AIAAJournai A Publication of the American

VOLUME 1

Institute of Aeronautics and Astronautics

JULY 1963

NUMBER 7

~\

Jet NoIse
M. J. LIGHTRILL Royal Aircraft Establishment, Farnborough, England

1. Introduction
i f

i i

i ,

.UI honored that this great association, devoted to the sciences that permit vehicles heavier than air to follow controlled paths away from the earth's surface, has asked me to visit the countr)' of the Wright Brothers, who first achieved this aim, and to speak in their memory. i am for a second time so speaking, having given the Wilbur Wright :\Iemorial Lecture in 1960, on the subject of ":\Iathematics and Aeronautics," to the Royal Aeronautical Society.! i spoke then, \'o'ithenthusiasm, of the many-sided abilities revealed by the documents of the Wright Brothers' career, a man)'-sidedness that made them, in a \Va)',typical of the greatest men of this cE'ntury. :Morerecent1)', in Michigan, visiting Henry Ford's admirable reconstruction of the Wright Cycle Shop, with it:; u,-;efuiwind tunnel in the smaIl back room, i was reminded how their career contains alsa, for this century, a lesson-that individua! initiative in modest surroundings can spark off a transformatian of the human condition. This lecture is also related to yet another that i gave in my own country, the Bakerian Lecture2 of 1961, to the Royal Society of London, on the subject of "Sound Generated Aerodj'Ilamically." In that lecture, to an audiencerepresenting almost all the sciences, 1 attempted to sketch in its broadest features a new science of the middle of this century-that of sound radiation fields that are by-products of airflows. Having given such a broad outline of that new science to a general scientific audience, i am glad to be able to deseribe in more detai! the part of it which i have explored most myself, namely, its application to the problem of the noise of jets. This topic alone is a very complexone and deserves,in iia 0"'11 right, the extended treatment that can be given appropriately to an audience concerned professionallywith the cacophon~-

that is, to a greater or lesserextent, inseparable from turbojet and rocket propulsion. The noiseof a jet can be regarded as a separn.teentity, e\'en whennoise emergesalsa from the exhaust nozzle, arising from conditions in the nozzle or upstream of it. For, whate\'er those conditions may be, a new, very intense turbulence is created in the shear myer wherethe the jet fluid, after leaving the nozzle,exchanges momentum with the atmasphere. The noise that this generates can be thought of as, to ii.large extent, addith'e to combustion noise or turbomn.chinery naise. Furthermore, to a good approximation, it is uninfluenced by the presence of the nozzle and can be visualized DOS souna radiated into free space b)' a limited extent of turbulent gas ftow. lt is simply the radiated sound,representing energy actuall)' extracted from the jet and propagated away through the atmosphere, that I shall discuss. My concern, in other words. will be with the atmasphere pressure ftuctuations in the "radiation field" or "far field," where those pressure ftuctuations fall offin proportion to the inverse first powerof the distance. Within a particu1ar narrow band of 'frequencies,J;h.e. far field of an}' noi~p~Ol1r,.p t.hpreKionwhose ilistanci' rro"l i" Lt substantiallv e""pprl" onp "',,,'plenii:th.2 Bv contrast. the near field "ithin a wavelen th or so of the 'ncludes not nlv outwardly propaii:atinii:WM'es hnt alsa ]01,,,1rp. 'ciprocating motions and pressure fluctuations, such as mav be inilnced ilirectly bv ftuctuatiniz:vortex movements in the ~urbulent ii:asflow. This near sound field is of concern to the designer of struetures to be placed near jet orinces, but the body of knowledge whieh has been sought after most by aireraft and roeket engineers comprises the acoustic power ragiated br a jet, its freQuencvspectrum. and its ilirp~tionaI distribution.

Dr. )-lichael James Lighthill is the Director of the Royal Aircraft Establishment, the premier establishment of its kind in Europe. Born in January 1924, Dr. Lighthill was educated at Winchester College and at Trinity College, Cambridge University. He joined the Aerodynamics Diviiiion of the National Physical Laboratory in 1943. From 1946 to 1950, he was Senior Lecturer in l\ilathematics at the University of Manchester, where, in 1950, he was named Beyer Professor of Mathematics, a position he held until hiii present appointment. E1ected II. Fellowof Trinity College, Cambridge Unh'ersity, in 1945, Dr. Lighthill also waii honored with a Fellowiihip in the Royal Society in 1953. The Royal Aeronautical Society awarded him its Bronze :\ledal for work leading to advances in AerOOynamics in 1955, "for his contributions to theoretical high speed aerodynamics which recently had included outstanding work on the source of noise, from jets particularly." Dr. Lighthill was made a Foreign Member af the American Academy of Arts and Sciences in 1958. He was elected Fellow of the Ro~'al Aeronautical Society in 1961 and is an honorar~' Fellow Member of AIAA.

Received April25, 1963. In essence, this is the Wright Brothers Lecture presented at the 31st Annual ~Ieeting of the Institute of Aerosp~ce Sciences, January 21, 1963. The author is deeply grateful to P. O. A. L. Davies, J. E. Ffowcs Williams, and Alan PoweJ[ fo.r their kindness in putting unpublished material at his disposal and drawing his attention to numerous published reports in connection '\10' .s preparations for this lecture. fo11ow8rom integrating the meiisured fluxes of meo.n-fiowand turbulent-flow energy. f
.

i.)
1

.-

150i

". .0
I., !":,,,

M. J. LIGHTHILL

\ i

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

ALU. JOURKAL

JULY 191

includingthe much higher speeds of rockets exhausts~-io

the aco.ustic efficienc:rBattens ofi to a maximum betwe~n 5 and 7 times 10-3. These are still "alues so law in relation to viscous sources of dissipation that the efiect of acoustic ener~'loss on ~urb~lence~evelopment is probably negligible. Admittedly, with mcrea.sing}.!ach number, compressibilit~"

Fig. 1 SubsoDlc Jet .,-

. .

".;,..c. ''-\.t:~ e1Iectsare found to alter the turbulence lowermgits intensit" <:..}o~ :-,..~...,... ~nd the con~equeIitrate f sprescl oJ th~ turbuIent shear Isyir (1..v ~ \q!.: ~compsre Figs. 1 and 3). Biit the radiation of scoustic energy ~ ~ ()'\~ \ ~icr-i.\:::-~ 18too. small to acco~t for this efiect, which perhaps should be 'i., '\r. -:iit-cx'"'-n""" L ~ .y c,;$ a.ssociated,rather, with energy redistributions within the near .. \ ~ ,'go dfi Id \( ~ O ~4:.'" 4.-:.(\ ' soun e . . \ '> 't" ~ ,."" ~~~.(Zi We have, th~, s. continuous.de~"elopment acous~ic in Now, i propose to regsrd this acoustic power ont.put. R power output of Je~,. fro~ subs~>nic .through the undI5Jets by- roduc f the main .et Bow not a henomenon thst has m~grated supersonicJe~ (mcludmg turboJet engine exhausts)

~~elfany markesl.

"iew is wellSbstantiated for subsonic jets ig. ,which radiate, at most, 10-4 of their power as souna, mostly from a region within 4 or 5 diam of the orifice,namely, the intensely turbulent shear la~'er between the central jet core and the atmosphere. Yet YlSCOUS damping of the turbulence in that short lengthof jet e~-iractsalready over a quarter ofthe normal jet power,. so that e\ident1y the additional damping of less han 0.0001 by sound radiation is negligible by comparison. A-dmittedly,c~ jP.t.c; ver" lowRe"nolds nUm1)ers, at which !ileonh' maridiialh"unstable. showa sensith"einteraction with ui " soundi; 2 ut' ts the much hi her Reynolds :iumbers chara.cteristic of enidiieermii:app cations are fullv
tQrhiili:nt.indp.pp.ndently of an)" acoustic stimulation.

"e o men .

