1、I J RESEARCH MEMORANDUM PRELIMINARY INVESTIGATION OF THE DELP1Y OF TURBULENT FLOW SEPARATION BY MEANS OF WEDGE-SHAPED BODLES By George B. McCullough, Gerald 5. zzberg, and John A. Kelly NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS WASHINGTON _- * xfEs%m ai;i-if%r March 1, 1951 I+.s-ry A,ti b wing spa
2、n, feet .- - C wing chord, feet cd average section drag coefficient, corrected for jet+oundary effect by the methti of reference 3 Acd incremental section drag coefficient (cd for airfoil with wedges) - (cd for airfoil without wedges)- . Cl average sectian lift coefficient, corrected for jet4oundary
3、 effect by the n a previously calibrated four-prouged yaw head for determining flow directicm. A photograph of the yaw head is shown in figure 2. The flov velocity was determined by means of a total- and a static-pressure tube mounted parallel vith the axis of the yaw head. The offset of the static
4、tube was taken into account in the cal- culation of the local flow velocity. Most of the tests were made with free-stream dynamic pressures of 25 and 50 pounds per square foot. The smoke observations, however, necessitated a much lower speed. Multiple Wedges Mounted o.auAirfoil c The airfoil model e
5、laployed in the ixmestigation of multiple wedges was a 5-foot-chord, WAC% 63 2-z 18 airfoil. When mounted in the wind tmnel,themodelspmnedt T-foot dimension. Attached to the ends of the model were circular plates, 6 feet in diameter, which formed part of the tunnel floop and ceiling. The model was p
6、rovided with a rov of pressure orifices along the midsm section and a 27l/L Wfth the ramp angle reduced to 4O, only a portion of the smke was entrained fn the vortex; the remainder drifted over the region occupied by the vortex and mf with the general flow. Pressure Distribution The distribution of
7、static pressure on the inclined rsmp and on the oblique face of the wedges, as well as on the wall downstream of the wedges, was deterh bJ23 (ded 7 30 4 I 30 6 15 2 V 0.86 0.29 .45 .18 .34 .l2 Pressure-drag coefficient It will be noted that both the strength of ths discharged vortex aud the drag wer
8、e lowered by reductig either the ramp angle or the angle of divelc- gence . IncreasFng the dFsplacement thickness of the bouudary layer on the wall immediately ahead of the 10 wedge from 0.2 inch to 0.5 inch had little effect on the vortex strength or the drag. Surveys of the flow in the vicinity of
9、 several wedges were rcade wfth a rake of total+ressure tubes which was moved laterally through the wake. Because the flow dfrection varied with distance away From the wall, some of the tubes of the rake were so oblique to the flow as to be unable to fndicate the true total pressure, but, since the
10、region inmedIately adja- cent to the surface was of greatest interest, the rake was alined in the direction indicated by a tuft attached to the wall at each of the several positions occupied by the rake. In som3 locations, therefore, the sur- veys cannot be considered as boundary-layer surveys, but
11、serve only to give a qualitative representation of ths nature of the flow. Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-RACA RK A5OU2 7 In figure 6 are shown contour mpe derived from surveys made behind a.mall wedge of 6 ramp angle and. 15O angle
12、of divergence. The SurV8yS were made at four stations, ont the correspanddng height on the bare wall was 2 inches, Reducing the ramp angle from To to 4 had little effect on the minimum height of the lsyer of reduced total pressure in spite of the reduced strength of the trailing vortex, but did redu
13、ce the lateral extent of the t-d-out layer. Reducingthe angle of divergence from 30to15* approxix thus giving a maximum nurdber of trailing vortices (all rotating in the ssme sense)t Effect of Wedges on Maximum LHt Chordwiselocation.- The variations of nmximum section lift coef- ficient with chordwi
14、se location for wedges 1 inch and 2 inches high are %ata for the rIght4mnd half of this statlon were obtained by tit- lation between data obtained o.ne4al.f and two wedge lengths downstream. Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-8 - - lWC!A
15、 RM A5OLl-2 I shown in figure 7. -The greatest average section lift coefficient obtained in this series of measurements was 1.85 for wedges 2 inches high with their leading edges-at 2Fpercent chord. (The maximum section lift coefficient of the basic airfoil was 1.33.) . A similar series of measurems
16、nts was made using right- and left- hand wedges alternately. The number of trailing vortices was the same as for the previous-arrargement, but the sense of adjacent vortices alternated. The results of these nreasurements are also shown in figure 7. The maximum lifts obtained with the l-inch-high wed
