XFoil Modelling of Surfboard Fins

Hi Folks,

I am interested in modelling various airfoils for centre and side fins on surfboards in xfoil or XFLR5. I wanted to get some opinions on an appropriate testing methodology. There is very little information in surfing forums around this topic so I thought I would ask you guys.

My thoughts were around selecting efficient foils with maximum lift and minimum drag or maximum CL/CD. Is this analysis accurate enough or do I need to consider other factors?

A typical fin foil would have a base chord of 100 mm and I would be calculating Re based on speeds of 1, 3, 5, 7, 10 and 15 m/s. Reading some of the forums I am calculating my Re based on the formula Re=V*c/V. Is this a correct formula?

I would also assume I need to run Xfoil in viscous mode?

Do I need to use an Ncrit value or a mach number?

Am I missing anything else?

Any advice would be appreciated.

BTW has anyone had any experience using 3D CFD software such as OpenFOAM?

Thanks in advance.

TK

 

Surfboard fins are typically moderate to low in aspect ratio and swept. This calls for a 3D code, and preferably one that can handle vortex shedding from the leading edge. XFOIL is not capable of representing this kind of flow. XFOIL will calculate the attached flow around two-dimensional sections on surfaces that are of moderate to high aspect ratio.

There's a lot of uncertainty about what value of Ncrit is appropriate for use in open water, and the people I've talked to that have tried to create laminar flow bulbs and the like indicate that the value should be quite low because transition appears to happen at a much lower Reynolds number under these conditions than in air. Probably Ncrit = 1 to 3 would be a good guess.

Mach number should be quite low, too, because compressibility effects are negligible in water.

 

Actually, Tom, modern surfboard foils are quite high AR - see for yourself the olympic RS-X. Funny thing is that they only use the centerboard in light airs upwind - in planing conditions, they sail with the permanently fixed aft fin only.

 

Actually Mikko, what you are referring to and have pictured is a sailboard/windsurfer, which you are correct in pointing out have higher aspect fins. A SURFBOARD is a completely different kettle of fish, and Tom is absolutely spot-on in his assertion that surfboard fins are typically low to medium aspect and swept.

 

Then again you don't get much surf in Finland, so I suppose we can give you the benefit of the doubt with the semantic misunderstanding! Lol

 

Typical windsurfer fins have "laminar flow" sections that perform well at Reynolds numbers of 3 million but not so well at 100-800k. They stall easily and tend to stay stalled untill the angle of attack is drastiacally reduced. They also do not have much of a low drag bucket in common windsurfer fin conditions.

Wind tunnel tests, Xfoil, and real life use will show that the NACA 4 digit series foils are superior to the laminar sections on windsurfer fins. They have higher max lift and recover from a stall more quickly in addition to having lower drag. I built a replacement fin using NACA 0009 on my windsurfer a few years back. I was able to get the board up on plane sooner and point higher than with the original fin. Speed did not seem to be affected though in beam reach or downwind.

 

Have you decided on a planform? I thought that might be just as important as a
starting point as selecting an appropriate section shape.

 

Leo,
What can you tell me/us about your avatar?
PM or start a new thread please as I'm sure there is a good history on it, no?
OP, sorry for the interjection.
Thanks

 

First to boring, on-topic matters...

I asked about the surfboard fin planform because I came up with a
relatively simple equation to create a reasonable shape for a fin,
however, I haven't had time to modify my program to handle "half-wings"
like real fins or keels.

I posted a first draft of a compendium of planar wing results a fair while ago:
http://www.boatdesign.net/forums/at...ut-inclined-underwater-hull-form-lsp_flat.pdf
See page 27 of the draft report for Flukes I use in my "avatar" and
page 29 for my "fin-like" wing.

It would be interesting to see how VLM or Mark Drela's codes compare
with the predictions of induced drag. I know Mark is not convinced
that near-field methods do well in predicting Di, but I'm not
convinced yet that the method I use is all that bad for low-aspect
ratios (above about 1.0).

When John Hazel learns OpenFoam, maybe he can compare his results too.

I created it because it has elements of many things I research.
The swept curved wings are a difficult case for the lifting surface
equation solver I have been developing because of the very pointed
tips.

The central body is an airfoil, but shaded to look like an axi-
symmetric body. The "creature" could be a bird, or a fish, or a
hybrid. The shape of the tail means I can use two critters as
parentheses. I also have another one that looks like an abstract
"wink" or a whirly-gig.

You must be so happy that you asked me to talk about myself!

 

Leo and OP,
Thank you.

