Lift and drag coefficients for sails and foils

[poul]

Hi all
I have some different ideas for various wing sails combinations, and I would like to know some comparable lift and drag coefficinets for the follwing.

normal sail Cl =1.5 Cd = 0.15 Cl/Cd = 10 ??

Wingmast-sail

rigid symetrical foil

rigid asymetrical foil

rigid symetrical foil with flap

You might want to pick up a copy of "Sail Performance " by Marchaj. It has a lot of good information on lift to drag ratios of soft and rigid sails and explains why the rigid sail isn't going to replace the soft sail.

[tspeer]

Originally posted by poul
Hi all
I have some different ideas for various wing sails combinations, and I would like to know some comparable lift and drag coefficinets for the follwing.

normal sail Cl =1.5 Cd = 0.15 Cl/Cd = 10 ??

This will depend greatly on the specific design of the sections, the Reynolds Number (wind speed compared to the chord length), the planform shape, and the type of rig. For example, a sloop rig will probably have a higher CLmax and may have more drag than a cat rig of the same height. And adding a jib to a well designed wingmast may not improve the performance at all over putting the same area into the mainsail.

Another problem you have is you must compare the different types of sections using data from similar sources. It is unwise to compare wind tunnel test data for one section with computational data for another section, for instance, or even between computational codes. XFOIL will do a better job of predicting maximum lift and drag than the older Eppler code, and even XFOIL may be optimistic with regard to maximum lift compared to test data.

Another huge factor is the maximum lift for a two-dimensional section is very different from a three-dimensional wing sail. So you have to know quite a lot about the whole configuration to make a valid comparison.

Wingmast-sail

I've been able to use XFOIL to compute the characteristics of tear-drop type wingmasts. The maximum lift varies radically depending on how the mast is trimmed. For example, a 10% chord wingmast with the sail shaped so the lee side has the same contour as the Clark Y airfoil is predicted to have 2D maximum lift coefficient of 2.1 with ideal trim at a Reynolds number of 1 million. The exact same mast and sail shape is predicted to have a maximum lift coefficient of 1.2 when the mast is rotated to zero degrees (in line with the boom). Rotating the boom past the point where the lee side contour is smooth results in a significant drop in Clmax.

However, even drastic changes in mast design can have surprisingly little effect. Changing the mast size from 5% chord to 40% chord is predicted to have a change in Clmax of about 0.06 so long as the lee side contour is smooth and of the same shape.

No attempt was made in these wingmast designs to achieve the highest lift possible. One could do better with a specific design or by basing the wingmast and sail shape on the lee side contour of a high lift airfoil. The real question is, "What sail shape is achievable in practice?" With today's molded sails and high modulus materials, it may be possible to come very close to the design shape.

So for a wingmast + sail, the most important thing is to know the shape of the mast and sail, and to trim the mast so the lee side has no sharp kinks. I have never seen the precise mast and sail shape tested in the wind tunnel reported in the literature.

rigid symetrical foil

Two dimensional airfoil data may be found in the literature. The classic reference is NACA Report 824 (http://naca.larc.nasa.gov/reports/1945/naca-report-824/), also published in expanded form as "Theory of Wing Sections" by Abbott & von Doenhoff (Dover Books). There have been few symmetrical sections that have been designed for high lift. There are definite limits to what one can do in this area because all you have to work with is the thickness distribution.

The FX 71-L-150 has a maximum lift coefficient of 1.2 at a Reynolds number of 1 million (Wortman, Franz Xaver, "Stuttgarter Profilkatlog I, Friedr. Vieweg & Sohn, 1981).

However, XFOIL is very capable of designing sections of this type. 2D maximim lift coefficients will probably be in the range of 1.2 - 1.4. Less at lower Reynolds numbers.

rigid asymetrical foil

The maximum theoretical lift coefficient I've seen for a two-dimensional, single element section is 3.06 at a Reynolds number of 5 million. This is not a practical section as it has almost no thickness, but it gives a useful upper bound. More practical cambered sections have been designed with maximum lift coefficients on the order of 2.2 - 2.6. (Liebeck, R. H., "A Class of Airfoils Designed for High Lift in Incompressible Flow," AIAA Journal of Aircraft, Vol. 10, Oct. 1973, pp. 1 - 16)

For example, the Eppler E423 is predicted (Eppler code) to have a maximum lift coefficient in excess of 2.0 at a Reynolds number of 1 million (Eppler, Richard, "Airfoil Design and Data," Springer-Verlag, 1990).

