Um NOUNCED 'NAVORD REPORT 2376

NOL HYPERSONIC TUNNEL NO. 4 RESULTS 11:

DIFFUSER INVESTIGATION

V/

LIBRARy OF C GRESS

5 May 1952 EERENCEEP

U. S. NAVAL ORDNANCE LAORATORY

WHITE( OAK, MARYAND

OW)

XAVORD Report 2376

Aerob.llistic Research Report 90

NOL HYPERSONIC TUNNEL NO. 4 RESULTS II:DI ?TFSER I NVESTI GATI ON

Prepared by:

Peter P. Wegener and R. Kenneth Lobb

ABSTRACT: Results of a diffuser investigation in the continuous12 x 12 c Hypersonic Tunnel No. 4 are presented. A brief introduo-tion describes previous supersonic diffuser work. The diffuserinvestigated and the experimental techniques are then discussed. Theresults show first that air condensation has little or no effect ondiffuser performance. Data on diffuser throat areas and overall pres-sure ratios needed to start and maintain hypersonic flow are given fora Mach number range from 5.9 to 9.6. The effect of two differentthroat locations and different diffuser configurations on tunnel per-formance is investigated. A peaked throat diffuser with 30 walldivergence aft of the throat was selected for a more detailed study.The pressure recovered by this optimuz diffuser in the rangs5.9 ' M ' 9.6 varies from 1.8 to 2.3 times the value of the pres-sure recovered by a pitot tube operated at equal Mach number. Thebest performance with 2.3 times pitot recovery is achieved at M = 7.2.Spark schlieren photographs taken throughout the test section and dif-fuser show the shock waves and boundary layers. Also tunnel startingrequirements were measured and are discussed. It is found that if thediffuser throat is opened muffidiently, the tunnel can at all times bestarted at an overall pressure ratio somewhat lower than the pitotpressur ratio for the same Mach maber. Quantitative comparisons ofall results are made with data previously given in the literature andone-dimensional theory.

UNTO: /~~ CENTER/

Q o,-CONGRESS /%G I fTh , . 25, D.C. /

U. S. NAVAL ORDNANCE LABORATORYW& TE OAK, MARTLAND

i

-5-

IAVORD Report 2376 5 May 1952

This is the seoond KAVMRD Report on an investigation oarrifed out in theU. S. Naval Ordnance Laboratory Hypersonic Tunnel No. 4. This windtunnel was sponsored by the Bureau of Ordnance and was f.irat put intooperation in May 1950. The experiments in this report were carried outin the redesigned and improvd version of the equipment which was putinto operation in Pall 1951. This investigation was partially spon-sored by Sverdrup and Parcel, Inc. and the U. S. Air Force. Specif-ically the diffuser investigated is a scale model of a diffuserconsidered for the hypersonic working section of the Gas DynamicsPacility of-the Arnold Engineering Development Center at Tullahoma,Tennessee. The present report contains an account of the investigationwithout models and supports in the tunnel. The work with models andsupports is presently being oarried out and will be reported in aforthcoming NAVORD.

K. Staab and R. Garran participatea in the tests. Messrs. E.Stollenwork (now with Sverdrup and Paroel, Ino ), C. White (N.O.L.),and R. Waiter (S & P) wre responsible for the mechanical design ofthe diffuser.

W. G. SCHINDLERear Admiral, USXCommander

H. H. KW1ZWBI, ChiefAroballistis Research DepartmaetBy direction

• ii

I/

NAVORD Report 2376

CONTENTS

Pape

I. Introduction . 1Ii. Description and P~erformance of Equipmtent. 3III. Experimental Technique .... .. 4IV. Effect of Air Condensation or. Diffuner Performanc. 5V. Pressure Recovery of Operating Tunnel for Various

Diffuser Configurations ............. 5VI. Tunnel Starting Requirements . 8VII. Comparison with Other Diffusers ..... .......... 9VIII. Summary .... .............. ...... . 9IX. Notation. . . . . ....... . .......... 0X. Referenem. . . . ................. 11

V1

Lii

xA

IAV01W Report 2376

SOL HYPERSONIC TUNNEL NO, 4 RESULTS II:DI FUSER I NVESTI GATI ON

I. hE1WODUCTIOI

1. A major component of a supersonic wind-tnnel working sectionis the diffuser. The diffuser decelerates the flow from supersonic tolow subsonic speeds. This transfer process is not free of losses.Due to shock waves and friction the stagnation pressure decreases andthe entropy of the flow increases. A general discussion of the super-sonic diffuser problem is given in chapter 9 of rererenoe (a).

2. Diffuser "efficiency,, "pressure recovery," and "overall pres-sure ratio" are quantitative expressions for the potential energy orprestture reoovered. It is sufficient to discuss pressure alone becausethe diffuser end-temperature is approximately equal to the tunnelsupply temperature. High-pressure recovery is desirable in order tooperate wind-tunnel power plats at a minim overall pressure ratiofor a given Mach number. This operating pressure ratio is widelydifferent for different types of wind-tunnel diffusers or test sectionconfigurations. The latter may have open, half open, or closed Jets,depenling on whether none, two, or four test section walls bound thGflow. For a fixed type of test section and diffuser, the pressurerecovery is also a function of test section Mach and Reynolds number.Finally one oommonly distinguishes between fixed or variable area dif-fusers. Only the latter type where the diffuser throat area or"second throat" can be closed down after supersonic flow has beenestablished is of interest here.

