Boats and foils comparison

MaxHugen

Super Anarchist
@MaxHugen We have a picture of TNZ with the top of the sail inverted providing RM.  So they can use far more power (which I think we agree they have?) to overcome increased drag from cavitation etc??
I don't know why this is not considered and evaluated more?  It seems the analysts on here have written the inverted top off for some reason?

Am I missing something?
Have you a link to to that pic, or is it just the one where the entire top half is luffing?

I've kept an eye on every overhead shot during races with at last mid strength wind and haven't seen any inversion. Nor on the Volvo Ocean 65s during in-port races etc in strong wind conditions.

At this point I'm inclined to disregard inverted sails.

 

MaxHugen

Super Anarchist
Does chord length (and Re) have any effect on the onset of cavitation or on how cavitation affects drag?

Just wondering if the real benefit of the T foil may have been the simpler flap mechanism allowing reduced thickness and therefore chord length -> less adverse effects from cavitation.
A reduced thickness definitely helps, as it can have a flatter top surface and more even low pressure distribution. Probably more important than chord length.

A profile I tried has a decent distribution over the top, but may be too thin for structural reasons. It would also help if it produced some more lift from the aft half of the lower surface, but that increases drag significantly.

This profile is at 60 knots with a decent CL of 0.64 and L/D of 37, but the Cpmin is rather low at -0.52, so I have to check if this is still below the water vapour pressure.

image.png

 
Does chord length (and Re) have any effect on the onset of cavitation or on how cavitation affects drag?

Just wondering if the real benefit of the T foil may have been the simpler flap mechanism allowing reduced thickness and therefore chord length -> less adverse effects from cavitation.
Reynolds number has to do with the boundary layer and cavitation is not fundamentally a boundary layer phenomenon (although it can affect the boundary layer).

Chord length, per se, does not affect cavitation - only the pressures do.  However, for a given physical thickness, it is easier to design a foil with a higher cavitation onset speed if the chord is larger because then the thickness ratio is smaller.  More chord for the same span also means more area and less loading, which makes it easier to avoid cavitation at takeoff.  So while chord by itself does not directly affect cavitation, a longer thinner section will help.

 
A reduced thickness definitely helps, as it can have a flatter top surface and more even low pressure distribution. Probably more important than chord length.

A profile I tried has a decent distribution over the top, but may be too thin for structural reasons. It would also help if it produced some more lift from the aft half of the lower surface, but that increases drag significantly.

This profile is at 60 knots with a decent CL of 0.64 and L/D of 37, but the Cpmin is rather low at -0.52, so I have to check if this is still below the water vapour pressure.

View attachment 435545
If you want to use Xfoil to design a base ventilated section, take a look at this thread in which Prof. Mark Drela responded to my first attempt at a base ventilated section with his own approach and design.  He has some good advice on how to set some Xfoil parameters to better represent the cavity.

Here is the design he came up with.  It could be good for 65 to 80 kt (with sweep)!

bv02.png

View attachment bv02.txt

 

MaxHugen

Super Anarchist
If you want to use Xfoil to design a base ventilated section, take a look at this thread in which Prof. Mark Drela responded to my first attempt at a base ventilated section with his own approach and design.  He has some good advice on how to set some Xfoil parameters to better represent the cavity.

Here is the design he came up with.  It could be good for 65 to 80 kt (with sweep)!

View attachment 435547

View attachment 435546
Thanks, will check the links out! :)

 

MaxHugen

Super Anarchist
If you want to use Xfoil to design a base ventilated section, take a look at this thread in which Prof. Mark Drela responded to my first attempt at a base ventilated section with his own approach and design.  He has some good advice on how to set some Xfoil parameters to better represent the cavity.

Here is the design he came up with.  It could be good for 65 to 80 kts
I've been searching for a while, but I haven't found any info on what the "Kdl parameter" is, in relation to XFoil or XFLR5.

Could you provide me with some info - or a source - pls?

 
I've been searching for a while, but I haven't found any info on what the "Kdl parameter" is, in relation to XFoil or XFLR5.

Could you provide me with some info - or a source - pls?
I had to get Dr. Drela to help me out finding it, too, in that thread to which I linked!  

In Xfoil, it's in OPER->VPAR->LAG.  Then step through the parameters.  Kdl is the last (third) one.  (You have to use OPER->VPAR->? to get the menu that shows the LAG command.)

