Aerodynamics of High-Performance Wing Sails

Aerodynamics of High-Performance Wing Sails J. otto Scherer^ Some of tfie primary requirements for tiie design of wing sails are discussed. In particular, ttie requirements for maximizing thrust when sailing to windward and tacking downwind are presented. The results of water channel tests on six sail section shapes are also presented. These test results Include the data for the double-slotted flapped wing sail designed by David Hubbard for A. F. Dl Mauro's lyru "C" class catamaran Patient Lady II. Introduction The propulsion system is probably the single most neglected area of yacht design. The conventional triangular "soft" sails, while simple, practical, and traditional, are a long way from being aerodynamically desirable. The aerodynamic driving force of the sails is, of course, just as large and just as important as the hydrodynamic resistance of the hull. Yet, designers will go to great lengths to fair hull lines and tank test hull shapes, while simply drawing a triangle on the plans to define the sails. There is no question in my mind that the application of the wealth of available airfoil technology will yield enormous gains in yacht performance when applied to sail design. Recent years have seen the application of some of this technology in the form of wing sails on the lyru "C" class catamarans. In this paper, I will review some of the aerodynamic requirements of yacht sails which have led to the development of the wing sails. For purposes of discussion, we can divide sail requirements into three points of sailing: Upwind and close reaching. Reaching. Downwind and broad reaching. Some of the requirements of upwind and downwind sailing will be discussed here. Reaching will be recognized as an intermediate condition with the sail design requirements falling between those of upwind and downwind sailing. There are many criteria which can be used to judge the quality of a sail design. For our discussion of high-performance wing sails, thé primary criterion will be to maximize the available driving force (thrust) for a given total sail area. Such a sail will maximize the speed of a racing yacht with a given measured sail area or minimize the size of the rig required for a cruising yacht. Definition sketch of force and velocity vectors in upwind sailing Upwind and close reaching Aerodynamic requirements for sailing upwind and close reaching are basically the same. These points of sailing are characterized by having the apparent wind at a small angle to the course of the boat. This situation is shown on Fig. 1. The angle between the apparent wind and the course is denoted by (3 and may vary from about 25 deg for a fast, close winded boat to about 35 deg when close reaching. We might consider 30 deg as typical for this point of sailing. It can be seen from Fig. 1 that the resultant force R, is nearly at right angles to the course so that there is only a small thrust force, T, compared with a rather large heeling force, H. The resultant, R is composed of an aerodynamic Hit lorce, L, normal to the apparent wind and a drag force, D, in the direction of the apparent wind. It is convenient to express these forces in terms of coefficients, nondimensionalized with respect to the apparent wind velocity, V, and the sail area, S. Thus: resulting coefficient (1) Cr T/y,pV's, thrust coefficient (2) CH H/%pV% heeling coefficient (3) L/y,p}r-S, lift coefficient (4) c D/Y>pV-S, drag coefficient (5) ^ Head, Ship Propulsion and Research Division, Hydronautics, Inc., Laurel, Maryland. Presented at the Chesapeake Sailing Yacht Symposium, Annapolis, Maryland, January 19, 1974. The thrust and heeling force are related to the sail lift and drag by the expressions C.;. C,, sin/3 - Co cos/? (6) 270 MARINE TECHNOLOGV

(7) A e r o d y n a m i c Load A e r o d y n a m i c Load Upwind and close reaching, i t is clearly desirable that the sail should develop the highest possible thrust up to wind speeds where stability or increased h u l l drag due to heeling force become i m p o r t a n t. Thereafter, the highest possible thrust w i l l be obtained when the. ratio of thrust to heeling force (CT/CH) or thrust to heeling moment (CTVS/CM) is maximized. Heeling moment is simply the product of heeling force and the height of the center of effort: M Hh Sail A r e a _ / Fig. 2(a) where H heehng force h height of center of effort b height of sail X h/b A aspect ratio S sail area b^/s Thus, for any given sail geometry, CM is proportional to CH. hispection of equations (6) and (7) (or Fig. 1) indicates that reducing drag w i l l b o t h increase the available thrust and decrease the heeling force. The question then becomes: What type of sail geometry w i l l provide the m i n i m u m drag, and thus m a x i m u m thrust, for a given l i f t coefficient, and what is the importance of increasing the available l i f t coefficient? Drag comes f r o m two sources: viscous or f o r m drag