3. Most supersonic wind tunnels have diffusers whose oonflgu-rations are largely patterned after subsonic experience. These giveoverall pressre ratios of the order shown in Figure 5.16, page 85of reference (b). One may conveniently compare this pressune recoverywith that of a pitot tube placed into the test section at the sameMach umber. In the pitot case the flow ia decelerated through anormal shock and a subsequent isentropic compression. A comparisonof the pitot tube as a prelsure-recovery device with smpersonicwind-tunnel diffusers shows that the latter are ordinarily lessefficient. This is due to the fact that for a tunnel diffuser vi-.cosly effects enter the picture in addition to losses corresponding

to a normal shook. Since a diffuser employing a system of oblique*hoeks must be more efficient than a normal shook diffuser, effortswere made to impreve supersonic wind-tunnel diffusers. In fact,such a device may theoretically have perfect efficiency. Furthermore,it was long known from r=aJet-diffu"r studies (reference c) thatefficiencies above those of a pitot probe could be obtained inpractice. An above pitot-recovery diffuser for a supersonic windunnel was demonstrated by Kuruweg (reference d) at M = 2.9. This

KAVORD Report 2376

diffuer had a single, peaked throat, variable area, and straight walls.Detailed investigations in the VOL 18 x 18 cm continuous tunnel werepresented by Diggins (reference e). Also Neumann and Lustwark die-cussed (references f and a) diffusers of above pitot tube recovery.Diggins (reference g) then extended the Mach number range of the NOLinvestigations to 4.4. Finally investigations at M = 6.9 in the NACALangley U1' hypersonic tunnel were made by Bertram (reference h).Recoveries up to two times pitot pressure were achieved by him. Further-more, the Langley U" hypersonic tunnel diffuser poessed a paralledduct as a second throat. (Similar configurations are also shown at lowerMach numbesin reference f.) Based on considerations by Kantrowitz(reference i) on the stability of channel flows, it became evidentthat such ducts night be useful to stabilize the final shock transi-tion to subsonic flow. In this arrangeent a stable shook will permitthe back pressure to reach a maximum before the shock "Jumps" upstreamthrough the diffuser throat and the test section.

4. The present investigation covers a Mach number range from5.9 to 9.6 with a variable area diffuser whose throat could be locatedat two distances from the nozzle exit. Furthermore, aft of the dif-fuser throat, the diffuser plates could be adjusted at any angle frogzero degree divergence (compare reference h) to large angles and tosingle peak configurations (compare references a and g). Also thethird or last diffuser plates were adjustable.

5. The results show that simple configurations with one peakedthroat give diffusr pressure recoveries above pitot tube recovery inthe entire range of Kaoh numbers. The peak recovery is higher thauany previously reported data. (2.3 pitot recovery at V = 7.2).

6. Aside from efficient operation of a supersonic tunnel afterstart, the overall pressure ratio needed for this starting is ofimportance. This pressure ratio determines the maximus performancerequirement for the power plant. Very little is known to date onthis subject (references e, g, and J). Fbr this reason startingrequirements were investigated and the results inlicate that hyper-sonic tunnels may be started at pressure ratios considerably belowthose previously anticipated. In fact the tunnel may at all timesbe started at overall preasure ratios which are about equal to thepitot pressure to stpply pressure ratio at the corresponding Eachamber. This result is of particular importanoe since it eliminatescostly additions to power plasts for the momentary attaiment of veryhigh-pressur ratios.

7. It is interesting to note that the open jet diffuser typeis amenable to theoretical treatment. Hereman (references k and 1)has correctly predicted open jet tunI diffuser-perfomanoe. Onthe other hand the closed jet diffuser of interest here because ofits high performance possibilities cannot yet be treated by a

2

)1

KAVORD Report 2376

unified theory. This is due to the fact that the losses to be computedare governed by shock waves and friction. Calculation would requirethe understanding of turbulent boundary-layer oharacteristics in con-verging and diverging channels and shook wave-boundary layer interaction(compare Figures 5 and 13). These phenomena are inadequately known atpres3nt.

II. DESCRIPTION AND PERFORMANCE OF EQUIPMENT

8. The investigation s conducted in the oontinuous NOL 12 x 12cm Hypersonic Tunnel No. 4 described in reference (m) and shown inFigures 1 to 4.

9. Supply temperatures ranging from room tmperature to 5000Cand supply pressures from 1 to 30 atmospheres can be accurately main-tained during the blow in the supply section of the tunnel. A newsteel wedge-type nozzle with a built-in cooling system expands the airto a desired Mach number in the range from 5 to 10. Preheating of theair is needed to avoid air condensation in the test section (refer-ence n). The Mach number distribution in the exit plane of the wedgenozzle at M - 7.2 is uniform to ± .01 M (outside of the about 2 cnthick boundary layer). The Mach number increases somewhat along thecenter line throughout the test section.