As to what it is, here is what Dr. Drela had to say:

"Re: Ventilation bubble modeling.
You can reduce the turbulent wake dissipation via the Kdl parameter in the VPAR menu, LAG command. Increasing Kdl from the default 0.9 value will make the wake recirculation bubble mix out more slowly, which I think better mimics a ventilation bubble. Second PDF corresponds to Kdl=4.0 which gives a much longer bubble."

''Klag and Kdl are two different parameters.

Klag = 5.6 always.
Kdl = 0.9 for a normal wake. Make a lot bigger for a non-dissipative wake, which is closest to a base ventilation bubble."

"Now that I think about it, a ventilated base bubble open to the atmosphere must have Cp=0 or nearly so. In reality, the bubble will assume whatever shape it has to in order to achieve this Cp=0 over its length.

The closest way to simulate such a bubble in XFOIL is to essentially turn off the wake viscous dissipation with a very large Kdl (I'm using 8.0 now), and shape the rear of the airfoil to get very nearly Cp=0 in the near wake.

I played around with this idea, and came up with the attached BV02. It's heavily based on transonic airfoil technology, with a deep center "spar" section with little loading, and loading in the front and back. The L/D is a ridiculously huge 20:1. This is even with a generous 6.8% thickness, and min(Cp) = -0.18, which should be good to 65 knots, or 80 knots with some modest sweep. Yowsah!"

 
Last edited by a moderator:

MaxHugen

Super Anarchist
I had to get Dr. Drela to help me out finding it, too, in that thread to which I linked!  

In Xfoil, it's in OPER->VPAR->LAG.  Then step through the parameters.  Kdl is the last (third) one.  (You have to use OPER->VPAR->? to get the menu that shows the LAG command.)

As to what it is, here is what Dr. Drela had to say:

"Re: Ventilation bubble modeling.
You can reduce the turbulent wake dissipation via the Kdl parameter in the VPAR menu, LAG command. Increasing Kdl from the default 0.9 value will make the wake recirculation bubble mix out more slowly, which I think better mimics a ventilation bubble. Second PDF corresponds to Kdl=4.0 which gives a much longer bubble."

''Klag and Kdl are two different parameters.

Klag = 5.6 always.
Kdl = 0.9 for a normal wake. Make a lot bigger for a non-dissipative wake, which is closest to a base ventilation bubble."

"Now that I think about it, a ventilated base bubble open to the atmosphere must have Cp=0 or nearly so. In reality, the bubble will assume whatever shape it has to in order to achieve this Cp=0 over its length.

The closest way to simulate such a bubble in XFOIL is to essentially turn off the wake viscous dissipation with a very large Kdl (I'm using 8.0 now), and shape the rear of the airfoil to get very nearly Cp=0 in the near wake.

I played around with this idea, and came up with the attached BV02. It's heavily based on transonic airfoil technology, with a deep center "spar" section with little loading, and loading in the front and back. The L/D is a ridiculously huge 20:1. This is even with a generous 6.8% thickness, and min(Cp) = -0.18, which should be good to 65 knots, or 80 knots with some modest sweep. Yowsah!"
Thanks for contacting Dr Drela!  I noticed in his tests with BV03 that increasing Kdl to 4.0 slightly reduced CL, but there was a major reduction in drag - 22.5% less!

If I understand all this correctly, due to the small difference in pressure between the bubble and the free flow, the rate of dissipation is low, so the bubble wake is longer.

Although the pressure variation is higher over the surface of the foil, it is still relatively low, so turbulence influences the boundary layer less, hence less surface drag and the much lower Cd?

Makes me wonder if a profile that started cavitation at the LE and created a very thin "layer" over the top surface of the foil would reduce drag even further, without drastically reducing lift. Or maybe the lift can be obtained primarily from the lower surface, although I guess this would bump up surface drag there, perhaps negating the reduction on the top surface?

 
The problem with base drag is you can get any answer you want, depending on just what the base pressure is.  Using the wake lag parameters in Xfoil to model a ventilated wake is really a  kluge as the ventilated foil is way outside of the fundamental assumptions behind Xfoil's methods.  I wouldn't put much stock in the drag numbers you get.  I think the lift and pressure distribution results are probably more accurate.  So Xfoil is a good way to design the section shape, but you're going to need a more sophisticated CFD method to get a good estimate of performance.

