Deck sweeping sails and effective aspect ratio

Is there a way to estimate the error and add a correction factor?

I seem to remember that some of the theories also do not handle low aspect ratios very well.

It might be necessary to draw up a chart with various aspect ratios, AoA and planforms and find the correction factor with cfd, check if there are any trends worth fitting a curve to and create an empirical formula from that.

 

Really? then why did you make so many other changes beyond the sail gap?
For example the shallower keel reduces heeling moment. The additional area of the gap seal increases the side force and now you have to sort out which causes what, where and whence.

BTW one thing that indicates the spreadsheet is not designed to work with low AR sails and keels is that you are indicating needing only 12 degrees of angle of attack to get Cl=1.2 That is too close to the theoretical value for an infinite AR wing.

 

Thanks Will. I didn't mean to declare the basic theory wrong (I'm in no place to do so) but merely the way it is being applied in this instance. As you say, the theory is well proven but something appears to be wrong with its application in this area.

I have a "baby Ragtime" with an even boxier cabin, and it's interesting looking at the flow along the cabin in your simulation.

One thing I notice is that the flow up the hull in these simulations looks pretty clean, but in reality there would normally be a rough fence of humans sitting on the rail, and in dinghies there is normally someone hiking out or otherwise acting as a bluff body. Perhaps these shapes are the thing that harms the flow in this area and therefore (IMHO) causes the issues with the theory matching the practise?

Those are very interesting queries about the effects of rig height and CL. "Closing the gap" in boards is considered important across a wide range of boards, from the skinny 30kg 12'9" Raceboards that race upwind in light winds and have centreboards, all the way to the tiny speed boards and the very short, fat Formula boards that just do WW/LW racing. I don't know whether that is significant. One side issue is that normal wave/bump and jump/fun sails are not designed to have such a closed gap, because of the various handling issues that it creates.

I can't think of any way to isolate the effects of closing the gap in boards. Not even historical data seems to be available, as there were many major changes in rig and board design that occurred at almost the same time as sail foot designs were changed to close the gap. Secondly, the pace of development meant that the new sails, with features like foot battens that closed the gap, were made available to pros first, and guys like Robby Naish were so much faster than guys like me that the effect of closing the gap was submerged in a welter of other factors related to skill and gear.

BTW - looking at the Ragtime illustrations leads me to wonder why, if closing the gap is so important, boats don't have even higher, bulkier cabins that could act as an end-plate under the boom.

 

When I used the phrase "just how much", I meant "over and above other odds stacked against it". I did not try and isolate its effect.

Those additional changes were to "simulate" the typical design features found on low performance boats that normally prevent them from pointing well. You raised the issue of poor hydrodynamic L/D on a cruiser and seemed to imply that unless both aero- and hydro surfaces have good L/D ratios, good pointing is impossible. I simply showed that a shortcoming underwater can be overcome by improvements on the aero side and that there exists an interaction between the two (i.e. the increased magnitude of the keel lift) to further reduce the harmful effect of a poor underwater L/D ratio.

There is no additional area to seal the gap. The entire sail and mast is simply assumed to have been lowered by the exact height of the gap such that the boom ends up lying flat on the deck. All that extra side-force is due to the higher 3D CL values possible with a higher effective aspect ratio.

The shallow keel actually contributes slightly more heeling moment (60lb-ft vs 58 for the original.) The benefit of a shorter lever arm is offset by the much larger force.

I would be happy to run a scenario with only the modifications you deem necessary. Mine was to illustrate a kind of worst case scenario.

 

Remember that those values were for a very different case study. The sail in question was a symmetric wing with an AR of 8 and almost no gap. That is equivalent to an aircraft wing with effective AR of 15 to 16. So, yes, the CL would be very close to that of an infinite wing.

I went through some old textbooks to see if I can find the practical lower limit of AR for using lifting line theory with confidence, but could not find actual values. Somewhere in my memory the figure 6 keeps popping up. If indeed the case, it means that a sail with an AR of 3 and a sealed foot should give reasonable results, but that same sail with any significant gap would not. I do not remember anything about AoA causing any errors, but then again it has been 15years since I worked with the nitty-gritty part of the theory.

Setting aside any influence of AoA, and assuming my reasoning regarding AR has any merit and that the theory is indeed to blame for the disagreement with practical results, it would suggest that the theory is overly pessimistic about the potential of sails with gaps rather than overly optimistic about sails with sealed gaps.

 

I think it would be a non-trivial task. I'm not an expert on alternatives, but I suspect it might be easier to use a more complex model than to try to "fix" this one.

Just keep the AoA under 10 deg and the related error should be pretty insignificant.

 

A gap model based on lifting line theory is hardly suitable for the sail-deck sealing problem. Lifting line theory places all the vorticity on the quarter chord line (lifting line), while in reality vorticity is spread along the foot chord. It assumes a high aspect ratio, non-swept and planar wing... additionally, viscous effects are condiderable as the gap gets smaller. And the deck is rather an end-plate or a winglet than a symmetry/reflection plane as is assumed in the lifting line theory - the sea could possibly be considered as a (soft and moving) reflection plane.

When I did sail simulations based on the VLM-method (MacSail), I concluded that a gap of 50% of the freeboard height was appropriate for a main + a deck sealed jib, to yield a more or less correct effective aspect ratio.

Your RANS simulation should clear those uncertainties, but will still require quite a good resolution over the deck and the foot of the sail. You should definetly model the wind shear, there is a lot less wind at the deck level.

