JOURNAL OF PROPULSION AND POWER Vol. 19, No.

2, MarchApril 2003

Conceptual Development of Quiet Turbofan Engines for Supersonic Aircraft

Dimitri Papamoschou University of California, Irvine, Irvine, California 92697-3975 and Marco Debiasi The Ohio State University, Columbus, Ohio 43235
This is a joint thermodynamic and acoustic study of engines for next-generation supersonic aircraft. It explores xed-cycle concepts with potential for quiet takeoff and ef cient cruise. The owpath is simple, without mechanical suppressors. The strategy is to take a representative state-of-the-art military turbofan engine and increase its bypass ratio to a moderate value. The engine core stays the same. Three exhaust con gurations are considered: mixed ow, coaxial separate ow, and eccentric separate ow. Engine-cycle analysispredicts thermodynamic performance and nozzle exhaust conditions at takeoff and Mach 1.6 cruise. Subscale experiments duplicate the static exhaust conditions at takeoff power and measure the far- eld sound. Flyover perceived noise levels are estimated for a twinengine aircraft in the 120,000-lbclass. In terms of effective perceived noise level (EPNL), the eccentric arrangement is 6.5 dB quieter than the mixed- ow arrangement and 5 dB quieter than the coaxial con guration. Spectral- and time-domain analyses indicate that the eccentric exhaust is free of strong Mach wave radiation. The acoustic bene t of the eccentric arrangement, combined with faster climb afforded by the modi ed engine, leads to a reduction of 14 dB in EPNL. Compared to the baseline engine, the speci c fuel consumption of the modi ed engine is about 13% less at subsonic climb and 3% less at supersonic cruise.

Nomenclature
D f M m P r T t U x y = = = = = = = = = = = = = = = diameter frequency Mach number mass ow rate distance from jet exit thrust time from liftoff velocity horizontal distance from brake release altitude geometric angle of attack climb angle polar angle relative to jet centerline azimuth angle relative to vertical plane polar observation angle of airplane

tot 1

= total = ight conditions

Introduction

Subscripts

com eng exp fan LO p s TOM

= = = = = = = =

compressor full-scale engine subscale experiment fan tip liftoff primary (core) exhaust secondary (bypass) exhaust takeoff monitor

Presented as Paper 2002-0368 at the AIAA 40th Aerospace Sciences Meeting and Exhibit, Reno, NV, 1417 January 2002; received 6 May 2002; revision received 21 November 2002; accepted for publication 4 December c 2003 by Dimitri Papamoschou and Marco Debiasi. Pub2002. Copyright lished by the American Institute of Aeronautics and Astronautics, Inc., with permission. Copies of this paper may be made for personal or internal use, on condition that the copier pay the $10.00 per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923; include the code 0748-4658/03 $10.00 in correspondence with the CCC. Professor, Department of Mechanical and Aerospace Engineering. Associate Fellow AIAA. Postdoctoral Researcher, Mechanical Engineering. Member AIAA. 161

OMMUNITY noise from aircraft has profound environmental and economic consequences.First-generation subsonic jetliners were very noisy because of the high exhaust velocity of their engines. Efforts to suppress noise using mixing enhancement had only moderate impact.1 It was not until the introductionof the highbypass-ratio turbofan that noise was reduced remarkably, by 20 30 dB. This was simply achieved by the same thrust being produced with a larger mass ow rate, hence lower exhaust speed. The associated gains in propulsive ef ciency led to much lower fuel consumption, making the high-bypass turbofan the only choice for commercial aircraft developed in the 1980s and beyond. The increase in bypass ratio was enabled by development of high-temperature materials for the turbine blades. For given size of the gas generator, the power that can be delivered to the bypass stream is directly related to the turbine inlet temperature (TIT). Development of economically viable supersonic transports hinges on solving the problem of community noise without penalizing aircraft performance. The same issue affects to some extent military high-performance aircraft because communities surrounding military bases are becoming increasingly sensitive to noise. So far, the bulk of the supersonic noise suppressioneffort has encompassed mixing enhancementand ejectorapproaches,2;3 which typicallylead to large and heavy powerplants.4 One may wonder if supersonic engines will follow the same evolution as subsonic engines, leading to supersonic high-bypass turbofans. However, the issue is not as simple. High-bypass ratio generally causes worse, not better, ef ciency at supersonic speeds. Figure 1 shows calculations of thrust speci c fuel consumption (TSFC) and fan diameter vs bypass ratio (BPR) and fan pressure ratio (FPR) at cruise Mach number of 1.6. The calculation, based on an engine-cycle analysis mentioned later in the paper, assumes TIT D 1600 K (2400 F), a value close to todays limits of turbine materials. It is seen that the TSFC slightly dips and then rises withp increasing bypass ratio. The fan diameter increases roughly with .1 C BPR/ meaning increased drag and weight of the vehicle. The quantitative information shown in Fig. 1

162

PAPAMOSCHOU AND DEBIASI

thus making the technique very effective at suppressingMach wave emission toward the ground. More generally,the MWE results illustrate the potential for noise reduction by shaping the mean ow of the primary and secondary streams. Our study represents the initial steps of a broader effort to reduce supersonic and subsonic jet noise by mean pro le shaping of realistic engine ows. This paper examines, at a fundamental level, the thermodynamic and acoustic performance of a xed-cycle, moderate-bypass supersonic engine.It will be shown that signi cantnoisereductionrelative to todays military turbofan engines is achievable with an eccentric separate- ow exhaust that reduces the convective Mach number of the core stream.