There have been supercavitating sections with the contour of the upper surface near the edge of the cavitation bubble.  The upper surface pressure will be at vapor pressure, of course, and will remain so as the angle of attack increases from the ideal angle of attack.  I don't know what the drag implications are of cavitated flow vs a wetted surface without cavitation.  Sailrocket reported a lot of pitting of the surface of their base ventilated foil, so it must have been experiencing some form of cavitation, possibly bubble cavitation in addition to the air ventilating the base.

I believe you'll see these kinds of sections have a very narrow range of operation.  That may be acceptable for a record-breaker that only needs to be concerned with performance near the speed record, but for a course-racing yacht performing brilliantly when it gets just the right conditions while being a dog otherwise is not the way to win regattas.  To have a high probability of win, you'd rather win by a little in nearly any condition rather than win big in only some conditions.

 

The Advocate

Super Anarchist
Have you a link to to that pic, or is it just the one where the entire top half is luffing?

I've kept an eye on every overhead shot during races with at last mid strength wind and haven't seen any inversion. Nor on the Volvo Ocean 65s during in-port races etc in strong wind conditions.

At this point I'm inclined to disregard inverted sails.
He has posted an image on the twin skins thread. Just looks like over powered luffing to me. Not buying into an increase of RM due to reversed forces.

 

MaxHugen

Super Anarchist
He has posted an image on the twin skins thread. Just looks like over powered luffing to me. Not buying into an increase of RM due to reversed forces.
Yes, saw the post. That was the image I also alluded to.

@Mikko Brummer posted this excellent pic that he took with a masthead mounted camera on a Finn. It showed an incredible "S" curve in the upper section, which the Finn sailors do in fact use for negative force in high wind strengths for RM, possibly way beyond their original design limits. Mikko was talking about sailing the Finn in 30 knots TWS, IIRC !

image.png

I mocked up an S shaped profile and tried it in XFoil:

image.png

Even this rough attempt produced a surprisingly good CL of 0.44 and L/D of 52, so would certainly contribute a lot to RM.

However, the AC75's mast rotation would play havoc with using this technique. In the above pic, imagine the forward ~10% pointing "up" - ie rotated clockwise. That would be the angle of mast rotation as used lower down in the "power" section of the sail.

I haven't seen a vid or pic showing this being tried by the AC75 while racing, but then there's been a dearth  of race days with strong winds, so maybe?

There's a difference between active luffing for RM like in the Finns, and just, well, luffing...

IMO :rolleyes:

 

MaxHugen

Super Anarchist
The problem with base drag is you can get any answer you want, depending on just what the base pressure is.  Using the wake lag parameters in Xfoil to model a ventilated wake is really a  kluge as the ventilated foil is way outside of the fundamental assumptions behind Xfoil's methods.  I wouldn't put much stock in the drag numbers you get.  I think the lift and pressure distribution results are probably more accurate. .... I believe you'll see these kinds of sections have a very narrow range of operation.
Yes, I'm going nowhere fast down this track.

I reverted to using your methods for reducing the onset of cavitation. My latest design can get to 50 knots without Cpmin reaching the water vapour pressure threshold - if I did the calcs correctly - but at just over 6% width it's probably not feasible from an engineering standpoint.

Tried a different way of hinging a flap, hoping to improve lift for such a thin foil. Thinking along the lines of a wingsail and the slot between the wing and flap... do you think there's any merit in pursuing this?

image.png

 
Last edited by a moderator:
Yes, there's a lot of merit in considering a slotted flap.  INEOS actually used one on on their AC75.  

I'd put the pivot location in a different place, however.  It would be better to put it under the leading edge of the flap.  That is the classic location for a Fowler flap.  It moves the flap aft as it rotates down, increasing the planform area.  

You're going to need something other than Xfoil to design a slotted flap because Xfoil can only handle single element sections.  The only free code I know for slotted flaps is Javafoil.  There are some things that Javafoil can't do, like predict stalling by means of wake bursting, but maximum lift is probably not a big issue for a hydrofoil that is constrained by cavitation.

 

MaxHugen

Super Anarchist
Yes, there's a lot of merit in considering a slotted flap.  INEOS actually used one on on their AC75.  

I'd put the pivot location in a different place, however.  It would be better to put it under the leading edge of the flap.  That is the classic location for a Fowler flap.  It moves the flap aft as it rotates down, increasing the planform area.  

You're going to need something other than Xfoil to design a slotted flap because Xfoil can only handle single element sections.  The only free code I know for slotted flaps is Javafoil.  There are some things that Javafoil can't do, like predict stalling by means of wake bursting, but maximum lift is probably not a big issue for a hydrofoil that is constrained by cavitation.
I wonder if it was the W foil where Ineos tried the slot, it has some unusual protusions on the underside which could be offset pivot points. My attempt to get some aft movement didn't work out really, will have a look at the Fowler flap mechanism. Being restricted to one rotation axis makes it difficult.