10. The diffuser has three sets of plates, the first of which islinked directly to the nozzle end. (Due to the high Math number inthe test section, the shock angle caused by the flow deflection atthis point is small, and long models can be used without having theconventional paralled wall test section.) Two pairs of motor drivenjacks connect to the junction of the first and second, and the secondand third plates respectively. Sliding gaskets (silicon rubberstripping) fastened to the diffuser plates seal the diffuser at allpositions. A flexible pressure seal between the first plates and thetunnel walls keeps test section pressure behind these plates. The twosets of acks operate together so that the second plate angle #8 (seenotation) is fixed during the run. This angle can be changed in betweenruns by offsetting the jacks. Angle q of the third plates can bechanged during the run by a third set of hand operated Jacks connectedto the diffuser end. The diffuser throat opening is reorded on amechanical counter by use of a selsyn system, and is known to within± .005 inch. Steel side walls with pressure taps on the centerline,or alls with 1" thick circular comercial plate glass windows atsome points enclose the diffuser. Photographs are taken with a con-tinuous or short duration (0.5 x 10- 6 sees) light source. Details canbe seen on the Figures 1 to 3 and dimensions are given on Figure 4(1 caliber refers to the 12 cm nozzle-exit width).

11. The diffuser leads into a 120 and then a 36N pipe connectingthe tunnel to vaowm pumps via a 2000 &3 vamuu sphere. Overall

3

vORD Report 2376

pressure ratios up to 50,000 are available. Close control of the supplypressure by the pressure regulators and the "sphere pressure" by a ventvalve permit the establishment of an arbitrarily chosen accurate overallpressure ratio. Th3's accuracy of po/ps is of the order 1o/o. On thisbasis a continuous blow down tunnel isldeally suited for diffuser per-formanoe studies.

ill. EXPERIMENTAL TECHNIQUE

: PI'LOSE= reaome-r or operatng ftmnfl.

12. If a tunnel operates supersonically for s given Mach nimber,supply pressure, and temperature (Reynolds nuber), there exists a mini-mum diffuser throat area for a -iven diffuser configuration below whichno supersonic flow can be maintained even for an infinite pressure ratioacross the tunnel. This minim=w running diffuser area is determined bya test called "area breakdown." During this test the tunnel is startedand operated at a very high overall pressure ratio. The diffuser throatarea is then slowly reduced until the supersonic flow "breaks down" inthe test section as observed with the schlieren system. Minimu run-ning diffuser areas are noted on the counter and indicated as a verticaldashed line in figures depicting pressure recovery..

13. The minimm overall pressure ratio needed to operate the tunnelfor a given Mach niube rf supply pressure, and temperature, and diffuserconfiguration is determined as a function of diffuser throat area forall values of the latter, larger than the minimzm running area. Thistest is called a "pressure breakdown." After starting the tunnel, thediffuser throat area is set on a given value larger than the minimmrunning area. Then by increasing the vacuum sphere pressure slowly,while watching the flow in the test section, the pressure ratio at whichthe supersonic flow "breaks down" can be determined. T hs pressureratio is indicated in the figures as po/pe. For actual application,it must be borne in mind that to maintain supersonic flow a slightlyhigher pressure ratio than Po/Pe must be provided by the power plant.

14. Minim starting diffuser areas ar determined by operatingthe tunnol at a very high overall pressure ratio and opening the dif-fuser. At some critical minimum diffuser throat area, supersonic flowis established in the test section. The identical test can be made bysetting the diffuser throat area on different openings and startingthe tunnel at very high pressure ratios by use of the fat acting valve(1/500 secs). Some diffuser throat will then be just sufficientlylarge to permit establishing mpersonic flow.

~4

kI

NAVORD Report 2376

15. Similar to the oPerating pressure ratio, minimum startingpressure ratios can be found for all diffuser throat ar3am equal toor larger than the minimum starting diffuser throat area determinedabove.

IV. EFFECT OF AIR CONDENSATION ON DIFFUSER PERFORMANCE

16. If a wind tunnel is operated from a room temperature reservoirat a nozzle area ratio corresponding to a Mach number higher than about5, then a fraction of the air will condense at or shortly after reach-ing saturation in the nozzle. Subsequent to condensation, the furtherexpansion may adequately be described as saturated and iaentropic,(references n and o). This condensation of air affects the commonlymeasured flow parameters differently. In particular, static pressuremeasurements are very sensitive indicators of air condensation whilepitot pressure measurements are nearly insensitive. It can now alsobe shown experimentally that the overall pressure ratio of a given dif-fuser configuration is nearly insensitive to air condensation asevidenced by Figures 5 and 6 taken for X5 = 7.6. In this comparisonReynolds effects on diffuser performance are minimized by keeping thesupply density equal in the two cases of (1) air condensation, and(2) no condensation. In the latter case To is high enough to avoidthe coexistence region altogether during the expansion and no conden-sation is possible. Overall pressure ratios in the case with conden-sation are slightly lower due to the actually lower Mach number.(Compare reference o). The shock angles are slightly smaller in thecase of no condensation as expected and previously demonstrated forshock waves on simple bodies in references (n), (o), and (p). How-ever, this angle change does not significantly alter the shock pattern.In general, the diffuser pressure ratio (similar to that of a pitottube) is remarkably insensitive to condensation. The following tests(with the exception of the sensitive static pressure measurements andchecks on starting) were therefore made without preheating the air andthe Mach numbers indicated as M(po'/po) are to be taken as those derivedfrom pitot measurements with the aid of a flow table (reference q).This procedure of running the tunnel "cold" makes it possible to photo-graph extensively without the danger of cracking w.ndows and it gener-ally simplifies testing. The "pitot Mach numbers" may, however, betaken as actual Mach nuambers for the diffuser study without appreciableerror.

