Wind Tunnel Design Studies and Technical Evaluation of Advanced Cargo Aircraft Concepts

SCHOO~OF ENGINEERING

OLD DOMINION UNIVERSITY

NORFOLK VIRGINIA

(,' Technical Report 76-Tl1 o .::

WIND TUNNEL DESIGN STUDIES AND TECHNICAL EVALUATION

'~ '*'

OF ADVANCED CARGO AIRCRAFT CONCEPTS

By

D.M. Rao

(1'1 t. S~\-CH-14;;1l.lt9)id NIJ rUNNEL ot.:s I GN S TV II If~S N'l6-2 ?P56 MH) TI:CHNIC/'\l tV~LUI\TIDN UF ADVANCl:.O CAfHJ[J AIRCMf-T C,J~{"l::PTS FH~f\L Rt;PdRT (ClO mH4I~Iqt~ lY\lIV. rd:Sf::Af~CIt FHJnDATUJI'i) 36 P HC ONCLAS $4.00 CSCL ViC G3/0~ 4214·6 Final Report 11,\ P~epa~ed fo~ the National Aeronautics and Space Administration Langley Research Center Ha,mpton, Virginia

o

Unde~

d

,o;Gritnt NSG 11..35 '. i, \ ,;, May" 1976 [J t)

PREFACE This report describes the work performed (June 1975 through May 1976) under NASA Grant No. NSG-1135 at Langley nesearch Center in support of the National Transonic Facility Project Office.

The report is in three parts, as under: .I!!!;rU: Estimation of Aerodynamic Losses in the Tunnel Circuit.

Part II: 2nd-Turn Model Studies.

Part III: Proposed Circuit Modification for LNz Economy and Shell Cost Savings.

This report emphasises the basic motivation behind the probl~m5 tackled, and the main results and conclusions obtained. A more detailed presentation of the experimental data and analysis is deferred tv a subsequent document.

ACKNOWLEDGEr.mNTS The author is pleased to acknowledge the contributions of the following personnel at NASA Langley to various aspects of his work: L. W. McKinney (Grant Monitor) and members of the Aerodynamic group, NTFPO -- for valuable discussions and helpful comments; I J. B. Adcock and the Pilot Cryogenic facility staff Ie: carrying out a diffuse:r-loss test program and providing valuable data; W. G. Johnson Jr. and Ann B. Bell for executing the 2nd-turn and rapid diffuser experiments with skill and patience, and supplying results under a demanding time schedule.

The author would also like to acknowledge the assistance of Dr. Gene L.

Goglia, professor and chairman, Derartment of Mechanical Engineering and Mechanics, Old Dominion University for administration responsibilities with respect to this grant.

StJr.fMARY In support of aerodynamic studies relating to the design and performance prediction of the National Transonic Facility (NTF), the following main tasks were accomplished: 1. Estimation of aerodynamic losses of the tunnel circuit, 2. Refinement of the high-speed diffuser loss prediction meth0d utilizing experimental data generated for the purpose, 3. Model studies of flow in the 2nd-turn and measurements of the fan inlet distortion and overall pressure loss, 4. Development of a shortened fan nacelle configuration of improved aerodynamic performance, and 5. Evolution through model studies of an efficient rapid-diffuser system as the key to a circuit-modification proposal to reduce volume and minimize liquid-nitrogen consumption, at the same time saving on the shell cost.

PART I ESTIMATION OF AERODYNN-tIC LOSSES IN THE TUNNEL CIRCUIT 1. Introduction Reliable estimates of the aerodynamic losses in the tunnel circuit are needed at the project stage in order to: a. Establish maximum fan power ::"'cquirements, b. Define the range of fan pressure ratio covering the tunnel operating envelope, c. Identify circuit components hav:tng significant individual contributions to the total loss, where aerodynamic improvement would therefore be well worth the effort, and d. Indicate the potential for achieving further economies in tunnel cost and- energy requirements by aerodynamic refinement.

It was noted early in the NTF circuit loss calculations that the high-speed diffuser by itself was responsible for more than 50% of the total energy loss.

