Wake Wash

I'm talking about the classical aerodynamic definition of induced drag. I realize there's a lift-dependent profile drag as well.

 

Tspeer,

I think that a specific machine and control configuration needs to be looked at in order to address Jehardiman's concerns and generally get at how such sensors would be integrated into the sailor's data set and the designer's validation process.

In a perfect world, I suppose there would be four controls to a wingsail. Ideally, 2 ,3, and 4 would not require adjustment across a tack.

1. Spar angle. Aiming the spar at the AW. (Zero lift at zero flap in uniform wind.)

2. Skew. Aiming the main element at the skewed apparent wind along the span. (zero lift at zero flap in skewed wind.)

3. Flap. Adds nominal camber via flap angle and also adjusts the main element AoA on the spar so as to avoid suction peaks.

4. Twist. A differential control that changes the flap gearing along the span to adjust vertical CoE. Can be force balanced to require very little control effort and also accommodate gust response. Gust response is a nasty issue when you start wanting to lower the CoE.

For a general sailing condition, a change in any one of the controls or a gust will change the Lift distribution along the span. So the question becomes - how do you integrate the span load data into the rest of the data stream?

On a soft sail with a simple vang and sheet control, the situation is very different. A single wake rotor positioned at the point where the spanwise shear ought to be zero should suffice for light air sailing. Just zero it out and sail. I think regular old tell-tail ribbons should work fine for this.

 

Twist will have a major effect on spanwise lift distribution. A more complex wingsail could have multiple twist controls to refine the lift distribution.

Spar angle, "skew" (second element angle) and flap angle will primarily alter the chordwise pressure distribution and overall lift. Their effect on spanwise lift distribution will be limited and depend the relative spanwise area distributions of the spar, second element and flap.

 

DCockey

But to what end? for a target CoE, there is basically one best lift distribution.

A couple of qualifications to that. It is really RM arm, not CoE, that is the target. If the hydrodynamic center of resistance is changeable (dagger board extension, for instance), then the heeling moment arm associated with a sails CoE can change. That does cause a small change to the optimum lift distribution. Heeling creates a sweep back effect. That basically changes the planform of the entire rig. But most boats want to be sailed at a fairly constant amount of heel over a wide range of conditions, so I don't think that is worth worrying about it should be able to be taken as a constant.

The basic premise I'm working from is that you have modeled the crap out of the wing and integrated all the controls so that they are independent of each other. Changing the twist only changes lift distribution, not Cl. Changing flap only changes Cl, not lift distribution. It may turn out that complete control independence isn't desirable. It might take more control power or it might not provide enough feedback to each control station. But it seems like a decent place to start from.

Um, it looks like I mispoke in my earlier post. Where I said "a change in any one of the controls or a gust will change the Lift distribution along the span", I should have excluded the flap control.

 

It's true that there's one best spanwise lift distribution, but there are multiple ways to achieve that lift distribution. The more mechanisms available for twist, the more secondary objectives can be satisfied.

As a first approximation, what matters most is the angle of leech to the apparent wind. If you twist the entire wing, or you twist just the flap, you get much the same spanwise lift distribution as long as the leech has the same angle of attack. For example, the wing on 17 was 50% main element and 50% flap. If the main element was held fixed and the flap deflected to a given flap angle of attack, the lift produced was 80% of the lift that was produced by keeping the angle between main element and flap fixed and rotating the entire section. So although the main element was 50% of the area, its angle to the apparent wind only counted for 20% of the lift. The wing on Aotearoa had approximately 60% of the area in the flap and 40% of the area in the main element, so the angle of the main element would have counted for even less.

The fact that the lift is largely determined by the the angle of attack of the flap should not be confused with where the load is carried. The majority of the lift may be carried by the main element, depending on angle of attack and flap deflection. If the main element has a tab, changing the tab angle will shift the proportion of the lift carried by the main element vs the flap, but the total lift will not be significantly affected (when the wing is operating in the linear range).

If the control system ties the flap deflection to the main element, twisting the main element will twist the entire wing. At each spanwise station, the flap chord would maintain its angle relative to the main element chord. It is also possible to anchor the flap control to the non-twisting spar, thus maintaining the flap angle of attack as the main element is twisted about the spar. So one has to be careful as to the implementation of the control system when discussing the merits of main element twist.