caused by the skin f r i c t i o n and turbulence of the air flowing over the sail and its supporting structure, and induced drag which results f r o m lost energy i n the wake of the sail due to its f i n i t e span. These two types of drag are referred to as parasite drag, CD(P), and induced drag, Cnn), respectively: (9) hiduced drag is a f u n c t i o n of the vertical distribution of aerodynamic loading not sail area. W i t h constant w i n d velocity f r o m top to bottom, the load distribution w i t h least i n duced drag is i n the f o r m of a semi-ellipse as shown in Fig. 2(a). For this case, the induced drag is expressed as (10) where A is the aspect ratio: A^tys (U) Semi-elliptic load distribution Fig. 2(b) \ < Distribution Load distribution with triangular sail The presence of the h u l l and water surface under the sail, plus the vertical velocity gradients above the water surface, tends to distort the o p t i m u m vertical d i s t r i b u t i o n of loading away f r o m the semi-elliptical shape. I n principle, i f the sail could be sealed to the h u l l along its foot, loading could be carried right down to the foot and the induced drag would be just cut i n half. The sail could be said to have an effective aspect ratio AE of j u s t twice its geometric aspect ratio. In reality this benefit is d i f f i c u l t to achieve for two reasons. First, i n practice i t is d i f f i c u l t to obtain a really tight seal between the foot of the sail and the deck. Even a small gap of only a few percent of the sail height w i l l ehminate much of the potential reduction in induced drag. Second, the reduced w i n d velocity over the surface of the water, plus the disturbed air flowing over the hull, w i l l tend to d i m i n i s h the benefit of sealing the foot of the sail to the deck. I n addition, it is rarely possible to obtain the o p t i m u m vertical distribution of loading for the existing conditions of vertical w i n d gradient and gap under the foot. I t m i g h t be noted t h a t i t is a physical necessity t h a t the vertical d i s t r i b u t i o n of loading go to zero at the free ends of the sail, t h a t is, the head and the f o o t ' i f the foot is not perfectly sealed to the deck. The load d i s t r i b u t i o n for a triangular sail m i g h t appear as shown i n Fig. 2(b). In any case, the sail w i l l produce the same induced drag as an equivalent wing operating in u n i f o r m flow with, an optimum load d i s t r i b u t i o n. The aspect ratio of this equivalent wing is called the "equivalent aspect r a t i o, " AE, and may be either slightly greater or smaller t h a n the geometric aspect ratio of the sail. Thus, equation (10) can be w r i t t e n more generally as: Thus, equation (10) can be w r i t t e n : C ^ C.^TTAE Cfl,, Ct^S/7r62 (13) (12) from which i t can be seen t h a t the induced drag is proportional to l i f t coefficient squared and inversely proportional to the square of the l u f f length or span b. Induced drag is of p r i m a r y importance i n u p w i n d sailing and w i l l actually l i m i t the m a x i m u m obtainable thrust for most conventional sail configurations w i t h effective aspect ratios between two and three. To understand this, i t is i n - Nomenclature A aspect ratio b^/s AE effective aspect ratio b sail span ( l u f f length) CR resultant force (6) c sail chord CD drag coefficient, equation ( 5 ) CH heeling force coefficient, equations ( 3 ). and (7) CL l i f t coefficient, equation (4) CM heeling m o m e n t coefficient, equation (8) JULY 1 9 7 4 coefficient, equation (1). CT thrust coefficient, equations ( 2 ) and D H h L M iï S sail drag heeling force height of center of effort sail l i f t heeling moment resultant sail force sail area T thrust U boat speed V apparent w i n d speed W true w i n d speed X defined as h/b P angle between course apparent Sh. h u l l drag angle atan 8s sail drag angle a t a n p wind. and (H/T) (L/D) air mass density 271

LIFT COEFFICIENT, C _ EFFECTIVE ASPECT RATIO, AE Fig. 3 Variation of ttirust coefficient with lift coefficient for various Fig. 4 Variation of thrust coefficient with effective aspect ratios effective aspect ratios, for various lift coefficients structive to rewrite equation (6) in terms of induced and parasite drag: Cr CL sin/3 - (CI^/TTAE) COS/J - C^^ cos/3 (14) This equation is plotted in Fig. 3 and 4 for the upwind case of /3 30 deg and no parasite drag. It can be seen in Fig. 3 that for any effective aspect ratio, a maximum thrust coefficient, CT, is achieved at some given lift coefficient, CL, and that further increasing the lift coefficient actually reduces the available thrust. This maximum thrust coefficient can be expressed as: Cr,,, \CL* sin/3 - C^^ cos/3 (15) where Ci* is the lift coefficient at Crtmax) and is given by Ct* (7rA /2) tan/3 (16) From equations (15) and (16) it can be seen that the maximum thrust is proportional to the effective aspect ratio. To