Engine Con gurations

We consider a supersonic twin-engine aircraft with maximum takeoff weight of about 540 kN (120,000 lb) The assumed lift-todrag ratio is 5 at takeoff and 10 at supersonic cruise, values roughly 20% better than those of the Aerospatiale Concorde.16 The study starts with a representative military turbofan engine for this kind of airplane, increases its bypass ratio to moderate value, and assesses noise and performance of three con gurations: the mixedow turbofan, the separate- ow turbofan with coaxial exhaust, and the separate- ow turbofan with eccentric exhaust. The comparison basis is the following. 1) All engines have the same supersonic cruise thrust. 2) All engines have the same core characteristics(mass ow rate to within 10%, overall pressure ratio, and turbine inlet temperature). The baseline engine is a military turbofan with BPR D 0:3, FPR D 5:0, static thrust of 126 kN (28,000 lb) and cruise thrust of 30 kN (6700 lb) at Mach 1.6 and altitude of 16,000 m. The static thrust is dictated by the federal requirement for the aircraft to climb at an angle of 1.4 deg with one engine inopertaive.17 The modi ed engines are increased mass ow rate derivatives of the baseline engine, with bypass ratio 1.6. The size, speci c fuel consumption,and exhaust conditions of the engines are derived from thermodynamic analysis of a Brayton cycle with component ef ciencies and speci c heat ratios listed in Table 1. (See Ref. 18 for more information on the cycle analysis.) For all engines, 2530% of the compressor air is used for turbine cooling, 1% of the compressor air is bled to systems outside the engine, and 1.5% of the turbine work drives auxiliary systems. These gures are representative of the operating conditions of modern engines. Total pressure loss due to turbine cooling is estimated at 7% times the mass fraction of cooling air.19 For the mixed- ow design, the core and fan streams mix at constant pressure, constant total enthalpy, and Mach number 0.4 before expanding to ambient pressure. The overall pressure ratio (OPR) and TIT for maximum static thrust were selected at 30 and 1800 K, respectively. The requirement that all engines have the same supersonic cruise thrust sets the size of each engine. Speci cally, it makes the fan diameter of each engine dependent on the OPR and TIT chosen for cruise. Because the fan diameter is a constantthroughoutthe aircraft mission, the OPR and TIT values at takeoff de ne the static thrust. They were chosen such that the baseline (smallest) engine provides enough thrust for climb with one engine inoperative, that is, T D 126 kN. The engine con gurations are summarized in Fig. 2. For convenience, we adopt a notation that gives the bypass ratio and type of exhaust. B16-MIX, for example, describes the bypass ratio 1.6,
Table 1 Component Inlet ( M1 < 1) Inlet ( M1 1) Fan Compressor Combustor Turbine Nozzle
a

Fig. 1 TSFC and fan diameter, normalized by turbojet value, vs BPR and FPR for Mach 1.6 cruise: , baseline engine and , modi ed engines.

will change somewhat with the assumptions of the cycle analysis, for example, component ef ciencies, but the qualitative trends will not. Figures 1 indicate that, at todays limits of TIT, bypass ratios beyond 3.0 would lead to poor performance at supersonic cruise. Designers of supersonic engines, therefore, face two con icting requirements:high-bypassratio on takeoff/landingfor reducednoise and low- or zero-bypass ratio for ef cient supersonic cruise. One possibility is the variable-cycleturbofan engine, but it entails complexity far greater than that of todays engines. Another approach is to seek an intermediate bypass ratio that satis es both requirements in a xed cycle. Because the bypass ratio would be moderate, it becomes crucial how one uses the bypass stream to reduce noise. One con guration is the mixed- ow turbofan, currently used on all military engines, in which the bypass and core streams mix before exiting a common nozzle. The other option is the separate- (unmixed-) ow turbofan, which is very common on subsonic transports. The unmixed design allows shaping of the bypass exhaust so that the bypass stream substantiallyreducesMach wave emissionfrom the core stream. Previous work on the Mach wave elimination (MWE) technique showed signi cant gains in noise reduction by changing the shape of a dual-stream exhaust from coaxial to eccentric.5 Mach wave radiation is considered the principal source of mixing noise in supersonic jets.6 11 It could also play a strong role in noise emission from high-subsonicjets due to the growth-decay nature of instability waves, which creates a spectrum of phase speeds part of which is supersonic.12;13 In the MWE method, generation of Mach waves from a primary stream is suppressed by owing a secondary parallel stream adjacent to the primary stream so that all relative eddy motions become subsonic.14 Speci cally, MWE seeks to minimize the convective Mach numbers of turbulent eddies throughout the jet ow eld. This includes the end of the potential core, a region of vigorous mixing and strong noise generation. In a coaxial arrangement, application of the secondary ow reduces the growth rate of the shear layer between the primary and secondary streams, thus stretching the primary potential core. The end of the primary potentialcore can easily extend past the reach of the secondary ow, thus reducing the effectiveness of the technique. The eccentric arrangement has been shown to prevent signi cant elongation of the primary potential core.15 It also doubles the thickness and potential core length of the secondary ow in the downward direction,

Engine cycle assumptions Ef ciency 0.97 0.85 0.85 0.85 1.00a 0.90 0.97
b

Speci c heat ratio 1.40 1.40 1.40 1.37 1.35 1.33 Calculationb

With 5% total pressure loss.

From internal mixing calculation.