On comparing estimates also made for Langley 8-ft and 16-ft transonic tunnels with available fan power measurements, it was concluded that diffuser losses were being grossly over-estimated (in relation to the precision needed for the present purpose) by the methods commonly in use (such as those described in refs. I and 2). A search was therefore made of the available diffuser literature for information and data from which to assemble a simple, more reliable method :J for diffuser loss estimation.

2. Diffuser Loss Calculation Even a cursory look at the vast amount of published data reveals clear that the inlet boundary layer plays a vital role in determining diffuser performance. Sovran and Klomp (ref. 3) demonstrated that this was true for different diffuser geometries; and also that the pressure recovery of near,..

I.

optimum diffusers correlated on the basis of the area blockage due to boundary layer displacement thickness (0*). The material presented in Reference 3 however does not readily lend itself to accurate calculation of the diffuser loss. The data reviewed earlier by Henry et al. (ref. 4), although rather sparse, was found to be more convenient to use for the present application.

1-2 The total pressure loss factor

K= x---

where

H = Average total prc~5ure

p = Static pressure

AR = Diffuser area ratio (>1)

for conical diffusors of total-angle 26 around 5 to 6 degrees is plotted in figurc I-I as an empirical function of the inlet displacement thickness ratio o*/Rl (from the data analysis of ref. 4). With this function, the loss in a diffuser preceded by a test-section length can be calculated using the inlet boundary layer thickness obtained (in the simplest approximation) from flat-plate turbulent boundary layer relations. In the calculation scheme (fig. 1-2) the test-section and the diffuser are necessarily treated as a coupled system. Note that this procedure admits both Reynolds nwnber and compressibility effects via the boundary layer thickness calculated from test-section stream parameters.

The pressure ratio (HIiHz)* thus calculated is compared with 'data measured in the Pilot Cryogenic Twmel and the Diffuser Test Apparatus at Langley (figs. 1-3 and 1-4) from low subsonic speeds to ~I = 1. The agreement is

good up to M = 0.8 beyond which the calculations increasingly fall short

of the data (by about 10% at ~I = 1). The K-function in figure I-I was based on incompressible (M <: 0.2) data; although compressibility effect on K has been indicated in reference 4, a reliable factor to quantitatively ~~ccount for it near M = 1 is lacking. Using a factor

yields good agreement with the present M = 1 data. Pending further experimental

verification, this factor is therefore adopted in the tunnel loss estimates.

* Hz was evaluated from exit pitot survey data integrated by the "mass- momentum" method described in NACA TN 3400.

1-3 The ability of the above method to predict the Reynolds number effect on circuit loss was tested against the Pilot Cryogenic tunnel data at M liZ 1 (fig. 1-5). The calculated fan pressure-ratio gives a proper Reynolds number trend, but under-estimates the absolute values. In the calculation, the loss contribution of a cylindrical LN2. injector bar mounted across the high-speed diffuser was based on two-dimensional uniform-fll)w drag coefficient data.

Although it is expected that the cylinder drag will be magnified in the adverse pressure gradient of the diffuser flow, there is no ready means of accounting for it theoretically. An empirical adjustment by multiplying the cylinder drag coefficient by a constant to match the lowest Reynolds no. data point is found to yield good agreement with measurements over the Reynolds no. range.

3. NTF Circuit Loss Estimation The component-wise break-up of the estimated NTP circuit losses for M = 1 and ambient stagnation conditions is shown in Tabl€) I-I.

4 • Concl usion A simple method has been assembled to calculated the aerodynamic loss of the test section and high-speed diffuser combination, ~s the most important item in the tunnel circuit accounting for nearly 60% of the total loss at M,~ 1. Comparison with experiments show that Mach number (0.2 to 1.0) and Reynolds llumber effects are reliably predicted by this method.

5. References "I. Pope, Alan. Wind Tunnel Testing. John Wiley, NY.

2. Pankhurst, R. C. and Holder, D. W. Wind Tunnel Technique. Sir Isaac Pitman, London.

3. Sovran, G. and Klomp, E. Experimentally Determined Optimum Geometries for Rectilinear Diffusers with Rectangular, Conical or Annular Cross-sections.