get an appreciation for the relative importance of the parasite drag, we note that the parasite drag coefficient (Cu{p)) might be between the values of 0.01 for a very clean rig and 0.05 for a rig where little attention has been paid to unnecessary windage. This would in turn reduce the available thrust coefficient, Cr, by an amount from 0.00866 to 0.0433 respectively. While this is not an insignificant amount, it can be seen by comparing this value with CT values at typical effective aspect ratios in Fig. 3 that the parasite drag is of much less importance than the induced drag losses. Figure 4 is a crosspiot of Fig. 3 which again shows the importance of aspect ratio. In viewing this figure it should he kept in mind that the aspect ratio of conventional slooprigged boats usually lies between two and three, while even the tall narrow rigs of the "C" class catamarans typically have aspect ratios of about four and never exceed values of 5.5 for the most extreme designs. In reality, a boat will achieve its maximum speed at lift coefficients somewhat less than those which yield the maximum thi-ust coefficient. This is because the heeling force also increases with increasing lift coefficient which, in turn, increases the hull resistance. In addition, the preceding equations apply only to a sail in an upright position, and heeling will also degrade the sail's perforrnance. Nonetheless, it appears that the primary requirement for improving upwind sailing is to design a sail with the minimum possible induced drag. This is more important than either the reduction of parasite drag or attainment of, high lift coefficients. Before leaving the requirements for upwind sail design, it is of interest to examine the requirements for minimizing the angle /3, that is, minimizing the angle between the apparent wind and the course of the boat. Looking again at Fig. 1, can be seen that the angle /3 is equal to the sum of the two angles 5s and 5h. The angle 5s is the arctangent of the sail lift" 272 MARINE TECHNOLOGY

drag ratio and the angle 8h is similarly the arctangent of the hulls side force-resistance ratio. Thus; /? ATAN ( L / D ) + ATAN { H / T ) 5, + 5* Thin Section (17) This relationship is often réferred to as the "course theorem." Stated i n words, i t says that the angle between the apparent wind and the boat's course is equal to the sum of the drag angles for the sail (6s) and h u l l (5h). T h i s relationship is true for a l l points of sailing b u t holds its greatest significance for the case of u p w i n d sailing. I t points up the necessity of achieving high l i f t - d r a g ratios for the sail as well as the h u l l if a close-winded boat is to result. Downvi^ind s a i l i n g Boats which sail straight downwind and depend on aerodynamic drag of the sails for thrust are h m i t e d to about half the speed of the w i n d. T h i s is because, as the boat sails faster, the apparent w i n d over the sails is reduced. Since the driving force is reduced at a rate proportional to the square of the apparent w i n d speed and the h u l l resistance typically increases at a rate between the square and cube of the speed through the water, i t becomes very d i f f i c u l t to obtain any significant gains sailing straight before the w i n d. However, no theoretical H m i t to speed made good exists for boats which tack downwind, and indeed iceboats, w i t h their very low resistance, can easily exceed the speed of the w i n d when tacking downwind. While the high-performance catamarans have not yet achieved a speed made good downwind which exceeds the w i n d speed, they do perform best when tacking downwind. Under such conditions the boat typically sails at about 45 deg f r o m straight downwind and the apparent w i n d is just about abeam; that is, the angle is about 90 deg. This condition is the same as broad reaching and is the one of concern i n the design of wing sails. Sailing downwind is conceptually much simpler than the upwind problem. Sail drag is of secondary importance as i t acts normal to the course of the boat while the sail l i f t force, acting i n the direction of required thrust, is of prime importance. Since an unstalled sail can produce m o r é t h a n twice the l i f t coefficient of a stalled sail, the downwind problem is basically one of designing a sail t h a t w i l l produce the highest possible l i f t coefficient before stall. There are two considerations i n achieving a high overall l i f t coefficient for a sail. The f i r s t is to provide a sail section shape which is capable of producing high l i f t coefficients. The second is to provide a sail p l a n f o r m and vertical load distribution such t h a t this high l i f t coefficient can be achieved over the entire span. H i g h section l i f t coefficients w i t h low f o r m drag are achieved by the proper shape of the sail cross section. L i f t coefficients are l i m i t e d by the phenomenon of " s t a l l. " A