The higher the design lift coefficient, the thinner the section becomes. Ultimately, the section becomes very close to that of a wingmast & sail.

rigid symetrical foil with flap

This depends on whether you are considering plain flaps (no gap at the hinge line) or slotted flaps formed by two or more symmetrical airfoils in tandem. A 30% chord flap will raise the mzximum lift coefficient of the FX 71-150 to 1.8 at a Reynolds number of 1 million.

I know of no published data besides that on my website for symmetrical sections with slotted flaps designed to be used as rigid wing sails. The methods used to calculate these data are VERY crude andthe maximum lift coefficients are HIGHLY questionable. I was also interesed in fairly low Reynolds numbers. At the higher Reynolds numbers of the data quoted above, the maximum lift coefficients would be substantially greater.

One of the best combinations from landsailing competition has proven to be two NACA 0012 sections, with the forward element 60% of the chord and the flap 40% of the total chord. The hinge line is normally located at about 90% of the forward element chord, and the links extending from the flap leading edge (through slots in the forward element trailing edge) sized so the flap just clears the trailing edge of the forward element. A flap deflection of 20 degrees will typically provide best performance. Higher flap deflections may increase CLmax, but the drag also increases rapidly from, say, 30 degrees to 40 degrees flap deflection.

In general, one can use a thick section for the forward element, but the flap should be a thin section. The thin flap section would normally produce sharp pressure peak at the leading edge that would lead to leading edge stall. However, when used as a flap, this pressure peak can be suppressed by moving the flap closer to the trailing edge of the forward element, making the gap smaller. There is an advantage to having the flap and forward element overlap somewhat when the flap is deflected. However, this adds complexity to the mechanism to keep the two elements from interfering when tacking.

Maximum lift coefficients for multi-element rigid wing sections are probably in the range of 2.5 - 3.0+. One must do considerable tuning and adjustment of the slots and flap deflections to optimize the performance.

As a practical matter, I belive it is possible to design a wingmast & sail that is competitive with a rigid wing under most conditions. The A-class catamarans seem to have settled on soft sails while the C-class cats have settled on rigid wings. From what I've seen, a single slotted flap is probably comparable in maximum lift and a double slotted flap will probably have a higher maximum lift than a wingmast & sail. However, the wingmast may have less drag at high lift and will most likely be lighter than the rigid wing and better able to accommodate a wide range of conditions. In landyachts, the best of the rigid wing rigs and the best of the wingmast rigs are neck-and-neck with each other.

As important as the section design is, once one has a reasonably well performing section the more important thing is to get the distribution of lift correct along the span of the rig. A major factor in Cogito's success in winning the C-class Challenge Trophy ("Little America's Cup") was the ability to twist the wing to optimize the spanload distribution. This reduces the induced drag and also allows the different sections from top to bottom to perform near their best angles of attack. This is more important than the differences between one section and another.

[poul]

Thanks Mr. Speer for your reply. I've checked out your HP and it is the only site that I've found sofar that provides some serious infomation on the subject. Even though my fluid dynamics theory is a bit rusty, it seemes to me that the few commercial sites I've have seen have some rahter strange conclusions on how and why thier concept works, in some areas atleast.
I know my initial post was a bit short, but the idea is to have a 360+ rotating mast or bi-pod, this allows the mast/wing to be big without the windage problem when lying i harbour. But it also means that the mast/wing have to be unsupported . The mechanical problems in having 360+ degrees rotation is relativily easy to solve, but the structual and weight of the rig problems is more difficult to answer. therefore I have to make a feasibility study to see gains and the loses. Right now I think that a 2 masted catamaran is most likely to succeed, but in a few months when I get more time, I will start to play with X-foil, and I also have acces to ansys-flowtran to check the 3d effects, and hopefully discover a genious concept or maybe just forget the hole thing

[tspeer]

Originally posted by poul
...it seemes to me that the few commercial sites I've have seen have some rahter strange conclusions on how and why thier concept works

I agree! So many explanations of fluid dynamics sound like they are right out of Rudyard Kipling's "Just-So Stories".

...the idea is to have a 360+ rotating mast or bi-pod, this allows the mast/wing to be big without the windage problem when lying i harbour.

I suspect you can do quite well with a stayed wingmast that turns +- 90 degrees, if a tail boom is added to the mast with a stabilizing surface, and possible a counterweight forward to ensure the ensemble is mass-balanced at or ahead of the pivot axis. This would keep the mast turned into the wind as gusts hit the boat, and the boat would probably swing on its mooring enough to keep the wind within the forward quarters.