Fluid ~techanics of Internal Flo\-,' Symposium, 1965, pp. 270-319.

4. Henry, J. R. et a1. Suwmary of SubsoniC-Diffuser Data. NACA RM L56FOS, 1956.

1-4 Table I-I NTF Circuit 110ss SUllunary eM :: 1, Stgn. Pro ::: 14.7 psi, Stgn.

Temp. ::: 635°H) till psi 1. Test Section + High Speed Diffuser 0.995 2. ~~de1 Support Strut 0.125 .3. 1st 'fum 0.057 4. 2nd Leg 0.077 0.040 S. 2nd Tum 6. Nacelle (Annular Diffuser) 0.184 3rd Leg Diffuser 0.087 7.

8. 3rd Turn 0.009 9. 4th Tum 0.009 0.049 10. Rapid Diffuser 11. Screens (3 x lq) 0.048 12. Cooling Coil (12q) 0.192 Total llH psi 1.680 (without cooler) 1.872 (with cooler) 1.129 (without cooler) fan Pressure Ratio 1.146 (with cooler)

0.4 0.3 K 0.2 0.1 oL---------~--------~--------~--------~---------L--------~ .01 .02 .04 .05 .06 Figure 1-1. Loss Factor for Conical Diffusers.

Equivalent I--CylindriC~l ~14 Conical Diffuser

I Test-sect~on Total Angle (28)

Radius (Rl) Area Ratio CAR) . ~ ,. ·;· ... :i.·~·:··:;tf·'~~ ....

I 0\.

Station i 15* B. L. Displacement

t

Thickness Station 2 (2-D Flat Plate Turb. B. L.)

Figure 1-2. Calculation of Diffuser Loss.

1-7 1.10

R. N. = 22 X 10 per ft

1.08

o

o

1.06 1.04 1.02 1.0 0.8 1.0 0.4 0.6 o 0.2 Figure 1-3. Comparison of Estim::tted and ~faa.sured Loss Data- Pilot Cryogenic TI.lnrtel.

1-8 1.10 Ambient Stagnation Pressure and Temperature.

o

1.08 1.06 1.04 1.02 1.0 . 0.2 0.6 0.8

o 0.4 1.0

Figure I-4. Comparison of Estimated and r.feasured Loss Data- Diffuser Test Apparatus.

~f :: 1.0 1.20 ,'easured Fan PeR.

o

!

.J (

I

)..

! ....

From Circuit '9, Loss E3timates ' ...

,~

1.18 ---- Adjusted Estimate ' .....

'" "-. _0

I ..........

-

.

""'Ii

~i

~ P. R. 1.16 .......

..........

~ .....

-

1.14 R. N. Based on Test- Section Diameter

V

6 6 10 x 10 100 x 10 Figure 1-5. Prediction of Reynolds No. Effect on Pilot-Cryogenic Tunnel Pressure Ratio.

PART II NTF SECONn-TURN ~tODEL STUDIES 1. Introduction The 2nd-turn presents a complex geometry, incorporating a center-body which through a 90-degree turn becomes the fan-nac~lle; an airfoil fairing ovpr the fan shaft also turning through 90 degrees; and four struts supporting the nacelle from the outer shell (fig. II-l). With so many complicated and aerodynamically-interfe~ing surfaces prosent, the resulting flow distortion in the fan-annulus is of concern, to be determined and controlled if possible in order to limit fan noise> vibration and dynamic loading on the blades.

Tests were conducted on a 1/12 scale model rig (fig. 11-2) for: a. Velocity survey around the fa.n annulus, b. Tuft visualisation of flow, c. Surface pressure measurements, and d. Total pressure survey across a downstream station to evaluate overall loss.