18

(such as rocket exha.usts),and all of these it remains a minor by-pro~uct of the powe! of the Jet. )ly main purpose in this lecture is to show ho.w.one ~heoreticalapproach, developedrecent1yby F~~,,;~s ~illia.ms~n the basis o.ften-year-old work ?f. ~y ~wn!' giv;s a g?od understanding not only of the initial rise m aco,!stic efficienc)'with jet speed, but also of its B';1bse9uent~ev~lin~ofi and of the associated changes in directional distribution and frequency spectrum.
2. Parts of General Aerodynamic Noise, Theory Relevant to Jet Xoise

to Jetsof speedsmanytimesthe atmosphericspeedof sound

ii in the :in' pl:ice(

:E'9rsunersoniciets. this distinction is somewhat less clear:i.it. since jets can be generated in the laboratory (Fig. 2) at "espectableReynolds numbers (though usuallr less than 108) :hat . edal kind of ins . 't" f acoustic origin. ?owell first described and explained this usually asymme ric lisintegration (see the Bakerian Lecture,2 Sec. 4.5 and ref~rencesthere giyen, together with arecent major study by )ayies and Oldfield3)in terms of a mechanism involving the Jattem of shock ~'a"es in a supersonic jet. Whfm an eddy J" shock waves an acoustic UISeis tional ea' in t e ups ream direciw:i. vorable conditions a" on a.ssin t e ,rifice.e:eneratea new eddy that can extract sufficientener~ it "~Q"n""he shock, to rega.inthe am~lit

The theory begins2. 12 (as a well-beha,"ed theory should!) with the fundamental equations of motion, e>...pressed terms in of the ma.ss density p and the momentum density PV" where Vi sta.nds for the velocity vector (Vi,t'2,V3).The equations are C>p +
c>t

i- 1

C>(pv,) C>x,

(1) (2)

C>(pv,)

c>t

+ j-l

C>(PV;t'i

C>Xi

+ P'j) = O

PhY$ic: llSil turb :;:imcftuc cal s/a/ioi plicd st re atmosph, he'suitabl hy-produ nificantl~. fraction i !;eDeratio lcnt appH separatel: On elir. and (i), (

. " an abnormall hi h uanti 'o acoustic ower sents direct convection of the momentum component Pv,by .ith a spectral pea" at a certain cha.racteristic frequency. th~ velocity component Vi, whereas Pii is the stress between .owever, any substantial irregularities created at the orificeadJ~cent el~ments of gas, which equally transfers momentum. ;ere found by Powell to reduce the amplitude gain in the :rhis stress is the sum of a.pure pressure term P~ii (where ~ii hear la~'ersufficientlyto destroy the e1Iectand to restore the ').~ 1 w:heni = and othe",ise is zero) and a viscous stress; arrow sprea.d characteristic of most practical superson~~in an ideal ~as, it represents simply cOD\'ectionof momentum ~ts.. .. .. .. . L- ~~.o-~ by the motion of.the molec~es relath'e to the gas velocity v,. These inte rated supersonicJets (Fig. 3), with ",hicn No~, one special.approximate form for momentum translone i shaH be concerne in tim ecture, a\"e acous lc e _ port yi?lds the classical equations of acoustics; this is a pure ciencies (that is, acoustic power output di"ided by jet i~otr~picpressure, the variations in which bear to the variaower) rising rapidly from about 10-4 for jet speeds equal tion in ?ensity a constant ratio CliI2the square of the at( ) the atmospheric speed of sound to 2 or 3 times 10-3for jets mosphenc sound speed). if this "acoustic" approximation to f t~ice that speed (Fig. 4).i-~ At higher speeds, ho\\"ever the momentum transport is made, Eq. (2) takes the simple , form

ude

. .

nd so mainta.in cycle. TheJet the

and they relate rate of change of mass or momentum in a small volume to the total mass transport or momentum transport out of the volum~ (PViappears twice because it is both the mass transport density and the momentum density).

The I?omentumtransportappearsin twoparts:

PV'Vi repre-

C>(pv,)/C>t

CliI2(C>P/C>Xi)

(3)

so that, from (1), the density p satisfies the classical \\"ave equation
c>2p

C>t2

- - i: _
3
CliI2

C>2p

C>Xi2

=O

(4)

However, without making an)" approximation, one can divide the momentum transport into two parts: first, this Hacoustic" approximation, CliI2p~ii, and, secondly, the remainderi which can be written as
Fig. 2 Disintegrated 8upersonic jet (5)

Xo\\", be( the soune mospheri continu01 Tii per u: quadrup( (Fig. 5) equal ani acting OJ: and so = l'qui"aler quadrup( dina)" qi poles; F amplitlid Xo\\", portant, their soll .though t composiJ: the dista of distan radiation square. radiation complete
times

if t

the radia

thequad
is propor .L _ i-

,.
-LU m.') ci to i:;tic ,ble. ility sity iyer rgy d be ciear istic iclisFig. 3 UndiBintegrated 8uper80nic jet
L(f'

Acoustie Power Output Jet power

ACCllsticEfficicncy"

Fig. " Acoustic eJl'iciency oC jets (excluding disintegrated 8upersonic

jets)

Icf' -I
LO LO" Lif' 0'3

H O.,

ists) !Und
iinor this

i
,

-.
i

Jet V~ Atmosphct1c: 2 3'"

.,.

Spccd

1 i

i
f

ii in the momentum equation the terms involving this part :ire placedon the right-hand side, one obtains these equations:
()p
()t

: the
)f its

d re.york

+ i-I
ao2 ()p

()(PVi)

()Xi

=O
~ '"

(6) (7)

.
t

()(PVi) +
()t

()Xi

- " /.

:s in

th~ quadrupole, small compared ""ith an acoustic ,,'aY.rlength js far greater, and it is this property of quadrupo~e radiation= that local fluctuations are so much in exeess of fluctuations iD .tion field-that is the fundamental explanation o ustic efticiencv of jets. The actual solution of tq. (8) can be written as ao2(p

Po) =

Jory

[Til] LL ()xi ()Xif v -dr i-ij-l 41M'

()2

(9)

i uld!)
.erms re Vi re (1) (2)
,

where V stands for the region of turbulent flow, and r represents the distance from a general point in V to the point (Xi,%2'X,) where ao2(p Po),which is approximately the pres-

sure fluctuation, is being mea~ured. T-he brackets around Tif

ean that it is evaIuated at the retarded time t

that

in a

(8)

ntum e it is .sity). repre)fJiby tween citum.

e Oil

~athad to mstantthe quadrupoletravelin~ at the speed of sound t e leave when a waye to reac the Doini of observ~ ,tion at the current time t. The integral in (9) represents what the solution would be if the right-hand side of Eq. (8) for the density were simply Ti; itself; and, from this, it evident1y followsthat the full Eq. (8) has the solution (9). i have said that V stands for the region of turbulent flow; admittedly, it would be more exact to take it as the ",hole of space, but outside the jet both terms in Til are smaIl. (pvt1/j alls ofi like the f
square of the disturbances, and so does Pil

- Q02pOis.) Ac-

;tress; mtum :itYVi. transvariahe at;ionto 3imple (3)

3.pure

cordingly,in this formulatian it is a good approximation to integrate onlyover the region of inten...c:e turbulent jet flowj indeed, by neglecting contributions from outside it, one is neglecting only how the sound emitted decays by attenuation and finite-amplitude propagation efiects (combined), which are better left to be estImated as a separate exercise. Now, in the radiation field, the difierentiations in Eq. (9) can be carried out very simply; no terms due to difierentiating the reciprocal of r with respect to Xi or X; are present, since they faIl ofi at least !ike 1/r2. However, [Ti;]represents the value of Ti; at a time t r/ao, which depends on r and hence alsa on the Xi; indeed, the rate of change of retarded time with x. is

~
GlXi

(t _ !..) = _
Qo

Xi aor

(10)

Using this in (9), one obtaIns in the radiation field

, W:l\'e
STRESS

STRESS
T1ilOl T1i

(4) ie can ;t, this .he re(5)

Fig. 5 Equivalence oC appIied stress to acoustic quadrupole

Tii--GTiL is cquivalcnt ta
QUADRUPOLE

-++DIR~

is cquivalcnt to QUADRUPOLE

... :!:. -

J!.ISTRIBUTIONS

-.'
i

-- - -

--- .

1\1. J, LIGHTffiLL
Measurements ai Mai:h number0.3 and Reynolds number 6 x LO' (Laureni:e 1956)
o-IX
X~'"

AlA.o\. JOURNAL

JUI

_---- -~.OF DttA'

-=~~'l4X

.. ) x-o .Ji.!!II!:..~. Xa4r--____(vtLDtmuSC X-.d --. mj~~ngrm

-u -- -

ADJUSTMtIT Tunui,UT
UGlOII
I

region where mcan veloi:/iy== 0.5U

i veloi:iiy

reai:hCl.

maximum

== o.i,ur

mately as v/bi (JoO.o'r', where Tb is a typical mean square Buctuation of ij'i;,which also can be written w4T', where w is a typical radian frequency of fluctuation, and T' is a typical mean square value of the quadrupole strength Tii. This gives the useful approxiIDateformula that the total acoustic power output of one eddy divided by its yolume V. (in other words, the acoustic power output per unit yolume of turbulence) is proportional to V.w4T'/pgaol (13)

,., po..' output

o"rOL conitoii' btbltot.d ocoulUc per unlt len,'h

~ o

Fig,6

Stational')'

cold jet

PO

= i-ij-i L: L:

~ . [2"ij]dr 411"ao2rJT
X'X .

