1、s)WlTECHNICAL NOTE 3996INVESTIGATION OF A SHORT-ANNUIAR.-DIFFUSER CONFIGURATIONUTILIZING SUCTION AS A MEANS OFBOUNDARY-LAYER CONTROLBy StafEord W. Wilbur and James T. HigginbothazuLangley Aeronautical LaboratoryLangley Field, Va.WashingtonJune 1957Provided by IHSNot for ResaleNo reproduction or netw
2、orking permitted without license from IHS-,-,-TECHLIBRARYKAFB,NMNATIONAL ADVISORY KMmTTEEDClL7DhbTECHNICAL NOTE 3996INVESTIGATION OF A SHORT-ANNKGAR-DR?FUSER CONFIGURATIONUTILIZING SUCTION AS A MEANS OFBOUNDARY-LAYER CON2ROIIBy Stafford W. Wtlbur ad Jsmes T. HigginbothsmSUMMARYA straight outer-wall
3、annular diffuser having a center-body lengthof one-half the outer-bdy diameter and an area ratio of 1.9:1 has beeninvestigated for mean inlet flow sagles of 0 and El.5 in order todetermine the effect of area suction applied on the inner wall. Theentrance shape, number, and 10cation of the p-gs ttim
4、w tair was removed were varied. The auxiliary air flow was varied from Oto approximately 4 percent of the main stream air flow; the mean inlet.Mach number was approximately 0.26.For most of the configurations, significant hprovement in perform-a71 ante was obtained over no control when a suction flo
5、w rate of as littleas 1 percent was utilized. Increased rates of suction were responsiblefor some additional improvements dependiag on the configuration of suc-tion openings. Rounding the entrance of the suction holes and increasingthe srea through which suction was alied effectively decreased theau
6、xilisry flow losses and thereby produced higher vslues of diffusereffectiveness. The diffuser-exit velocity distributions were alsoimprovedby the increase in suction sxea and by an increase in thesmount of suction.INTRODUCTIONThe genersl purpose of this investigationwas to develop a shortdiffuser de
7、sign that is applicable to turboet-engine installations.Specifically,the obective was to achieve a minimm total-pressureloss and a uniform exit velocity distribution within a diffuser lengthof 1.0 outer-body Usmeter or less. Previous research has indicated that.-this objective can be accomplished on
8、ly through the use of boundsry-layer controls.a71Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-2 lWICATN 3996The effects of vortex generators in diffusers with center-bodylengths vsrying from zero to 1.0 outer-body dimneters sre reported inreferenc
9、es 1 to 5. Suction and injection controlby means of slots onvery short diffusers is described in references 6 and 7. From a studyof the results of the investigationsof references 1 to 5, it was evidentthat the most favorable velocity distributionswere obtained at the down-stream station correspondin
10、gto a length-dimneterratio of 1.0 when thecenter-body length was 50 to 60 percent of-the outer-body dismeter. Inaddition, references 6 and 7 indicated that designs with good aerodynamicshapes should be used in conjunctionwith suction control in order toreduce to a minimum the auxiliary flow quantiti
11、es and pumping require-ments. Whirling flow at the diffuser inlet also must be removedbystraighteners (see ref. 2) before efficient diffusion can be accomplishedin the type of design under study.The present investigation employs suction from an appreciable sur-face area of the center body in contras
12、t with suction from a discreteslot as in reference 7. The center-body configurationis about thessme length as the longer configurationo? reference 7, its lengthbeing percent of the outer-body diemeter. The effect of whirlingflow at the inlet, with and without a flow-straightenerinstallation,was inve
13、stigatedbecause of the possible application as a turbine dis-charge diffuser in which areciable whirl exists under some operatingconditions. The investigationwas conductedwith fully developed pipeflow at the inlet of the diffuser, which had a 21-inch constant outer-wall diameter and an area ratio of
14、 1.9:1. The mean inlet Mach numberwas maintained nesrly constant at 0.26 with a correspondingReynoldsnuder, based on inlet hydraulic diameter, of 1.6 x 106. The meaangles of flow at the inlet were 0 and 19.5.DPPt44tSYMBOLSdismeterstatic pressuretotal pressurestatic-pressurerisetotal-pressureloss-.va
15、71cProvided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-NAC!ATN 3996 3. qc compressible impact pressure, - pw u local velocity within boundary layerY radial distance from the diffuser outer wall.M Mach nuniberw mass flowR ratio of auxilisry air -s flow to
16、 main-stream mass flow,percentNR Reynolds rnmiberbased on the inlet hydraulic diameterCp pwiw-pow= c=ff icientn diffuser effectivenesse flow angleSubscripts:0 1 diffuser inlet station- la reference static pressure station2,3 diffuser tailpipe stationsx diffuser stationt total velocity of whirling fl