V. PRESSURE RECOVERY OF OPERATING TUNNEL FOR VARIOUS DIFFUSERCONFIGURATIONS

17. The pressure recovery (in terms of pitot pressure) over therange of Mach numbers tested is generally of the same order (1.8 to 2.3)with a maximum around Mach number 7. A detailed investigation to deter-mine diffuser performance for various operating conditions, configura-tions, etc. was therefore carried out at this Mach number only.

5

MAVORD Report 2376

18. The pressure recoveries at a Mach number 7.2 for three diffuserconfigurations, different only downstream from the diffuser throat, aregiven in Figure 7. (The Reynolds number based on tunnel width for thecomparable case free of air condensation is about 3.4 x 100). Thearrangement with the second diffuser plates parallel give-3 a slightlybetter recovery for the larger diffuser throat openings. No signifi-cant improvement is obtained when the angle between the second platesis increased to 10 to allow for boundary-layer growth. Recovery is alsopractically unaffected if the angle between the third plates is increasedfrom 30 to 120.

19. It appears that the configuration beyonm the diffuser throat hasno significant effect on the pressure recovery. This indicates that theimportant pressure ratio increase is already obtained in the convergingsection of the diffuser. In the present tests the length of the con-verging section could be changed by locating th3 throat at 3.62 or 6.64calibers from the nozzle exit. These throat locations correspond to thefirst and second Jack positions of Figure 4. The recovery comparison givenin Figure 8 shows the optimum recovery is considerably better for theshorter converging section. The schlieren photographs in Figure 9 showthat in this case the initial oblique shock is reflected twice before theflow enters the throat. Apparently this design criterion of two shockreflections optimizes pressure recovery while further shock reflections in

a longer duct do not improve the re.ovs'ry due to increasing viscouslosses in the Reynolds number range oif the tests (compare reference g).

20. During all tests it was found that small mechanical changes,such as different seals, slack in diffuser plate suspension, etc. affectthe recovery appreciably. For one series of tests the third diffuserplates were free to move by about, ± 1/164 and 'rattled" during the run.The pressure recovery for this czise was less than that for a Orattle-free" diffuser as seen on Figure 10.

21. Schlieren observations for the "rattle-free' case show arapidly fluctuating f i in the diverging section of the diffuser.From noise measurements it is found that, as the diffuser throat areadecreases towards the optimum, the predoain w t sound frequency increasesfrom about 1,000 to 5,000 cycles per second.

22. After breakdown of supersonic flow is achieved, e.g. by closingthe diffuser throat below the permissible minimm area, the noise fre-quency changes abruptly to about 10,000 cycles per second. Pressuredistributions to be given later reveal that the final transition tosubsonic flow occurs through an unsteady shock systew as is to beexpected from results of other investigators, (references e, f, g, i).

23. For a given Mach msber and diffuser oonfl.uration, increasingthe supply pressure improves the pressure recovery, (compare Figure U

6

NAVORD Report 2376

with Figure 7). Here the higher Reynolds number reduces the boundarylayer thickness and relatively decreases the viscous losses. This alsoresults in a smaller minimum running area as shown in Figure 12. Thesedata show that for application of these results to larger tunnels athigher Reynolds numbers, better overall pressure ratios may be expected.Most data presented here were obtained at Po - 10 atm to shorten therunning time of the tests.

24. Spark sohlieren photographs of the flow in the convergingsection of the diffuser for the optimum throat area at various supplypressures, are shown in Figure 13. ihotographs of the shock system inthe converging section of the diffuser for four throat areas are shownin Figure 14. As the first plate angle decreases the number of shockreflections traversing the flow decreases, thus lowering the pressurerecovery.

Prssure distriUtion.

25. A static pressure survey along the centerline of the tunnelside wall for the optimum diffuser throat area is shown in Figure 15.The overall pressure ratio was set at a value where supersonic flow couldbe just maintained in the test section. Beyond the nozzle exit thepressure decreases slightly as a result of the diverging flow in thewedge nozzle. The pressure then increases through oblique &hock waves(compare Figure 13) up to the diffuser throat where the flow is stillsupersonic. The transition from supersonic to subsonic flow occursjust beyond the diffuser throat. (Breakdown occurs when this transi-tion moves into the diffuser throat.) A similar survey for a dif-fuser throat area large enough to start the tunnel is given inFigure 16.

26. In comparison to a diffuser with parallel or nearly parallelthroat plates, a single peak diffuser is simpler to construct, isshorter and provides approximately equal pressure recoveries. Forhypersonic tunnels operating at high-supply temperatures, it offersan additional advantage. Such tunnels exhibit maximum rates of heattransfer from the flow to the wall at the nozzle (first) throat andalso at the diffuser (second) throat. To operate continuously, cool-ing systems must be installed at these points and a short, singlepeak diffuser presents smaller overall cooling requirements.

27. The broken 1" thick windows shown in Figure 17 are examplesof the effect of this localized heat transfer near the diffuserthroat. Prior to the breaking, the tunnel ran roughly 5 minutes atN = 7.2, Po = 21 atm and To = 33PC with the diffuser throat area at theoptimum setting. In both oases the cracks in the glass originated nearthe diffuser throat.