17、ges was about the same as with the arrangement employing right4an.d wedges only, but with the alternating 2-inckhigh wedges the maximum lifts were less than vith the 2-inch-high right-hand wedge.s. A few tests were made with wedges 3 inches high, but in each case the maximum lift was less thsn with
18、the correspondfng arrangement of wedges 2 inches high. Wedge spacing.- The next variable investigated was that of wedge spacing. Lt was found that greater maximum lift was obt+ired with an and that an om.space equal to one 1 -z open space between adjacent wedges , wedge width was about optimum for t
19、his type of wedge. Since the data fof figure 7 showed that it was advantageous to use a more forward loca- c tion of the wedges, the tests with spaces between the wedges were made - with the leading edges at lO-.and 25-;percent chord only. The greatest maximum average section lift coefficient obtain
20、ed was 1.93 for 2-inch- high right others produced as much-maximum lift with less drag, but the results were not consistently repeatable. It was concluded that they were too sensitive to small random flow disturbances to merit fur- ther consideration for-this application. - Effect of Wedges on Drag
21、_ - The difference b-the .dlrag coefficient (based on wimeasured for a geometrically similar wedge on the dummy wall; Doubling the height of Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-2 XACA RM A5OLl2 9 I .- the wedges more than tripled the incr
22、emental drag. This would be expected from the drag data obtained for the individual wedges which showed that the drag coefficient based on frontal area a8 nearly pr- portional to the ramp angle or height of the wedges. Thus, doublfngthe wedge height would quadruple the value of an incremental drag c
23、oefficient based on wing area. The rapfd rise of drag with forward movement of the wedges may be, in part, caused by the forward movement of transition from laminartoturbulent flow. Removing every other wedge reduced the incre- mental drag of the model nearly by half, and, as previously mentioned, a
24、ctually benefited the maximum lift of the wing with 2-inch4Lgh wedges. Lift, Drag, and Pitching4oment Characteristics of Configuration Adopted for Detailed Study In figure 9 are shown the lift, drag, and pitching+nomant character+ iatics of the atifoil w5th 2-inch-high wedges spaced one wedge width-
25、 apart across the span 8t the l&perc8n+chord StatiOn. Data are BhoWn for the model with the trailing-dge flap set at various deflections from o“ to 400. Simi.l data for the model without wedg,es are also pre- sented. It should be rem8&ered that the section drag coefficient includes the tare drag of
26、the circular end plates. The max3mum sectim lift coefficient with the flap zmdeflected Was fncreased from 1.33 to 1.93, an increase of 0.60. With the flap deflected bO“, the increase was from 2.07 to 2.39, or an increment of 0.32. The effect on the 1Ut curve was to 8xtend its nearly linear range to
27、higher angles of attack. There Was little effect on the ane for zero lift or onthelift-curve slope. With the flap deflected 20 , the shift of the lift curve Caused by stalling of the flap wa6 delayed to a higher angle of attack. Withthe flapdeflectedtiO, the flapwas always stalledin the positive lif
28、t r-e, which probably accounts for the reduced effec- tiveness of the wedges. The drag of the airfoil intheloandmoderate liftrangewas, of course, greater with the wedges than withoti. In the high lift range corresponding to separated flow on the basic airfoil, however, th8 drag of the airfoil with w
29、edges was less than the drag of the basic airfoil. The zero-lift pitching moments were not significantly affected by the presence of the wedges, particularly for the airfoil with the flap undeflected. For a flap deflection of 20 the airfoil Without Wedg8B suffered a reduction in longitudinal atabilt
30、tg, but this reduction wa8 delayed to a highsr angle of attack by the addition of wedges. c Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-10 Flow Studies NACA RM A5OLl2 Tufts.- Observations of tufts attached to the airfoil without wedges showed tha
31、t the basicairfoil stalled from separation of the turbulent boundary layer. The separated area appeared Ut$ally at the trailing edge for an angle of attack of about go, and progressed steadily forward to about midchord at maximum lift. With the wedg8s in place, the Initial appearance of separation w
32、as delayed to an angle of attack of about 20. At higher angles of attack the flow was unsteady. The ,area of separation swept forward Intermittently from the trailing edge to the position of the wedges, causing the airfoil model to lunge aB the flow separated and real+ tached. At no time did the flo
33、w ahead of the wedges separate from the SLITf 8C8. Pressure dFstribution.- In figure 10 are shown chordwfee distribu- tions of pressure on the airfoil with and without wedges. The angle of attack was l&.7, corresponding to cl, of the basic airfoil. Flow separation is indFcated over the rear half of
