| University of Central Florida Document Delivery FTU Q1 .J6875 Volume: 57 Issue: 2 Studies in applied mathematics Flow over a Wing with an Attached Free Vortex Date: 1977 TN: 736722 IN Saffman, PG Pages: 107-117 | | 4 Samik Bhattacharya Faculty 412112017Flow over a Wing with an Attached Free Vortex By P. G. Saffman and J. S. Sheffield Exact solutions are constructed for two-dimensional inviscid potential flow over a wing with a free line vortex standing over the wing. The loci of positions of the free vortex are found, and the lift is calculated. It is found that the lift on the wing can be significantly increased by the free vortex. 1, Introduction It was claimed several years ago by W. Kasper that the lift on a large-aspect- Tatio wing could be significantly increased, so that controlled flight at extremely low forward speed would be possible, by designing the wing so that there would be an extensive region of vortex flow over the upper surface. Films demonstrat- ing the possibility were shown in 1973 by O. Sviden & W. Kasper (see [5]), and a Popular description of the wing has been given by Cox [I]. Sink rates were Teported of 200 fpm at 30 mph and 100 fpm at 20 mph, corresponding to lift-drag ratios of L/D=13.2 and L/D=17.6, respectively. There is nothing unusual about such values at high speeds, but at 20-30 mph they appear remarkable, and are presumably concomitant with significant increases in the lift coefficient without corresponding change in the drag coefficient. The fact that the vortices produced by separation at sharp leading edges can increase the lift on an airfoil is well known for delta wings and similar low-aspect-ratio airfoils, and there is a fairly extensive literature on vortex lift for slender wings (see, for example, [3]). The concept of using the vortices for control has also been discussed [2]. However, to the best of our knowledge there have been no calculations of a similar nature carried out for high-aspect-ratio wings, to determine if a free vortex could stand over the wing and increase the lift. There have been studies of two-dimensional flow past cylinders and flat plates with a pair of counter- Totating line vortices lying downstream symmetrically and at rest relative to the STUDIES IN APPLIED MATHEMATICS 51, 107-117 (1977) 107 Copyright © 1977 by The Massachusetts Institute of Technology Published by Elsevier North-Holland, Inc.108 P. G. Safiman and J. S. Sheffield body. For references and corrections of earlier errors, see {6]. These calculations are relevant to the slender-body theory of flow over delta wings. But although it is obvious that, particularly at high angles of attack, leading-edge separation will produce vortices and increase lift on a wing of large aspect ratio, we have been unable to find references to flow past such shapes with attached free vortices. Perhaps this obvious idea has not been explored theoretically (or work has been forgotten) because of the assumption that such flows would in practice always be associated with large values of the drag, as the flow around the vortex would not reattach to the body but form an extensive wake, It should be noted in this connection that a prototype “Kasper airfoil” was tested in a wind tunnel as a student project [7] with discouraging results; the lift increased at high angles of attack, but so did the drag. Nevertheless, the claims of Kasper suggest that it may be possible to gain the advantages of increased lift by creating a vortex over the wing, without necessarily paying the price of increased drag, and the purpose of the present paper is to present a simple idealized solution describing such a flow. We consider two-dimensional flow of an incompressible inviscid fluid over a two-dimensional airfoil at an angle of attack. For simplicity, we shall do the calculation for a flat plate. The flow will be assumed to be steady, and it will be supposed that there is a line vortex in the flow at rest relative to the airfoil. We shall show that a solution exists—in fact there are infinitely many—and calcu- late the increased lift on the wing. Of course, this calculation is no more than 2 small first step in the study of theoretical feasibility. It remains to consider stability (to both two- and three-dimensional disturbances), effects of viscosity or compatibility with boundary layers, how the vortex is to be created, effects of finite aspect ratio, and so on. It is hoped that if the idea remains promising, these subjects will form the basis of future work. The result of a preliminary incomplete analysis of stability to two-dimensional disturbances is given in Sec. 5, and there are indications that for small angles of attack some of the configurations may be stable to such disturbances. 