f o i l stalls when the angle of attack becomes so high t h a t the flow can no longer adhere to the low-pressure side and separates from the f o i l, leaving a t u r b u l e n t wake and causing the l i f t produced by the low-pressure side to be destroyed. There are two types of stall: leading-edge stall and trailing-edge stall. Leading-edge stall occurs when the flow cannot adhere to the leading edge because of severe pressure gradients at the leading edge. T h i s type of stall can be delayed by selecting the proper amount of camber and a good leading-edge shape. Sail camber has probably received more attention than any other factor i n sail design and therefore w i l l not be discussed further here. However, a good leading edge is very i m portant i n both achieving high l i f t coefficients and low f o r m drag. The conventional fixed mast provides a very poor leading edge. The t u r b u l e n t wake shed f r o m the lee side of such a mast leads to b o t h high f o r m drag and leading-edge stall at rather low l i f t coefficients. Even the t h i n leading edge of a JULY 1974 > Fig. 5 Wing Mast Section Foil sections tested j i b is poor f r o m the standpoint of early leading-edge stall and is very sensitive to small changes i n angle of attack as would occur when sailing i n a seaway. Rotating, a i r f o i l - s h a p é d masts provide improvement. The large wing masts currently i n vogue on the l Y R U " C " class catamarans provide substantial improvement i n leading-edge shape and have rendered the conventional mast nearly obsolete on these boats. Trailing-edge stall is more d i f f i c u l t to deal w i t h. T h i s type of stall is caused by the i n a b i l i t y of the low-energy air i n the boundary layer to move f r o m the low-pressure region near the mid-chord of the lee side to the higher pressure at the trailing edge. T h i s f o r m of stall usually l i m i t s the m a x i m u m l i f t coefficients of well-designed airfoils to values of two or less. To achieve higher l i f t coefficients, s o m é f o r m of boundarylayer control is required to remove the low-energy air f r o m the surface of the f o i l. The most common f o r m of boundarylayer control is the slotted f l a p used on nearly a l l large mode m aircraft. I n the spring of 1973, A. F. D i M a u r o sponsored a series of tests at Hydronautics, Incorporated to determine the feasib i l i t y of using a double-slotted f l a p to achieve high l i f t coefficients on a new wing sail for his " C " class catamaran Patient Lady IL T h i s very ingenious wing was designed by D a v i d H u b b a r d. I t consists of three panels: a leading-edge panel of 45 p e r c é n t of the chord which serves as the mast, a 15-percent chord central p a n è l, and a 4 0 - p e r c é n t trailing-edge panel. The after two panels are hinged on a four-bar linkage f r o m the leading panel so t h a t i t can be cambered to sail on either tack. T h i s section was tested i n the Hydronautics 273

2.6 Z LU Ö O Z O u 10 20 30 40-0.4 l 0.2 0.4 0.6 0.8 O 0.2 0.4 0.6 0.8 ANGLE OFAnACK,a (degrees) DRAG COEFFICIENT, Fig. 6 Variation of lift coefficient witti angle of attack and lift coefficient versus drag coefficient for ttiln section high-speed water channel in both a high-camber and lowcamber configuration. For comparison, high- and low-cambered models of a conventional wing mast and a thin section similar to a jib were also tested. Sections of the six models are shown in Fig. 5. The models have a 6-in. cord and a 15-in. span, giving them an aspect ratio of only 2.5. The results of these tests are presented on Figs. 6 through 8. These results have been corrected for tunnel boundary effects and converted to an aspect ratio of 4, the aspect ratio of the full scale rig. Each figure presents lift coefficient, Ci, as a function of angle of attack, a, and lift coefficient, CL, versus drag coefficient, CD, for each foil type. The drag coefficient plots also contain a reference induced drag line for aspect ratio 4 as given by equation (13). In principle, the parasite drag is the difference between this reference induced drag line and the measured drag data. However, it will be noted that in Figs. 6 and 7 the measured drag is actually less than the theoretical induced drag. It is beheved that this is caused by the aspect ratio correction procedure so that, in reality, these data actually represent an aspect ratio of slightly greater than 4. However, this does not detract from the relative comparison of the foil types. The unusual behavior of the highly cambered thin section (Fig. 6) is of interest. This behavior is the result of separation occurring on both the pressure and suction side at the same time an indication of excessive camber for a section with such a thin leading edge. However, of greatest interest are the results of the Hubbard slotted section which exhibited a maximum lift coefficient of 2.4. This is a very high value compared with 