 

It is quite amazing how well your LSIE predicts forces for low aspect, swept & delta-type planforms. I guess it has to do with the fact that flow over this kind of lifting surfaces is always separated, dominated by vortex lift whether it comes from the LE vortex or the sides. I don't think that a VLM, or vortex lattice code, or a more general panel code either could predict well these planforms, unless they have extensions for vortex flow - they are basically potential flow codes. On the other hand I believe Lamar's VLM did quite well, with its edge suction corrections.

So you need something like Open Foam. I remember I did a RANS code prediction for the circular planform, which was not quite as good as your LSIE prediction, but the LB-code was good all the way to complete stall. Of course, these codes can take days to run, while yours probably does it in fractions of a second!

Likewise, I believe Michlet is hard to match in accuracy for predicting a very thin ship, while free surface RANS codes will do a fairly good job for most any shape of hull.

Can you give equations to easily produce your flukes, for instance?

 

The Olympic surfboard RS-X high-aspect fin, modeled from a laser scan of the real thing. I haven't looked closely at the section, but it does have max thickness further aft as in laminar sections, but no inflections in the rear part.

 

The flukes are just simple cosine curves meeting at the tips.
The "Sine-Swept Elliptical" wing is described by a sine wave along
the centre-line with an elliptical chord-length distributed around
it.

I'm not sure that I agree with you that the LSIE I use is very
accurate in general, but it is quite good for small AOA, which is just
what it was devised for. It also does quite well for delta wings
using the vortex extensions, but those extensions are not on very
solid theoretical ground. It's an engineer's hack, IMO, but not bad
for some wings.

I seem to recall that your modelling of the circular wing was much
better than LSIE for larger AoA. My only question with your CFD work
is how the results would change if you used different edges, e.g.
squared-off, bevelled, rounded etc. I guess you would have to spend a
lot of computer time to go through all the varieties! My code takes
several seconds, or a few minutes for fine meshes. I'll try to release
another version where users can input a leading edge curve and a trailing
edge, but I am trying to finish of Flotilla for cats and SES which
is taking up all my time at present.

VLM is "remarkably accurate" for some planforms, but it suffers from
some peculiar behaviour near the LE which we have discussed
previously. See:
http://www.boatdesign.net/forums/hydrodynamics-aerodynamics/circulation-real-46025-2.html#post610721

The kinks in the loading shown in the graph are difficult to see
unless you take out an inverse square root factor.
They occur even on simple rectangular wings, and they don't disappear
with a finer mesh. See:
http://www.boatdesign.net/forums/at.../78142d1359173252-circulation-real-lesing.gif

When integrating the loading to get the lift coefficient, those kinks
tend to cancel each other out, so CL is still accurate. They can,
however, affect Di and the leading-edge suction. I guess that's why
Mark Drela is very critical of near-field methods for estimating Di.

The rectangular panelisation of swept wings I use must seem very
peculiar to some people (squaring the circle AdHoc called it), but
there is some method in the apparent madness.

You can try to fit horseshoe vortices with a slanted bound portion, but
that involves some level of approximation. They are not exact for each
panel, and errors can be amplified when inverting the large matrix to
get a solution. Some problems can also arise when the bound portion (or
a line through them) comes close to a collocation point. Mark Drela
came up with an ingenious way to avoid those difficulties in some of
his codes.

Alternatively, you can use rectangular panels for which there are
exact formulas, but the panels don't fit the planform exactly.
Integrating the LSIE once in the x-direction avoids the difficulty
with collocation points I mentioned earlier. The kinks in the loading
can be eliminated using another ingenious technique developed by
D.W.F. Standingford in his PhD thesis.

I should also mention that Mark Drela's codes are far more ambitious
and much more useful for actual wings, i.e. ones with dihedral and
other important practical aspects. I am only interested in infinitely
thin, planar wings which only exist in geek heaven (where I live)!

 

2-3 decades ago there was a craze in the aerodynamics world regarding the "crescent planform." Other names are "Schuman planform" and "swept tip planform." Interesting stuff but watch out. Some of the conclusions are not so good. Some guys were applying codes that were known to be poor performers in the vicinity of wing tips and drawing conclusions about the magic of swept tips. (VSAERO) There was also some other voodo going on with wind tunnel results by what was pretty shakey data massures whose goal seemed to be the next research grant. So look at that stuff with a grain of salt.

 

The results first looked very promising but, IIRC, as finer and finer meshes
were used their advantage evaporated. And, as you said, wind tunnel results
were not particularly accurate.

Nature might favour very fine wingtips on swallows, swifts, and some other
creatures, but their wings also flex and twist, so steady results with planar
wings are very suspect. They are, however, a good severe test for many
lifting surface codes.