PAPAMOSCHOU AND DEBIASI

163 Engine characteristics at subsonic climb: y = 10,000 m and M1 = 0.8 B03-MIX 24 1400 55 85 51 0.3 4.5 1.04 0.71 1.9 832 B16-SEP 24 1400 63 163 55 1.6 2.2 1.43 0.62 1.78 819 1.40 425 B16-MIX 24 1400 60 156 54 1.6 2.2 1.40 0.61c 1.53 584

Table 2 Quantity OPR TIT,a K m P com , kg/s m P tot , kg/s T , kN BPR FPR Dfan ,b m TSFC, 1/h Mp U p , m/s Ms Us , m/s
a b

Engine characteristics at takeoff: y = 0 m and M1 = 0 B03-MIX 30 1800 125 160 126 0.30 5.0 1.04 0.67 1.55 770 B16-SEP 30 1800 120 303 162 1.60 2.6 1.43 0.50 1.20 640 1.20 430 B16-MIX 30 1800 115 290 159 1.60 2.6 1.40 0.49c 1.20 530

Table 3

Characteristic OPR TIT,a K m P com , kg/s m P tot , kg/s T , kN BPR FPR Dfan ,b m TSFC, 1/h Mp U p , m/s Ms Us , m/s
a b c

For turbine blade cooling, 30% compressor ow. M D 0:5 at fan face. c Does not account for mixer losses.

For turbine blade cooling, 25% compressor ow. M D 0:6 at fan face. Does not account for mixer losses.

Table 4 Characteristic OPR TIT,a K m P com , kg/s m P tot , kg/s T , kN BPR FPR Dfan ,b m TSFC, 1/h Mp U p , m/s Ms U s , m/s
a b c

Engine characteristics at supersonic cruise: y = 16,000 m and M1 = 1.6 B03-MIX 24 1600 55 72 30 0.3 4.5 1.04 1.00 2.10 890 B16-SEP 24 1600 53 137 30 1.6 2.2 1.43 0.97 1.80 820 1.95 610 B16-MIX 24 1600 51 131 30 1.6 2.2 1.40 0.94c 1.90 700

For turbine blade cooling, 25% compressor ow. M D 0:7 at fan face. Does not account for mixer losses.

Fig. 2

Engine con gurations.

mixed- ow engine. The suf xes COAX and ECC indicate a coaxial or eccentric exhaust, respectively, of the separate- ow engine. For the eccentric con guration, an additional numerical suf x indicates the azimuth angle at which noise was measured. Tables 2, 3, and 4 summarize engine characteristicsand thermodynamic performance at takeoff, subsonic climb, and supersonic cruise, respectively. Exhaust conditions are pressure matched for the primary stream and secondary streams. Because of their increased mass ow rate, the modi ed engines have 28% higher static thrust than the baseline engine. This is an important consequence of the thermodynamic cycle that needs to be dealt with as one increasesthe bypass ratio of supersonic engines. In other words, for the same cruise thrust, the higherbypass engine will be inherently overpowered on takeoff. Later we take advantage of the excess thrust to enhance takeoff performance and, thus, reduce the noise footprint.At subsonicclimb, the modi ed engines have 10% more thrust than the baseline engine, which translates into improved climb performance and faster reach of cruise altitude. The takeoff FPR of the modi ed engines is 2.6, a value that presently may require a two-stage fan but in the future may be achievablewith a single-stageaspirated fan.20 Table 2 shows that, at

takeoff, the TSFC of the modi ed engines is 25% lower than that of the baseline engine. At subsonicclimb (Table 3) the TSFC improvement is about 13%. At supersonic cruise (Table 4) the TSFCs of the modi ed engines are marginally lower than that of the baseline engine. As mentioned in the Introduction, increased bypass ratio at supersonic cruise yields only small bene ts in fuel consumption. The mixed- ow turbofan has slightly better fuel consumption than the separate- ow turbofan. This calculation, however, does not include mixer losses or the added weight of the internal mixer on overall engine performance. The velocity ratio of the separate- ow exhaust at cruise, Us = U p D 0:74, is very close to the ef ciency of energy transfer between the core and bypass ow (the product of turbine and fan ef ciencies, in this case, 0.76). This indicates that B16-SEP operates at optimal cruise conditions.21

Noise Measurement
Facilities

Noise testing was conducted in the University of California, Irvine, Jet AeroacousticsFacility.5 Single- and dual-streamjets with ow conditionsmatching those given by the cycle analysis (Table 2) were produced. The jets were composed of heliumair mixtures, which duplicate very accurately the uid mechanics and acoustics of hot jets.22 Jet nozzles were fabricated from epoxy resin using rapid-prototypingtechniques. Two primary (core) nozzles were designed with the method of characteristicsfor Mach numbers 1.5 and 1.2, matching approximately the takeoff exit Mach numbers of the baseline and modi ed engines, respectively. Both primary nozzles had the same exit inner diameter (14.8 mm), lip thickness (0.7 mm), and external shape. One secondary (bypass) nozzle formed a convergent duct in combinationwith the primary nozzle and terminated in a diameter of 21.8 mm. The pipe that fed the primary nozzle was

164

PAPAMOSCHOU AND DEBIASI

a)

Fig. 4

Experimental setup with set of polar angles covered.

b)

Fig. 5 Far- eld spectra in the direction of peak emission, scaled to equal thrust: baseline engine ( = 40 deg), mixed- ow derivative ( = 25 deg), and eccentric separate- ow derivative ( = 40 deg).

