Assymetrical daggerboards

Any idea on a good section for an assymetrical daggerboard ?

Design goals are :
Re : 1,000,000 to 3,000,000
Cl @ 3° : higher than 1
Cd @ 3° : lower than 0.01 (bucket between 0 and 6°)
Not prone to cavitation
Highly tolerant to AoA changes
High stalling angle (>10°)

I know those are quite stringent requirements, and there's probably no published foil meeting them.

I got Xfoil, but don't have enough experience to start from scratch. Any indication on how to attain the design goals would be much appreciated.

Thanks

 

You've laid out some challenging requirements. You want a Cl at 3 deg angle of attack to be > 1, but still want to go to 10 deg. This would imply a maximum lift coefficient greater than 1.7. You want the drag bucket to go from roughly Cl = 0.7 to 1.3. These imply a highly cambered foil that won't work well if the speed varies by 20% from the design condition. I guess you plan to closely regulate how much foil area is exposed in order to keep it in the groove.

You've specified a minimum drag coefficient, but not the Reynolds number. If the Reynolds number is too low there's no way to meet Cdmin = 0.01. You can get just about any drag coefficient you want if you make the Reynolds number high enough, but that probably won't meet your system requirements.

You might start with something designed for human powered aircraft, like the Eppler E395 and add more camber, since needs to be shifted to a bit higher lift range.

 

Tom, Tom, Tom....

Admittedly, quite a range to play with, but it is there.

We used a NACA 6412 on the Black Sea 40's, which worked very well all things considered. Not the latest whizz-bang (Hah!) foil in the world, but capable of taking the conditions you are going to find offshore - pitching, yawing, rolling, etc. If you fine-tune the foils to a fare-thee-well, then any perturbation (sorry, big word for the day) will kill them. Any foil that has to experience wave action must first and foremost be foregiving, which is why so many still come back to the NACA "00" series.
II know there are some loverly techie foils out there, hopefully someone can come up with an alternative to the plain vanilla we used...

Steve

 

Thanks guys ! Very interesting answers. Cl=1.7 at 10° seems a bit far fetched, indeed... And I do agree with the necessity of a forgiving profile.

BTW, one more constraint I forgot : For structural reasons, I'm aiming at a t/c of approx. 14%.

Tom, the E398 (14% version of the E395) cavitates too much... I was taking a look at your H105, which sounds interesting from that point of view, but the design point seems quite different...

Got a polar for the NACA 6412 ? How would you advise to increase it's thickness ? Lateral scaling seems too simple to be trusted...

 

Guest,
Polars for the 6412:
**************************************************
! Airfoil Polar from PANDA
! Re = 1000000.00 USgrit = 1.0000 LSgrit = 1.0000 Mach = 0.000
! Alpha CL Cm CD
0.000000 0.675852 -0.155406 0.008307
0.500000 0.734937 -0.155821 0.008358
1.000000 0.793666 -0.156265 0.008974
1.500000 0.852023 -0.156736 0.009260
2.000000 0.909988 -0.157236 0.009378
2.500000 0.967545 -0.157763 0.009184
3.000000 1.024676 -0.158318 0.008344
3.500000 1.081364 -0.158900 0.008613
4.000000 1.137591 -0.159510 0.008745
4.500000 1.193341 -0.160147 0.009042
5.000000 1.248595 -0.160810 0.009355
5.500000 1.303338 -0.161501 0.009536
6.000000 1.357552 -0.162218 0.009696
6.500000 1.411222 -0.162961 0.009882
7.000000 1.464331 -0.163730 0.010629
7.500000 1.516862 -0.164525 0.011954
8.000000 1.568800 -0.165345 0.014250
8.500000 1.620129 -0.166191 0.016614
9.000000 1.670834 -0.167062 0.017639
9.500000 1.720898 -0.167958 0.018814
10.000000 1.770306 -0.168878 0.019653
10.500000 1.819044 -0.169823 0.022339
11.000000 1.867096 -0.170791 0.023433
11.500000 1.914449 -0.171783 0.024616
12.000000 1.961087 -0.172798 0.025905
12.500000 2.006997 -0.173837 0.027328
13.000000 2.052164 -0.174898 0.028874
13.500000 2.096575 -0.175982 0.030188
14.000000 2.140214 -0.177088 0.000000
14.500000 2.183072 -0.178215 0.000000
15.000000 2.225133 -0.179364 0.000000
end
************************************************

