What is a significance of a wing thickness

[CT 249]

I have seen a few wingmasts, but do not recall a lot of detailed discussion on advantages of specific sections.

While not in any way trying to tread on the toes of those who know their sections, I seem to recall that Steve Clark has mentioned here (or on SA) that section shape was not a particularly significant consideration in real world performance; getting it working was what made the difference.

 

[P Flados]

He was referring to the airfoil sections for a two element slotted wing, not a wingmast. He was also talking in the context of 'any old average approx naca 0020 in front of any old approx naca 0009' where the 'any old' are decent non-laminar airfoils known to be pretty good.

For a general purpose slotted two element wing, a little more "blunted" at the front of the lead element may be good as it makes it more forgiving for the rotation that may be needed for max lift. I have come to wonder if the need for "blunted" airfoil shapes for boat wings has anything to do with how a well known individual came to be known as Blunted on SA.

I have seen a lot of difference in selecting 'wrong' shapes for given sections in what I have done.

Note however, that I am limited to a single wing profiles. I would love to be able to compare my stuff directly to the stuff Steve is talking about, but XFLR5 will not do it and I do not have a lot of reference material to compare to.

 

[Erwan]

Flados,

For 2 elements wing, you should have a look at:

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

Thanks for sharing your research, I would have one question if you don't mind:

When you mirror the S1223,
which point do you choose at the leading edge to start the symetric axis?
Do you choose it because y=0 after de-rotating, normalizing and so on...?
or for instance do you search for the theorical stagnation point for high AoA?

Best Regards

EK

 

[P Flados]

I have done several methods to try to pick a mirror point, but just keeping the original 0,0 point seems to work as good as anything and it is my first choice. I did some manual shifting a couple of times (a fixed vertical offset as step 1) and some manual "blunting" efforts, but these never really ended up with a shape that worked well.

For the standard approach here is a summary:

I take the dat file for the starting foil and load the starting two columns of my spreadsheet.

I then have two columns where I keep the X values but displace all Y values down based on a rotation angle. This is kind of like rotating the starting foil, but it actually gets "stretched" a little bit.

The next pair of columns requires some manual manipulating.

I pick a string of entries of the rotated starting foil to represent the Top forward portion of the main element.

I do a straight vertical mirror (same X values, make the Y values negative, then re-sort to get increasing X values) to get the Bottom forward portion.

I have string of cells to generate an upper flat rear half that connects to the front section and terminates at a some distance below the chord line of the front element.

I have another string of cells to build a similar lower flat rear half.

I usually end up shuffling things around to adjust the location of where the the "front" connects the "rear".

The layout allows me to go back and change the trailing edge end points without affecting anything else. Much else and I end up manually "fixing" it to get the front and rear stuff all mirrored and/or connected up properly.

For the "double flap" stuff I am actually using XFLR5 to create the rear flap effect.

FYI, I will attach some text files of my "best" shape. One will be straight from the spreadsheet and one will be after I did some clean-up and tweaking in XFLR5.



Thanks for pointing to Java foil for an option to evaluate the slotted foil thing.

Not sure at this point if I want to dive in and try to start using it.

 

[P Flados]

Ok, here are some comparisons.

First lets look at a slotted wing with a Steve Clark Open Wing front element (as best I could extract from the PDF from SA) with a NACA 0009 behind it.

Some details are:

I used the 'painted fabric' surface for these. Smooth does not make any real difference in polar shape but does shift the results to the left some. My builds will probably not involve mold making, so less smooth seems more appropriate,

I found that the JavaFoil software can do the rear element rotation. I have attached a txt file with no rotation on the rear element. I selected a gap of 0.001 to just give clearance.

I played around with the pivot point and 0.49 seemed to work best. This really minimizes the gap when the rear foil is rotated.

The pictures below show what looks like a decent minimum rotation operating point for upwind and the rotation that produced max lift.

 

[P Flados]

When I switched software, I got some noticeably different results from XFLR5 in some areas. I like XFLR5 output better, but eventually got used to JavaFoil.

Here are pictures showing my double flapped S1223 based reversible camber foil in the max lift configuration. I also included one polar at 0 rear flap deflection for upwind sailing and a polar with a 14% flap deflected 26 degrees for max lift.

 

[P Flados]

And now for a summary post.