The annulus dO\'/nstream of the fan station around the tail cone essentially acts as a diffuser, as opposed to the cOlrunon practice of designing the tail cone as a "streamlined body" as in ext.ernal-flow aerodynamics. From an annular diffuser viewpoint, '* the ori~'inal NTF tailcone length was fOlUld to be overly conservative, and it appeared that P. shorter shape might be adopted for its practical advantages and for possibly impro~ed aerodynamics. The opportunity offered by the 2nd-turn model pr()gram was accordjngly utilized to develop a new tailcone geometry.

2. Discussion of Results a. Fan Stat.ion Surv~: The circumferential variation of the core-flow velocity in the fan annulus is shown in fi~tre 11-3\ The salient features in the distribution are the sharp velocity drop locally in tile strut wakes. The wake of the fan-shaft fairing is more diffused at this station, as expected. Also noted f is a shallo\~ region of reduced velocity'** between the inside-struts, attributable * See reference -3 of Part I.

'It'* This feature is also noticeable in the downstream pressure-loss surveys, see figure II-S.

11-2 to the increased resistance of the more closely-spaced turning-vanes on the inside of the turn.

b. Nacelle Tailcone; The static pressure recovery in the tailcone annular

------/,..-

diffuser (from outer wall moasurements) is shown in figure U-4. The original as well as the 20-foot-shortened tailcone shapes arc also indicated (the outer wall was unchanged). The tailcone geometry \~as con$trained by the preference for single-curvature surfaces for ease of fabl'ication. The lJlodified shape eml~loyed a single-row of vor~ex-generators just upstream of the break in the cone.

The data comparison of figure II-4 shows that tho shortened tailcone achieved a faster rate of pressure recovery in the annular diffuser, as intended. Note that a higher maximum pres.;ure occurs at the exit of this tail cone (in accordance with incl'eased area-ratio); to take advantage of this feature, the outer shell should be suitably modified to eliminate an area contraction downstream of the;! tailcone. Comparison of the static pressure at the downstream station (where the area ratio is constant at 1.7 for both cases) indicates a 6% improvement in static pressure recovery ldth the shortened tailcone, which translates to a lOgo decrease in loss approximately (using one-dimensional flow relations).

The downstream total-pressure 1055 profil~s (typical plane-of-turn surveys shown in fig. 11-5) are also considerably improved, primarily due to the reduced wetted area (and friction-lOSS) of the shortened tailcone. A more uniform flow is produced at the entry to the 3rd-leg diffuser, \vhose performance is expected to improve as a result. The vortex-generators, whose effect is confined to the body wake, provide rel~tively little further benefit to an already attenuated wake.

From integration* of the downs ;~ream profiles (measured along four 45 - spaced diameters), the overall loss of the 2nd-turn model with the original ta.i..i Gone was 0.784 qo' ** The shortened tailcone reduced the loss to 0.706 qo' i.e., a 10% improvement, as already inferred from static pressure data.

* Performed by B. B. Gloss ~ of NASA I,angley.

** q measured at Station 0 indicated in figure I1-2.

o

II-3 3. Conclusions A shortened tailcone for the NTP fan nacelle, amounting to a 20-foot (or 40%) reduction over the original length was designed and tested. A 10\ reduction in the total pressure loss of the 2nd-turn ,."as thereby achieved. The accompanying improvement in the downstream flow uniformity is anticipated to benefit the 3rd-leg diffuser performance also.

I,

-, Tail-Cone Fan-Shaft Fairing Center-Body. ____ ~, Figure II-I. N. T. F. 2nd-Turn Geometry.

r48"~

Fiberglass Bellmouth

MS 0 = p.2

tao 94" 16.82"--~ ~-+-~I-- Sta. 0 ....

....

I Plexiglass Shell --------~

'"

\'lood Nacelle--_ (Replaceable) Aluminum Turning Downstream Vane Assembly Survey Stn. 16.5" 82" Fan Stn.

Survey Rake (Rotating) Figure 11-2. N. T. F. 2nd-Turn MOdel (1/12 Full Scale).

----, ----, / , / ~" / .

/ , / , / , / .

, / Cross-Sect1on , / at Fan Station , , // (Looking , I Downstream) I , I I I I I

"

I

Strut Strut "

o I I I I I I ....