(ll)

and one recognizesthe appearance of the second time deriva'iye 'i'ij, resulting in this calculation from the yariabi1ity of retarded time, a dependence on rapidity of variation of quadrupole strength to be ~xpected from the physical argumentI ga\'e earlier. Equation (ll) shows that the coefficients of the characteristic quadrupole terms in the directional distribution of pressure amplitude are obtained by integrating the quadrupole strength Tij oyer the jet volume V, after double differentiation and e\'alua.tionat retarded times. In order to estimate this integral contribution from the wholejet, it is essential to take into account the statistical randomness of turbulent Bo,,"."li It is a general principle of wave theolJ' that, with well-correlated sources, amplitudes combine linearly, but that, with uncorrelated sources, energy intensities combine linearly, ?o",", tl.lrbulent Bow measurements taken simultaneously at nearby points are well correlated, but those at points more remote from one another are almost uncorrelated, and, in the simplest form of the theory, one merel~'-uses:tlSresult, neglecting that the [1'ii] are not exactly simultaneous ,'alues. This yields a pil'tllTP a wrbulent Bowdh;ded into reinons, of such that strenltths of quadrupoles mthin am' one reinon are cor . stren at oints in different re-,ons are uncorrelated, the exten~ eac ID epen ent IRdMlpolp distrihution beinl/:rouii:hlythe size of a typicBI'
pnpT'lQ,_hpiiT';n~ eddy, For the output of a single region o

StrictIy speaking, this estimation should be done separately for different parts of the turbulent frequency spectrum, but in practice the '\"ariationof the product Y.w4 over the spectrum is rather small, because high frequencies go "ith smaIl eddy volumes. Evidently, an essential assumption in this estimation was that variations of retarded time "ithin an edd~'of size Lcoiild be neglectedj this, for Buctuations at radian frequency w, wouldrequire wl/aoto be smail, so that the eddy sizeLis small compared ",ith the wavelength of the sound it generates, This assumption, that wl/aobe smail, is satisfied at low Mach numbers, since the product wl of a characteristic frequency and eddy size is a characteristic \'elocitr. Furthermore, i shall show that the Mach number range \dthin which the assumption applies can be extended at least ta unity by a modification insuring that w is a frequency in a frame of reference moving with the eddies, for then wl is approximately a root-mean-square velocity Buctuation, which is a mere fraction of the jet velocity.

E par fluc leni x-i fall pro x "' uni tat 1 bay fre, hs: wa~ st::. cc fr ".
Ct.

pn to ine To

3. Turbulent Jet Aerodynamics

Fi
BD

dji

kind, oh'olume V.,

(12)

i fre tu: ed pe be tb, tli ftx qu dc cb ta D to \'f T: in di ix at ra in tl La

since mtliin r. the yalues of the integrand can be taken equal to their value at the center; on the other hand, the outputs of different regions are perlectlr uncorrelated, so that their energy intensities, < (p - 1'0)'>/poao, must be added to gi,'e the int~nsit" in the radiation field, From'Eq. (12) one can get a good idea of ho,," the acoustic energy intt'nsity generated by one such eddy at distance r yanes with the main parameters in"oh'ed. it yanes approxiWotU rodlotcd

i 'per holl-octo.c

Vorlotlan os .....
'

0'11,

0'01 i

. Frcqo.i.:~cy (ci:c1cs/~cc) Z50 500 1000 2000 4000 8000 l6OO

~
.

Vorfotlonas ",i , (iiiconswotti/cyclc _..)

~~

~/CYCJC

Fig. 7 aeousUc from an 3.9 cm m/see

Spectl'um of total power output ajr jet of diameter and velocih' 290 (Fitzpatrick. and ueli)