Slotted Wing

The slotted wing has a drag bucket with low drag values for both upwind and max lift configuraitons.

The upwind bucket has a lower Cl than I can get on my double flapped sections but it has a small range of good AoAs (a.k.a: angle of attack or alpha) and very sharp break points where performance suddenly gets terrible. Since we can not control AoA shifts, it looks like trimming should focus on getting a rear flap rotation so that overall AoA can keep you in the middle of a bucket.

The max lift setting also has a nice drag bucket region. However, when you need max lift, you do not need a wide bucket, you pretty much want to stay at the max lift point in the bucket region. Again, since we can not control AoA, the best operating point is probably closer to a Cl of 2.0 instead of the 2.2 where you are on the edge of flow separation.

Double flap Reversible Camber S1223M-10 foil

Before comparing it to the slotted OW_N9 above, I am going to compare it the the low RE / high lift foil that gave me the curvature for the front element, the Selig S1223 shown in the pictures below.

When I have the rear flap of the S1223M-10 foil set for max lift, the shape of the polar is pretty close to the S1223. The S1233M-10 is a little more rounded at the top of the bucket region and it actually has a higher max lift value. The min drag value for this setting not as good as the S1223, but the mid drag value for the S1223M-10 with no rear flap deflect is better than what the S1223 does.

When comparing the slotted OW_N9, the S1223M-10 will give a higher max lift but it comes with more drag. A really good high lift RE: 2e5 operating point for the slotted foil is the Cl: 2.0 that gives Cd: 0.027. At this Cl, the S1223M-10 has Cd: 0.04. However, the S1223M-10 has the advantage that it can get up to Cl: 2.4 at CD: 0.047.

For the upwind stuff, the slotted OW_N9 middle of the RE 8e5 bucket is a great Cl: 1.0 and Cd: 0.008. I not sure how to keep the wing operating in the middle of a bucket region if you have to get well below Cl: 0.8. With an upwind bucket Cd of around 0.017 the S1223M-10 does not do as good as the slotted OW_N9. However, given the extra thickness of the front section, I am not surprised.

General Notes

The slotted wing has proven popular both on the water and on the hard for competition sailing. My attempts at producing polars may be off due to my inexperience in these things, but they tend to show the ability to get good operation in both high lift modes and in low drag modes. I found the slotted wing to be very sensitive to gap setting and the results indicate that more than normal understanding and/or experience may be needed to get a slotting wing "trimmed" where it should be.

I think the double flapped reversible camber wing concept definitely has merit. Based on a number of discussion I have seen elsewhere (wing section drag such a small percentage of total drag) the extra drag at max lift may not hurt boat performance at all. If so, you either get up to the "powered up" operating point at a lower wind speed, or you can use a smaller wing to "powered up" at the same wind speed. If you use the same wing, the upwind operating point has a lot more section drag. Again this may not matter as much as for aircraft. If on the other hand, you go with a smaller wing the difference in drag is reduced. One real advantage, is the potential to go with a higher aspect ratio due to the thicker first element. The overall drag could be very comparable.

 

[CT 249]

He was referring to the airfoil sections for a two element slotted wing, not a wingmast. He was also talking in the context of 'any old average approx naca 0020 in front of any old approx naca 0009' where the 'any old' are decent non-laminar airfoils known to be pretty good.

For a general purpose slotted two element wing, a little more "blunted" at the front of the lead element may be good as it makes it more forgiving for the rotation that may be needed for max lift. I have come to wonder if the need for "blunted" airfoil shapes for boat wings has anything to do with how a well known individual came to be known as Blunted on SA.

I have seen a lot of difference in selecting 'wrong' shapes for given sections in what I have done.

Note however, that I am limited to a single wing profiles. I would love to be able to compare my stuff directly to the stuff Steve is talking about, but XFLR5 will not do it and I do not have a lot of reference material to compare to.

Mea culpa

 

[tspeer]

...When you mirror the S1223,
which point do you choose at the leading edge to start the symetric axis?
Do you choose it because y=0 after de-rotating, normalizing and so on...?
or for instance do you search for the theorical stagnation point for high AoA?

When I mirror a section to form a wingmast, I anchor the trailing end of the mast chord and then pick the point on the leading edge that is perpendicular to the mast chord. This will be a little to the windward side of the leading edge, and results in a smooth leading edge when reflected.