I ",e:.==--..,.ql..::-- _____ -r_~l~-_..~l-. O ___ ""'i ....

• Q\ 0.95 V/V = 0° <P 0.9 (Inside of Tum) ........... -------1 Fan-Shaft Fairing -180° -90 o Figure II-S. Fan Annulus Circumferential Velocity Variation.

Modified 0.6

........ -------()

--~

0.4 Original From

c Static Probe

p on Survey Rake 0.2

-

I

-

'-I Distance Along Outer Wall • o Fan Stn. Outer Wall Modified Original Nacelle v.

I

I

(on l-fod. - r----..

.

• Nacelle)

-'--~_J __

- -"---

------~~

+-

f...-lO,;....-i.~I.~---- 21. 5 ---- ... 0011

flH-Survey Station Figure 11-4. Outer Wall Pressure Distributions with Original and Modified Tailcones.

Modified l-fodified Original Nacelle (with V. G.' s) (without V. G.'s) NOTE: Surveys shown are in the plane of turn.

t-C ....

•

o O.S 1.0 1.5

o 0.5 1.0 1.5 o O.S 1.0 1.5

MI/q o Figure 11-5. Total Pressure Loss Profiles Downstream'of Tailcone.

PAR'r III TUNNEL CIRCUIT HODIFICATION FOR LN2 ECONOMY AND SHELL COST SAVING 1. Introduction A unique feature of cryogenic wind tunnels is that the energy supplied directly to the fan is a rather small part of the total energy requirements.

The major energy input is in the form of liquid nitl'ogen (LN ), initially to cool the circuit to the low operating temperature and then to maintain this condition during a test program, by balancing the heat of compression at the fan and also the thermal losses through the shell. To achieve high energy efficiency requires not only an aerodynamically clean circuit design (in order to reduce the fan compression ratio, and for low turbulence and noise levels).

but also a minimum circuit volume on which LN2, consumption directly depends.

Volume reduction without unduly compromising circuit aerodynamics poses a challenge, and an attempt is made here to propose a circuit modification for LN2, economy and to evaluate its merits.

2. Proposed Circuit Modification Inspection of the current NTF circuit, which is already a fairly compact one by conventional standards, suggests additional possibilities for volume reduction. For example, by substituting a constant diameter section for the 3rd-leg diffuser and maintaining this diameter al1 the way up to the rapid diffuser, a reduction of about 17,000 cubic feet results (fig. III-I). Calcu- lations* for a typical NTF test program indicate that this 6% volume reduction is worth about 4% saving in LN2, consumption. Additionally, there is a like saving in the tunnel time to complete the test program (taking into account the time for changing the stagnation conditions). The test time saved yields a direct saving in electrical energy of 4%, and contributes an increment to the annual productivity of the facility, The more compact shel1 resulting from this modification is estimated** to cost less by about $1.8 million (including savings in supports, foundation, insulation, etc.).

* By E' S. Cornette of NASA Langley.

** By G. A. Wentland of NASA Langley.

111 .. 2 Th~ aerodynamic consequences of the proposed circuit modification will bo: a. Reduced loss through the constant diameter section replacing the diffusor, b. Incl'eased losses through 3rd and 4th turns due to increased mean velocity (although some alleviation may be expected due to a more uni£prm velocity profile produced at the 3rd turn by the cylindrical section), and c. Increased loss through the rapid diffuser of larger area-ratio (viz. 4.5 versus 2 originally).

While items a and b can be reasonably well estimated, the rapid-diffuser loss is not easily calculated. There is also the question whether the desired level of settling-chambel' flow uniformity is achievable with an area-ratio 4 rapid diffuser, without incurring undue loss penalty incurrod through flow treat- ment. In order to establish with confidence the feasibility of the proposed circuit modifica~ion, experimental data on rapid-diffuser performance in a configuration pertinent to the present study is needed.