JCL Y 1963

JET ~OISE

15i1

~~~
~rew is typical I
.

Be\'ond :i: = 4<1,similar dimensiona! considerations apply,

Dsrti~ularly to the total sound emitted, a!though,

little..!!!> to ~l""lUs suggests a distribution of sound emissionper:init length as in the lo~'er half of Fig. 6, with about half the (L3) total comingfrom the mixing region. Different parts of the jet emit sound in di1Ierent frequency -arately. hands; in particulB.r,in the mixing region, where a typical , but in : irequency w a: x-i, the sound emitted uniformly in:i: should ,ectrum i ha\'e a frequency spectr,um per cycle !ike w-i, since ek a: .11eddy ; ..,;-~. a spectrum varying as i , :is w2or In the region x > Bd, one gets!ike x or remains conw. accordingly as Lincreases ,on was ~ ,;unt. E:\"Perimentalspectra (see, for example, Fig. t) are 1coiild ,:nn:;i~tentwith the view that about half the sound comes mcy w, ~ ::.)111he mixing region and has the w-2 type of spectrum, t ',\'hereasmost of the rest has a spectrum most !ike w thAn!ike i,;sm:i.ll :ierates. i ...;& probably comesfrom the adjustment region. :ind \' ~Iach Experiments agree well, alse, as will be shawn, with the ~<:uency i prediction that acoustic power output should be proportiona! more, i to {Joeld2/ail, lthough i must emphasize that i so far have a .ch the indic:ited how it can be derived only for law :\Iach numbers. ;y by a I To understand why it is also a good approximation at higher

is gI\'es : power words, !nce) is

~ :'Iroportionalto x for large en~.!!!~arieiu:ath~r

lik~ T.-ei-i, where the cprrelation

fiuctuations in Til fall e ~t Teng!ClJidee~Wld off, so doe_~.!t'Q:!!.ndmitte~ir~o-1[e x = Bd where velodties f r':i the..m!!l~mission pei::Jmit leQ.~ would be expected to radi~L..Jh.Q!!gll'

.2 o.e

.!.. d

1'5

O'Z

o -o-t

0'1 iladial distancc

o-Z of of

.
i

Fig. 9 Radial distribution of meaD veloeity ii, and convection velocity U. in the mixing region of a jet veloeit~' U (Davies, Fisber, and BarraU20)

.
,

~ of ref1ately

. 're frac- I i

ribe the ent jets Ayer, or Jecomes

~ orifice.

blstanc~ from nozzl~ ii -"Sd Dlstan from j~t aiiis r O'Sd

...J ZOO

rly with reen 0.2 ;ion, the .mum of ls.tionof ;ream of 1 such a :Telatian ~relation it O,04x, result of n ceases et enters = Bd,it !s faliing perimenh should t \'olume e mixing adrupole e ,yith :i:, ~reasthe ~eshould t ...olume mgth inet should 4d being

rii:. 8 Curves of constant correlation (Davies, Fisher, and Boirr.att%O)between velodties measured at two wires, ODe .ifi,'ectl~. dowDstreain of the other, in a l-in.-diiun jet at M"" 0.45

~I:ichnumbers, or to prediet directional distributions or peak irequencies,one must, as i mentioned earlier, consider the turbulent fiuctuations in a frame of reference moving with the edrlies. In my paperu of i954, i assumed, without conclusive ex;>eriment:il e\idence, that at each point of the jet there would be such n frame of reference, roughly speaking ..mo ing with theeddies,"in whichfiuctuations wouldbe considerablyslower than in a fixed frame. in other words, time variation at a fi..wd point would giye a misleaoing impression of high frequencr, refiecting merely the speed of comoectionof the randam space pattem past the point, rather than the speed of changesin that pattem as it is swept dow1lStream. ~n the past two :rears this assumption has been given det.'iiledexperimental erifieation, particularly in the work of Da\ies, Fisher, and Barratt of the University of Southampton%O (but see their later paperu for corrected numencal o;aluesof certain expressions caleulated 'from their data). :nieycarried out a systemAticseries of hot-"ire measurements

variable time delay in the secondhot-~ire circuit. By these means they were able to infer how, in a frame of reference movin at any chosen s eed e ual to hot-wire separation di\ided y time e y, t e correlation etween one-turbulen~ velociw value and alater onefalls oif ,,"ith time. Their results are plotted best as cur es of constant correlation (Fig. B), that is, eurve~in a diagram of wire separation against time delay on which the correlation between the measured velocities takes constant values. These show thAt the frame of reference in which the eorrelation falls ofi most slowly with time is rather well defined, and that the rate of fall in this moving frame is onlyabout a quarter of that measured in a fi.."ed frame. This impIies that typical frequencies in the frame moving at the eddy eonvection velocity U. are about a quarter of those that would be measured at a fixed point. Davies, Fisher, and Barratt20made similar me:isurements at sever:ilstations in the mixingregion and at variOU5 velocijet ties. Theyfoundthat the ,'elodtyof eddycom'ection varies across the mL'\:Ingegion (Fig. 9) but not nearly so widely as r does the mean velocity in the jet; values of U. from 0.2U to 0.7U were observed, an ayerage value being 0.5U. As a measure of rates of change in a frame mm'ing with the eddies, they used the time scale T, defined as the time t:iken for the correlation in that fra.meto fall to a :ilue ile; the reciprocal of T would be a typica.l radi:in frequency of turbulent fiuctuations in that fra.me. By plotting the time sca!e T against the reciprocal of the loc:il mean shear (Fig. iO), they discovered the interesting approximate relationship T ::: 4, mean shear (15) at eii.chpoint of the jet, This means that correl:ition falls to the alue ii e in the time t:iken for a sphere of fiuid to be stretched br the me:in she:ir into an ellipsoid with ::i.xesn the i ratios 1:4t :20. Dayies, Fisher, and Barr:itt%O obt:iin a number of other also
+ , u ISO Ha_ard 11- 0.34

.. .

. H-0045 H-004s" .-I'Sd .

}H- 0.Z2or-0'5d

_~ 1001

~ the

mi"<Ing region of a jet, using two wires, one placed

directlydownstream of the other, the distance between them being ariable. B:r correlating simultaneous va!uesofvelocity at ~hetwo wires, they evaluated the longitudinal correlation ~3dius an eddy, obtaining good agreement with previous of I!l,oestigators.However, theyalsa eorrelated the velocity at the upstream point \\ith the velodty observed a short time !:iterat the downstream point, by ineorporating a sman and '1

-11

50~

/:
oV

/
200

.~

400 600 Time scale .,. (ii see:)

800

Fig.

io

Variation of time scale with ioeal (Davies, Fisher, and Barratt20)

rnean

sheiir

.-'
,512

M. J. LIGRTHILL

ALU JOURXAL

JtjL Y 1963

Fig. II Inerease oCemittil1g vohime and emission time differenees, because oC convection, b~' the CaetOI"1/(1 - M. eosB)

into which we earlier imagined the jet as divided must be thought of as moving quadrupole sources of sound. But ac.. coring to well-established theory (see the Bakerian Lecture' for full references), a moving volume 1'. in which the quadrupole strength per unit yolu me is Tij has the sound radiation field

p _ po)2) ;;;;.--pooo

- PO = i-1 j-1 tt

411'ao2r3(1 M. COS8)3 ~

xixjTT.[T'i)

(18)

results of importance to turbulent-shear-flow theory as well as to jet-noise theory. These include the follovringrelationvi)2>}1I2, the ships between the rms velocity vi' = It'i longitudina,l correlation radius l, the time scale T, and the mean shear:

,,'here the e: of 1954. F: for a jet in it i:; ,"er: directional : niade "alid
!'C\lnd spee<

4.

Subsonic Jet Noise Theory

where M. is the velocit\' of convection U. di"idcd bv the atmospheric sound speed-ao. The effect of con"ectio~ appears factor, in which 8 is merelyin the additional (1 M. COS8)-3 the angle between the direction of emission and the jet direction. This factor, whichincreasesthe sound emitted forwardsul:>staDtiallymore than it decreases that emitted backward, can be understood from a diagram (Fig. 11) in which time is plotted vertically and distance in the direction of emission horizonta1ly; the figure shows asound signal being emitted with velocity ao from unconvected eddies and from eddies whoseconvection velocity component in this direction is ao.lf. cos8, It shows that, for a given quadrupole strength per unit volume, the emission is increased when this velocity component is positive, because the total emitting volume is in-

t'<idies are'
b:isI5 of th, ,'cloeit:-', De i:,. howe\'e: fuii:-- what

ii:imdy, it fur sma11a of emi5sioi jrl ,"clodt <lirectiona: 300 m 'see using :in ~

,'rlodt\' .

creased by the factor 1/(1

After this descriptionofjet aerodynamics, canreturnnow i to the soundradiation fieldand explainwhy, for calculating
it, the tnrh111pnt prlr1ie!!ml1!!t hp viewed a!! a pattem of Quadru-

poles which is moyini downstream.2.13 i hinted earlier at one reason for this, namely, that, by confiningconsideration to frequenciesin a frame of referencemoying with the eddies,the range in , hich wl/ aois small is e~..tended to at least M = i. up This is necessary in order to justify the neglect of difierences in retarded times within an eddy, and the results i have just quoted on frequency in the moying frame show that in a sulisonic jet wl/aois indeed everywhereless than one-sLth. A more basic reason, and one that applies even at low jet Mach numbers, why in estimating

Iv TijdT