Coordinates for the 6414

**************************************************

! Airfoil Coordinates from PANDA
! x yup ylo
0.000000 0.000000 -0.000000
0.001541 0.008484 -0.007561
0.006156 0.017586 -0.013921
0.013815 0.027235 -0.019090
0.024472 0.037325 -0.023091
0.038060 0.047716 -0.025966
0.054497 0.058242 -0.027771
0.073680 0.068716 -0.028580
0.095491 0.078938 -0.028482
0.119797 0.088699 -0.027584
0.146447 0.097790 -0.026007
0.175276 0.106011 -0.023887
0.206107 0.113173 -0.021369
0.238751 0.119106 -0.018607
0.273005 0.123665 -0.015761
0.308658 0.126731 -0.012989
0.345491 0.128214 -0.010443
0.383277 0.128057 -0.008267
0.421783 0.126332 -0.006490
0.460770 0.123520 -0.004751
0.500000 0.119729 -0.003063
0.539229 0.115028 -0.001490
0.578217 0.109495 -0.000082
0.616723 0.103222 0.001121
0.654508 0.096312 0.002097
0.691342 0.088873 0.002834
0.726995 0.081025 0.003333
0.761249 0.072892 0.003608
0.793893 0.064604 0.003679
0.824724 0.056296 0.003574
0.853553 0.048102 0.003327
0.880203 0.040162 0.002973
0.904508 0.032609 0.002548
0.926320 0.025577 0.002085
0.945503 0.019191 0.001618
0.961940 0.013568 0.001174
0.975528 0.008812 0.000777
0.986185 0.005014 0.000449
0.993844 0.002247 0.000203
0.998459 0.000564 0.000051
1.000000 0.000000 0.000000
end
****************************************************

Enjoy!

Steve

 

I agree completely. A sailboat has to work well over a wide range of conditions, including low Reynolds numbers, and the NACA 4-digit sections are eminently suitable.


Here are the XFOIL predictions for the 6412. I've overlaid your Panda polar - XFOIL and Panda agree in the linear range, but XFOIL looks to be more realistic about predicting Clmax. I've also included the NACA 4412 for comparison at the same Reynolds number as the Panda data and again at a higher Reynolds number for comparison with the published NACA test data.

I disagree with you about the sensitivity to perturbations, though. You just have to design for the desired range of conditions.

Instead of the NACA 4-digit camber line, I think I'd reshape the camber line to get a bit of aft loading. Something like this:
Here's the comparison of the stock NACA 6412 and the modified version:
I think it meets the specs pretty well.

Here are the coordinates of the modified section: http://www.boatdesign.net/forums/attachment.php?attachmentid=1425&stc=1

 

oops - missed the post about the 14% thickness. That'll make the minimum drag a bit more challenging.

 

Here's a new section that meets the requirements although not quite at the specified angles of attack. It has the 14% thickness but still has lower drag than the NACA-based sections. I think this shows the potential of inverse design codes like XFOIL that let you design to a precise set of requirements.

 

Awesome ! Thank you so much !

Two more questions for my personal information :
1) The modified NACA, with more aft loading will be more prone to cavitation, right ?

2) To obtain it, you modified the camber line, while you designed the H110 directly from a velocity distribution, right ?

 

Tom,
Thanks - now I know why we keep you around
Panda is a great little beast, but does not do any inverse stuff.
Steve

 

And Panda doesn't have the viscous-inviscid coupling capabilties of XFOIL. That's why Panda doesn't predict maximum lift or handle separation bubbles. I think Panda is a nice academic code, but for real work, give me XFOIL. Not to mention, XFOIL is the free one.

 

You're welcome.

Normally, I'd expect the foil with aft loading to be better in terms of cavitation. With more of the lift carried by the aft part of the foil, less has to be carried by the front part, so the maximum velocity can be lower. However, in this case, I've used the extra camber to shift the whole foil to a higher lift range, so I've actually made things worse. You've put the emphasis on high lift, so that's where I went with the mod.

You realize that cavitation performance and high lift are mutually exclusive, right? In order to have high lift, you have to have high suction side velocities. This means earlier cavitation.

The plot below shows the incipient cavitation boundaries for all three foils. Since this is a dagger board, it's harder to get a grip on just how it's loaded. With a horizontal wing, you know the lift has to equal the weight. So the foil loading tends to stay constant and the lift coefficient drops off with speed.

But with a dagger board, it may be the faster you go, the harder you're driving the boat, and the greater the load on the board. So it may tend to operate more like a constant lift coefficient. But then again, maybe when you go fast, it's because you're sailing more off the wind and the dagger board is less loaded. In that case, you'd be moving more toward the bottom corner of the velocity envelope where you get the best cavitation performance.

If unload it too much, a velocity peak forms on the pressure side of the leading edge, and the cavitation speed starts to increase. So you don't want to drop below your stated range of Cl=0.7 at high speed.

At your design condition (Cl = 1), you can get maybe 23, 24 kt without any cavitation. Call it 25 kt in practice, since the actual cavitation speed is typically higher than the calculated incipient cavitation speed.You haven't said what the application is, so it's hard to know if 25 kt is a good number or not.

If you want 30 kt or higher, you need to revise your requirements, dropping the lift range down and increasing the chord on the daggerboard.

Yes. I actually started by trying to increase the camber on the H105, but it stalled out way too soon. So I started modifying the velocity distributions and it very quickly became something completely different.

I don't really consider the H110 to be a useable section because the trailing edge is so thin. The next step in the design would be to give it a square-edged trailing edge to boost the structural thickness while not changing the contours that much, especially on the suction side.

With some rework of the lower surface, it might be possible to get that little drop in drag extended to a lower lift coefficient. You'll notice it happens when the transition point moves to the trailing edge on the lower surface. After all, it's good for 20 counts.

 

Panda was "free-to-me"
I don't use it much, but it's handy for semi-techie checks.

Steve