2. Equilibrium of the free vortex The flow picture is sketched in Fig. 1. Dimensionless variables will be used exclusively. The airfoil, of tength 2, lies along the x-axis from —1 to 1. The flow, of unit velocity, is at angle of attack a. There is a line vortex of strength « at rest at the point (xp,Y9). In addition, there is a circulation I about the airfoil. We use complex variables z= x + iy, and w=®+/¥ is the complex potential. Then w is an analytic function of z, with ¥=constant on the airfoil and wre Ms Le (x+T)logz as 2-00. (2.1) Note that for convenience we are departing from the usual convention and taking the circulation positive when in the clockwise sense. At this stage, «, I’, Z9= Xo+iyy are unknowns. There is no loss of generality in supposing 0< a< w/2.Flow over a Wing with an Attached Free Vortex 109 x Creare) Figure 1, Sketch of (a) physical plane z= x+y, (b) transform plane {=£+ i. The circulation about the airfoil is referred to as being due to a bound vortex. A force must be applied to the wing to maintain the flow; this is the Kutta lift, and it is perpendicular to the direction of flow at infinity. The vortex at zo is free, i.e., it is not subjected to any force, and is not produced by inserting a small wing at Zo. The Helmholtz laws require for steady motion that the free vortex be at a stagnation point, i-e., lim (#-#-4)-0 22) eaol dz 2x z-zZg However, the presence of the free vortex will affect the force on the body, and it follows readily from considerations of the momentum flux at infinity or directly from the Blasius formulae that the Kutta lift on the wing is Lact. (23) (In dimensional units, multiply by pU, where p= density, U=velocity of free Stream.) If «=0, the indeterminacy of the picture is removed by imposing the Kutta condition at the trailing edge, that the velocity is finite there. The result is T=2zsina. (2.4) If «#0, we can obtain three further equations, giving four in all for the four real unknowns, by also imposing a Kutta condition on the leading edge and using Eq. (2.2). However, it will be shown in Sec. 3 that these equations have no solution and that it is not possible to impose a Kutta condition at both the leading and the trailing edge, even though in principle there are sufficient degrees of freedom (cf. [6}.) We obtain the velocity potential by mapping the airfoil into the unit circle by the transformation 2=4(s+4), faz4(2-1)'7, (25)110 P. G. Saffman and J. S. Sheffield where (2?— 1)!/?=|z?—1|'/? when z > 1, and the z-plane is cut from ~1 to +1. Then | ia ‘ watent (s+ =)+ slo oe 4 + (e+ T)logt (2.6) is an analytic function with the appropriate singularities and satisfying the boundary conditions, where {, is the image of Zp. The condition (2.2) that the free vortex is at rest gives, after some algebra, Ry 2~S5- [bol ik iT . eo (52 — 6) 4 6 79 En, (2.7) 7 (83=1)(Ko?=1) Put {)=pe'*, and break into real and imaginary parts to obtain 3 . Kp’ sin2@ cos(¢— a) = —— ——__—__ 2.8) eo (9? — 1)(0*— 2p? cos29 + 1) ee) 4(1+cos2$) — p?(1+3.c0s2$) +2 sin(¢—a)= =“ pl( +0526) ~p'(1+3cos26)+2 Tp (29) (0*=1)(e*— 2p? cos2¢ + 1) (p+ 1)" 3. The Kutta condition In order for the inviscid flow to be compatible with thin boundary layers on the airfoil surface, we assume (as is customary) that the velocity is finite at the trailing edge z= 1. This Kutta condition requires that dw/d{=0 at {=1, because of the singularity of the transformation. One real equation then results: 2 . —1 sina= ctT_ ok iy la a p?—2pcoset! . G.) The velocity at the trailing edge is p(p?—1)sing ymcosa— £0 se (3.2) (p?+1—2pcos¢) The physical considerations which led to the Kutta condition will only be applicable if V >0, which is a necessary condition for (3.1) to apply.Flow over a Wing with an Attached Free Vortex a A Kutta condition at the leading edge {= —1 gives the real equation 2-1 sina= — S40 4 # PN (3) 2a 2a p 2+ 2pcospt I It is now not hard to verify that Eq. (2.8), (2.9), (3.1), (3.3) have no solution. We solve (3.1) and (3.3) for x and I and substitute into (2.8) and (2.9) to obtain (e?- @-1y sinasing _ sinasin?6 2p? cos(G—a) —cos@sin(g—a) This equation obviously has no non-trivial solutions. Hence, there is no non-tri- vial flow field of the type being considered in which a Kutta condition is satisfied at both leading and trailing edges. 