The key is to ensure that the natural frequency of the mast's weathercocking is high compared to the roll and yaw response of the boat. I believe this will inhibit the boat's tendency to sail back and forth at anchor.

...Right now I think that a 2 masted catamaran is most likely to succeed...

There has been some recent discussion of biplane rigs on the Multihulls mailing list recently (http://www.steamradio.com/mailman/listinfo/multihulls). Actual experience of biplane rigs has been very mixed.

[sigurd]

Tspeer said:

The maximum theoretical lift coefficient I've seen for a two-dimensional, single element section is 3.06 at a Reynolds number of 5 million. This is not a practical section as it has almost no thickness, but it gives a useful upper bound. More practical cambered sections have been designed with maximum lift coefficients on the order of 2.2 - 2.6. (Liebeck, R. H., "A Class of Airfoils Designed for High Lift in Incompressible Flow," AIAA Journal of Aircraft, Vol. 10, Oct. 1973, pp. 1 - 16)

So this Javafoil estimate is totally bogus then? Could there be something in the settings that I did wrong?

[daiquiri]

Originally Posted by sigurd

So this Javafoil estimate is totally bogus then? Could there be something in the settings that I did wrong?

From the pics included it looks like your software uses potential flow field to calculate the velocities and then somehow estimates the friction drag only. I see no flow separation over the upper surface of the airfoil, which is totaly unrealistic given the huge angle of attack of the section.
The potential field methods, without estimates for boundary layer thickness and separation, are valid only for relatively small AoA's (+/- 7-8 degrees around zero), where lift curve is linear and no significant boundary layer separation is happening yet.
I would absolutely not trust results from that software for angles as huge as 30°.

P.S.
How come it reads -30° when the foil is clearly at a positive AoA??? And how come the maximum CL is at 0° ?

P.P.S.
I've just tried the online version of Javafoil and it does include boundary layer thickness and separation. And the angles of attack are correct and consistent with the visualization of the airfoil. Maybe if you can attach the .txt file with the coordinates of your foil, I could try to repeat your simulation. I'm really curious about the results you have obtained...

[Tcubed]

It would appear that S rotated the foil on the 'modify' tab and so that skews the apparent AoA. It seems the program uses the original frame of reference to calculate AoA, not the actual chord line.

I myself would be interested to know what are the particular limitations to the javafoil program, in particular how much error to expect when modeling low aspect ratios in the region of 1 to 3 , say.

So far i do not get as much of an increase in stalling AoA as i would expect from such low aspect ratios.

[Guest625101138]

Originally Posted by Tcubed

It would appear that S rotated the foil on the 'modify' tab and so that skews the apparent AoA. It seems the program uses the original frame of reference to calculate AoA, not the actual chord line.

I myself would be interested to know what are the particular limitations to the javafoil program, in particular how much error to expect when modeling low aspect ratios in the region of 1 to 3 , say.

So far i do not get as much of an increase in stalling AoA as i would expect from such low aspect ratios.

It gives good results for low to moderate RE# foils in 2D for their normal range of operation. I have compared it with Selig test data on a number of foils and I could not pick a significant difference. I rarely use the correction JavaFoil has for aspect ratio so cannot comment on its accuracy.

There used to be a Japanese site with a great deal of foil data but it no longer offers access to the database. I got the impression it may have breached copyright.
http://www.nasg.com/afdb/list-polar-e.phtml

You can get co-ordinates from here:
http://www.ae.uiuc.edu/m-selig/ads/c...atabase.html#N
These can be pasted into JavaFoil. If you search around you can find polar data on certain sections. The MA409 foil is a popular low Re# foil and you might find data on this. I have Selig wind tunnel data for it.

I found these crude polars for a Clark-Y:
http://marlongofast.tripod.com/zippe...tes/clarky.htm

So you can do your own analysis and see how JavaFoil performs. I have found it to be quite reliable for the things I do but I am long past the maths involved. Mark Drela or Leo Lazauskas may be able to comment. You could also ask Martin Hepperle. He has some detailed notes on the basis of derivation and likely limitations. I really appreciate the work LL and MH have done and the utility of the software they offer. I do feel guilty for shamelessly using it without any requirement to pay.

From my comparisons with measured data the curves are more jagged in JavaFoil than the test data but this could simply be that JavaFoil gives more precise determination of discontinuities that cannot be measured in real life or real life irons out somewhat through perturbations around the discontinuity or measurement system inertia (smoothing). Typically JavaFoil will give slightly higher L/D before stall than what comes out of the test data. This could also be a function of how well the foils are cut and/or the resolution of data used to enter the foil section. The detail around the leading edge of thin foils is critical to this of course.