34、the basic airfoil by the regfon of relatively constant pressure, but for the airfoil with wedges the flow is attached as is shown by the continual recovery of pressure, A localized area of low pressure occurred in the vicinity of the wedges. (The line of pressure orifices passed through the center o
35、f an open space between wedges.) c For h&her angles of attack the peak negative pressure near the nose of the airfoil continued to rise. For an angle of attack of 19.4 the pressure coefficient P attained a value of at least -XL.5 without indi- cation of flow separation at the trailing edge. Because
36、of unsteadiness of flow, satisfactory pressure measurements could not be made at maximum lift. Total-pressure surveys.- Total-pressure surveys were made at several chordwise stations downstream of the wedges. In figure 11 are shown the results of surveys made at the 95ipercentihord station for four
37、angles of attack. These data are shown as contour maps of the parameter (l- &/q) similar to the maps in figure 6. The outline of the wedges in the figure appear distorted because of the magnffied vertF- cal scale. Similar data for the basic airfoil (except for 14.7 angle of attack for which angle th
38、e flow had separated from the surface) are also shown. The result of the action of th8 wedges as injectors of high- energy air into the thick turbulent boundary layer is apparent. Test With Multiple Small Vanes , A brief investigatfon was made of vortex generstors. These devices consisted of small v
39、anes made of flat, l/s-inch sheet brass aB shown in the following sketch. i .z. - Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-The vanea were installed 442 inches apart across the span of the airfoilmodelatthe lOpercent- chord station. The angle o
40、f attack of the v5ne5 wa5 22-l./2 with the 5ense of the angle of attack alter- nated between adjacent vane5 50 a5 to produce oppositely rotating trailFng vortfces. Liti, drag, and pitchingeom3 &data forthenodel with the mm-type vortex germ+ ators are 5hown Pn figure 12. Also shown are 5hilar data (f
41、rom fig. 9) for the basic airfoil and the airfoil with wedges. ThemaxGnum1W-t coef- ficient with the vortex generators was 1.89, only 0.04 less than the maximum obtained with wedges. -siona in inches The tests of wedge-shaped bodies demonstrated that they did delay separation of the turbulent bounda
42、ry layer from the surface of an air- foil. The relative importance of the roles played by the simple lateral spread- of the flow engendered by the diver- face of the wedge and by the inductionof high-energy air Vito the bow layer downstream of the wedges by the circulation of the trailing vortex was
43、 not made clear. The effectiveness of this latter lnechanlem depend5 on the di5- tame of the axis of the vortex above the surface and an the d&meter of the core9 as well as on the cticulation of the vortex. It is appment that greater mixing action in the bozmdary layer would be realized if the axis
44、of the vortex were brought down close to the vicinity of the outer edge of the boundary layer, and ff the core diameter were reduced. In the present tests, the vane-type vortex generator5 were superior to the wedges in regard to both these effects. The lesser effectiveness of the l-inch-high w-edges
45、 as compared to the 2-inchhigh wedges (fig. 7) may be accounted for by the reduced ramp angle of the forwardpartofthe wedge cau5ed by contourFngthe lower surface to fit the.surface of the airfoil. When mounted well for- ward onthewing, the average ramp angle oft& foruardhalfof the 1-inch+igh wedge5
46、was about 3O. The testa on-the dummywall showedthat the wedges were less effective when the.rw angle WELB reduced to 4O. The addition of the second wedge to make the total height 2 h&e5 fncrea5ed the raq angle of the forwar & portion of the wedge to about go. , Provided by IHSNot for ResaleNo reprod
47、uction or networking permitted without license from IHS-,-,-12. -*. I- NACA RM A5OLl2 s The reason for the faflure of the maximum lift to increase when the closely spaced, 2-inctiigh wedges were moved forward from the 2.5-percent- chord station to the X&-percent-chord station was not made clear. It
48、is believed that the sU.ghtly blunt edges of the closely spaced wedges, when placed in the thJn,boundary layer near the leading edge, may have caused a sufficiently large local disturbance to precfpitate flow sepa- ration. The lesser effectiveness of the adjoining wedges as compared to the open-spac
49、ed wedges,. iqspite of the fact that the former arrangement produced twFce as msny vortices per unit span,may be due to the presence of a dead-air regicm in the angular space between the adjoFning wedges. Such a dead-air region would accelerate boundary-layer growth and the occurrence of flow separation. The drag of the wedges was shown by the tests on the duJmqy wall to be high. When