17、owA axial component of total velocityA bar over a symbol indicates a weighted average.APPARATUS AND PROCEDUREGeneral ApparatusThe main air flow was sucked through the general test apparatus(fig. l(a) by a fan. The air entered a inlet bell that was covered%with a fine mesh cloth. Before entering the
18、diffusing region and down-stream duct, the air passed through an annular-approach duct which wasapproximately 27feet long and which had a constant inner diameter. - Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-4 NACA TN 3996of l= inches and an out
19、er dianeter that varied from 21 to 25 inches.2The center body was used for the auxiliary air duct and contained anorifice designed according to the specificationsof reference 8. Forthe whirling-flow tests, a set of fixed guide vanes was installed justdownstream of the inlet bell to give a whirling m
20、otion to the main-streamair flow. The resulting weighted mean flow angle at the diffuser inletwas 19.50.Diffuser ModelThe diffuser center body shown in figures l(b) and l(c) was verynearly elliptical in the profile cross section and had a length thatcorresponds to a 31 equivalent cone sngle. The sha
21、pe was identical.to that of configuration3 of reference 5 in order to obtain comparableresults. Rows of tsuctionholes were drilled normal to the center-bodysurface at center-body diameters of 13.08j 12.54, 11.88, 11.14, snd10.24 inches for rows 1 to 5, respectively. Each row containedeighteen -inch
22、equispacedholes, indexed o around the circumference8 3from those of the previous row. The holes were tested with both squareand rounded leading edges.For part of the whirling-flow tests, flow-straightenervanes weremounted on the outer wall 2 inches downstream from the inlet station(measuredfrom the
23、30-percent-chordpoint) an angle of attack of 0.These flow straightenerswere symmetrical rectangular NACA 0012 air-foils with 3-inch chords and 3-inch spans. Twenty-four were equispacedaround the outer-wallperiphery.InstrumentationA single row of static-pressureorifices was installed longitudiXfalong
24、 the diffuser outer wald fi?omthe diffuser inlet station to a pointapproximately 3 diameters downstream. Four equlspaced static-pressureorifices on the diffuser outer wall were installed at stations 1, l(a),and 3. Surveys of stagnationand static pressures and flow angle weremade at stations 1, 2, an
25、d 3 by using two probes sPaced Iw” ax ateach station. Two shielded reference total-pressuretubes were instilledpermanently 180 apart in the center of the annular passage upstreamfrom the diffuser inlet. A shielded total-pressuretube was alsoinstalled inside the center body to measure the recovery of
26、 the suctionair.The investigationfor axial flow with not- -.aTest Procedure .was initiatedby obtainingpressure measurementssuction holes (no suction boundary-layer control).Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-NAimTN 3996. After these meas
27、urements were obtained, the origin ofregion was determined by visual observations of small. to the center body and was found to coincide with the5the sepsrated-flowwool tufts attachedrow of suctionholeslabeled row 2 (see fig. l(b). This row of suction holes was drilled andperformance measurements we
28、re taken with vsxying suction rate. The holeswere then rounded on the outer edge to reduce the suctionpower required.Successive rows of holes were drilled, the edges were rounded, and testswere made in order to determine the effect of varying the nrmiberof rowsand of their position. Amixture of oil
29、and lsmpblack was usedto studythe flow along the centerbody when rows 1, 2, and 3 were used for suctionof appruimately 2 percent. When this method was used to observe theflow, thin bands of oil and lmnpblack were painted at critical locationson the center body to clarify the flow phenomena in the vi
30、cinity of thesuction holes and imediatel.y downstresn of this region.After the sxisl-flow tests, the whirling-flow tests with varyingrates of suction through rows 2, 3j and 4 were initiated. Data weretaken with and without the flow strdghteners installed.Bases for Comparison of ResultsThe velocity d
31、istributions across the duct at stations 1, 2, and 3.are presented in terms of U/El, a ratio of the local velocity to theaverage velocity of the fully developed pipe* The angular distribution across the duct eity distribution for the whirling-flow inletstatic-pressuredistribution along the outerof ,
32、 a ratio of the difference the%,1flow at the diffuser inlet.is presented with the veloc-condition. The longitudinalwall is presented in termslocal wall static pressureand the-wall static pressure at station la to the mass-weighted com-pressible impact pressure at station 1. For sxi flow, the static-
33、pressure rise between stations 1 and 3 obtained with various amounts of3-1, which is similar to theauxiliary flow is presented in terms of%1term used in presenting the longitudinal dis correcting forthe suction pumping power reduced the 63-percent reduction to 30 percent.PerformanceWith Whirling Flo