Rick W

[sigurd]

Tcubed is right, except I did not rotate, but camber it each 5% step from TE to 70% from TE.

One thing I have noticed is that if you scale a section, the coefficients change. It appears it does not use the new chord in the CL calc. (this section was not scaled)

I did not save the foil but I think I have created one that is the same. Here it is.

[Guest625101138]

Originally Posted by sigurd

...
One thing I have noticed is that if you scale a section, the coefficients change. It appears it does not use the new chord in the CL calc. (this one is not scaled)

......

Probably worthwhile advising MH that it does this so it can be corrected.

Rick W

[sigurd]

oki I did now.

[sigurd]

oki I got an answer. The scale thing is as should be, that is how you do in aerodynamics apparently, if you put a flap on a wing the area for calculation does not change. So the multi element foils' coeffs gets inflated. Me I think it is stupid since when considering a jib and main for instance, you calculate from total area, so several definitions of CL and CD coexists and you would have to know which one is used.

For the CL he says CLmax is troublesome and a hot topic in cfd, but maybe xfoil can give a more accurate answer.

For the 3d thing you asked, he says it's a 2d program and very low AR will probably not be accurately calculated.

Maybe he allows me to quote his email, his explanation is much more detailed.

[sigurd]

Sigurd,

yes, JavaFoil has some limitations and some design features
which may not be optimum for all applications.

All coefficients are using the given geometry, usually the
chord length is equal to 1.0. In case of a flapped airfoil
however, with a flap extending like a fowler flap, the overall
chord length increases above 1.0. Nevertheless the aero
coefficients are referred to the original unflapped chord
length - this is the accepted usual way to handle such cases.
Of course you can always scale to whatever chord length you
need. So this behavior is by design and will not be changed.

One must also be clear that JavaFoil can only indicate the
onset of separation and not really calculate stall.
This can partially be done with Xfoil, but at the cost of
being less robust and requireming more analysis time.
Even more accurate predictions can be made with Navier-
Stokes methods, but being even less robust and requireing
much higher computational cost.
Even with these "state of the art" CFD tools, Clmax is still
a big problem and not predictable for general cases.
Usually people tune their codes using experimental data so
that they can simulate a certain class of cases, e.g. for
the wing of a transport aircraft quite well. But when it comes
to a different configuration the result often is still quite
poor. So: Clmax prediction is still a very hot topic in hichlgy
sophisticated CFD research and while there are promising
results there is not yet a general model which works like
the "real worls". Finally: stall usually has a 3D-pattern,
so 2D section analysis is only part of the truth.

If it comes to thin airfoils or sails, one must be very
careful, as almost all panel methods run into numerical
difficulties when upper and lower surfaces are close together.
For a real sail simulation one would need to write a special
panel code which can handle thin, single sheet airfoils
accurately. With the xisting codes like JavaFoil, Xfoil, etc.
one should model a sail as a quite thin, airfoil, but not
having zero thickness.

Finally it seems to be that if you use too many panels, the
results do not become better - good numbers seem to be as low
as 60-120 coordinate points. If the number is set to high
values like 200 or more, as it is sometimes required in
other lower order codes, the drag calculation seems to become
unrealiable. I will look into this behavior, but this may
take some time.

Concerning low aspect ratio wings, it is clear that a simple
2D-tool cannot produce accurate data on complex 3D- wing
shapes, especially for low aspect ratio. For such a purpose
one would have to use more complete 3D models which require
more computational effort. The attached draft document sheds
some light on the aspect ratio corrections as implemented in
JavaFoil.

Concerning the performance of rigid 2D multi-element wing
sections of airlines, these achieve Clmax of about 3-4,
whichb yields a Clmax for the 3D wing of about 2.5. These
are slat-main-flap configurations.

Regards,
Martin

The forum software requires me to write something, but I cannot think of anything, but I like this graemlin:

[Tcubed]

Great ! I was going to write to him myself, but you did already.

So it is as i suspected, just an approximate aspect ratio correction.

I know that as aspect ratio decreases so does Cl max but at greater AoA, and it gets very high . For example with AR of 0.2 say it becomes almost impossible to stall, because the whole thing is essentially operating within its own tip vortex and this makes the flow "wrap" around.

Does anyone know anything about "Flow-3D"? It's another Cfd app.