c) Fig. 3 Nozzle coordinates (inches) for a) B03-MIX, b) B16-SEPCOAX, and c) B16-SEP-ECC-0; nozzle for B16-MIX was the inner one alone of (b)/(c).

able to ex, enabling coaxial or eccentric secondary ow passages. The coordinates of the nozzles, are plotted in Fig. 3. For all of the nozzles, the contraction (before any supersonic expansion) was described by fth-order polynomials and was 4:1 for the core nozzle and 15:1 for the bypass nozzle. The jet Reynolds number was on the order of 0:5 106 . Noise measurements were conducted inside an anechoic chamber using a 1 -in. (3.2 mm) condenser microphone (Br uel and Kjr 8 4138) with frequency response of 140 kHz. The microphone was mounted on a pivot arm and traced a circular arc centered at the jet exit with radius of 71 core-jet diameters. Earlier experiments have determined that this distance is well inside the acoustic far eld.23 The polar angle ranged from 20 to 130 deg in intervalsof 5 deg for 20 50 deg and 10 deg for the rest. Figure 4 shows the overall setup and the range of polar angles covered.For the eccentricjet, azimuth angles D 0 and 45 deg were investigated.The sound spectra were corrected for the microphonefrequency response,free eld response, and atmospheric absorption. All spectra were referenced to

r = D p D 100. Comparison at equal thrust was done using geometric scaling.23 In our facility, repetition of an experiment under varying temperature and relative-humidity conditions (typically from 20 to 50%) yields spectra that differ by at most 0.5 dB. Comparison of our single-jetspectra with those from NASA large-scalejet facilities and with the Tam et al. similarity spectra,24 shows excellent agreement both in the spectral shape and in the value of overall sound pressure level (OASPL). 25
Spectra

Sound pressure level spectra are compared at equal thrust T D 50 N unless otherwise noted. A frequency range of great relevance to perceived noise level is 2575 kHz, which on a full-scale engine correspondsroughly to 5001500 Hz, that is, the scale factor is around 50. Below this range,the human ear becomes insensitiveto noise; above this range, sound gets attenuated very rapidly by atmospheric absorption. We show only spectra in the lower hemisphere of the acoustic eld. As has been shown in a previous study,5 noise emission of the eccentric case in the upper hemisphere matches that of the single (core) jet. The spectraof B03-MIX, B16-MIX, and B16-SEP-ECC-0 at their respectiveanglesof peak noise emission(aft quadrant)are compared in Fig. 5. The eccentric, separate- ow case has dramatically lower noise levels than the baseline case: about 8-dB reduction at low

PAPAMOSCHOU AND DEBIASI

165

Fig. 6 Far- eld spectra in the direction of peak emission ( = 40 deg for all spectra shown), scaled to equal thrust: baseline engine, core stream alone of separate- ow derivative, coaxial separate- ow derivative, and eccentric separate- ow derivative at azimuth angles of 0 and 45 deg.

Fig. 8 Far- eld spectra at = 90 deg: baseline engine, mixed- ow derivative, and separate- ow eccentric derivative.

Fig. 7 Far- eld spectra in the direction of peak emission ( = 40 deg for all spectra shown), not scaled to equal thrust: eccentric separate- ow engine and its components.

frequencies and 20-dB reduction at moderate-to-high frequencies. The mixed- ow case is much louderthan the separate- ow eccentric case, its spectrum exceeding that of B16-SEP-ECC-0 by 23 dB at f < 5 kHz, 8 dB at f 10 kHz, and 1012 dB for f > 20 kHz. Note that an actual mixed- ow exhaust would be noisier than that shown here due to exit nonuniformitiesand internal noise from mixing. The spectra of B03-MIX, B16-SEP-CORE (core stream alone), B16-SEP-COAX, B16-SEP-ECC-0, and B16-SEP-ECC-45 are compared in Fig. 6. The core stream of the modi ed engine is almost as loud as the exhaust of the baseline engine, except at low frequencies where it is about 3 dB quieter. (The drop in OASPL is about 2.5 dB.) The reduction in exhaust velocity from 770 to 640 m/s did not bring appreciable noise reduction. This is consistent with the sound intensity following a U 3 power law, rather than U 8 at lower speeds.26 The separate- ow, coaxial exhaust is on average 68 dB louder than the eccentric exhaust. The spectra of the eccentric case at azimuth angles D 0 and 45 deg practically coincide, indicating that the eccentric con guration has good sideline bene t. Figure 7 shows the spectra of B16-SEP-CORE, B16-SEP-BYP (eccentric bypass stream alone), and B16-SEP-ECC-0. The spectra

are presented without equal-thrustscaling to get a clearer picture of the modi cation of acoustic sources. Because of its lower velocity, the fan stream is much quieter than the core stream. At frequencies above 20 kHz, the spectra of B16-SEP-ECC-0 and B16-SEP-BYP coincide, that is, the combined ow emits the same noise as the bypass stream alone. This indicates that the Mach wave emission from the core stream has been suppressed so well that the core stream is practically silent compared to the bypass stream and that the bypass stream now constitutesthe noise oor. The effect of forward ight should lower this oor because the relative velocity of the bypass stream will reduce from 430 to about 310 m/s at climb speed. Note that the B16-SEP-BYP experiment does not represent accurately the uid mechanics and noise emission of the bypass stream in a dual-stream jet. Instead, it attempts to establish the lower bound of noise emitted by the dual-stream jet. Experiments not covered in this paper show that the noise of B16-SEP-BYP is practically identical to that of a round or annular jet at the same velocity and Mach number. Figure 8 shows the spectra in the lateral ( D 90 deg) direction. The modi ed engines are 68 dB quieter than the baseline engine. The spectrum of the eccentric exhaust presents small bumps of 23 dB magnitude relative to the fully mixed exhaust. These bumps could be the result of broadbandshock noise that occasionallyarises in dual-streamjets,5 althoughfurtherinvestigationis neededinto this issue.
OASPL and Skewness