7

NAVORD Report 2376

28. The pressure recoveries for Mach numbers 5.88 and 8.49 aregiven in Figures 18 and 19. As the Mach number Increases the overallpressure ratio required to maintain supersonic flow increases greatly.However, again the recovery in terms of pitot prossure remains nearlyconstant.

29. A further check at Mach number 9.6 gives a minimm overall

running pressure ratio of 142. This corresponds to 1.9 times pitotrecovery.

VI. TUNNEL STARTING REQUIRPJ)ENTS

30. The overall pressure ratios required for starting supersonicflow, free of air condensation, in the test section for various difniserthroat areas are given in Figure 20 (M - 7.2, To = 330C). The minimumstarting area shown as a vertical dashed line is roughly three timesthe minimum rnning area. However, for all areas larger than thisminimum, there is little difference between the pressure ratio requiredfor starting supersonic flow and the pressure ratio required to maintain

supersonic flow. The optimum starting pressure ratio (at an areaslightly greater than the minimum) is somewhat lower than the pitotpressure ratio.

31. Figure 21 gives a comparison between "fast" and "slow" startsfor M(po/po) = 7.2. A "fast" start refers to the case where there issufficient mass and overall pressure ratio available to start thetunnel quickly by opening a valve (within 1/500 sec), while a "slow"start means a start achieved by slowly increasing the overall pressureratio until steady supersonic flow is established. No difference inthe minimum starting area or in the starting pressure ratio could befound in these two tests.

32. During starting of the tunnel, a supersonic jet detaches fromthe nozzle walls and passes into the diffuser. Spark schlieren photo-graphs, (see Figure 22), show the jet in an unsymmetrical position inthe test section (compare with photographs in reference J). When acertain area ratio or pressure ratio is reached, the jet suddenly occupiesthe whole test section and steady supersonic flow is established.

33. The starting requirements for M(po/p0 ) = 8.49 are given in*Figure 23. An in the previous cases, there is little difference between*the starting pressure ratio and the running pressure ratio for the same

diffuser throat area.

34. The optimum starting pressure ratios and the pitot pressureratios for the range of Mach numbers tested are given in Figure 24.In all cases the tunnel can be started at a lower pressure ratio thanthe pitot pressure ratio. The minimum starting area ratios are compared

8

NAVCi.D Report 2376

to th minimum running area ratios in Figure 25 for this same range ofMach numbers.

35. Tests with other diffuser configurations showed no noticeabledifference in the minimum starting area. The change in the startingpressure ratios corresponded to the change in recovery pressure.

VII. COCIPAKISON WITH OTHER DIFFUSER$

36. In Figure 26 optimum pressure recoveries of four closed jettunnels are given in terms of oitot pressure. This recovery increasesup to M = 7.2. (The drop at the two highest Mach numbers might be dueto the relative decrease of Reynolds number or the increasing departurefrom optimum first diffuser plates length.) However, the pressurerecovery in terms of supply pressure decreases greatly as Mach numberincreases. Figure 27 gives the Jnverse overall optimum running pres-sure ratio for the same tunnels. To every Mach number one may alsoread test section and pitot pressure. Although the diffuser recoversabout 100 times the test section pressure at the higher Mach numbers,the final diffuser pressure is only about 1/100 of the supply pressure.Figure 28 finally gives "efficiency" values as defined on the graph.The three previous figures show that at the higher Mach numbers nosignificant performance difference exists between the diffuserscompared.

37. Figures 29 and 30 compare starting data from several tunnelswith the simple one-dimensional non-viscous theory. (Compare ref-erence a). The minimum diffuser throat area and minimum overall pres-sure ratio needed for starting are calculated assuming that a "normal"shock at test section Mach number must be "swallowed" by the diffuser.It can be seen from Figure 30 that the tunnel will start at one-halfthe predicted swallowing diffuser throat area. If the starting werecorrectly described by the above theory then an actual diffuser, withviscosity entering the picture, should only start at larger areas thanpredicted. Since this is not the case, one must assume that a systemof oblique shocks is built up during starting. Therefore, as shownon Figure 29, the starting pressure ratios are even slightly smallerthan the pitot pressure ratio although viscous effects (compareFigure 22) must play an important role in the actual process.

VIII. SUMMARY

38. A diffuser investigation carried out in the NOL 12 x 12 cmcontinuous Hypersonic Tunnel No. 4 in the Mach number range from 5.9to 9.6 is described. It is first shown that air condensation doesnot affect diffuser performance. A detailed investigation of anumber of diffuser configurations and diffuser throat positionsresulted in the following observations:

9

NAVORD Report 2376

39. A single peak, plane wall, variable area diff'user withconverging plates of 3.6 calibers length (one caliber equals nozzlee:it width) and a 30 plane wall divergence aft of the throat givesoptimum performance at M - 7.2. The pressure recovery ranges up to2.3 times that of a pitot tube measured at equal Mach number. Inthe range of Mach numbers investigated, pressure recovery is neverbelow 1.8 times that of a pitot tube. It is found that a parallelall center portion of the diffuser does not significantly improveefficiency, but lengthens the diffuser undesirably. Detailed dataon all configurations, pressure distribution measurements, andspark schlieren photographs of the flow in the test section anddiffuser are given. The effect of increasing Reynolds number,resulting in decreasing minimum diffuser throat area, is demonstrated.Comparisons with other high efficiency diffusers and the one-dimen-sional theory are given.