4. The free vortex locus The Kutta condition at the trailing edge and the two equations for equilibrium of the free vortex provide three equations for the four unknowns k, I, p, ¢. A locus of positions is therefore expected. From (2.8), (2.9) and (3.1), we obtain cos(— a) (p?— 1)(p*— 2p? cos2¢ + 1), (4.1) p inde 2+] +T= =e [t= eosgcasat (ptt tsngsing} (4.2) F(od)=| (0+ 1) | eorpsoue—a)~(o+ 4 )sinasing +2sinasingcos¢=0. (4.3) This last equation determines for given angle of attack a the locus of the Possible positions of the free vortex. Only real roots of the quadratic for p+1/p which are greater than 2 are relevant, as we require p>1. In addition, the Condition V >0 is equivalent to 2 et ; —2cosd 3{(r* 4) 4} 0+anatung<—t—. (44) 6 pt—+2cose Pp By inspection, one sees that (4.3) has one positive root for p+1/p greater than 2 if 0<¢ $Y pricey oF ’ (4.7) sin(9—a)+sina= 7 and the root @=0 is spurious (it is the Kutta condition). For ¢ close to 7, the additional stagnation points are at @= 7+ ($tana)'/?, When ¢ is close to zero, they are at 0= 9+ $(3¢ tana)!/?, 5. Results In Fig. 2, we show the free-vortex locations (x9,¥9) for a=0.1 and a=7/6. The lift, x +T, is shown in Fig. 3 as a function of yy for the same angles of attack. The maximum lift occurs when the trailing edge is a stagnation point and theFlow over a Wing with an Attached Free Vortex 113 one 8 -1 0 t % Figure 2. Loci of free vortex for a=0.1 (5.1°) and a=! (30°). Trailing edge is a stagnation point when vortex is at end of locus. Point A marks position where leading-edge bubble lifts off (see Fig. 6). For a=0.1, this happens too close to leading edge to be marked on figure. Point B marks transition from instability to stability according to the criterion of Eq. (5.5). 20 a LIFT K+P 6 LEADING EDGE Locus 5 314 63 4 ° 05 10 Lift on the wing as a function of height yo of free vortex above the wing for a=0.1 and Fore “ordinate intercepts are 2nsina.114 P. G. Saffman and J. S. Sheffield 50, 40 30} MAXIMUM LIFT 20) TRAILING EDGE LOCUS oe sina 1 4 a 1 ° % 3 2 Figure 4. Maximum lift as function of angle of attack for leading-edge and trailing-edge loci. Also shown is the Kutta lift. Note non-zero lift of V6 = for a=0. condition (4.4) is violated. Figure 4 shows the maximum lift plotted against the angle of attack. A finite lift is obtained at zero angle of attack. Also shown on this figure is the lift without the free vortex (27sina); it is clear that large increases in the lift can be obtained. The streamline patterns depend on whether the vortex is on the locus coming from the trailing edge or on the one coming from the leading edge. In the former case, as the free vortex gets stronger and moves away from the wing (the angle of attack being constant), the rearward stagnation point on the upper surface moves backwards, the forward stagnation point on the upper surface moves forward (giving an increasing region of reverse flow), and the stagnation point on the lower surface moves rearward. In Fig. 5 we show streamlines for this case when near the limiting flow. This flow pattern is qualitatively similar to that reported by Kasper and Walton, who notice by means of tufts that there is forward flow over the rear part of the wing. In the case when the free vortex is on the locus coming from the leading edge, the stagnation point on the upper surface moves rearward and those on the lower surface approach one another, as the free vortex gets stronger and moves away from the wing, with « kept constant. Eventually, the stagnation points below the wing merge and move off the wing, there being subsequently a stagnation point in the flow. Examples of these two cases are shown in Fig. 6. This happens before the upper stagnation point reaches the trailing edge. In Fig. 7 we show a typical plot of g? (= —2p) on the airfoil. The case shown is for a=7/6 and maximum lift on the trailing-edge locus. The increased lift is due to additional suction under the vortex and near the leading edge.Flow over a Wing with an Attached Free Vortex 115 Figure 5. Streamline pattern for vortex on trailing-edge locus for a=0.1 and maximum lift. There is a close correspondence between this figure and Fig. 3.83 of [4] showing a stalled airfoil. Finally, we present a partial investigation of the stability of the flow. Even if the disturbances are restricted to be two-dimensional, the