Figure 9 shows the directivity of OASPL for the baseline engine and its three derivatives. The bene t of the separate- ow, eccentric exhaust is again evident. The maximum level of OASPL is reduced by 4.3 dB in B16-MIX, 5.6 dB in B16-SEP-COAX, and 10.5 dB in B16-SEP-ECC-0. In the lateral direction, for 90 deg, the eccentric exhaust is 23 dB louder than the fully mixed exhaust. This increase in noise was noted in the spectrum of Fig. 8. To gain further insight into the modi cation of the noise sources, we examine the acoustic signal in the time domain. Figure 10 shows the time traces of B03-MIX and B16-SEP-ECC-0 at the angle of peak emission. The signal of B03-MIX is highly skewed on the positive side and exhibits strong, irregularly spaced positive spikes. This phenomenon is associated with nonlinear formation of Mach waves in the vicinityof the sourceand, in a full-scaleengine, is heard as crackle.27 The signal of B16-SEP-ECC-0, on the other hand, is nearly symmetric without any spikes. The skewness of the acoustic signal allows quanti cation of this feature of noise, which cannot

166

PAPAMOSCHOU AND DEBIASI

Fig. 9 Directivity of OSPL for the baseline engine and its derivatives.

Fig. 11 tives.

Directivity of skewness for the baseline engine and its deriva-

Fig. 10 Microphone time traces in the direction of peak emission for the baseline engine and the eccentric separate- ow derivative.

Fig. 12 Directivity of skewness for the eccentric exhaust and its components.

be captured by spectral analysis. Early work by Ffowcs-Williams et al.27 has shown that high skewness is associated primarily with Mach wave emission. This was further evidenced in recent experiments by Papamoschou and Debiasi.5 Ffowcs-Williams et al.27 also noted that, in certain cases, high skewness was observed in the forward arc, a possible result of shock-inducednoise in an imperfectly expanded jet. Figure 11 shows the directivity of skewness for the baseline and derivative engines. The skewness of the baseline case is very high, about 1.3, at the angle of peak sound emission. The skewnesses of the mixed- ow and coaxial-exhaust cases are substantial, exceeding the threshold of 0.4. As shown by Ffowcs-Williams et al.,27 this indicates that both of these cases emit signi cant Mach wave radiation. The skewness of B16-SEP-ECC-0 is very low in the direction of peak sound emission, consistentwith eliminationof Mach waves. However, it gradually rises with increasing emission angle, reaching a peak of 0.4 at D 100 deg. The reason for this increase is not yet understood; it may be related to broadband shock noise, as was noted in the discussion of Fig. 8. It is also instructive to compare the streams of B16-SEP-ECC separately and in combination. As shown in Fig. 12, the skewness of the core stream alone is very large, reaching 1.05 at D 45 deg. The skewnesses of the

bypass stream alone and of the combined ow are both very low in the direction of peak sound emission, indicating that neither ow emits strong Mach waves. This is further evidence that Mach wave emission from the core stream has been substantially prevented by the eccentric bypass ow.

Perceived Noise Level

The regulatoryde nition of a quiet airplane is deceptivelysimple. To be able to operate, an airplane must not exceed certain noise thresholds at three set locations: takeoff, sideline, and approach. Noise is quanti ed in terms of the effective perceived noise level (EPNL), a metric that incorporates human annoyance to sound and its duration.28 EPNL thresholds are based on the con guration and weight of the airplane and stem from a mix of scienti c and political considerations.As the ultimate goal of this research is development of quieter airplanes, it is essential to obtain estimates of perceived full-scale noise. Here we attempt to process our spectra into EPNL for obtaining an assessmentof perceivednoise reduction.We feel that this a meaningful exercise because it includes crucial effects that are typically left out of academic studies of jet noise: distance from the source, atmospheric absorption, and human perception.The absolute levels of EPNL are not accurate because the effect of forward ight on jet

PAPAMOSCHOU AND DEBIASI

167

Fig. 13

Takeoff ight path with key geometric parameters.

acoustics is not present in the experiments. Other sources of noise, such as fan/compressor noise and airframe noise, are obviously not included in our assessment. We calculate noise recorded from the takeoff monitor for a full-power takeoff. Future studies will address takeoff with power cutback and noise recorded by the sideline and approach monitors.
Flight Path

