflex in the 231 mean line. BTW the word "trim" is key here, it really is a trim issue. When I'm making a new airfoil I start with a seed that has some other characteristics that I want, thickness, nose radius, wide low drag bucket, etc. Then I add negative flap, smooth with mixed inverse design and analyze. Repeat until the the pitching moment is what you want and other characteristics aren't too degraded over the seed. Yes the way I do it is trial and error, so su me. Anyway you can only have one flap on an airfoil in XFLR5 so I add one flap and smooth over the hinge. When you save this smoothed version it becomes an airfoil that you can now add another flap to. I used to start with a flap hinged at 75% but lately I've been starting at about 90%. That last 10% of the chord makes the biggest difference so why not start there?
 

rotax618

Well-Known Member
Joined
Oct 31, 2005
Messages
989
Location
Evans Head Australia
The one in the foreground with the red control surfaces flew so well that my friend was able to win several competitions against convention aircraft at his model club. At the time everyone wanted plans but I had family and business matters to attend to.
 

rotax618

Well-Known Member
Joined
Oct 31, 2005
Messages
989
Location
Evans Head Australia
If the aspect ratio is low and the planform is conducive to the formation of vorticies then there is no stall, even a very sharp leading edge will not induce a stall, the 2D values you are concerned about do not exist in the LAR 3D environment.
 

Norman

Well-Known Member
HBA Supporter
Joined
Nov 28, 2003
Messages
3,047
Location
Grand Junction, Colorado
But again the rule that vortex starts at 50-55º sweep isnt written in stone either. Some papers show that at 40 eddies show up in the oil tests.
Any wing will produce a leading edge vortex if the pitch rate is high enough. It's called dynamic stall and produces a lag between the onset of stall and the loss of lift. The trick is getting the LEV to stay attached to the wing. For unswept wings the vortex only stays attached for a few seconds and produces a big boost in lift which can be useful for tricks like jumping over trucks or the airport boundary fence. As sweep increases the stall vortex stays attached longer until you get to leading edge sweep angles greater than 55º, then it's a permanent feature above AoA of 7 or 8º.

As i plan to sweep the center section 50º and the outward panels 60º a compromise would be to install stallstrip at the beginning of the outer panels to trigger there stronger eddies that stabilise the wingtip stall tendencies.
Actualy a 10 degree kink in the leading can be a vortex starter however for it to work the inboard sweep should still be 55º or over. Double deltas are tricky. It's possible to get into an attitude where the aft section is really stalled and the inboard section still has a healthy LEV. The SAAB Draken could do this and the pilot's orders were "if you haven't recovered by 16,000 feet, eject". Likewise the trapezoidal wings with strakes that are common on fighters now have a similar problem if the strake is more than 18% of the total area.
 

Norman

Well-Known Member
HBA Supporter
Joined
Nov 28, 2003
Messages
3,047
Location
Grand Junction, Colorado
As i plan to sweep the center section 50º and the outward panels 60º a compromise would be to install stallstrip at the beginning of the outer panels to trigger there stronger eddies that stabilise the wingtip stall tendencies.
Oops, your LE kink is not what I was perseverating on. The kink on a normal double delta needs to be 10º to trigger co-rotating vortices. You're talking about an increase of sweep on the outboard panel? If the LE sweep angle out there is 60º you've got enough sweep, all you need is a feature to stabilize the starting point so it doesn't oscillate.
 
Last edited:

berridos

Well-Known Member
Joined
Oct 10, 2009
Messages
1,024
Location
madrid
Actualy a 10 degree kink in the leading can be a vortex starter however for it to work the inboard sweep should still be 55º or over. Double deltas are tricky. It's possible to get into an attitude where the aft section is really stalled and the inboard section still has a healthy LEV. The SAAB Draken could do this and the pilot's orders were "if you haven't recovered by 16,000 feet, eject". Likewise the trapezoidal wings with strakes that are common on fighters now have a similar problem if the strake is more than 18% of the total area.
Wouldnt such a stall be prevented with an all moving tip elevon like the one shown in the rc model?
 