1.9 for the wing mast and 1.7 for the thin section. Even more encouraging is the fact that the parasite drag is not greater than that of the wing mast in the low-camber configuration. This means that this section will provide improved downwind sailing with no penalty to upwind performance. As a result, this section has been adopted for the new wing sail on Patient Lady II. In order to achieve a high overall lift coefficient for the entire sail, the entire sail should be uniformly loaded. Referring to Fig. 2(b), it can be seen that for a triangular sail the top portion tends to' be overloaded while the foot can never be fully loaded. As a result, the top will stall before the bottom and thus prevent the achievement of a high average lift coefficient for 274 MARINE TECHNOLOGY

-0.4! -10 0 10 20 30 Fig. 7 ANGLE OF ATTACK, a,(degrees) -0.4 I J L 40 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 DRAG COEFFICIENT, Variation of lift coefficient witfi angle of attack and lift coefficient versus drag coefficient for conventional wing mast secti( C^ the entire sail. Obviously, it is desirable to distribute the sail area in a more or less elhptical fashion so that it tends to match the load distribution. This will not only provide a reduction in induced drag, but higher overall lift coefficients. The roach which can be supported with full-length battens does a great deal to improve sail area distribution in the upper portion of the sail. The wide chord of a wing mast further helps to provide a good distribution of sail area. However, the foot remains a problem unless it is sealed against the deck or cut back to a rather small chord. The control of sail twist is also an important factor in obtaining high overall lift coefficients. The presence of a vertical gradient in wind velocity over the-water surface results in a twist to the apparent wind approaching the sail. Sailing upwind this twist is very small, while on a board reach or tacking downwind the twist in apparent wind may be 20 to 30 deg or more between the top and bottom of the mast for a fast boat. The reason for this can be seen by comparing Figs ^(a) and 9(b). When sailing to windward, Figure d(a), the wind generated by the speed of the boat makes an angle of only 45 deg with the true wind. When combined with the true wind, only a small angular twist in the apparent wind occurs even for a fairly large difference in true wind speed between the top and bottom of the mast. However, when broad reaching as in Fig. 9(b), the wind generated by the boat's speed makes an angle of about 135 deg with the true wind. The geometry works out to cause a very large twist in apparent wind direction between the top and bottom of the sail. Therefore, to operate efficiently over a wide range of courses, the sail must be capable of operating with both small and large amounts of twist while still maintaining good cross-sectional shape along the entire span. In the case of the doubleslotted flap section on Patient Lady II, this has been accomplished by dividing the flaps into three sections so that the bottom flaps can be set at higher angles of attack than the upper flaps, thereby effectively creating a twist when sailing downwind. Summary The future of sail development holds great promise if we JULY 1974 275

z LU O u CO) Fig. 8 O io 20 30 40 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.6 ANGLE OF ATTACK, a ( degrees) DRAG COEFFICIENT, C^ Variation of lift coefficient witti angle of attack and lift coefficient versus drag coefficient for Hubbard double-slotted flap section ALOFT TRUE WIND ALOFT ANGULAR TWIST IN ALOFT TRUE WIND ALOFT NEAR WATER TRUE WIND NEAR WATER ANGULAR TV/IST IN WIND FROM BOAT SPEED NEAR WATER TRUE WIND NEAR WATER WIND FROM BOAT SPEED Fig. 9(a) Velocity diagram showing twist in apparent wind when Fig. S(b) Velocity diagram showing twist In apparent wind when going to windward broad reaching can advance from the old soft-cloth sails and fixed masts. The technology already exists for major improvements in sail design and performance. The adoption of aerodynamically clean wing sails will eventually improve boat performance as well as lead to smaller, more manageable rigs. It appears that the major requirement to improved upwind sailing is the reduction of induced drag, while for downwind sailing the most important problem is to achieve high hft coefficients. Substantial room for improvement still exists in both cases. Acknowledgment The author wishes to express his sincere appreciation to Messrs. A. F. Di Mauro, owner, and David Hubbard, designer of Patient Lady II for permission to publish the experimental data from their sail tests. This is an especially welcome addition to the available yacht sail data. Theh willingness to make these data available for use by potential competitors stands in marked contrast to the general secrecy found among the designers and owners of highly competitive development boats. 276 MARINE TECHNOLOGY