34、wVelocity and flow-angle distributions.-The ratio of the radialdistribution of the local total velocity to the mean inlet total.velocityis presented in figure 13 along wtth the flow-angle distribution acrossthe duct at stations 2 and 3 for the diffuser without flow straighteners.The flow sngles were
35、 reducedby suction control in the region 5 900). Suction on the a71diffuser inner wsll inherently increases the flow angle new the ductcenter line because of the law of conservationof angulsr mcmentum. .The velocity distributionswere improvd by suction in the region5 to 9 inches from the outer wall
36、at station 2. The effect of suctionwas to produce less uniform total-velocitydistributions at station 3and to establish reverse flow near the duct center line.With rotation, with or without suction, the axial-flow componentof whirling flow at stations 2 and 3 was m-re uniform than the axialinlet flo
37、w condition except in the region near the duct center line(fig. 14). Suction with whirling flow improved the axial-velocitycom-ponent over most of the duct srea; however, it also intensified thereverse flow nesr the center line.Rotation had some favorable influence-onthe diffusion process;however, c
38、onsiderable energy is represented in the rotational componentat stations 2 and 3 and would have to be recovered to make the processefficient. Rotation could possibly be used to advantsge in some con-figurations in which the mount and distributionof rotation wascontrolled.The effect of flow straighte
39、nersin conjunctionwith whirling flowis illustrated in figure 15. The flow angles at station 2 were reducedby the flow straightenersfrom the outer WSU to a distance of 4 imihesfrom the outer wall; however, the straightenerswere less effective overProvided by IHSNot for ResaleNo reproduction or networ
40、king permitted without license from IHS-,-,-. NACAm 3996 SL. the remaining duct area because of the l/4-inch gap that existedbetweenthe inner wall and the tips of the straighteningvsnes and, also,becauseof the very low axisl velocities in this region. The flow angles at sta-.tion 3 were more uniform
41、ly low with the straightenersbecause of theincreaseduniformity of the ponent. The velocity distributionsat stations 2 snd 3 were nonuniform with or without suction.Performsace coefficients.-The static-pressure-risecoefficientbetween stations 1 snd 3 is presented in figure 16j low values sre imdi-cat
42、ed with no flow straightenerswith or tithout suctionbecause of thehigh flow angles (mean flow angle of about l+l”). Without flow straight-eners, suction increased the static-pressurerise by only a slight amunt.This result is the net effect of a reduction in loss due to the fact thatsuction is almost
43、 counteractedby an increase in the flow angle. Withflow straighteners, suction increased the static-pressurerise m appre-ciable smount. This result is principally due to the large reductionin loss caused by suction.The diffuser effectiveness with straightener vanes (fig. 16)did not increase as rapid
44、ly as the static-pressure-risecoefficientbecause of the pumping-power correction. The effectivenesswas muchlower than that for the s.xislinlet flow condition with or without suc-tion because of the losses through the vsnes, because of the lack of.effectiveness of suction in impro the velocity distri
45、bution, smd,also, because of the unrecovered energy represented by the rotation. The mean flow angle (fig. 16) doubled between the inlet and exitstations without flow straighteners. Flow straightenerswere respon-sible for a reduction in the mean flow angle of about 30 at the exitstations; however, f
46、or good performance the flow angle should have beenreduced even more. Suction, in general, was responsible for an increasein the flow angle.The loss coefficient for whirling flow with and without straight-ener vsnes is presented in figure 16. Whirl reduced the measured losscoefficient *out 30 percen
47、t without suctioribut the use of flow strtit-eners doubled the value obtained without flow straighteners. Part ofthis increase is beeved to be due to the increased diffuser losseswith the straighteners. The measured-loss curves indicate much lessreduction in loss produced with suction without straig
48、htenersthan withstraighteners. This result shows that, tith high flow es and noflow straighteners,the total-pressure deficiencies along the inner wallare very smsll because of the mre favorsble pressure gradient. The losscoefficient corrected for pumping power produces optimum suction qumti-ties of
49、about 1 percent. This low value is causedby the relativelyhigh pumphg-power coefficient u compared to the improvements obtained*through suction.Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-I-2 NACATN 3996 “The performance ofouter wall and a centerthe outer-body diameterSUMMARYOF