40. An investigation of starting overall pressure require-ments was also carried out. It is shown that if the diffuser isclosed to an empirically determined minimum starting throat area,the tunnel can at all Mach numbers be started at an overall pressureratio about equal to that of pitot to supply pressure for the sameMach number. The minimum starting diffuser throat is considerablysmaller than that predicted by one-dimensional theory and the com-paratively low starting pressure ratios required eliminate costlypower plant additions for the momentary attainment of extreme pres-sure ratios.

41. The effect of a model and support on pressure recovery will

be discussed in a forthcoming report.

IX. NOTATION

42. The symbols used represent the following physical quantities(see sketch-on page 13)

Ms Mach number setting from nozzle area ratioM(p;/po) Mach number determined from measured pitot

pressure, using isentropic flow tables,reference (q) (insensitive to air condensation)

To supply temperaturep static pressurepo static pressure at nozzle exitPo supply pressurep 0 pitot pressurePap diffuser end pressure

Pe Pt when supersonic flow breaks down in test sectionA th section areaA* nozzle throat areaD diffuser throat area

total angle between second diffuser platesW total angle between third diffuser plates

10

NAVORD REPORT 2376

Note: Jacks 1 and 2 move together so that angleg remains constantfor all D*. The angle is adjusted by offsetting one set ofJacks before the test begins.

X. RFFERENCES

(a) Ferri, A., "Elements of Aerodynamics of Supersonic Flows," TheMacMillan Company, New York, 1949

(b) Liepmann, H. W. and A. E. Puckett, "Introduction to Aerodynamicsof a Compressible Fluid," Calcit Aeronautical Series, John Wileyand Sons, Inc.. New York, 1947

(c) Oswatitsch, K., "Z7eutrale fuer wissenschaftliches Berichtswesen,"Report No. 1005, 1944

(d) Kurzweg, H. H., "A Few Aspects of Future Supersonic Wind-TunnelDesign and Test Techniques," U. S. Naval Ordnance LaboratoryNOLR 1133, 29 June 1949

(e) Diggt-ns, J. L., "Diffuser Investigation in a Supersonic WindTunnel," U. S. Naval Ordnance Laboratory NAVORD Report 1570,3 January 1951

(f) Neumann, E. P. and F. Lustwerk, "High Efficiency SupersonicDiffusers," Jour. of the Ae. Sci. 18, No. 6, pp 369, June 1951

(g) Diggins, J. L. and A. H. Lange, "An Improved Diffuser forSupersonic Wind Tunnels," U. S. Naval Ordnance LaboratoryNAVORD Report 2421, in preparation

(h) Bertram, M. H., "Investigation of the Pressure-Ratio Requirementsof the Langley 11-inch Hypersonic Tunnel with a Variable-Geometry Diffuser," NACA RI L 50 1 13, Oct. 6, 1950, Declassified,MACA Change in Classification Form No. 750, 21 March 1952

(i) Kantrowitz, A., "The Formation and Stability of Normal ShockWaves in Channel Flows," MACA TN No. 1225, March 1947

(j) Ball, G. V., "Starting Processes in an Intermittent SupersonicWind Tunnel," Institute of Aerophysics, University of Toronto,UTIA Report No. 12, February 1951

(k) Nermann, R., "Diffuser Efficiency and Flow Process of SupersonicWind Tunnels with Free Jet Test Section," USAF Air MaterielCommand, AF Technical Report No. 6334, December 1950

(1) Bermann, R., "Diffuser Efficiency of Free Jet Supersonic WindTunnels at Variable Test Chamber Pressure," IAS Preprint No. 349,Presented at IAS Meeting, 28 January, 1 February 1952, New York

11

Uu

11AV(IRD Report 2376

(m) Wegener, P., S. Reed, E. Stollenwerk, and G. Lundquist, "NOLHyperballistic Tunnel No. 4 Results I: Air Liquefaction," U.Naval Ordnance Laboratory NAVORD Report 1742, 4 January 1951

(n) Wegener, P., S. Reed, E. Stollenwerk, and G. Lundquist, "AirCondensation in Hypersonic Flow," Jour. of Appl. Phys., Vol. 22,No. 8, pp 1077, August 1951

(o) Wegener, P., "Hyperballistic Wind Tunnel Research at NOL,"Bureau of Ordnance Symposium on Aerobllistics at the Univer-sity of Texas, November 1950

(p) Wegener, P., "Summary of Recent Experimental Investigations inthe NOL Hyperballistics Wind Tunnel," Jour. of the Ae. Sci.,Vol. 18, No. 10, pp 665, October 1951

(q) Burcher, M. A., "Compressible Flow Tables for Air," NACA TNNo. 1592, August 1948

(r) Eggink, H., 'Stroemungsaufbau und Druckrueckgewinnung inUeberschallkanaelen," Forschungsbericht No. 1756, ZWB 1943.English translation by the University of Michigan: FlowStructure and Pressure Recovery in Supersonic Tunnels,ATI 3570, CGD 808