stability of the configuration is a non-trivial problem, as the Kutta condition requires that in unsteady flow a vortex sheet of variable strength exists downstream of the wing. However, if the Kutta condition is ignored for unsteady flow, and it is supposed that the free vortex and bound vortex have the strengths x and I’, respectively, of the steady state, then the calculation is completely straightforward. We shall describe the analysis briefly. From the complex potential given by (2.6), it follows after some algebra that the complex velocity of the free vortex, u— iv, is given by -ip= — ~iap2_ pia ict? tLe rjt— ing u-iv fa eke — @' e8) ai 9 We) =0(5.5), say, (5.1) where { is the position of the vortex. In the equilibrium position $=, and Q (So So) = 0 gives Eq. (2.7). If now the vortex is displaced to the point [)+¢’ in the transform plane, it can be shown that « = al + OF, (5.2) _fao & (22 & (Fa); 4 (22): e where116 P. G. Saffman and J. S. Sheffield ' / Figure 6, Streamline pattern for vortex on leading edge locus showing separation bubble, a= 1/6. (a) ¢=3.0, separation bubble is attached. (b) ¢=2.8, separation bubble has lifted off. and the suffix 4 denotes evaluation at {={). Equation (5.2) has solutions proportional to e, where (54) o?~0(b+6)+bb—aa=0. It is found that b+ 5=0. Hence, there is instability if aa> bb. (5.5)Flow over a Wing with an Attached Free Vortex 7 20) ° Figure 7, q? vs @ on the airfoil. a=7/6, maximum lift, trailing-edge locus. No free vortex—~. If (5.5) is not satisfied, linear theory predicts stability, but non-linear effects may destabilize. It is found that there is a range of values of a and ¢ or yp for which (5.5) is violated. The angle of attack a must be less than 0.137 (8°), the vortex must be on the trailing-edge locus, and the values of ¢ lie in a range depending on a, which includes, however, the value for maximum lift. The possible positions for a=0.1 are shown in Fig. 2. There are no stable positions on the leading-edge locus. We emphasize that this conclusion of stability is tentative, and the problem needs to be analyzed using the methods developed for unsteady wing flow. Acknowledgements This work was supported by the Energy Research Development Administration (AT 04-3-767) and the U. S. Army Research Office, Durham (DAHC 04- 68-C-0006). References 1. J. Cox, Soaring 37, 20 (1973). 2. M. T. Lanpast and S. E. WIDNALL, in Aircraft Wake Turbulence (A. Goldberg, J. Olsen, and M. Rogers, Eds.) Plenum, 1971, p. 137. 3. E. C. Potuamus, J. Aircraft 8, 193 (1971). 4. L. PRaNDTL, Essentials of Fluid Dynamics, Blackie, 1952. 5. N. Ruey, J. Fluid Mech. 62, 33 (1974). 6. J. H. B. Sarr and R. W. CLarK, AJAA J. 13, 1114 (1975). 7. D. Watton, Soaring 38, 26 (1974). CaLiFoRNia INSTITUTE OF TECHNOLOGY (Received November 15, 1976)Studies in Applied Mathematics Editorial Board H. Cheng S. A. Orszag H. Chernoff S. A. Papert H. P. Greenspan E. Reissner L.N. Howard G.-C. Rota D. J. Kleitman P. G. Saffman E.H. Lieb C. E. Shannon Cc. Cc. Lin W. G. Strang W. V. R. Malkus J. T. Stuart A. Toomre D. J. Benney (Managing Editor) Studies in Applied Mathematics (formerly The Journal of Mathematics and Physics) is published bimonthly by Elsevier North-Holland, Inc. It publishes research papers originating from or invited by members of the Editorial Board. Editorial correspondence should be addressed to Professor David J. Benney, Studies in Applied Mathematics, c/o Department of Mathematics, Room 2-341, MIT, Cambridge, Massachusetts 02139. Subscription rates are: $52.00 per year for institutions and $26.00 per year for individuals. Foreign subscribers: Please add $3.00 for postage and handling. Claims for missing numbers can be honored only up to six months from date of issue. Ad- dress orders and correspondence to: Subscription Department, ELSEVIER NORTH-HOLLAND, Inc., 52 Vanderbilt Avenue, New York, NY 10017. For back volumes 48-53, address inquiries to the MIT Press, 28 Carleton Street, Cambridge, Mass. 02142. For Volumes 1-47, con- tact the Johnson Reprint Corporation, 111 Fifth Avenue, New York, NY 10003. A Five-Year Index, covering Volumes 47 through 51 (1968-1972) is available from the MIT Press. Advertising inquiries should be addressed to: Leni Ellinger, 3133 Connecticut Ave., N. W., Washington, D. C. 20008. Copyright © 1977 by the Massachusetts Institute of Technology. Second class postage paid at New York, NY, and at additional mailing offices. Manufactured in the United States.