over an eddy the time rate of change must be taken in a frame of reference mo,'ing with the com'ection velocity, is as follows. if instead one were to inflate that time rate of chan~e by adding to it a space rate of change times a convection velocity, the extra contribution to the volume integral from the space rate of change could be transformed at once into a surface integral oyer a surface outside the eddy, Its contribution to the intensity of aeoustic radiation would then be zero, because values on that surface would be uncorrelated with values within the eddy; so, in the end, one would be left simply with the time rate of changein the moving frame. These eonsiderations indicate that the eddy-sized regions
tent of the Guaclrupoles whose signols arrlve togcther X(I-M.cos e). QUadrUpaies convected wlth veloclty component

the radiation field of a quadrupole is due entirely to incomplete canceling of signals from posith'e and negative source8, ovringto differences in times of emission, two ex'tra factors 1/(1 - M. cos8) appear, representing the increases in those time differences, These are related closelyto the Doppler increasein the observed frequency. No\\",this incorporation of a directional factor (1 - .lle' ity Ii\lt ii witli jet' cos8)-i, in the pressure amplitude generated by each eddy in Indeed its radiation field,callsfor a directional factor (1 M. cos8)-C i-hould c in the intensity of sound generated by the eddy. When one (i - .11 considersthe sound intensity generated by the jet as a whole, that, :<\11 bowe,'er, or e\'en by a gi\'en restricted extent of jet, this ferred rir direetional factor is modified, as Ffowcs Williams fir:"t tutal rad showed.n Figure 12 depicts the mo,"ingeddies within a giwD kiiicl. 11\1 extent of jet X and showsthat this extent must be multipEe,d.. ari:lil'd i: by (1 - M. cos/J)to inelude only that number whose radiasl1l'ar, tl tion arri"es simultaneously. it foiio\\'s (multipIying this by ilr~ wi the directional factor on intensit)') that the aeoustic intensity hi" fol field per unit \'olume of jet takes the form tt'-of-;

- M. cos8).

Furthermore, because

rraiiy goo the orifief; other so\l region. ( ('r:i.l eddi! praeiical intensit\' that not t

100

..
-

o\lld 1

5db

. .

Lee(955)
LGilitir O-Hc and Hubbard (1952)

. '0.111 (1951)

cos.).,

r aciM.cas 9 \ _.,,,6~
.~~~ 1'&
-50

Number of quadrupoles whoie signols arrille togethcr . '?;I has factor(I-H. cos 9) TIME -EXTENT-Intensity of eacn has OFJET factor(I-H.cos e)". Distan X

'n direct!,>n

ol CINSlion

iHen factor

total

(I-Hecos e)'>

intenslty

has

-100

Fig.]2

Stationar~'

jet; acoustie intensit)' volume oCjet is

field per unit

Fig. 13 Direetional distribution of sound radiation aIr jets at 300 m/see. in decibels relath'e to B =

75

from

ri;;. ]

---

--

JET JeL)" 1063

':\UI;::L
LO' LO'

.~t be Lut ac~cture~ nadrn. :liation

. p - pop) = ~
;

[:ofIo

poa{)

i_1j

tt

_1

x,x;[2\~]
411'ao 2r

))

x
(19)

:: io'

(1 - M. cos8) -s

ioi

7i io'

,:irectional factor, a basicaIly lowMach number theory can be ;~:~Je\'alid for jets at speeds at least up to the. atmospheric ..)und speed~, The reason, as has been shown, is that, when the at~ .ppears 5 ,'<idicsare viewed in a mo\-ing frame, the assumption at the ich8is o !135is the approximations becomes that the rms turbulent of ','l'lodh', not the mean \'elocity, is smaIl compared with ao. it :he jet ~ j:,. ho\\:ever,stilI more satisfactory that this factor explains rd sul). .!. :ulh' hat is one of the most noticeable features of jet noise, :1:i~1l'h',ts marked directionality, that is, greater intensity i rd, can . ., :ime15, :..r :;n;al1angles 8 between the jet direction and the direction .. . 11l~~ion .' ,ii l'mL--sion, variation that becomesmore pronounced as the a ~mitted 'l't vclocitv increases. For example, in Fig. 13, showing the eddies ' :!irectionaI distribution of sound for stationary cold jets at i~110.1[, :300m./sec,the cur\'e represents the. (1 - M. cos8)-S factor, : Jer unit ~ u:,ingan a\'erage \'elocity of com'ection equal to half the jet \' com- ~ ','('Iocit\'. The measured directional distributions are in geni~ in- ~ "raiiy good agreement; they fnlI off somewhat simyerbehind >eca~e i thc orificeth:in the formula.would predict, probably because . u mcom- ~ other sources of sound are additionallv detectable in this :ource~,t ~I'gion. (These include the sound fields~f the weaker periphfactors . ..ral eddies with lower nlocities of convection and also, in n tho~e 1 :iractical experiments, some reverberation of the much higher intensitv sound transmitted forwards.) i shaIl show later that not only the directional distribution at a giyen jet velocpler init~' but also the measured change in directional distribution with jet velocity are in good accord with theor)'. 2ddy in Indeed, theory indicates only how directional distribution cos8)-6 i 1enone i ~hould change with jet velocity, through changes in the !1 .ll. cos/J)6 factor, whileleavingopen the possibility . whole, that, superimposed on this factor, there may be some preet, this ferredorienta.tionof the quadrupoles Tif. Experimentally, the 1~ first total radiated sound does not exhibit any marked effectof this a given kind. Imt the high-frequency sound does; to explain this, i iltH:"o 'f :irgiiedin my paperu of 1954that, in any regionof large mean if ~hear,there would be some predominance of lateral quadruif poles with directional peaks at 45 to the direction of flow. }a Thi.i followedfrom an expression for Tr'i as the product of Erin r:ite-of-strain and pressure, plus other terms whose effect ~houldbe smaIler. This relation would lead one to expect a ti!n p.cz, dominant lateral quadrupole associated with the mean rateof oC-strain, ith directional peaks at 45 to the flow direction w <'Ombined with randomly oriented quadrupoles (associaterl }J(' be;es \\;th fiiictuatingrates-of-strain :ind other terms in Tif).

(i~)

where the exponent 5 replaces the value 6 given in my paper13 ; L!)54 Ffowcs Williamsu has made a simils.r modification ~:'.r:i je~ in fiight which there is not time to discuss here. , 1t i:; \'ery satisfactory that, by incorporating this simple
i

1"1 o

Ret.7 ~ol<l-oirJet lu - ---iici.26ond 27 . (nozzle-exit diometer._ turbojet .engillos!~\ . Dt3 ,".j68convergence i i ongle)- ___ __ __n., Ret.7 cold oir jet \ (O-Sin}------.... (0-3in) , " ! i '
Prolt

.'

.~ 10-1

hat-and (old'oir jets --~

, lItIitney

oireratt

i
~

.d.iZ

_--'
~-Ref. 28 cold-oir jet

io-z 10-' LO.. LO.., i

'-Rc!.26 rm-aiLJets W-I'

j
i

!
;
~

LO

LO

io'

LO.

lO

106

i ' LO

11,u' "/0: . woits (,,-oreo ot nozzle)

Fig.15

Aeoustie

power

radiated 600 m/see

by jets

at speeds

up to

- .ll,

i
(

tli,?rs tli" it fj.,)ng 'lU! d'J/"encha,uld pme tail(cing Da \pole ton::h i" l'aM the nip.ould !1 iIlume irNh in?in,10uld ; thheing diu ,'e..,
O~

_
10db

sin i 28 (I-M~~ose)'

(pure lateral \quadrupoie:t:j:

(purc loteral quodrupole. plus randomly oricnted distribution of quodrupolu)

Figure 14showsthe type of directionaldistribution that would be expected, ",ith a peak around 30 and a subsidiary peak around 120, .ind it is significant that. the upper-frequency part of the jet noisespectrum, whichis regarded as emitted by the region of high mean shear, does ha\'e this type of directional distribution. s. 24. 24 i comenow to the rate of increaseof acoustic inteusity with jet velocity, \\'hich is expected to be in proportion to the product of the factor r,w4T2/poaos per unit yolume) which i ( discussed earlier and the directional factor (1 - .ll. cos8)5. At lowMach numbers the formerfactor should, as i expIained, increase as the eighth power of the jet velocity i;. However, turbulence measurements at the higher subsonic ~Iach numbersu. is indicate a definite reduction in turbuIent intensit~-, that is, in the ratio of rms velocity to mean velocit)', as the Mach number increases. With this goes a small reduction in rate of spread of the mixingregion (which, being much easier to measure. is valuable collater:il evidence for the intensity reduction), preparatory to the inuch greater reduction in rate of spl'ead that occurs, as i mentioned earlier, at still higher Mach numbers (compare Figs. 1 and 3). The work of Davies, Fisher, and Barratt20suggests that a reduction in rms velocit)' also should reduce the product wl of the eddy size L:ind the frequency w in a moving frame and so reduce also the product V,w4. if the rms velocity goes up as the tth power of the jet velocity U, as the limited e\'idence at the higher subsonic jet speeds seems to indicate, then V.w4p should be proportional to U6,rather than to Usas it is at lo\\' l',lach numbers, Now, this checks well with measurements of acoustic intensity at /J = 90, where the directional factor (1 - ii[., cos/J)-6has no effect because its value is unity; such measurementsS.25show close proportionalit)" to [-a between jet Mach numbers of 0.5 and i. Furthermore, measurement;; at other angular positions show dependence on i.' equal1y consistent with the idea of a basic U6term. modified by the directional factor (1 M. cos8)-&. For example, at 8 = 20, measurements of acoustic intensity5. 25 roughly proportional to U' are between M = 0.5 and 1, where the factor (1 - .ll. cos8)-i does vary in close proportion to U3. Similarly, the integral of

(1 M. cos/J)-6 over a sphere va.ries in close proportion to U2, which superimposed on the ca variation at /J = 90 gives a

Sdb

~ U (14)

ig. 14 .',

Direetional

distributions

