1、TECHNICAL NOTE 4309 cn 0 m d z E-l 4 NATIONAL ADVISORY COMMITTEE 3 FOR AERONAUTICS USE OF SHORT FLAT VANES FOR PRODUCING EFFICIENT WIDE-ANGLE TWO -DIMENSIONAL SUBSONIC DIFFUSERS By D. L. Cochran and S. J. Kline c. Stanford University ,.+ ;- *, g-; 2 ,- ; ; g; ; ,. . *: , i;$?.$.; , , X _ , . ,_ Wash
2、ington September 1958 Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-NATIONAL ADVISORY COMMITTEZF: FOR AXRONAUTICS TECHNICAL NOTE 4309 USE OF SHORT FLAT VANES FOR PRODUCING EFFICIENT WIDE-AWGLE TWO-DIMENSIONAL SUBSONIC DIFFUSERS By D. L. Cochran and
3、 S. J. mine SUMMARY An investigation of the use of flat vanes in two-dimensional sub- sonic diffusers was made. Using optimum designs of vane installations, high pressure recoveries and steady flows were obtained for diffuser- wall divergence angles up to 42. Criteria for optimum configurations were
4、 developed which indicated that the vanes should be symmetrically arranged in the vicinity of the diffuser throat, that the vane-passage divergence angle should be approximately 7.0, and that the vanes should have a certain predictable length dependent upon the diffuser geometry. INTRODUCTION In vir
5、tually all systems involving the motion of fluids the need arises to accelerate or decelerate the flow. In general, a flow accel- eration (or nozzle process) results in a smooth flow with high effec- tiveness, low loss, and good velocity profiles. On the other hand, unless considerable care is taken
6、 in the design of the passages involved, a flow deceleration (or diffusion process) almost always results in unsteady flows with large pulsations, large losses, and highly nonuniform exit velocity profiles. Such a condition is not only undesirable in itself but it also frequently creates even more u
7、ndesirable effects on the performance of downstream components such as compressors and burners. This problem becomes particularly aggravated when a space limitation is present as, for example, in jet engines. Since a given deceleration of the flow requires a given area ratio, by continuity, a compac
8、t sub- sonic diffuser must necessarily incorporate large angles of divergence of the walls, but such geometries are precisely those that create the most unsteady flow, the largest losses, and the most unpredictable behav- ior. Thus, the discovery of a means for production of wide-angle dif- fusers o
9、f high performance, stable flow, and predictable behavior is a problem of considerable concern. The study of a promising solution to this problem by the use of well-designed, short, flat vanes is one of the primary purposes of the present investigation. Provided by IHSNot for ResaleNo reproduction o
10、r networking permitted without license from IHS-,-,-NACA TN 4309 The basic cause of the poor flows found in diffusers has been known for many years to lie in the effect of an adverse pressure gradi- ent increasing in the direction of the flow) on the boundary layer which inevitably occurs on the wal
11、ls in any real fluid flow. Since f2 diffusion, by definition, involves an adverse pressure gradient and since in almost all applications of interest the boundary layer on the walls is turbulent in character, the problem of diffuser flows is inextricably connected with the problem of the flow of a tu
12、rbulent boundary layer in an adverse pressure gradient and, hence, with the problem of stall or separation. These problems have been the object of a great deal of the- oretical and experimental effort over the past w years, but despite this, sufficient understanding has not been gained. Consequently
13、, no means are available for advance theoretical prediction of even the over- all basic flow pattern that will occur in a given diffuser, and it is therefore necessary not only to measure performance but also to study the basic flow mechanisms if any real understanding of diffuser behavior is to be
14、obtained. Thus, the second principal objective of the present investigation is to add to the available knowledge concerning the flow mechanisms in diffusers, that is, flows with adverse pressure gradients. A Both of these objectives are being pursued as a portion of a con- tinuing research program i
15、n the Mechanical Engineering Laboratory at Stanford University. The experimental work on vanes in wide-angle dif- Susers in itself constitutes a rather extensive research program; there- fore, the emphasis of the present report has been placed on the presenta- tion of the experimental results; descr
16、iption of the flows, and discussion of a means for the design of vaned wide-angle diffusers having good flow characteristics and high performance. This report is based primarily on the work of Cochran (ref. 1) . Certain additional information concerning performance calculations and future work is al
