Innovations in IACC Keel Design

3D Keel Hydrodynamics

Induced Drag and Winglet Systems

Lift-induced drag arises from imparting energy into the surrounding fluid as a result of generating lift. Since monohull sailboat velocities are limited by hull speed, and because induced drag increases with the square of the lift (sideforce) at a given velocity, induced drag quickly becomes dominant over foil profile drag as winds increase. Induced drag reduction schemes are therefore vital to an IACC yacht's upwind performance, particularly in a breeze. Thus it is that every competitive IACC yacht employs winglets to reduce induced drag.

Understanding Winglets

The principle behind winglets is familiar to every sailor. They function as "sails" operating in the rotating vortex induced by the differential pressures on the keel. Their lift vectors are rotated forward by an angle equivalent to the local induced flow vector, resulting in a forward thrust on the winglets as they extract energy from the vortex and put it back into the boat.

Planar and Non-Planar Winglets

While practically all nautical and aeronautical winglet systems to date have been of the "non-planar" variety - that is, not in the plane of the lifting surface - it is interesting to note that nowhere are non-planar winglets found in Nature. Many land soaring birds employ planar winglets, however - a series of separated surfaces (feathers) extending from the wingtip, roughly in the same plane as the wing.

Planar winglets work on fundamentally the same principle as non-planar winglets: they operate in a local induced flow field, and thus their lift vectors are rotated forward, generating a net forward pressure component on the winglets. The difference is that planar winglets have a further primary function - they constitute an integral part of the lifting surface itself.

This difference is a double-edged sword. On the positive side, total surface area is reduced over non-planar winglet systems because rather than adding winglet area one "removes" surface area from the existing lifting surface. Thus, the compromise of hurting downwind performance by adding winglet area goes away. A carefully designed planar winglet keel will have little more drag downwind than a bare keel with no winglets at all. For this reason alone, planar winglets are potentially ideal for racing sailboats.

The challenge arises in designing planar winglets which maintain an optimum spanwise lift distribution over the length of the lifting surface (keel), while creating the correct conditions for the winglets to operate in an induced flow field at an optimum angle of attack, and to maintain this relationship throughout the keel operating range.

Reference: Dr Ulrich la Roche and the WingGrid system

Planar winglets are an elegant yet subtle technology requiring a degree of lateral thinking and mental dexterity. The researcher who has done most to experimentally dispel the idea that they are "white man's magic" is Dr Ulrich La Roche of Zurich, Switzerland. If the reader wishes to fully understand what follows in this section, I strongly recommend that he begins by reviewing Dr La Roche's "WingGrid" website at www.winggrid.ch

Dr La Roche's Planar Winglet Design Rules

The wing (keel) and winglets should be rectangular and parallel.
The sum of the winglet chord lengths should not exceed 0.7 times the wing (keel) chord length (La Roche calls this parameter "overlap").
The surfaces should be positioned relative to each other at an angle relative to the wing chordline not less than twice the maximum operating angle of attack (with each winglet, from forward to aft, positioned further towards the high pressure side). La Roche calls this parameter "stagger angle".
The total winglet lift coefficient at the wing/winglet juncture should equal the lift coefficient of the wing.

While these empirically derived design rules may not make a lot of sense at first glance, they can each be illuminated by a deeper look at planar winglet physics.

Understanding Planar Winglets

We begin with the fundamental premise that planar winglets can reduce induced drag only by operating in an induced flow field, and consequently at a higher angle of attack (AoA) than the parent wing (keel). I call this induced flow vector "beta". If we call the wing AoA "alfa", the resultant AoA of the winglets is therefore alfa + beta.

For this induced flow (beta) to occur at the winglets, if follows that the total lift per span of the winglet system must be somewhat less than the lift/span of the parent wing. For rectangular planforms, therefore, the winglet system must be operating at a lower lift coefficient than that of the wing.

This immediately explains Rule 2 (overlap): Since the individual winglets must operate at a higher AoA than does the wing, while the winglet system lift coefficient remains less than that of the wing, the sum of the winglet chord lengths must be significantly less than the wing chord.

Rule 4 arises because it is important to achieve an efficient resultant spanwise lift distribution. In this regard I disagree with Dr La Roche that the resultant spanwise lift distribution is rectangular - it cannot be, or there would be no local induced flow through the winglets (no "beta"). Rather, the rectangular parent wing maximizes winglet dimensions and Reynolds numbers while generating the necessary pressures to induce the required flow vector at the winglet system, which operates at an appropriately lower total lift coefficient due to the reduced net chord. Consequently, even though we have a rectangular planform, the lift distribution tapers off towards the tip, which in itself reduces induced drag over a plain rectangular wing.

The clue to Rule 3 (stagger) lies hid in the rule itself. This has nothing to do with the induced-drag reducing properties of planar winglets, but is required to prevent them from stalling. To function correctly, the winglets will always be operating at a higher AoA than the parent wing, and also at a much lower Reynolds number. Further, as the wing AoA (alfa) increases, so does the induced AoA at the winglets (beta), resulting in a disproportionate increase in winglet AoA (alfa + beta). Consequently, there is a real danger of stalling the winglets at some critical angle of attack.

The solution is to arrange the winglets relative to each other in such a manner as to create a slot effect, creating a weakly-coupled foil cascade to delay winglet stall. This further explains why the winglets should be rectangular and parallel - to optimise this slot effect.

A further insight into planar winglet physics is offered by La Roche's empirical observation that the effective aspect ratio of a planar-winglet lifting system will be approximately equivalent to the sum of the inboard wing aspect ratio and the mean winglet aspect ratio.

This is but a taste of the subtle fluid dynamics of planar winglets - what birds have mastered. Though birds control their tip feathers with a sophistication beyond our technology.

Keel-Specific Issues

Planar winglets have never (to my knowledge) been applied to sailboat keels, which present three unique obstacles to the effective utilization of the technology:

1. The need to structurally support a massive ballast bulb.
2. The requirement for port/starboard symmetry.
3. The IACC rule allowing only two moveable surfaces below the waterline.

While the most effective planar winglet systems typically employ at least three winglets, the above issues dictate that sailboat keels will likely employ just two.

Lateral symmetry is achieved through a key innovation over conventional planar winglet systems: the aft winglet is made an integral extension of the flap (trim tab). If the flap fulcrum is moved sufficiently far forward, the lateral movement of the aft winglet will create adequate stagger, along with the associated slot effect, in either direction. Further, the variable geometry aft winglet permits the creation of appropriate winglet loadings at zero keel AoA (leeway).

The ballast bulb must then be structurally connected to the keel strut through the forward winglet, forcing a larger forward winglet chord and thickness than would be optimum considering hydrodynamic issues alone.