ol intensity

in deeibels

UBvariation for the total power output, as is observed. A smaIl selection of the numerous data bearing on this eighthpower law for total power output is included in Fig. 15, taken from a report7of the XASA LewisResearch Center. People ah..-:iysfind it difficult, at first, to believe that the fact that the total acoustic power output of jets varies in the high subsonic region as Us. which happcn:; to be the same power as is gi..-enby low Mach number theory, is beeause of the canceling out of two corrections to that theory. However, both correetions are indisputably neces5arYi the directional factor is supported excellently by experiment, as well ra

~.-

M. J. LIGHTHILL as being theoretically essential, and the decrease in turbulent intensity with Mach number is also well substantiated.

AlAA JOUR~AL

when M. cos8 = 1 (so that the slope of the sound signal becomes equal to the slopeof the lines signifying the motion of the eddy), it appears inevitable that the emitting volume and emissiontime difierencesincrease to infinity. However, from the experlmental work of Davies, Fisher, and Barratt,20 one knows that this picture of a mo\'ing eddy is oorealistic, because even in the moving frame of reference it is changing gradually, and velocity values are losing correlation with earlier values. Figure 8 showed this elearly and suggests the revised version of Fig. 11 wmch is shown as Fig. 16,still with time plotted vertically and distance in the direction of emission horlzontally. This represents properly the decay of an eddy with time, by means of curves of constant correlation or, more precisel~' (taking into account the three-dimensional character of the correlation function), curves on whicb"its surface integral over planes at right angles to the direction of emission re'; mains constant. In the case when the component aoM.co;;8 of the convection velocitv in that direction is considerabh' less than the atmospheri~ sound speed Qo,one sees that th'e emitting volume and emission time difierences are still to a M. cos8). good approximation increased by the factor 1/(1 However, in the case when the two speeds coincide, neither the volume nor the time difierence becomes infinite; in particular,the emission timedifierence, hichwas formerlyl/Go, w becomes I/w, where w is a typical radian frequency in the moving frame. As for the emitting \'olume, its dimensions perpendicular to the direction of emission do not change, but its dimension parallel to the direction of emission increa"es from Lto ao/w. Thus, at the angular position 8 = cos-l(l/.lI,), both the emitting volume and the emission time difIerences increaseby the factor ao/wlinstead of b:r the factor 1/ (1 - .ll<' cos8). The deduction of the acoustic radiation fieldfrom this goes through exactly as before, with 1 li{. cos8 replaced throughout by wl/ao, and, in particular, one finds (wl/ao)-' appearing in place of (1 M. cos8) -I in the expression for

(lppO~iD~ .i tcn::ity \n' inrn'~i::e in

number of 1,.,con1r:, \. nrW tl'rm forc tl. (01

mtlii':"
:i;ii'rii:!diii i...'i:::

with it. T
tu zi-rii. :1'

n1t'nt~l\iy : :itin(l~i'h('r lin.' rnn

\i'iii:tli rii:ih'lY ti.,ti tlf ~itni' oi l'" tn t:'1' ,\.i.(.~~. '-, .Ii..:

J!:,..t::' t~ll": th:it tlll' rOll'kl't I'X!

J:,

i :ii:::rl' :\ ft'.'r t i )1:u'" nur

t ri Iiui j..i lo

i:i\','

:i

r.'

'1111'11":,. \\" r:ii li:iii i ri

nil':i-IIT\'c! di~t:iiii'i' . 11('~lk in'o of thi' ii; t~'Jlic':i1 o quitl. "i) i: :i maximi dillll'ii-ii,) w:iy- iii t'

acoustic intensity

FfowcsWilliamsu has carried out the calculation for a more general case, when the two slopes do not coincide but may be near each other in value, and he find" that (1 M, cos8) in this more general case should be replaced by

in the direction
- M.
COS8)2

= cos-i(l/M.).

Iiidll'r (.x IIl':iK ir.." (.i\ily in:

~(liii..\\" Iia frl'quI'I1/': i;i-ti'nt \\. :\i'xt. : ('iiin.. int. r.,~lI.i 011 ruiin' ra 1 infiiii'ii,';' ,...h :il ra:,i,lh' t

{(I

(WZ'GO)2}

1/2

(20)

In particular, the acoustic intensity field of unit volume of turbulent jet is proportional to

p -=--x - Po)2)
poaol
UNCONVECTED

F.w4T2 1 Poao 411'T2

Time

CONVECTED WITH VELOcin COMPONEMT a M cosll

o .i.c

{ (1

- M. CO;;8)2+ (ao) }

Wl

-1/2

(21)

DIstance When fi

In direction

ol emission

. cos., (I/Mc) slopu

(i.
volume
ol .minlon

In rlght-hand diagram colnclde j In Icft-hand diagraiii) becomes

diagram

emls~ion time dlHennce

dlmenilon of cmlttlng
iii to dlrectlon

~
~

( t In Idt-hand

) becomCl

w'

Fig. 16

Illustrating

the Ffowcs Williams sonic edd" eonvection

tlieory

of super-

if one neglects for simplicity the additional directional efiects arising from any preferred quadrupole orientations. in estimating the terms in this, it is fair to assume as a basic property of turbulence what Davies, Fisher, and Barratt found in their jets, namely [see Eq. (1i), that the product wl of frequencyand length-scale in a frame of reference moying v.'Ith the eddies is approximately the rms turbulent Yelocity. Then both the peak intensity and the total power output given by Eq. (21) must varyas va, once the Mach number of comoectionis well supersonic, say for jet speeds three or IDoretimes the atmospheric sound speed. Since the jet power also varles as va, this means that the acoustic efficiency, that is, the ratio of acoustic power output to jet power, becomes constant above about this speed, as I indicated earlier to have been found experimentally. For subsonic jets, it has been show'n that acoustic power output varies as UI, so that acoustic efficiency \'aries as ["6, experimental values being close to 10-4(U /ao)5 for subsonic jets (Fig. 4). In the upper half of the subsonic range, it was shown that the correctness of this law was maintained by tw~

,Ji-

!'iri,

iiii.r'.:!-....

tlll' 1"II~1 tli,. ~lii.;: nt tlil' i'n

f:i('t..r~

n,,!';III'r'
Exi"'riii'i di in"!
:.i.pprll:lf'!

Fiiiali
~"iinr.l 1'1 qii('nr\' ,

- .ll; <:'
tin\!' difi rddir~ r
:ilii,iit tl :~o(' /. \\'
!Jr tlii' (Jj tlil' li ..,

bilt it i,

"_. ,._- ..__.__..

".

JET XOlSE .'I " .aL "'IP"'~D " fnct01'3: the gradual reduction dinthturbul(,l1t inb i.:n--ity \\;th incre~in~ ~Iach. nu~ber an. e su stanti : , "i...:en forward emission with mcrease m JI., the Mach i .m~ber of com'ection. it is now clear that this latter effect ~U n1~... weakened as .ll. increases further, owinii;to the r..."1:'" . .
~ ft"

1515

ian of .e and
20

.Ifroni

r
i

one

c, he.nging t

with .

ts the 1with ! emisae, by :ciel\' of th .tegr:il :)11 re-

. w tNm (i.:l a~p in the denommator of (21). it fails theref:~re to iii;trun the reduction in turb~lent II!-tensity .~ut r:itilt'r hrltin5. as .ll. approaches .1 (that is, for Jet \'elocit~es a:,:irii:ii:hing twice the atmosphei:ic so~d speed), to conspire ;\'iih it. The rate of rise of acoustic efficiency then falls rapidly i., zi'ro. and the efficiency levels out at a constant (experi:i1<'n~llIyabout 0,006) at jet speeds around three times the
:itll1.,:,pheric sound speed:

. eo:,8
r:ihly

at the Lito a eo;;8). ieither n pary l/Qo, in the 'ni;ions ie, but :rea;;es 1,.ll.), 'renres

- .ll,' imthis f 'PInced

-l/ LIO) -I

ion for

i
i

na\" be ..

a more

,,8)(20) i in . ::

ume of

i .
i

Oill' consequence of this for rocket noise is that a \'err great ','iii~th of rocket emaust mn be radiating sound at approxi~::Itt'h' cnn~tant acoustic efficiency, that is, in direct propor:;.'11t:, the jet energy dissipated. This length should erlend ~.. WIll'rl' the jet \"elocity has fallen below three times the ..t:lili:,uhC'ril' sound speed, in other words, to where the jet O::L-":' tlil'Iu:,C'h'esare mo\ing subsonically, This conclusion, t!i:it tl1l' noi:!e sources e"..tend over a substa,ntia,l length of i' ,,'kC'ti'xliaust, perhaps as much as 30 nozzle diameters, is in i::IKKI ai;rcement ~ith obsen'ation.' .\ftC'r thi:, reasoned account of the changes, throughout the ~I:idi number range, in acoustic efficiency and directional disirilnnion, 1 should !ike to point out that the theory seems to ;:h'C' :i rr:'onable e:..-planation also of changes in peak fre'IIII'II('Y",ith jet \'elocitr. At low Mac.h numbers, a typical r:iili:iii fri'quencr w for a moving eddy in the mixing region is :11t':L"'lircdy Da\ies, Fisher, and Barratt20 as 1.3 i-/x at a. b di.-t:iiil'C x from the nozzle. This rises to 0.3 U i d where the jH':lk frequency is expected to be emitted, that is, at the end IJi the mi:..ing region. On the other hand, this frequency typii'al of the main energy-containing eddies mar not be quite ~olarge as the frequency of the eddies that make V",,4T2 :i maximum. The eddy yolume V. (proportional to an eddy .!imrn"ion cubed) and the frequency factor w4 pull different w:iy:,in this expression, but the frequency factor (carI'ying the iiii::hi'rcxponent) must be expected to win and to cause the ;1':lk frl'quency to be a multiple greater than 1 of the typical '1111Yfrequency 0.3 i- Id. E,''(perlmenta11y, the peaks are "JI11l'",hatfi:i.t, as i ha\'e already shown, but they occur at a. i'n.'(iucncyusually between 1 and 1.5 times Uld, which is con,i."'h'nt ",ith the argument i have just giyen. Xl'xt, as the jet yelocity increases, two opposing influences
romc into play, of ,,'hich one is the Dopplerfactor 1/(1

. .

a)

b)