Yet another modification to the current NTF circuit suggested in figure III~l consists of substituting a rapid-expansion (of a small area-ratio) for the conical portion of the high-speed diffuser' (i.e., between the end of transition and 1st-turn entry). This takes advantage of the shortened fan-nacelle tailcone (see Part II) to reduce the overall tunnel by about 20 feet. The additional saving in circuit volume is about 8,200 cubic feet, to bring the total reduction to 9%. The aerodynamic loss penalty accruing from the high-speed diffuser modification has been roughly estimated (taking credit for the loss- reduction in a shortened 3rd-leg) to be less than 5 6. However, for a more precise trade-off study experimental data for the ~articular configuration will be required; in its absence, this aspect \~ill not be pursued further in this report.

3. Rapid Diffuser Experiments A 1/12 scale test rig was used to obtain comparative data on rapid diffusers of area-ratio 2 and 4.5. The diffuser test configurations are shown in figure 1II-2.

The "baseline-bell" diffuser conforms to the geometry currently adopted for NTF. The "baseline-conical" model represents an alternate shape for ease of manufacture. The remaining three models are area-ratio 4.5 and consist of 40- and 50-degree total-angle conical diffusers fitted with radial vanes, and a

111-3 beU .... shaped geometry deoigncd* for constant wall pressure (following the theoretical procedure as used for the "baseline-bell" diffuser design).

The radial-vane concept wns originally propesed as a low-loss flow- spreader for rapid conical diffuSC1'S of large area ratio (up to 15) J as commonly employed in blowdown tunnels (sec refs. 1, 2, and 3). Briefly described, the diffuser is compartmented oyer its full length by means of several equally spaced radial vanes. A small disc (of about 2% area blockage) placed on the vane leading-edge intersection forces separation of flow in the inner corners of the compartments. The inlet flow is displaced radially outwards by the corner separation bubbles, remains attached to the diffuser wall and is divided equally into the vane compartments. With;i,n the compal'tments, the bubble closure is accompanied by continuous flO\\' diffusion as evidenced by a substantial pressure rise. A relatively flat exit velocity profile is attained follO\dng the merging of the flows leaving the compartments.

While the previous work on the vane-diffusers was conce.rned mai-n1y with effective flow spreading in large area-ratio expansions" the present object was to develop a combined sys tern of moderate area-ratio rapid diffuser and flO\~ treatment for a specified level of settling-chamber flow uniformity with minimum energy loss. A quantitative definition of this minimum loss was then needed to evaluate the feasibility of the circuit modification, as already discussed.

The geometrical variables tested for loss~minimisation were the di.sc diameter and the vane recess depth (fig. III-3). An optimum combination of the .

two was critical also with respect to the dO\vnstream flo\'1 Wliformity.

4. Discussion of Results The effects of radial vanes alone and with disc in the 40-degree diffuser are illustrated in figure 1II-4. Vanes by themselves ;\:lOW a relatively minor effect on the highly distorted ~H profile delivered by the 0iean diffuser across the settling chamber. Addition of an optimum disc to the vanes not only i.'llproves markedly the flow uniformity, there is actually a small reduction in the loss.

An increased static pressure recovery in the diffuser further reflects the beneficial corner~bubble type of flow produced within the vane compartments by the disc (fig. 111-5).

* By J. B. Peterson Jr. of NASA Langley.

111-4 Interestingly I with an optimum vane-disc combination the final prCSSl.lrO recovery actunlly exceeds the vnlue repol'ted in literature* for plain conical diffusers having the same angle as the compnrtment equivalent cone- nngle. The nppenrancc of a suction-peak ncar the start of the diffuser unmistakably indicates attached wall flow induced by the addition of the dhc to the vanes (even though the disc in this case was located 5 inches into the di£'£u5er).

trho distribution of total-pressure loss (relative to diffuser inlet total pressur¢) distl'ibution across the ~ettling-chamber, in terms of 61-1/ 6H* ~ (where the bar denotes an average value) is shown in figure III-6 for aU the test configl.lration~. The topmost curve for each case represents the best uniformity achieved. These comparisons show that the flO\i uniformity with 40- degree cordcal vaned-diffuser exceeds that of all the other mod<::ls. Nith com- parable effort devoted to vane-disc optimisa.tion in the SO-degree diffuser, it is believed that this diffuser will prove to be at least eq!.lally efficient.