(a) Neumann, E. P., F. Lustwerk, "Supersonic Diffusers for WindTunnels," Jour. of Applied Mechanics, Vol. 16, No. 2, pp 195,June 1949

12

Af

NAVORD REPORT 2376

ce 0

00

Iii--N<

00zz

13D

NAVORD REPORTr 2376FIG-

MV

14

NAVORD REPORT 2376FIG. 2

,a ?,fI' el'

y.,

NOL 12XI2 CIM HYPERSONIC TUNNEL No. 4

(COOLING SYSTEM EXPOSED ON ONE SIDE)

WEDGE NOZZLE AND ADJUSTABLE DIFFUSER

15

NAVORD REPORT 2376FIG. 3

FIT-

NOL 12XI2 CM HYPERSONIC TUNNEL No.4

DIFFUSER CONTROL MECHANISM

16

NAVORD REPORT 2376

FR(IL PRESSURE TANK FIG. 4

_ _ _ ' 0

SLIDE VALVE SUPPLY PRSSURLS

NOZZLE BLOK -- -

L Co

TEST E TI;ICo

PIAT

ist

DIFFUSER V CPIATE2nd JACK--

3rd DI FFUSR

OFF VVE ....

SI-E . FR( .. VIEW

ASSMBLY SKETCH OF TEST SECTIIM

SCHEMATIC LAYOU' OF NOL HYPERSONIC TUNNEL NO. h

(All dimensions in calibers: 1 caliber 12 cm)

17

NAVORD) FEPORT 2376

-IL

C)

00

0

0 0 0L

IC) C

0 1*

CC~J

00

C)

NAVORD REPORT 2376FIG. 6

00 0

uOu

-L - -m-- 0

cr Ii

>) o

*0

_ _ 0f* 0 ~

uJ 0I- 0

qu 6 w

M r0 0d

0 0

C) 0. z

I-0

0

0

191

NAVORD REPORT 2376FIG.?7

00

0

w000M 00

0 do

w5 V

W--- IT -3

0 p

c10 LL 00 0

0> 0

0 13

ci

0 t

0 0 0 '0

0

F-IaJ-- % 1 0- e

NAVORD REPORT 2376FIG. 8

PRESSURE RECOVERY VS DIFFUSER AREA RATIO

(FOR TWO DIFFUSER THROAT POSITIONS)

M (P/!P.) =7.2

Po 10 ATM, To= 150 C, RUNS 439,440, 442, 452

140SYMBOL DIFFUSER CONFIGURATION

-1-4 1 12

120 -ST JACK

2,

FLOW 0

0 0

00

0.

480 ~ Z OPITO PRSUEfi

WIQ:

IjJ

w>

0

MINIMUM RUNNING AREA AT I SrJACK20

4MIN IMUM THROAT AT 2r'ND JACK

00 01 0.2 0.3 0.4 0.5 0.6

DIFFUSER AREA RATIO Dw/A

21

NAVORD REPORT 2376FIG. 9

I0

t.iaIL

D o0 LL tC0-JLL Q

IL i-0U)

Tk0L -

00 r

22:

NAVORD REPORT 2376FIG. 10

EFFECT OF "RATTLING" DIFFUSERPLATES ON PRESSURE RECOVERY

M p./p.) 7

P0 10 ATIM, To= 15" C, RUNS 427 -430, 44440 _ _ _ _ _ _ _ _ _. - _ _

FLOW /~73

ST JACK

O3 RD PLATES RATTLED

(±6~ AT END OF PLATE)

o 3 RD PLATES RATTLE FREE

100- PITOT PRESSURE P/P

0L 0 0

80-._____

0

40

60"

-23

NAVORD REPORT 2376FIG. II

PRESSURE RECOVERY VS DIFFUSER AREA RATIO

(EMPTY TEST SECTION)

M(Po/Po) -7.22

Po:20 ATM, To=150C, RUN 453140

IS AGI(

I~jJ

oo

0

cc," - PITOT PRESSURE p Po,

60

.J

--MINIMUM RUNNING AREA

2040

0 O.1 0.2 0.3 O4 0'5 06

DIFFUSER AREA RATIO 0*/A

A24

NAVORD REPORT 2376

FIG. 12

MINIMUM RUNNING DIFFUSER AREA

RATIO VS SUPPLY PRESSURE

M ( pV po) =Z 2

TO= 16°C, RUN 482

30

250

i2o

0

w

0

5

DIFFUSER THROAT MEASUREMENT (MM)

0.06 0.08 0.10 0.120.4.1

DIFFUSER AREA RATIO D*/A

25

NAVORD REPORT 2376FIG. 13

LJ

00>.0

z Q i

0

LL- o-.

CC

26Icl

NAVORD REPORT 2376FI G. 14

DIRECTION OF FLOW -*

D*IA =0.6 D*/A=O.4

D*/A =0.3 D*/A =0.1 (MIN)

FLOW IN CONVERGING PART OF DIFFUSERFOR FOUR DIFFUSER THROAT AREA RA.T'OS

Mi(p, /p )7.20 0

Po :30 ATM, To =15 0 C

27

NAVOR) REPORT 2376FIG. 15

0

0

6~ ON3 a~sniiaI-O

ccki

_ _ (9

w

w

0 mQ/).