The rst step in assessing perceived noise is de nition of the takeoff ight path and attitude of the engines relative to the ight path. The airplanesare those de ned in the Engine Con gurations section, that is, twin engine with thrust given by the speci cations of Table 2. All aircraft must have the same weight because they share the same cruise thrust. The ight path of the baseline aircraft comprises a takeoff roll xLO D 1800 m followed by a straight climb at angle D 15 deg. The lift coef cient at climb is 0.6, which for a delta-wing aircraft corresponds to an angle of attack D 12 deg (Ref. 29). The engine exhaust is assumed to be inclined at the angle of attack. The modi ed airplanes have 28% more takeoff thrust than the baseline case. The excess takeoff thrust can be used in two fashions: faster velocity at the same climb angle or higher climb angle at the same velocity. Here we consider the latter option. Also, takeoff roll distance is reduced by roughly the amount of thrust gain. When well-known relationsfor takeoff and climb performance30 are used, the takeoff roll of the modi ed airplanes is x LO D 1400 m and their climb angle is D 23 deg. Figure 13 shows the generic ight path with key variables. The takeoff ight speed of all airplanes is 110 m/s ( M1 D 0:32). The Cartesian position ( x , y ) of the airplane is calculated at 0.5-s intervals from the time of liftoff. For each aircraft location, its polar coordinates (r , ) relative to the ight path and seen by the takeoff monitor are calculated. Here we distinguish between the apparent (r 0 , 0 ) and true (r , ) locationsof the airplane with regard to sound emission. The apparentlocation is the actual location of the airplane. The true location is the one from which sound reached the observer. It is easily shown that the true position is at a distance M1 r behind the apparent position along the ight path. From the geometry of Fig. 13, the apparent coordinates are r0 D

2) The spectra are extrapolated to frequencies higher than those resolved in the experiment (140 kHz) using a decay slope of 30 dB/decade. This is done to resolve the audible spectrum for a full-scale engine. The perceived noise level (PNL) results are very insensitive on the assumed slope. Changing the slope from 0 dB/decade to 1 dB/decade results in a 0.5 dB difference in EPNL. 3) The spectra are scaled up to engine p size by dividing the laboratory frequencies by the scale factor .Teng =Texp /. The full-scale enginediameteris the experimentaldiametermultipliedby this scale factor. 4) The spectra are Doppler shifted to account for the motion of the aircraft. The relations of McGowan and Larson32 are used. In those relations, the value of the convective Mach number M c is obtained from the empirical relations of Murakami and Papamoschou.33 5) For each observation time t , the scaled-up spectrum corresponding to .t / is obtained. This step requires interpolation between spectra and, for angles outside the range covered in the experiment, moderate extrapolation.To enhance the accuracy of interpolation or extrapolation, the spectra are smoothed to remove their wiggles. 6) For each t , the correspondingscaled-up spectrum is corrected for distance and atmospheric absorption. The distance correction is

20 log10

.r = D p /eng .r = D p /exp

The absorption correction is applied for ambient temperature 29 C and relative humidity 70% (conditionsof least absorption)using the relations of Bass et al.31 7) For each t , the correspondingscaled-up, corrected spectrum is discretized into one-third octave bands. The PNL is then computed according to part 36 of the Federal Aviation Regulations (FAR).28 8) The preceding step gives the time history of PNL, PNL( t ). From it, the maximum level of PNL (PNLM) is determined. The duration of PNL exceeding PNLM-10 dB is calculated and the correspondingdurationcorrectionis computed according to FAR 36. The EPNL equals PNLM plus the duration correction. The duration correction can be very substantial; hence, the importance of assessing PNL(t ). Our estimate of EPNL does not include the tone correction, a penalty for excessively protrusive tones in the one-third octave spectrum.
Results

y 2 C . x xTOM /2

0 D =2 arctan[. x x TOM /= y ] and the true coordinates are obtained from r D r0

2 1 M1

q
2 sin2 0 1 M1

M1 cos 0 C

sin D .r 0 = r / sin 0 The polar angle of the exhaust observed by the takeoff monitor is D When these relationsare used, the true distance r and emission angle are obtained as functions of time observed by the takeoff monitor.
Data Processing

The steps for processing the laboratory narrowband spectra into perceived noise level follow. 1) The spectra are corrected to zero absorption using the relations of Bass et al.31

We will compare PNL time histories, and resulting EPNLs, of the baseline and modi ed airplanes in two ways. One is a realistic comparison that accounts for the better takeoff performance of the modi ed-engine aircraft. The other is an academic comparison in which the ight paths are identical and the engines produce the same thrust. For the latter comparison, we scale down the modi ed engines so that their thrust equals that of the baseline engine. Obviously the scaled-down engines do not meet the Mach 1.6 cruise requirement; hence, the academic nature of this comparison. Figure 14 shows a comparison of PNL time histories of aircraft powered by the B03-MIX, B16-MIX, B16-SEP-COAX, and B16SEP-ECC engines. The ight paths are identical (baseline case), and all engines produce the same thrust, 126 kN. The superiority of the separate- ow, eccentric exhaust is evident. In terms of PNLM, it is 13 dB quieter than the baseline, whereas the fully mixed and the coaxial exhausts are 5 and 6 dB quieter than the baseline, respectively. EPNL is as follows: 108.5 dB for B03MIX, 104.5 dB for B16-MIX, 103.0 dB for B16-SEP-COAX, and 98.0 dB for B16-SEP-ECC. In other words, the eccentric separateow exhaust gives a 10.5 dB bene t in EPNL, whereas the mixedow and annular exhausts produce only 4.0 and 5.5 dB bene ts, respectively. Figure 15 shows a more realistic comparison that accountsfor the higher thrust of the modi ed engines and resultingimproved takeoff performance. The larger size of the modi ed engines causes a moderate increase in noise. This is more than counteracted, though, by

168

PAPAMOSCHOU AND DEBIASI

Fig. 14 Time history of yover PNL for identical thrusts and ight paths.