A static-pressure recovery of nearly 50% of the ideal was obtained in the settling chamber with the optimum 40-dcgree vaned-difit.1ser. For .;:ompar .. , ison, according to reforence 4, to produce uniform downstream flow with rapid Jiffusers utilising screens would require overall resistance corresponding to ~ pressure recovery. 'rhus, the radial vane concept represents an advance in rapid-diffuser technology by significantly reducing the associated energy- loss penalty.

'lhe overall performance of the various diffuser models is summarised in figure llI-7. From this comparison it is concluded that: a. The baseline-bell diffuser was not superior to the baseline-conical model, both requiring exit-resistance to l'un full.

b. Flow uniformity better than the baseline case was achieved with area-ratio 4.5 radial-vaned conical diffusers. The penalty was an increased loss by a factor of 2.S.

* e.g., see reference 4 of Part I.

** Since~H/Hl in all cases was close to n, ~II/Hl = .01 x 611/~H.

111-5 To put the above loss comparisons in proper· pel'spectivo, it may bo pointod out that to uttain the sarno quality of downstream flow unifol~mity ,dth tho baselino ~~iffusers will require increased exit-resistance. At the same time, since the .

VllllOU-diffuscrs are not dependent on exit resistance for flow spreading, the loss ponalt.y in this case can be partially alleviated by using less rosistance Idthout significantly compromising flow quality (as already noted in the case of 50-degree diffuser, fig. IIl-6). Also, the "tuning" of the disc-vane combination was not carried to the limit due to timo limitations in the lll'oscnt tests, and some additional improvement is considered possible. Finally, the vane- diffuser loss being largely made up of skin-friction, a favornble Reynolds no., effect may be expected.

5. Conclusions The experi.mental data discussed in the foregoing were used to esti,nate the NTF drcuit losses in the baseline and the "low-volume" alternate configurations (fig. 8). The alternate circuit shO\~s a loss increase by 6% over the baseline when operating at M:= 1 and ambient stagnation conditions.

Under cryogenic (viz., high Reynolds no.) conditions I its impact on the fan-· energy is diminished by the time-saving due to reduced circuit volume, as discussed in Soction Il, The final effect of J~N2 consumption taking into account the opposing influences of vo.1wne l'cduction and fan-energy increase (due to increased circuit loss) idll depend heavily on the pnttern of facility utilisation. This aspect requires a more detailed study of the energy trade- off and costs, which is outside the scope of this report. HOI"cyer, if the proposed circuit VOlume-reduction is found to be cost-effective, then the resul ts of the present study provide confidence that by the iil.pplication of optimised radial-vane rapid diffuser, no penalty need be paid in terms of tunnel flow quality.

6. References 1. Rao, D. M. itA Method of Flow Stabilisation with High Pressure Recovery in Short, Conical Diffusers." Aeronautical Journal (Royal Aeronautical Soci~ty), May 1971, pp. 336-339.

2. Bclyanin, B. V. et a1. "Study of Flow Characteristics Behind Diffusers with Large Angles of Flare." FTO HT-23-586-73.

3. Rna, D. ~t. nnd Seshadri, S, N. "Application ('If Radinl Splitters for Impl'oved Wide-Angle Diffuser Performance in r.i Blowdown TunneL"

ero be pubHshod in AIM Journal of Ail'craft.)

4. ewald, B. "Low Speed Tunnels with Tandem Test Sections, A Contribution to Somo Dosign Problems." AGARD CP 174, Paper no. 7, 1975.

[Length Saved by Short Nacelle] 3Td Leg Diffuser Replaced by Cylindrical Section I1V = 16,800 ft

2(1 ft1 /

.,-----/---

, ---~~--- I I I I I I I I I I

-~. "I

) ~

~V = 8,200 ft

Rapid-DiffuseJ: A-rf!2 Ratio Increased from 2.0 to 4.5 Conical Portion of High-Speed Diffuser Replaced by Rapid-Diffuser (20 ft Shorter) Total Circuit Volume Reduced About 9% Figure III-I. N. T. F. Circuit Volume Reduct~on Potential.