If you pick a point that is not far enough around the leading edge, you get a sharp crease. If you pick a point that is too far around the leading edge, you get a heart-shaped indentation.

I imagine P Flados does something similar.

 

[P Flados]

I imagine P Flados does something similar.

Yep.

FYI, I have been plugging away and have started to get more convergence between JavaFoil and XFLR5 for analysis of my foils.

Javafoil results change a lot if the tail does not taper to a point and the drag Bucket tends to have sharp edges. However, I am now getting about the same results when I use a sharp TE and juggle the flap settings around.

I am actually a little confused about results that may seem "too good" to believe.

For downhill / low wind, max lift is important and XFLR5 shows believable results. This is based on the XFLR5 evals of the real S1223 and my foil, and comparing these to the published wind tunnel stuff for the S1223. Javafoil shows a lot more max lift & I do not see this as real. Both Javafoil & XFLR5 show S1223M getting around a Cl of 1.8 with a decent Cd (0.03) even with low apparent wind dropping RE to 200,000. Javafoils shows more rotation giving higher lift but XFLR5 pretty much flattens out. Again, I am leaning toward the XFLR5 result.

For upwind or anytime you are powered up (have to back off on the main sheet due to righting moment limitations) I am starting to see that Cl/Cd really does not matter as much as just being in the middle of a drag bucket with a minimum possible Cd value. If possible you can twist off (reduce flap deflection for me) the upper part of the wing to eliminate lift up top and go for more lift down lower.

This all means that the minimum achievable Cd in a bucket from Cl of say 0.2 to 1.0 should probably be what I should use to measure upwind performance. I started playing with no small flap rotation (see "S1223M low drag" data and the picture on the left below) and tweaking around the main flap to see what I could get.

It looks like I can reliably get a Cd down below 0.01 from near no lift up to a CL of around 1.0 (XFLR5 says 0.8 and JavaFoil says 1.2). The "too good to believe" part is that this is not far off of what you would get with a ClarkY or a NACA 0012 and yet S1223M is 18% thick!

I am thinking that a low angle of convergence for the air behind the foil is a really big part of the puzzle to get good minimum Cl values.

 

[Petros]

you are saying you can get an L/D of about 100 to 1? with 18% thick section? I doubt it.

 

[P Flados]

you are saying you can get an L/D of about 100 to 1? with 18% thick section? I doubt it.

At RE = 1,200,000 I get values over 100 for L/D with quite a few foils.

For the results above, XFLR5 results are not quite Cl = 1.0 at just over Cd = 0.01 (L/D of mid 80s as shown on the output in the previous post) for this RE. JavaFoil is showing crazy high L/D values, up over 150.

Again, the Cl values under 0.01 surprised the heck out of me for a 18% foil.

I went ahead and generated up the XFLR5 pressure / BL picture for the above. The pressure distribution and air flow look pretty good to my untrained eyes. If I am misreading or misinterpreting some results, I am hoping for some coaching from the experts (again, thanks in advance).

 

[Petros]

at first glance I would say this shape is likely to suffer flow separation from about .25 to .5 of cord on the low pressure (upper) side. The slope looks a bit too steep, if it was a perfectly smooth shape and the air was clean, maybe not. But when the pressure drops that fast on the low pressure side, real airfoils tend to get large separation bubbles. If you intend a hinge there with an abrupt edge, it will certainly separate.

Also, the bump on the high pressure side would also likely separate, but not as severely as on the upper surface, the flow should reattach at about 0.55 or .6 on the lower surface.

You want to create a shape that generates a lot of low pressure on the upper surface, with preferable a fairly flat or rounded top (not a sharp spike), and than a long uniform ramp down to the TE. The area under the pressure curve is the total lift generated by the surface. The area where the pressure goes from max to the TE is called the pressure recovery zone, because the pressure is returning to free stream pressure. This is a very sensitive area because the air flow wants to detach. consider on the front 20 percent, the air is moving toward lower pressure, so it wants to stay attached, as if it was being "sucked" along the surface. But after the low pressure peak, the air pressure is increasing as it heads back to the TE and the free stream pressure, it wants to pile up and tumble kind of like a steam of water running into a rock in the stream. On the underside if it was flat like a clark-y, it would all be high pressure and heading toward the TE would be a drop in pressure, and also not prone to separation. but your irregular shape on the high pressure (lower) side also shows a low pressure region where when the air climbs out of it will likely separate.