Norman

Well-Known Member
HBA Supporter
Joined
Nov 28, 2003
Messages
3,047
Location
Grand Junction, Colorado
Wouldnt such a stall be prevented with an all moving tip elevon like the one shown in the rc model?
Sorry, I was half asleep when I wrote post 49# and somehow thought we were talking about double deltas with high sweep on the inboard panel and lower sweep outboard, like the SAAB Draken. Then I read your previous post where you said "I plan to sweep the center section 50º and the outward panels 60º" and realized my mistake and said so in post #50. What you describe there is not a traditional double delta and won't have the co-rotating vortices that are characteristic if planforms like Draken. What you describe ( 50º inboard and 60º outboard) is a segmented gothic arch. That planform can definitively have LEV on the outboard panels but probably not on the center section. This is actually a good thing because it means that it will develop a nose down moment when the center stalls which is exactly opposite of the Draken superstall, which yanked the plane up to an AoA of about 70º so fast that it put a 4G load on the pilot. Even with speed jeans that was hard to deal with. The LEV on the outboard panels will increase the lift on that part of you wing by 50 to 80% so you may have enough lift to keep flying even with the center section stalled and since the elevons are out there you would still have control.
 

berridos

Well-Known Member
Joined
Oct 10, 2009
Messages
1,024
Location
madrid
I suffer strong sleep deprivation in my endless search for the appropriate root profile for my double delta vortex project. Requirements are RE13.000.000 and pitch neutrality.
After several days getting fond of XLFR5 i have to conclued that this software smells a lot like a black blox, at least the full inverse and mixed inverse features.
After breading several hundred prototypes i just dont manage to find patterns that guide me progressivly in the right direction. The process resembles more like a blind trial and error.

In the following chart you find the following foils.
Dark Blue - NACA63021A (Foil used by Verhees delta)
RED - NACA654221
YELLOW - NACA654221 but i just straightened the lower convexity of the last 10% of the chord and the CL increases more than CM. So i consider that move beneficial and even more so the power factor.
WHITE - NACA654221 flapped up 2% at 60% chord station.
Thick GREEN - (My poster child after hundreds of iterations) NACA654221 flapped up 2% at 60% chord station and 2% at 80% chord station.

Following Normans advice i tried initially to achieve pitch neutrality by tweaking up the last 10% of the foil, but from my subjective perception the loss of CL against gain of CM wasnt optimal.
Changing forward the max camber really never resulted in a wortwhile improvement but killed the overall performance. Again based on my limited perception and experience.

Currently, my best option is the green example (showing a huge 1,75 max lift - shall i believe that?). The CM in the graph could by considered tolerable for a 12 feet chord?
The laminar drag bucket suffered a down displacement on this foil and a 0,3 cruise CL seems to be at the upper extreme of the bucket. Could i move the bucket up somehow or is 0,3 CL on this specimen acceptable?
I would love to allow the airfoil to fly as much nose down as possible in order to improve visibility but that would require a higher max camber. The problem is that xlfr currently with the 2 flap kinks sets the maxcamber at the 90% chord station.

I suspect my message is too long and boring....
 

Attachments

Norman

Well-Known Member
HBA Supporter
Joined
Nov 28, 2003
Messages
3,047
Location
Grand Junction, Colorado
That's actually a surprisingly high CLmax for having started from an NACA 65xxxx (surprising to me anyway, I've never gotten that much lift out of an ultra long laminar flow reflexed airfoil). They sacrificed a lot of lift potential for low drag at high speed and getting more lift out of them usually means losing some high speed performance. I don't even try for anything above 40% laminar anymore. Reflexing the TE does reduce max camber and make it appear that it's in a different place but the front 2/3 of the mean line as actually still the same shape. I use <mixed inverse> to clean up the the kink in the pressure distribution because it only modifies a patch on the airfoil without changing the rest of the shape. <Full inverse> could also be used but it affects the whole spline which is often a good thing but it takes control out of your hands and puts it in that "black box". As far as how much negative pitching moment is tolerable the answer is "none". Any excess moment will have to be trimmed out with the elevator. Really that's all we're doing by reflexing the airfoil. The advantage of a reflexed airfoil over a non-reflexed airfoil with elevator deflection is that the reflexed airfoil section is cleaner than the elevator. Sometimes it makes a big deference in drag and sometimes it doesn't. One thing you should keep in mind is that panel codes are optomistic outside of the linear range ie where the CL/alpha curve begins to round over or the Cm curve has a pronounced kink that doesn't smooth out with forced transition. A neat characturistic of fat airfoils is that they can have pretty thick trailing edges (2%c or less) without suffering a drag rise (verified in wind tunnel tests). This suggests another experiment you could try. One of the options in <direct design> is <set T. E. gap>
 