Fig.17 Noise suppressors on the Boeing 707; a) wiih Pratt & Whitney 3C-6 engine, h) ~ith Rol1s Royce Conway engine

length-scale Lmay have risen as high as 3d, which would explain the observed figure.

6. Conclusion
To sum up, the theory gi\'es a reasonably clear explanation of the sound radiation fieldsof jets in terms of observations on the turbull'nce in those jets (although not pretending to deal with other noise sourees that may be present along with the jet or with losses associated with propagation of the jet noise), In particular, the observed yariations in acoustic efficiency, directional distribution, and frequency spectrum, with increasing jet velocit)" are accounted for adequately by the re resentation of a .et as a distribution of acoustic ~uacirupoes ID an ot ennse undistur ed atmos ere, e uadru ole stren s mg wl' corre te o y wit conv,e eddies of limite exten an i e e ime.
.

i ii
(21)

- M..

i effeets ne as a Barratt product ce mo\'lent \'e.i power .e Mach t speeds ince the ustic ef.t to jet s i indiie power :s as Us, subsonic , it waS :i by tw

ro~8) on the peak sound frequency, tending to ms.ke it rise ~ore rapi?ly than in proportion to U. The other opposing int1l1l'nreIS that the t)-pical frequency w of the eddies them>t.i\"('", t the end of the mbcing region tends to rise less a r:ipi.iiy than in proportion to U. For it has been shownthat ,:"ii~ "ruportional to the rms turbulent velocity, which itself ini'ri':!:,c"les;; rapidly than U. Furthermore, as U inereases, tlil' Ii'iigth of the mi'.:Ing region increases, and so stretching in ilii' ::hl'ar layer gi\'es the longitudinal dimension Lof an eddy :si thl' l'nd of it an opportunity to become greater. Both these !:ictors reduel' w and may be expected to outweigh the JI. cos8) in the radiated frequency. D(,p!>l~rfactor 1/(1 E.'<Pl'!1ment:il1r,S' 11.~~-:!5 7. one finds that the peak frequency dOl':!merease less than in proportion to jet speed as the latter :ipp~oaehesthe speed of sound. Fin:illy, for yery high speeds, the directional peak of the :'l)iind emission b:r eddies of frequency w should haye a fre'1l1l'ne)'equal to w multiplied not by the Doppler factor 1/(1 -:- .ll. cos8) but by the corrected factor am p lifvin g emission

ti

d'ff.

~1~ i erenees, that is, aolwl. if one assumes again that the ilics most effecth'e for sound emission have a frequency ~~t t~ree times as great, then the peak frequenc)' should b"e f ,wi~~ Lequal to the longitudinallength-scale at the end o the mixing region. Experimentally,S-IO the peak frequency 'bf ~~~~ound.radiation field of a. rocket exhaust is about 0.01 , d u i is not impossible that in 30 diam or so the longitudiniil

To go beyon the matter of physical understanding, and inquire concerning means for reducing the noise, elicits from the theory a somewhat pessimistic answer, namely, that the one parameter that affeets total acoustic power output to ani a,IDJreciable e:<tent is rms turbulent ,,~ity. which in turn depends principally on jet speed. By far the best way to reduce noise output is, therefore, to reduel' jet speed; and, indeed, if this is aIready less than twice the atmospheric sound speed, the factor by which noise output is redueed is the sixth powerof the factor by whichjet speed is reduced, for a constant value of the jet thrust. This has been one of the motive forces behind the trend to engines of high b:rpass ratio. Apart from this, only two successful methods of reducing ..jet noise radiation for a. modest weight penalty (with no reduction in jet speed) have been found (see the Bakenan Lecture2 for full references). One of these contrives to ~e rms turbulent velocity by diminishing thjLre1ii.tive velocitv of the jet an4 the air adjacent to it, as a result of nozzle shaping t.hl'lt.indtlC'e!!QIJYardmotion of much of that adiiicent air. f The other method makes u..i.e f the ma.rked directionality or o jet noise, from which it follows that the peak noise from a cluster of nozzles "iii be less than the sum of the peak noise from each separately. This is because noise in the peak direction from each nozzle will be redirected in some different directionon encounteringone of the others. Several good designs of noise suppressors for aircraft (see, for ex:i.mple, Fig. 17) h::l.\'emanaged to make use of both these principles,

.__._---

M. J. LlGRTRlLL and reductions of up to 10db (in peak sound) can be made with an overall performance loss, which, although severe, is not crippling economically. in the long run, however, to bring down the jet velocity is the only satisfactory solution. In contrast with all this, hopes of reducing the noise ofrockets, at any rate after theyare airbome, when even moderate penalties on thrust and weight are likely to be quite unacceptable, appear very slender indeed. I have tried in this lecture to build up a picture, by a combination of theoretical analysis and experimental evidence,of how the shearing motions in a turbulent jet manage to shed same of their energy as sound radia.tion. I hope that a fe\\"of the ideas that I have presented may prove useful to those who either may be faced with practical jet-noise problems or who have in mind e=-.-perIments throw new light on this imto portant and difficultsubject.
Appendix A: Infiuence of the Ratio Density to Atmospheric Density of Jet

AlAA

JOURNAL

This new equation is evidently the same as (8) but "ith T,i. replaced by the isotropic tensor ipOi/. The solution [compare Eq. (9)] is
{MJ2(p

.;,,-hai:

- Po) = V' f

like r!'tep in
peneli ,
tion. aJ ('()rrt'ct L-ni'\t rluil('~ tlir "'.:1 \\"hid: l>('("(lmi hilwr\-' u-"t'\1 ii n't:ir.1f iii tlll.
,;{'n'~ ri

~' 47!7'

[</>]

dr
(B3)

= .!. ~,
{MJ' >t' c

[</>]dr f V 4'11'7 = f V i4i/{MJ']dr 47!7'

A matter, not discussed in the lecture, on which experimental data have not reached complete agreement is the

efiect of dift'erencesbetween jet density

pJ

(defined as the

average density across the oMce) and atmospheric densit:r Po. Same good correlations of power output with PoU8A/{MJI, such as Fig. 15, have been obtained.7.32 However, other work, i. i. 23 using values of pJ considerably smal1er than Po, has demonstrated a resulting reduction in the acoustic power. This might at first sight, through its dependence on T'/ po [Eq. (13)], be expected to vary as PJ2/po rather than as Po. However, the peak sound is believed to originate in the center

of the mi,dng region,wherea typical density Pi shouldbe

intermediate between po and PJ, suggesting dependence on Pi'/ Po,wIiich does not fall ofi with decreasing pJ so rapidly as PJ'/Po. The situation is complica.ted further when, as often happens, the velocity of sound ai in the jet difi'ers considerably

fromaa. Thenthe termp,! -

T'I [Eq. (5)] represents sound generation approximatel;}'equivalent [by arguments like those leading to (B3) below] to that froma source distribution of strength (1 Qo'/aJ')o'p/()t' per unit volume. Our new knowledge (See. 3) on frequencies in the mixingregion in a frame maving with the eddies indicates that, when aJ'/{MJ's large (which general1~' oes with smaIl PJ/Po),this i g radiation will represent an addition t of around 20% to that represented by the term pM'j and therefore wili help to mitigate the decrease of sound output with decreasing PJ/Po. A bals.nced ,-iew of the existing data, and one reasonably consistent with the preceding argument, seems to be that some falling-ofi from proportionalit)" to POU'A/{MJ' in\\ith creasing PJ/po occurs, but that it is represented better by a factor PJ/po than by (PI!po) . This means that acoustic 2 power output for U/{MJ 1.5 is about 1O-.(!PJ,U8A/{MJI) < in agreement \\ith the law, stated several times in this paper, that acoustic efficiencyis about 1O-.(['/ao)8.

{MJ'P~'iin

The equality of the second and third terms in (B3) follows from the fact that the retarded potential is a solution of the wave equation. The fourth term shows that the acoustic radiation is equivalent to that generated by a distribution of sources of strength 41/00'per unit volume. it should be remarked that </> be regarded as a sart of incompressiblecan flowapproximatian to the pressure. Some mathematica.l criticisms of these theories in their earlier forms were made, particularly in relation to the treatment of source convection, but these criticisms appear to be met fully in Ribner's latest paper.if The correct directional factor [that is, for subsonic jets, (1 M. COS8)-I] is now dedueed. Furthermore, the author pro\"es conclusivelythat his formal expressionsfor the acoustic radiation field are equivalent mathematically to the theory set out in this lecture. The relative value of these two mathematica.lly equivalent formulations can accordingly be assessed only on the basis of whichgives more help in estimating how the noise generated in Yarious regions of a jet depends upon all the relevant variables, on the basis of approximations suggested by physical argument and experiments on jet turbulence. The simple-source theory has three main difficulties from this point of view: 1) One essential feature of Tij is that it is sina1l, of the order of the square of the disturbances, outside the jet itself (See.2), so that the integral for the sound field that has to be estirnated is merely an integralayer the region of the jet. This is not true of or of the pressure (to which it represents an approximation), so that, in the simple-source theory, noise generatian from outside the jet has, rather artificially, to be postulated and, presumably,estimatedo 2) Inside turbulent jets extensive measurements of the correlation of velocity ys.luesat difi'erent points, and of ho\\' it faUs away ",ith increasing distance between the points, have been made. These, ",ith further information on velocity spectra and space-time correlations, were used in the estimation (Sees. 3 and 4) of the sound generation associated ",ith the quadrupole distribution Ti;, with reason, because its fluctuations arise mainl~'from \"elocityfluctuations. By contrast, little is known of pressure fluctuations inside turbulent jets, whose correlations and spectra cannot be measured directly. Estimation of fluctuations in </> though fluctuations in the pressure (to which it is same sart of approximatian) is therefore not feasible, whereas direct estimation through Eq. (Bl) defining </> leads right back to the problem offluctua-