1II-8 AH Survey (Cruciform Ruke) .....-Buseline Bell

t

A. R. = 2.0 3 x lq SC1'eens __ _ Exit Honeycomb with 2 Double-Screens [Each Made from 2 )( lq Baseline Screens Crossed 45 ) Conical 12 Vanes 8 Vanes 40 Conical

t

A. R. =4.5

l

Alternate

_----'"~Bell

Figure IlI-2. Rapid-Diffuser Test Models.

llI-9 Disc Diameter ,.-~-~-~

" I

I I , , , I \ \ \ , \ \ , ..... -~--~----- Recess Length Figure 111-3. Splitter Geometry (40 Conical Diffuser) Showing Variables Tested.

III-lO All/l\Jf .95 1.0 1.05 ,

·

·

·

~t\H/qE = 14.4 Clean · -+- t.

~ l.IH/'lE = 14.4 'Hth Splitters ~--t.

Only ......

Ml/qE = 14.0 Splitters ~.-t.

and Disc.

OWith Double-Screens Behind Honeycomb Total Pressure Loss Profile in Settling Chamber Figure lII-4. Effects of Radial Vanes and Disc vn' Flow Unifo:rmity Downstream of 40° Conical Diffuser.

111-11 1.0

/~~

Recovery ~ 'Ideal' Data for ~ (l-Diml. Inviscid Flow)

A. R. = 4.5

0.8 Conical

/

Diffusers of Equivalent

/

Angle (15 )

/

~ Optimum Vanes + Dioc

/'

/

/

~ Vanes Only C

/

P

I

Clean Diffuser

I (No Vanes)

0.2

I

I

o

Disc Diffuser Location Exit I0I0- ..... Diffuser -0.2 Starts • Figure III-S. Pressure Recovery in 40° Conical Diffuser (Wall Pressure Measurements).

I1I-12 Baseline Baseline (Bell) (Conical) • 1. 05 llWllH ../'b Honeycomb h::::l'-l.o~"","","f>-Io.4-ti~~ 1.0 '....,-JJ + 2 Doubl e Screens 0.95 Honeycomb + 1 Double Screen Settling Chamber Dia. ----~

<!:=D Honeycomb Only

Settling-Chamber Ml Survey: - Vertical

o Horizontal

___ ..I Honeycomb + 2 Double Screens Honeycomb + 1 Double Screen Honeycomb On'ly 40 Conical Conical Bell Figure III-6. Settling Chamber Flow Uniformity with Various Rapid- Diffuser Models.

III-13 ~ .~

Performance Parameter ~ e

~ ....

I:: ::;, Test Configuration R.M.S.(AH) R.M.S. (U' AH

m

U qExit

l Percent Percent

Baseline 1. 26 0.68 5.40 (Bell) A. R. = 2.0 Baseline 1.43 0.53 5.45 (CoJ\ical)

-------- -------.-~ _ ................... _-- - --~~~~

Arrows Indicate Potential Exists Conical for Improvement 0.88 0.97

13.60 t

8 Vanes

' 500

I ._ ~ Conical A. R.

= 4.5 NOTE: Estimated 0.83 1.10 14.64l

! 12 Vanes

loss due to I

~ honeycomb and

all screens = Sq

approximately.

Bell 1.18 0.63 15.60 Figure 111-7. Rapid-Diffuser Experimental Study Summary of Results.

M = 1. 0, Ambient Stagnation

) Rapid-Diffuser + Screens

e~~~~~~~=:=~;::j~~~~~~~~ } 3rd + 4th Turns

-3rd Leg + High Speed Diffuser Mi psi + Strut ~ + ~ ~ 1st Turn I ....

+ .;.

2nd Leg + 2nd Turn + Nacelle

!

Baseline Alternate (Low Volume) Figure III-S. Aerodynamic Loss Penalty Due to Circuit Nodification.