Fig. 16 Variation of PNL with emission polar angle observed by the takeoff monitor; curves correspond to time histories of Fig. 15.

Conclusions
The possibility of a quiet, xed-cycle supersonic turbofan engine without mechanical silencers has been explored. The assessment is preliminary and comprises thermodynamic cycle analysis, subscale acoustic measurements, and estimates of EPNL. The approach was to take a typicalstate-of-the-artmilitary turbofan,increase its bypass ratio from 0.3 to 1.6, and investigatethe acousticand thermodynamic performance of several exhaust con gurations. The leading con guration has a separate- ow, eccentric exhaust, which is shown to prevent strong Mach wave radiation toward the ground. An equalthrust, equal- ight-path comparison shows that the engine with eccentric exhaust it is 10.5 EPN dB quieter than the baseline engine and 6.5 EPN dB quieter than the mixed- ow enginewith BPR D 1.6. When the better takeoff performance of an airplane powered by the modi ed engines is accounted for, the eccentric exhaust gives a 14-EPN dB bene t relative to the baseline engine. Speci c fuel consumption at Mach 1.6 supersonic cruise is about 3% better than baseline. The modi ed engines have a fan diameter 37% larger than the baseline engine. This will require careful integration of the engine with the airframeto avoid substantialincreasesin wave drag due to the increased nacelle size. In the case of the eccentric exhaust, internal losses due to the nonconcentricarrangementmust be carefully assessed. The study does not include the effect of forward ight on acoustics because the University of California, Irvine, laboratory is not equipped with a tertiary stream. Forward ight is expected to enhance the noise bene t of the eccentric exhaust. Examination of the spectra shows that, except at very low frequencies, noise from the eccentric exhaust is dominated by that of the secondary (bypass) stream (Fig. 7). Forward speed at takeoff will reduce the relative velocity of the bypass stream from 430 to 310 m/s, a 28% drop. Using the simple argument that, at this low velocity, sound intensity is proportional to relative velocity raised to the eighth power,26 we expect a noise reduction of 11.5 dB. In contrast, the relative velocity of the mixed- ow exhaust will reduce by only 23% (from 530 to 410 m/s), leading to a noise reduction of 9 dB. These preliminary arguments suggest that the noise bene t of the separate- ow, eccentric exhaust relative to the mixed- ow exhaust will widen with increasing ight Mach number. Moreover, classical relations for shear layer growth rate15 indicate that forward speed will stretch the potential core of the bypass stream further than that of the core stream, yielding better coverage of the core stream by the bypass stream. It is hoped that large-scale experiments in government or industry facilities will address these issues.

Fig. 15 Time history of yover PNL for actual (dissimilar) thrusts and resulting different ight paths.

sound attenuationdue to the higher altitude of the aircraft. The PNL histories of B03-MIX (baseline ight path), B16-MIX (enhanced ight path) and B16-SEP-ECC-0 (enhanced ight path) are plotted in Fig. 15. Comparing Figs. 14 and 15, we note that the enhanced takeoff pro le produces a 33.5 dB attenuation in the sound from the modi ed aircraft, a result of their greater distance from the noise monitor. EPNL is as follows: 108.5 dB for B03-MIX, 101.5 dB for B16-MIX, and 94.5 dB for B16-SEP-ECC-0. The aircraft powered by the B16-SEP-ECC engine is, thus, 14-EPNdB quieter than that powered by the baseline B03-MIX engine. It is also instructiveto plot PNL( t ) vs .t / to evaluate which range of polar angles is signi cant to EPNL for the ow conditions considered here. This is done in Fig. 16. First, it is observed that the angles of peak PNL are very close to the angles of peak sound emission as measured in the laboratory. The aft quadrant ( < 90 deg) is clearly very critical to EPNL because it contains all (in the case of the baseline engine) or most (in the case of the derivative engines) of the noise exceeding PNLM-10. The forward quadrant has no impact on EPNL for the baseline engine and has limited impact for the derivative engines.

PAPAMOSCHOU AND DEBIASI

169

Acknowledgments
The support by NASA John H. Glenn Research Center is gratefully acknowledged (Grant NAG-3-2345 monitored by Khairul B. Zaman). We also thank Erin Abbey for her work on rapid prototyping of the jet nozzles and for running the experiments presented here.