hi. Fi:: Ti", i.. ii..t

:'t,,\.,:~t' hri,.fiy ti..ii~ :i :tcrrirti

, L.i: ,\I'r"D~ : 1.ii

nil", :;., in: :ixisrm 'Lt: lIIl'trk ~till air

</>

'c"
Urui'lii: 'C,.. i i h:rii~ 111, TH 1:" 'Ci. I'iivir..! (i~I-T ) :\1 lil'\d :u r:l1u:'. , i.. ('. ali.1 H II"\".
i:

\'l'i'i",d
;;in l!

r:

tions in TH itself.

Appendix

B: Simple-Source

Theories

of Jet Noise

Ribner3f and others (cited by him) have deseribed jet noise and other aerodynamic-soundphenomena in an alternative manner using sirnple sources. The~-define a function </> such that
V'</>

i-i i-i

t t ox,ox, 02~i~.
v2</J C>Xi'

(BL)

in terms of which Eq. (8) becomes o'P

- ()t'

i_ i

3 o'P i: ao' ---;;=

(B2)

t The :iuthor's HJ,54paper,n on p. 22, estimated t.his proportion as.li- of the squ:ire of the ratio of radian frequencr to mean shear (a mtio now known, from Fig. 10, tu be around 0.22).

3) Furthermore, yalues of Ti; are expected to be well correlated only "ithin convected eddies of limited extent and limited lifetirne, and, for at any rate subsonic jets, the eddy size is very smaIl compared \\ith the wayelength of the sound that it generates, whose estimation is made much easier by this fact. By contrast, the correlation of ipat two points h expected to fall off much more slowl~' ",ith the distance between them. Indeed, it appears doubtful whether carrelatian ra.diIfor </>re smal1enough compared with the wa"elength, a for any jet Mach numbers, to permit the neglect of dift'erences between retarded times where \"alues of </> carrelated sigare nmeantly, for this neglect would lead to a uniform directional distribution of Ii'coU$tic radiation (exeept for the canvecthoefactor), a serious errorin eyery case when T'j departs signifieantly from isotropy. It i;;, in fact, easy to show (one exampleis given in a footnote to the Bakerian Leeture) that a localizedYariatian in Tij proom'e;; a ;;olution of Eq. (Bl) for
- - --

Jt"I. Y 1\163
Til pare

JET XOISE
12

151i

:B3)
10\Vs the istic n of ~reible,heir 'eatJ be Dna! . de~his .iva-

~ who:,e correlation with that local variation falls ofI in general :ike r-i. This ma.kes impossible in simple-source theory a :n~p like th:it from Eq. (ll) :o Eq. (12), whose validity de:1C."nd:' 1r; being a volume mtegral of the relevmt correlaon ~ion.:in integml that in this case diverges. Ribner points out ,-orrcoi.th'th!Lt the exact integral for the sound radiation field :..~ Mt di\'ergent, beeause for sound of each frequency it in,-lo.1,I.,:; sinusoidally ,'al1ing term (with wavelength that of :i t!:i' ,"",unditself) , !LSsociat~dwith difIerences in ret&rded times, :\'hidi renders it con"ergent as the region of the integration :,<"("()01C:; compai-ed "ith the wavelength. This mea.ns, l:irge howco\'cr, that ordinary eddy-size measurements cannot be :1."4."1 the estimation process, and also that difIerences in in n'urded times ne\'er can be neglected. it may be significant in the fje:ht of the foregoing remarks, that Moll-Christenn.:' nica,.urementsJ1 of pressures just outside a. jet show (see ~i,. Fig. 12) e~..tremely long tails to the correlation curve. Th,' threedifficulties otedhereseemto indicatethat q,&ii n ~" II'lt :i useful alternati\'e form of Ti;. There are not such

Lighthill,~i. J., "On sound generated aerodynainically,i:

"\'rl. litficultie:s the 'Piic ",ith

ao2p&ifterm

in Tif, discussed

:1~id1r in Appendb.: _\, because outside the jet itself its varia.tiiin:' :ire negIigible, and the large correlation spheres reduel', :ii'i'ordingly, to their intersections with the jet itself.

~nt is of ited Tant iysiThe this the tseU obe jet. ants oise >be the
tio,," nts, city ma\ith its :onlent ired ions ion) ugh ~ua-

References : Li~hthill, ~i. J., "~lathematics and aeronautics," J. Ray. .\i'r.,n:iut. Soc. 64, 3i5-394 (1960). , I.ighthill, ~'i. J., "Sound generated aerodynamicii.lly," Proc. Hii". 5ve. (London) A267, 147-182 (1962). i f>:ivies,:\1. G. and Oldfield, D. E. S., "Tones from a choked :i.u,ymmetric jet," Acustica 12, 257-277 (1962). Love, E. S. and Griaby, C. E., "Some studies ofaxisym. metrie free jets exhausting from somc and supersomc nozzles into
~till :iir :ind into supersonic streams,"

"Estimating jet noise," Rolls Royce (Derby) HrodiureA.P.11 (1961). 'Cvles, G. )OL., "The noise of high velocit)' jets," Rolls Royce !)~rliy) unpublished Rept. (1961). 7I1IJ\\'es,W. L., "Similarity of far noise fields of jets," NASA TH H-S:!(1959). 'ChIJbotov, V. and Powell, A., "On the prediction of acoustic ...n\'irrinmcnts from rockets," Ramo Wooldridge Corp. Rept. (;~I-TR-190 (195i).

, Coles, G. M.,

NACA RM L54L31 (1955).

, ~[ayes,

W. H., Lanford, W. E., and Hubbard,

H. H., "Near-

:i...iil:ind f:ir-field noise surveys of solid-fuel rocket engines for a ~ii:t. of nozzle exit pressures," NASA TN D-21 (1959). aCille, J. X., von Gierke, H. E., Kyrii.zis,D. T., Eldred, K M., ~.i Humphrey, A. J., ":Xoise radiation from fourteen types of T'oekt.ts the 1000 to 130,000 pounds thrust range," Wright Air in II'!". ('cntt'r Rept. 5i-354 (1957).
ii Ffiiwcs

WilIinms,J. E., "The noise from turbull'nce con-

...~.teJ.:it high speed," PbiL.Tr:i.ns.Ro". Soc. (London) A2ss, 469:.ixi

1%3).

General theory," Proe. Roy. Soc. (London) A2U, 564-587 (1952). li Lighthill, M. J., "On sound generated aerod:rnamically, II: Turbull'nce as a source of sound," Proc. Roy. Soe. (London) A222, 1-32 (1954). u Laurence, J. C., "Intensity, scale and spectrii. of turbulence in mixingregion of free subsome jet," NACA Rept. 1292 (1956). 11 Lassiter, L. W., "Turbull'nce in smail air jets at exit velocities up to 705 feet per second," J. AppL. Mech. 24, 349-354 (1957). 11 Townsend, A. A., The Structure of Turbulmt Shear Flow. (Cambridge University Press, London, 1956), Chap. 8. 17Ribner, H. S., "On the strength distribution of noise sources along a jet," Univ. Toronto Inst. Aerophys. Rept. 51 (1958). 18 Powell, A., "On the generation of noise by turbulent jets," Am. Sac. Mech. Engrs. Paper 59-AV-53 (1959). 11 Fitzpatrick, H. M. and Lee, R., "Measurements of noise radiated by subsonie air jets," Taylor Model Basin Rept. 835 (1952). . ID Davies, P. O. A. L., Fisher, M. J., and Barratti M. J., "The characteriatics of the turbull'nce in the mi:dng region of a round jet," J. Fluid Mech. 15,337-367 (1963). il Ffowcs Williams, J. E., "Some thoughts on the effects of aircraft motion and eddy convection on the noise of air jets," Univ. Southampton Aeronaut. Astronaut. Rept. 155 (1960). li Lee, R., "Free field measurements of sound radiated by suhsonic air jets," Taylor Model Basin Rept. 868 (1953). li Lassiter, L. W. and Rubbard, H. H., "Experimental studies of noise from subsome jets in stili air," NACA TN 2757 (1952). it Westley, R. and Lilley, G. M., "An investigation of the noise field from a small jet and methods for its reduction," College of Aeronaut. Rept. 53 (1952). u Gerrard, J. H., "An investigation of the noise produeed by a subsonic air jet," J. Aerona,ut. Sci. 23, 855-866 (1956). 21 Ca\laghani E. E. and Coles,'W. D., "Far noise field of air jets and jet engines," XACA Rept. 1329 (1957). 27 Clark, W. E., "~oise produced by aireraft during ground running-up operations," Wright Air Dev. Center TR 56-60 (1957). 21 Waterhouse, R. V. and Berendt, R. D., "Reverberation ehamber study of the sound power output of subsonic air jets," J. Acoust. Soe. Am. 30, 114-121 (1958). 21Rollin, V. G., "Effect of jet temperature on jet-noise generation," NACA TX 4217 (1958). io Eldred, K., "Xoise measurement of four rotor &ircraft model jet nozzles," Pii.ul S. Veneklasen and Assoc., Los Angeles, Rept. 238-11-1 (1958). . 31 MoH-Christensen, E., "Jet fiow and jet noise," Inst. Aero. space Sci. 31st Annual Meeting, New York (January 20-23, 1963). si RoHin,V. G., "Effect of jet temperii.ture on jet noise generation," XACA TX 4217 (1958). u Davies, P. O. A. L. and Fisher, M. J., "Statistical properties of the turbulent velocitv fiuctuations in the mbcing region of a round subsonil' jet." u;Uv. Southii.mpton Aeronii.ut. Astronaut. Rept. 233 (1963). u Ribner, H. S., "Aerod;rnamic sound from fiuid dilatations," Univ. Toronto Inst. Aerophys. Rept. 86 (1962).

well and dd\' und br ;s is be~ion ~th, ices sigrec:onirts one it il for