One might classify wing twist control into three general classes. Whole wing twist only, flap twist only, and independent control of both main element and flap twist. The same spanwise lift distribution would be obtainable with any of the three. For the two single-twist mechanisms, one could get the desired angle of the main element to the apparent wind at one spanwise station, but the main element would be off the optimum at the other stations. The difference is away from that matched section, the whole wing twist approach would have a twisted main element and the flap twist approach would not, resulting in different profile drags away from the matched section.

There are two sources of lift-dependent drag. One is induced drag, which is really an apparent drag due to the boat sailing in a header of its own making as a result of having deflected the wake to windward in order to produce the lift. The second source of lift-dependent drag is the variation in two-dimensional profile drag resulting from losses in the boundary layer as the pressure distribution changes with angle of attack and flap deflection. Of the two, the induced drag is several times that of the profile drag. Independent main element and twist control allows the crew to minimize both types of lift-dependent drag.

Since the induced drag is determined by the spanwise lift distribution, regardless of how the lift is produced, it can be minimized largely by the profile of the leech relative to the local apparent wind angle.

The lift-dependent profile drag is due primarily to the formation of suction peaks on the leading edge of the main element and flap. The local velocities just outside the boundary layer can be several times the freestream apparent wind speed near the leading edges. Since skin friction varies roughly by the square of the local velocity, a pressure peak can greatly increase the skin friction over a small region, and this local skin friction will dominate the profile drag of the whole section.

The key to minimizing the profile drag is to avoid forming a leading edge suction peak. The suction peaks are formed when the air has to negotiate a sharp turn at high speed. When the wing is operating at high lift the stagnation point can be well back from the leading edge on the windward side. The air then has to negotiate the hard turn around the leading edge after it has accelerated away from the stagnation point. The attached picture shows a good example. If you look closely at the orange telltales, you will see that the leading edge telltales are all streaming smoothly forward. The stagnation point is clear back on the black main spar! This is a high-lift situation, so the stagnation point location is unavoidable. But a similar situation could be experienced at lower lift levels if the flap was deflected very little and the angle of attack was high.

The optimum trim to maximize the section lift/drag ratio is typically obtained when the stagnation point is at or just to windward of the leading edge. If too much flap angle is used, then for the same low to moderate lift level, the stagnation point can be on the leeward side, and a suction peak will form as the air flows around to the windward side. If too little flap angle is used, the stagnation point will be behind the leading edge on the windward side, and a suction peak will form as the flow goes around to the leeward side. But when the stagnation point is just to windward of the leading edge, the flow can negotiate the turn while it is in the early stages of accelerating away from the stagnation point and a near-constant velocity can be obtained over much of the main element. Since the lift depends approximately on the average velocity, each increase in velocity is balanced by a decrease somewhere else. But the drag due to the increase is more than the reduction in drag where the velocity is lower, so it pays to minimize the variation.

This suggests a qualitative strategy for trimming a wing with independent flap and main element twist. Trim the trailing edge to minimize the induced drag, and trim the leading edge to minimize the profile drag by putting the stagnation point at the leading edge. Depending on the control system, this may require a lot of coordination between the main element twist mechanism and the flap deflection mechanism, or only minor adjustment of the flap as the leading edge is tweaked.

The problem with main element twist is it is typically implemented with a rigid spar surrounded by a flexible shell. The thickness of the shell structure results in a smaller diameter for the spar if the same outer thickness is kept the same so as to have the same profile drag as the wing without main element twist. This is a less efficient structure compared to the D-tube of the non-twisting main element that puts all of the stressed structure on the outside. That means there will either be a decrease in stiffness or an increase in weight.

In the end, the question of independent main element & flap twist vs a single twist control comes down to how important it is to minimize the lift-dependent profile drag. The choice of whole wing twist only vs flap twist only comes down to weight and the nature of the variation of off-optimum main element angle of attack along the span.

 

This is an idea I had for a wingsail control. It is a cartoon that shows the starboard tack controls for a wing rib. The flap isn't shown, just its tiller arm. I needed to sketch something out to figure out what I would do with the span loading info if I had it. I'm not a racing sailor and I started sailing so young that I can't remember not knowing how to sail. That makes it interesting to try to relate the physics I'm learning to what I've been doing all my life. One motivation for this design is to get the twist control function to have an extremely high gain so that gust feedback is gets through, but the control effort to mitigate it would be small. It also seems to handle the suction peak issue for all changes to flap and twist.

 

The book can now be pre-ordered on Amazon. Says it's available Feb 7.

http://www.amazon.com/Flight-Vehicl...?ie=UTF8&qid=1390272127&sr=8-1&keywords=drela