When we did some windtunnel tests in the late 90's with a 470 model, we also tried lifting the jib tack off the deck some 5% of the luff length (30 cm in the 470 case), to simulate the effect of a roller furler, as mentioned by Joakim earlier in this thread. The conclusion there was that while the side force (heeling force) was increased 8%, the drive was also increased a few percent... the overall effect on performance was therefore rather small. Attached images from those experiments.

There's a vortex behind the foot of the jib even if it's sealed to the deck - when you lift the jib up from the deck, the edge vortex is fortified, adding to the lift & drag of the sail. So what you gain in sealing you tend to loose in interference between the deck and the foot.

 

Would that baby be the Carter design?

We also have a local 30' Van der Stadt design called a Royal Cape One Design (RCOD) that literally looks like the spawn of Ragtime. It has the same slender hull, hard chines and most examples have an identical cabin shape to Ragtime. The most striking resemblance is however the dead straight, upward sloping sheer-line. It is also sailed at ridiculous angles of heel.





I deliberately included the crew in my Spindrift cfd simulations for exactly the reasons you mention. I will add a few blobs of rail meat ballast to some of the Ragtime simulations for comparison. Just want to get the "clean" rail scenario out of the way first.

There are a number of designs that have sail-cabintop gaps not suitable for crawling under, and it might would be well worth investigating if raising the roof to meet the boom is any good. If there is just enough aerodynamic benefit to offset the additional windage, you will at least have some more headroom! I will add it to the list of Ragtime scenarios to test.

 

Thanks Rasta,

For cambered sails I normally assume a -6deg zero-lift AoA. Should I then limit the actual AoA to 4?

 

You should limit the minimum AoA as well. Sails are not rigid airfoils and their zero-lift AoA is actually 0°, when they just flap (luff) without producing lift.
The equation

CL=(dCL/dAlpha)*(Alpha-Alpha,0)​

which is at the base of the linear lifting-line theory, is valid only in the portion of the lift curve where the sail profile can be treated as a quasi-rigid body, when there is a sufficient wind pressure difference between the windward and the leeward side to keep it in a correct shape.
I have used a term "quasi-rigid", but even that is a linguistic stretch because a sail is actually an elastic membrane. It's shape will depend on aerodynamic forces acting on it, and aerodynamic forces will depend on it's shape. So the CL-Alpha curve of a real elastic sail is actually never linear. For every angle of attack and for every wind speed the sail profile will conform to a different shape - and hence the linear relationship between CL and Alpha (expressed through the previous equation) is actually never valid, strictly speaking.

It is a big mess to model physically (a non-liner aeroelasticity), and the reason why it is nearly impossible to obtain a reliable and repeatable set of wind-tunnel aerodynamic data for a scaled-model sailboat. But since we need a starting point for our calculations, we accept to model the sail as a quasi-rigid body in a limited range of flow conditions. The important thing here is to be aware of these limits and inaccuracies when interpreting the results.

Cheers

 

Absolutely untrue Will. Directly from your spreadsheet output: "Total Hydro L/D" increased from 2.73 to 2.99, a 10% IMPROVEMENT

 

Mine's not a Carter, but a Spencer - as was Ragtime of course. In some ways the RCOD is closer to a baby Ragtime than my own boat, which is 28' LOA and has more beam (and a bigger cabin top) for rating and accommodation. With the mods I've done she'd (IMHO and from seeing the Aussie versions of the RCOD) beat a RCOD upwind and in the light stuff, but my boat can't plane and surf as well.

In Spencer's articles he mentioned the VDS boats like the RCOD/Zeeslang/Black Soo, but said that he used more flare for more stability. Others also said that Spencer had a more innovative approach to structural design than VDS, which appears to be right. Of course, Spencer was designing years after VDS created his amazing boats so it wasn't surprising that he developed some aspects further.

That pic looks like Hout Bay to me....a scary place to sail! I was in SA a while back and had the chance to look over the original Zeeslang, which was a real leader in design in many ways.

Cheers

 

Looking from the wind's perpective it's easy to conclude that the gunwhale plays a big role here and the sealing of the sails is in a side role only - not least because of the dihedral introduced by the heel.

 

Ragtime simulation

My solver can unfortunately not model a wind gradient as an initial condition, let alone twist. The best I can do is to model the actual see surface as a frictional surface and give the wind a long enough run-up so that some wind gradient is established by the time it hits the boat.

The mesh resolution is adaptive, i.e. it starts of very coarse and at predetermined numbers of iterations it looks for excessive gradients in flow variables between adjoining mesh cells. If the gradient is more than a defined minimum, the mesh is refined in that area for subsequent iterations. Detailed geometric features as well as downstream flow disturbances (such as a trailing vortex) end up with a much denser mesh in its vicinity.

I should perhaps explain my proposed approach to show whether heeling would reduce the benefit of sealing the deck gaps:
I initially plan to keep drive constant between scenarios (gaps vs sealed gaps) and then simply compare the heeling force values. But to be able to say whether heeling as such affects the amount of benefit their may be, I need to run the same scenarios at an upright attitude and compare the difference between the two values of benefit.

 

Which bit is untrue? I have already clearly explained where that increase comes from.

I am trying to figure out who is misunderstanding who now -
Are you under the impression that the L/D value is fixed for a given keel design?

Maybe I should have rephrased that "to further reduce the harmful effect of poor KEEL L/D ratio."