Z 'Tz z )D- 0-

LL- ~* .LIV08H 83sn=~i IQ- w00 z

0 -I.L 04.I~~ZOW----~ a)

C-,

-3-

LU X33ZZNU0

£01 X dd0LiI~fS~

28<

NAVORO REPORT 2376FIG. 16

p, 0N N3Sl0UO- 1

CL w

w \LLJ a.

(I)

Z nw0 it___ _ 0

H

0 w

z 0~ -J

w c I IX3 3'1ZZON---b 0ORL -ii

(1 ) 0r

_~j

0 0 ID0 ID0

U,W £01 lx'd/d oIjv a 3anSS3d

29

NAVORD REPORT 2376

FIG. 17

ir

A ct C0

o

00 0om ,,i

0.IAa

03

NAVORD REPORT 2376

() K -- 0

o U q:-~ 0 U

cc <

0) 0 aI <I- a. 1, (SWC5

I. z

00

xr 0I

l

DU-

-o 0

o 0 t'a 0

mLU

w a- 00 00tQr) U) .1 0

*r d -r 0LU zna

ct W z zD i-- n i a.-

- - CD 1

I -J _ _ _ _ _ _ _ _ _ _ _ _ _5_ _ _

D w1a , F _ U0l

Lii L

00

0 0 00 0 0 0

i/doiLv8 a~flS38d -1-V83AO

31

NAVORD REPORT 2376FIG. 20

00

D I a: a:

D LJIKy

Qw cr 0

w z

00

0L 0

0 0 I

0 N 0~/d 0~V~ WclSrI 1V~A

00 32

CL CCw1

NAVORD REPORT 2376FIG. 21

STARTING PRESSURE RATIO VS DIFFUSER AREA RATIO

(EMPTY TEST SECTION)

M (P.,/ P.) =7. 2

poz 10 ATM, To 15- C, RUNS 440, 442, 452

1?0 0-

10 0-_0

FPITOrT PRESSURE

0 o

06

2

wm60

wm-

MINIMUM STARTING AREA

40

0 IST JACK

LOW j

o "SLOW" START

0 "FAST" START

- RECOVERY PRESSURE______ (FROM FIG. 6)_ __

004050.6 0.7 0.8 0.9

DIFFUSER AREA RATIO 0 /A

33

NAVORD REPORT 2376

FIG. 22

OIREGTION OF FLOW -

p0/p~, :60.8

P": 62.8

SPARK SCHLiEREN PHOTOGRAPHS OF JET IN THE

TEST SECTION DURING STARTING OF TUNNEL

TUNNEL STARTED AT %o/psp6 4 .4 ,M(Ip / 0 ):z.2

Po 0 ATM, T,,25 C

34

NAVORD REPORT 2376FIG. 23I STARTING PRESSURE RATIO VS DIFFUSER AREA RATIO

(EMPTY TEST SECTION)

M(P, /P0 )x 8.49

P0 =3 0 ATM, To= 160 C, RUNS 433, 434, 438250__ _ _ _ __ _ _

200 - - 1.-T-q-MINIMUM STARTING AREA

0. PITMT PRESSUREVI)

0

00

0.--O 15D

WL

100

w0

50 ~=5002'

0 STARTING PRESSURE RATIO"SLOW"

RECOVERY PRESSURE(FROM FIG. 19)

00 0.3 0.4 0.5 0.6 0.7

DIFFUSER AREA RATIO D*/A

35

NAVORD REPORT 2376

fli

0

0 -o

040< t

01 Z 01q

>~ LLJ K (

<n 00 7-6 2

Lz F- (f M

6 o 04 0

F- 0 0

_ U a. in z

cy- < !U

22

00

o? 0

0o 0 0

d/40 0d 1V i V 38V 83fSS3 1103A

ItI

NAVORD REPORT 2376FIG. 26

____

0

I) U) 0

C U

Ulj

U )

0

-r

0_

0 L0

z1 0)LL

z - w w

U-

CL U

W

U 0

t~ri>~

. QxLI

0 4 a

Ld~ I

0C-0

CL

in N

0

37

NAVORD REPORT 2376

FIG 27

PRESSURE RATIOS FOR OPERATING TUNNEL

DIFFUSER END PRESSURE Peo

0.01

0

w(r

TEST SECTIO'N STA-riC PRESSUREPoP

0 NOL ISx 18 CM REF e00001 a MIT REF. f

'q NACA I I"REF. h

0.000011

MACH NUMBER M(p /p0 )

38

NAVORD REPORT 2376FIG. 28

z u

0 -11) 0 I)

CL

0 0 < /C L]

LL 0 O V /00 x

z

wLL

LL

d 0 6 5

U AON3IOI4.4

39

NAVORD REPORT 2376

FIG. 29

U)

U)

Q-

0

0-C

_ - U:0w D

U) w

LL ) LU D

0-LUUz w x L:c

090 a. aL.z O- LU______________

-- r

00

0 0~'ZZ 2a

04

NAVORD REPORT 237T6FIG. 30

0

I c

0 y

o arZ: ) :W-; 0

U, U .

0. W'L~ ~ *

wu -' <J~ 00

U) M0 :

I-

0

F-

U) /IL

0

0

CD t- to tO0i d6 c; di 0 c

V/*09NI18VIS 8~0J O11Vb V38V *NIA

41