17、so contained in reference 1. The present results cover a wide range of geometries but are limited in respect to inlet Mach number and variation in both inlet flow geometry and condition. Discussion of the flow mechanisms is given as needed, but it was beyond the scope of the present work to attempt
18、a presentation of their details and full implications. However, as an outgrowth of the experimental results and observations of the continuing overall research program, Kline has written a report on the topic of basic flow mechanisms (ref. 2). Reference 2 has been written to integrate with the prese
19、nt report, and Kline not only rationalizes the results found in vaned and unvaned diffuser flows but also discusses the entire problem of stall in terms of two new concepts including the introduction of a new flow model which has already been experimentally verified. The reader who is interested in
20、the more basic aspects of stall and of the flow mecha- nisms in adverse pressure gradients should refer to reference 2; however, it is suggested that the present report be read first since considerable reference is made to the material presented herein. Provided by IHSNot for ResaleNo reproduction o
21、r networking permitted without license from IHS-,-,-NACA TN 4309 This investigation was carried out at the Mechanical Engineering Laboratory, Stanford University, under the sponsorship and with the financial assistance of the National Advisory Committee for Aeronautics. SYMBOLS Nomenclature involved
22、 in diffuser geometry is illustrated in figure 1. A area, sq ft a minimum spacing between adjacent vanes, in. b minimum spacing between diffuser diverging wall and adjacent vane, in. CER energy-recovery coefficient, defined by equation (4) C pressure-recovery coefficient, defined by equation (11) di
23、stance from plane of diffuser throat to plane of vane leading edges, in. ft-lb specific heat at constant pressure, - lb- OF ft-lb specific heat at constant volume, - lb-OF ft-lb kinetic energy, - lb length of vanes, in. distance between parallel walls of diffuser, ft constant of proportionality in N
24、ewtons second law equation, lb-ft lb-sec 2 boundary-layer shape factor head loss, lb/sq ft or in. water nondimensional head-loss coefficient, H/ Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-NACA TN 4309 ft-lb internal energy, - lb correction facto
25、r for compressibility effects ratio of specific heats, cp/cv length of diffuser diverging wall measured from throat to exit, in. number of vanes static pressure, lb/sq ft volumetric flow rate, ft3/sec * dynamic head, p -, lb/sq ft or in. H20 2gc Reynolds number ft-lb gas constant, - lb-qi radius of
26、curvature, ft or in. temperature, OR free-stream velocity at edge of boundary layer, ft/sec local x-direction velocity in boundary layer, ft/sec local x-direction turbulence velocity or time-variant com- ponent of u, ft/sec velocity, ft/sec exit velocity for one-dimensional flow spec if ic volume, f
27、t3/lb distance between diverging walls normal to geometric axis of diffuser, ft n f t-lb flow work, - lb Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-NACA TN 4309 w weight flow rate, lb/sec X, Y length variables, ft or in. a total divergence angle
28、 between adjacent vanes, deg )=O2-o1 a change in a variable 6 boundary-layer thickness, in. boundary-layer displacement thickness, 1 - ;)w, in. 6 6- boundary-layer momentum thickness, J0 (1 - :); dy, in. qp presswe effectiveness, CpR/Cmideal 8 half of diffuser total divergence angle, deg 2 8 diffuse
29、r total divergence angle, deg P viscosity, lb/hr-ft P density, lb/cu ft If normal to a streamline, ft or in. Subscripts: 1 inlet 2 exit a ambient condition as appreciable stall av average dL center line L diverging wall length Provided by IHSNot for ResaleNo reproduction or networking permitted with
30、out license from IHS-,-,-max maximum o outer or stagnation W throat width w wall BACKGROUND INFQRMATION The work reported herein is a direct extension of the work of Moore and mine (refs. 3 and .4) . A summary of the status of diffuser research which is pertinent to the present report is given in re
31、ference 4. A summary of the available diffuser design data was given by Patterson (ref. 5), and a more recent summary of the data applicable to the pres- ent geometry was given by Reid (ref. 6). It therefore seems undesirable, in view of the length of the present report, to repeat these summaries he
32、re. It is, however, pertinent to make a few remarks concerning pre- vious investigations of the governing parameters and the overall patterns involved. Reid (ref. 6) found that the performance f a two-dimensional dif- fuser was strongly dependent on the total divergence angle. and the ratio of wall
33、length to throat width. He also found that, with length ratio held constant, flow separation and pulsations occurred as the divergence angle was increased beyond the point of maximum pressure recovery. Vedernikoff (ref. 7) obtained stable flows for included angles up to approximately 14.0 in a two-d