That is why abrupt changes in shape are always bad for foils. And this is why a shape like that would not produce L/D of 100/1. You might get 12 or 15, which is better than the 3 or 4 that most sails generate.

 

[P Flados]

Petros,

Your discussion of how pressure distribution leads to separation matches up with guidance from Tom Speer. I have read his stuff forward and backwards.

From 0.25 chord to 0.5 chord is probably the region of most concern for separation that would cause problems. There are no abrupt changes in this region. The software sometimes shows a need for a laminar to turbulent trip. Generally I set this as the default configuration with a location very close to 0.25 chord (max thickness).
At low overall AOA (alpha) there is a little concave indention on the top side at 50%. The pressure distribution for low this range does seems to keep flow attached pretty good.

The basic shape that provides the pressure distribution for the forward half is the S1223 shape. This shape was wind tunnel tested. I searched around and also found that it has seen some real use. I did not read any reports of it being prone to separation in the region of concern. The pressure spike & initial recovery for this foil does not look as bad as a number of "standard" foils that have seen quite a bit of use.

For mid/high lift conditions, the main flap is adjusted for a top side very close to the S1223. No real "abrupt change" in curvature, but definitely a "flap discontinuity" at 0.5 chord. Back at the rear flap, all of the high lift setups do show separation at the rear hinge line. This hurts with respect to drag, but there is a big lift gain.

I have intentionally introduced imperfections in the software and got a drag increase but no gross separation. I also ran JavaFoil with all of the surface finish choices (from smooth to "bugs and dirt") and got no surprises. In XFLR5, I did Ncrit = 9 some, but mostly Ncrit of 1.

I started with other shapes and quite a few got into separation problems. Others just did not perform as well due to big pressure spikes up at very front. The ones with simple NACA 0026 to NACA 0030 up front did pretty good. Again, the NACA shapes are very proven. After Erwin pointed me to look at the S1223 paper, I saw that the top side looked very interesting. The forward curvature did the kind of pressure recovery I was looking for. From there to ~0.9 cord it was very flat, and then the last 10% looked a lot like something I could do with a second flap.

 

[Petros]

yes, the use of flaps was always done to improve low speed lift, without concern for drag since it was for a low speed landing condition. the extra drag on an otherwise clean aircraft is actually helpful to slow the airplane right at landing and prevent it from bouncing back into the air (a problem sail planes have, that is why they always use spoilers and even drag brakes and small droge chutes). It is problematic when trying to use a hinged flap on a rigid wing sail to adjust the camber, messes up the performance of the sail, L/D goes to crap.

Perhaps your pressure gradient is not so bad for a low pressure sail set up, I just find it hard to believe you can get L/D near 100/1 with that shape on a real sail. It would be much better not to have the sharp discontinuities.

That is why I think that using some form of "wing warping" to adjust camber, as in your earlier tests of a naca shaped mast with a flexible sail off the TE. This form of camber control was used on some very early aircraft for flight controls. They could adjust the foil shape for the flight conditions. As aircraft started going faster with more powerful engines, it was abandon for the more structurally stable rigid wing and flap. The military experimented back in the '80's with a specially built wing with flexible composite upper surfaces that could change shape on an F-111B, the wings had no flaps or ailerons. Control was by warping the wing shape. I take it was not practical since no aircraft ever used it, but I sure liked the idea. Which might be feasible on a sailboat since the speeds and forces are much lower than a fighter aircraft, so it might be possible to do it with a simple lightweight batten and pulley system I want to try on my next dingy build. It would be even better to have a mast shape that is optimized for the purpose, but a standard shape looks like worked pretty well, and good enough to see if the idea is feasible.

The real trick would be to make it appear enough like a conventional sail so others would actually consider using it. If it looks too much like an aircraft wing, it will not get wide spread acceptance, no one will buy it. So I figure if it looks like a slightly more refined conventional sail, but with some very different control devices, and it out performs conventional sails, than I might have something that I could sell. Sailboat customers want something that looks like a sailboat, not a half an aircraft sticking out of the water. This is why it is so hard to sell a new idea.