berridos

Well-Known Member
Joined
Oct 10, 2009
Messages
1,024
Location
madrid
I guess those very high CL numbers are related to the huge RE in this psecial case. Always understood that the longer the chord the more stable the laminar flow and this case seems to guarantee extreme laminar flow.
Should cruise CL be set at 0,3 for a delta or should i take additional considerations in such a planform and get below 0,3?
I always considered 0,3 as a cook book recipe but havent dig into the nuissances.
quote:
Ideal lift coefficient also called Optimum Lift coefficient: lift coefficient at which the drag coefficient does not vary significantly with the minor variations of angle of attack, it usually corresponds to the minimum drag coefficient. Best to choose the cruise lift coefficient as close as possible to the ideal lift coefficient. For subsonic aircraft, it varies from 0.1 to 0.4.


This airfoil has a CD bottom flatted out at 0,2, reaching from 0 to 0,25.
What worries me is that at 0,21 the CM is 0 but passes steeply from positive to negative thru that point.
It hasnt a constantly flat CM=0 across the sorounding values of AoA like the 63021.
 
Last edited:

Norman

Well-Known Member
HBA Supporter
Joined
Nov 28, 2003
Messages
3,047
Location
Grand Junction, Colorado
Ideally you want the wing to stay in the low drag bucket in all phases of normal flight so you don't have to power through maneuvers. This is especially important for low AR airplanes with limited horsepower because the drag buildup can be very fast in a turn. So, yeah, cruise in the middle of the bucket or lower so you can get through a 2G turn without losing altitude. As for the slope of the Cm curve: don't worry about it. The only number that matters is Cm0... that's Cm at zero lift. The slope of that line just shows that the aerodynamic center is not exactly at the 25%c reference point. It's not uncommon for the AC to be up to 2% forward or aft of 25%c but you need a reference point to graph the data so 25% is the standard. There's a very simple formula to find the exact AC from the slope of the plotted curve.
 

Speedboat100

Well-Known Member
Joined
Nov 8, 2018
Messages
1,642
Location
Europe
I went for a 22% thick foil.....as I ran it in XFLR5 I got 1.7 Cl...at 7 degs of AoA ( at re 500 000 )...for an 24% thick I even got Cl 2.0+ ...at 9 degs of AoA.

I can fly a fairly small and lite wing at PART 103 regime....at huge AR numbers.

Anyone tried this before...I think 1/4 scale rc model will show benefits of any.
 

TFF

Well-Known Member
Joined
Apr 28, 2010
Messages
13,833
Location
Memphis, TN
At about 8 ft wingspan, models take on more man carrying traits than the model ones. Three foot wingspan planes have a certain characteristics, from there to five foot change some, five to eight moves some more, and eight and up start acting with their character. You have to get the plane big enough that the ambient surrounds are not amplified. Small model’s power available tends to hide a multitude of sins aerodynamically. A model airplane will fly; how it reacts for data collection requires a lot that would compromise it as a model airplane.
 

Pops

Well-Known Member
HBA Supporter
Log Member
Joined
Jan 1, 2013
Messages
8,534
Location
USA.
At about 8 ft wingspan, models take on more man carrying traits than the model ones. Three foot wingspan planes have a certain characteristics, from there to five foot change some, five to eight moves some more, and eight and up start acting with their character. You have to get the plane big enough that the ambient surrounds are not amplified. Small model’s power available tends to hide a multitude of sins aerodynamically. A model airplane will fly; how it reacts for data collection requires a lot that would compromise it as a model airplane.
Right on.
 
Top