References
M. J. T., Aircraft Noise, 1st ed., Cambridge Univ. Press, Cambridge, England, U.K., 1989, pp. 120134. 2 Nagamatsu, H. T., Sheer, R. E., and Gill, M. S., Characteristics of Multitude Multishroud Supersonic Jet Noise Suppressor, AIAA Journal, Vol. 10, No. 3, 1972, pp. 307313. 3 Tillman, T. G., Paterson, R. W., and Presz, W. M., Supersonic Nozzle Mixer Ejector, Journal of Propulsion and Power, Vol. 8, No. 2, 1992, pp. 513519. 4 Plencner, R. M., Engine Technology Challenges for the High-Speed Civil Transport Plane, AIAA Paper 98-2505, June 1998. 5 Papamoschou, D., and Debiasi, M., Directional Suppression of Noise from a High-Speed Jet, AIAA Journal, Vol. 39, No. 3, 2001, pp. 380 387. 6 Troutt, T. R., and McLaughlin, D. K., Experiments on the Flow and Acoustic Properties of a Moderate Reynolds Number Supersonic Jet, Journal of Fluid Mechanics, Vol. 116, March 1982, pp. 123156. 7 Tam, C. K. W., Chen, P., and Seiner, J. M., Relationship Between Instability Waves and Noise of High-Speed Jets, AIAA Journal, Vol. 30, No. 7, 1992, pp. 17471752. 8 Tam, C. K. W., and Chen, P., Turbulent Mixing Noise from Supersonic Jets, AIAA Journal, Vol. 32, No. 9, 1994, pp. 17741780. 9 Tam, C. K. W., Supersonic Jet Noise, Annual Review of Fluid Mechanics, Vol. 27, 1995, pp. 1743. 10 Seiner, J. M., Bhat, T. R. S., and Ponton, M. K., Mach Wave Emission from a High-Temperature Supersonic Jet, AIAA Journal, Vol. 32, No. 12, 1994, pp. 23452350. 11 Mitchell, B. E., Lele, S. K., and Moin, P., Direct Computation of Mach Wave Radiation in an Axisymmetric Supersonic Jet, AIAA Journal, Vol. 35, No. 10, 1994, pp. 15741580. 12 Crighton, D. G., and Huerre, P., Shear-Layer Pressure Fluctuations and Superdirective Acoustic Sources, Journal of Fluid Mechanics, Vol. 220, 1990, pp. 355368. 13 Avital, E. J., Sandham, N. D., and Luo, K. H., Mach Wave Radiation in Mixing Layers. Part I: Analysis of the Sound Field, Theoretical and Computational Fluid Dynamics, Vol. 12, No. 2, 1998, pp. 7390. 14 Papamoschou, D., Mach Wave Elimination from Supersonic Jets, AIAA Journal, Vol. 35, No. 10, 1997, pp. 16041611. 15 Murakami, E., and Papamoschou, D., Mean Flow Development in Dual-Stream Compressible Jets, AIAA Journal, Vol. 40, No. 6, 2002, pp. 11311138.
1 Smith,

16 Mair, W. A., and Birdsall, D. L., Aircraft Performance, 1st ed., Aerospace Series 5, Cambridge Univ. Press, Cambridge, England, U.K., 1992, pp. 260, 261. 17 Airworthiness Standards: Transport Category Aircraft: Climb: OneEngine-Inoperative, Federal Aviation Regulations Pt. 25.121, Federal Aviation Administration, Jan. 2002. 18 Debiasi, M., and Papamoschou, D., Cycle Analysis for Quieter Supersonic Turbofan Engines, AIAA Paper 2001-3749, July 2001. 19 Horlock, J. H., Watson, D. T., and Jones, T. V., Limitations on Gas Turbine Performance Imposed by Large Turbine Cooling Flows, Journal of Engineering for Gas Turbines and Power, Vol. 123, No. 3, 2001, pp. 487 494. 20 Lord, W. K., MacMartin, D. G., and Tillman, T. G., Flow Control Opportunities in Gas Turbine Engines, AIAA Paper 2000-2234,June 2000. 21 Guha, A., Optimum Fan Pressure Ratio for Bypass Engines with Separate or Mixed Flow Exhaust Streams, Journal of Propulsion and Power, Vol. 17, No. 5, 2001, pp. 11171122. 22 Kinzie, K. W., and McLaughlin, D. K., Measurements of Supersonic Helium/Air Mixture Jets, AIAA Journal, Vol. 37, No. 11, 1999, pp. 1363 1369. 23 Papamoschou, D., and Debiasi, M., Noise Measurements in Supersonic Jets Treated with the Mach Wave Elimination Method, AIAA Journal, Vol. 37, No. 2, 1999, pp. 154160. 24 Tam, C. K. W., Golebiowski, M., and Seiner, J., On the Two Components of Turbulent Mixing Noise from Supersonic Jets, AIAA Paper 96-1716, 1996. 25 Tam, C. K. W., Review of Noise Data and Recent Advances in Jet Noise Theory, Proceedings of the Jet Noise Workshop, NASA CP-2001-211152, Cleveland, Nov. 2002, pp. 143183. 26 Dowling,A. P., and Ffowcs Williams, J. E., Soundand Sources of Sound, Halsted, New York, 1983, pp. 159163. 27 Ffowcs Williams, J. E., Simson, J., and Virchis, V. J., Crackle: an Annoying Component of Jet Noise, Journal of Fluid Mechanics, Vol. 71, Pt. 2, 1975, pp. 251271. 28 Noise Standards: Aircraft Type and Airworthiness Certi ciation, Federal Aviation Regulations Pt. 36, Federal Aviation Administration, Jan. 2001. 29 Bertin, J. J., and Smith, M. L., Aerodynamics for Engineers, 3rd ed., PrenticeHall, Upper Saddle River, NJ, 1998, pp. 319321. 30 Shevell, R. S., Fundamentals of Flight, PrenticeHall, Upper Saddle River, NJ, 1989, pp. 265, 293. 31 Bass, H. E., Sutherland, L. C., Zuckerwar, A. J., Blackstock, D. T., and Hester, D. M., Atmospheric Absorption of Sound: Further Developments, Journal of the Acoustical Society of America, Vol. 97, No. 1, 1995, pp. 680 683. 32 McGowan, R. S., and Larson, R. S., Relationship Between Static, Flightand Simulated FlightJet Noise Measurements, AIAA Journal, Vol. 22, No. 4, 1984, pp. 460464. 33 Murakami, E., and Papamoschou, D., Eddy Convection in Supersonic Coaxial Jets, AIAA Journal, Vol. 38, No. 4, 2000, pp. 628635.