Reading the JavaFoil manual to see how to use a multi-element profile.

An ongoing issue I have is trying to convert a concept done in SVG to a set of x-y coordinates.  XFLR5 has a very basic design function, but with only 12 splines it's difficult - and sometimes impossible - to create the foil profile there.

Do you know of any way to convert SVG to coords, or a different spline-based app that can export to x-y coords?

image.png

 

amc

Anarchist
533
102
Auckland
I wonder if it was the W foil where Ineos tried the slot, it has some unusual protusions on the underside which could be offset pivot points. My attempt to get some aft movement didn't work out really, will have a look at the Fowler flap mechanism. Being restricted to one rotation axis makes it difficult.

Reading the JavaFoil manual to see how to use a multi-element profile.

An ongoing issue I have is trying to convert a concept done in SVG to a set of x-y coordinates.  XFLR5 has a very basic design function, but with only 12 splines it's difficult - and sometimes impossible - to create the foil profile there.

Do you know of any way to convert SVG to coords, or a different spline-based app that can export to x-y coords?

View attachment 435852
Max here is a photo of a Fowler flap hinge setup. 

67DA03E2-6BDE-4311-8337-3DA2B90AD0EE.jpeg

 
...

An ongoing issue I have is trying to convert a concept done in SVG to a set of x-y coordinates.  XFLR5 has a very basic design function, but with only 12 splines it's difficult - and sometimes impossible - to create the foil profile there.

...
I don't use XFLR5, so I can't help you there.  But here's how I generate coordinates for Xfoil.  It's perhaps an overly manual process, but it's not so onerous that I've been driven to automate it more fully.

My CAD program is Rhino.  I can import a vector PDF or an SVG.  If I have a graphic format it can't import, like a raster PDF or JPEG, then I often use Inkscape to trace it and create an SVG.

Once I have the shape defined as curves in CAD, I use Rhino's Convert To LInes command to convert the spline into a polygon.   Then I extract the points and export them as a points text file.  I find it helps to export the top and bottom contours separately.  I then read the point files into Excel. 

In Excel I normalize the coordinates by the chord length and sort them back to front on the upper surface and front to back on the lower surface.  (It's because of the sort that it helps to export the two surfaces separately.)  Then I delete everything but the normalized coordinates and save as a text file.  The final step is to use a text editor to change the tab characters to spaces because Xfoil hates tabs.

I load the coordinates into Xfoil and execute PANE at the top level to interpolate for the desired number of coordinates.  Then I'll get a quick pressure distribution to see how it looks.  This is often quite oscillatory due to errors in the original coordinates.  I use Xfoil's MDES mode to smooth through the pressure distribution and create at new foil that is fair.  The difference between the original shape and the new shape can often be too small to see with the eye, but it makes a definite difference in the pressure distribution.  If needed, I will use the GDES mode to set the thickness and camber if those are definitely known for the section.

When I want to take a modified profile back to CAD, Xfoil has a top level RSAVE command (which is undocumented, btw) that will write out a command file that can be executed in Rhino to create the shape.  Or else I will strip the name off the Xfoil coordinates file, import it into CAD as a points file, and pass a spline curve through the points.

 

MaxHugen

Super Anarchist
I don't use XFLR5, so I can't help you there.  But here's how I generate coordinates for Xfoil.  It's perhaps an overly manual process, but it's not so onerous that I've been driven to automate it more fully.

My CAD program is Rhino.
Rhino at USD 995 is way way beyond my budget, but I'll scout around for an alternative CAD app that can maybe do the same.  Thanks for outlining the procedure.  :)

 

coercivity

Member
63
6
Auckland
Wouldn't  the best way to reduce drag and increase lift low down thus allowing smaller foils etc be to allow the foils to vary in angle.

Quite hard to do from a strength perspective but it would allow smaller foils quite aggressive AOA at lift off maybe 15deg then reducing to near zero for the speed runs thus raking boat and mast back for lots of lift and very clean drag profile at speed also.

Flaps for trim and control in the run but small clean foils.

The discussion about super critical above 55knts is fine but for these boats to get around the track 5 or 10 knts faster its not higher top speeds you need its higher average speeds and they are closer to half the cavitation speed than not.  

 




Top