34、imensional diffuser having a constant length ratio of 10.0. At larger angles vortices formed first along one wall and then along the other, the intensity of the vortices increasing as the angle was increased. In addition, Vedernikoff found that the maximum pressure recovery occurred just prior to th
35、e formation of the vortices. Wts (ref. 8) experimented with an asymmet.ric two-dimensional diffuser of constant length ratio and also found that, with increase of angle, the occurrence of maximum pressure recovery very closely preceded the occurrence of separation. He found further that the angle at
36、 which separation occurred did not depend on the inlet Reynolds number but that the amount of separation present increased as the angle was increased. Jones and Binder (ref. 9) investigated the effect of inlet turbulence intensity in a square diffuser and found that the turbulence level influ- enced
37、 both the velocity distributions and the diffuser flow patterns. In addition, they found a complete lack of two-dimensional flow, although the total divergence angle of their diffuser was only 8.0. Kalinske (ref. 10) studied the effect of turbulence in conical diffusers of constant area ratio. He ob
38、served that the effect of turbulence was to produce a Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-NACA TN 4309 flattening of the velocity profiles and that large-scale turbulence was more effective than fine-grained turbulence. Moore and Kline (r
39、efs. 3 and 4) studied the basic flow mechanisms in diffusers by using two sizes of water-table units incorporating dye injection for flow visualization (see fig. 2) . They also investigated a large number of schemes for the production of efficient, wide-angle diffusers using various types of vanes a
40、nd mixing devices for boundary- layer control. With the exception of aspect ratio, which was shown to be unimportant, the large water-table unit had the same geometry as the air unit used for preliminary tests by Moore and KLine. Moore and Kline discovered several important new characteristics of un
41、modified, plane-wall diffusers. Holding all the inlet flow con- ditions, the wall length, and the throat width constant, they found that four regimes of flow are obtained as the diffuser divergence angle is increased from oO. These regimes are illustrated in figure 3 which shows typical dye trace ph
42、otographs (reproduced from ref. 4) and a schematic sketch for each regime of flow. The four regimes are as follows : (1) A regime of well-behaved apparently unseparated flow (fig. 3(a) . (2) A regime of large transitory stall in which the separation varies in position, size, and intensity with time
43、(fig. 3(b). his is the regime of highly pulsating flows noted by Reid, ref. 6, and others .) (3) A regime of fully developed stall in which the flow is rela- tively steady and follows along one wall of the diffuser (fig. 3(c). The major portion of the diffuser is filled with a large turbulent recirc
44、ulation region which extends from the diffuser exit almost to the diffuser throat and is roughly triangular in shape. The main flow stream experiences little expansion in passing through the diffuser; hence, little pressure recovery is obtained in this regime. (4) A regime of jet flow in which the f
45、low separates from both walls and proceeds through the diffuser similar to a free jet (fig. 3(d). Furthermore, it was found that the divergence angles bounding the dif- ferent flow regimesare not unique but, in fact, depend strongly on the ratio of wall length to throat width and in some instances o
46、n the entering free-stream turbulence level. This is clearly illustrated by figure 4 which is taken from reference 4. Figure 4 shows the zones of flow as a function of divergence angle, ratio of wall length to throat width, and inlet free-stream turbulence for constant thin inlet wall boundary layer
47、s and nearly constant mean inlet velocity profile. The effect of a change in free-stream turbulence intensity is shown by Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-NACA TN 4309 comparison of the dashed boundary curves with the solid boundary cu
48、rves. It should be noted that the definition of the line of appreciable stall, line aa in figure 4, is somewhat subjective. This line was determined experimentally by Moore and nine as the locus of angles for which a transitory stall was first visible using dye injectors of roughly 118 inch in diame
49、ter near the wall of the diffuser. However, later investigations led to the conclusion that small transitory stalls probably existed at much lower angles, and this conclusion has been extended and verified experimentally in the work described by mine in reference 2. Conse- quently, the name “appreciable stall“ is now used in referring to the li
