CHAPTER

COMPRESSIBLE FLOW-
SOME HISTORY
AND INTRODUCTORY
THOUGHTS

so fe\ rlzou.si;i L')uld Sukeed, un

It required an unltesitating boldryess ro undertake a L'enture
and ttrii:tt{tv'rt:' hut most of
alntost exuberant enthusi&nt to carm'across lllc man)'obslacle's
in departing so drasticall)',{ror'i rlie knott'tt tt'ttt"
oi,ll , ronrpt"trly unpreiudiced imagination
faced in designinu lhe Beil XS-l' thc
J. van Lonkhulzen, 1951, in discussing the problems
{irsi suPer-sonic airP}anc

1.1 HISTORICAL HIGH-WATER MARKS

has been opelle'l
The year is 1893. In Chicago, the world-columbian Exposition
thaa 27 miilion peopie *'ill
if ei"ria.rrt Grover Clevei-and. During the year, more
from
visit the 666-acreexpanse of gleaming white buildings, speciallr constructed
jute fiber to simulate rvhite marble. Located
a composite of plasier of paris and
adjacent to the newly endowed University of Chicago, the
Erpcsition commemo-
rates the discovery of America by christopher columbus '100 vears earlier'
and dornestic and liberal arts, as
L,xhibitions related to engineering, architecture.
are scattered over 150 build-
well as collections of all-modes of transpoftation.
lrrlr. ln the largest, the Manufacturer's and Liberal Arts Building. engineering
that will
exhibits from all over the world herald the rapid advance of technology
soon reach explosive proportions in the trventieth century' Almost lost in this
a roof of iron and giass. is a small machine of
massive 31-acre building,'under
 N'tODtrRN CONTPRE.SSIBLE Fl-O\\

Turbi ne

,-rrrrr'7

FIGURE I.I
Schc'matic ol dc Laval's turbtne incr-rrporatilrg a

Lon\ erscnl-di\ cJ g('nl nnTlla.

great importance. A single-stage steam turbine is being displayed by the Swedish

for
J1girr.".. Carl G. p. de La'al. The machine is iess tha* 6 'ft long: designed
*-ina use. it has tu,o independent turbine wheels, one for fonvard motion and
is that the
the other for the reverse direction. But rvhat is novel about this device
b1'a strearn. of hot, high-pressure stcam from a series of
turbjne biades are driven
nozz-les. As sketched in Fig. 1.1, these nozzles. rvith
unique convergent-diuergent
their convergent-divergent shape representir..g a complete departrrre from
previ-
ous engineering applications, feed a high.speed flo*'of steam to the blades of the
turbine whee1. The tleflection and consequent change in momentum of the steam
as it flo.ws past the turbine blades exerts an impulse which rotates
tire rvheel to
speeds prer;iously unattainable-over 30.000 r/nin. Little does de
Laval realize
that his convergent-divergent steam nozzle wili open the door to the supersonic
*,ind tur.rnels and rocket engines of the midtrventieth century'
The year is now 7947. Tlte morning of october 14 dawns bright and
beautiful over the Muroc Dry Lake. a large expanse of flat, hard iake bed in
the
Mojave Desert in Caiifornia. negrnning at 6:00 A.M., teams of engineers and
tecinicians at the Muroc Army Air Fieid ready a small rocket-powered airplane
for flight. Painted orange, and resembling a 5O-caliber machine gun bullet mated
ro u iulr of straight, stubby wings, the Bell XS-1 research vehicle is carefuily
instalietl in the bomb bay of a four-engin e B-29 bomber of World War II vintage.
At 10:00 A.M. the B-29 with its soon-to-be-historic cargo takes off and climbs to
an altitude of 20,000 ft. In the cockpit of the XS-1 is Captain Charles (Chuck)
Yeager, a veteran P-51 pilot from the European theater during the war. This
n o.iirg Yeager is in pain from two broken ribs incurred during a horseback
riding aicident the previous weekend. However, not wishing to disrupt the events
of the day, Yeager informs no one at Muroc about his condition. At 10:26 4.u.,
 TTIOU(iIITS 3
( oN{YRT-SSIBLE FI'OW' SOME HISTORY ANI) INTRODUCI'ORY

ataSpeedof250lfii/h(|12m/S),th€.brightlypaintedXS-ldropsfreelrorltire
g-zg. y"ug"i nr"r rri. n.u"iion Motors xLR-11 rocket engine
bomb bay of the
the sleek airplane accelerates and climbs
and, powered by 6000 ii ir tm,rrt,
diamonds from the four convergent-
rapidly. Trailing a, e*huttst jet of shock
XS-1 is soon flying faster than Mach
divergent rocket ,,o.,,f", oi ttr'e erigine' the
0.85,thatspeedbeyondwhichthereisnowindtunneldataontheproblemsof
regime' Yeager momentarily
transonic flight in tiql - Ynt"'ing this unknoun
and carefully tests the controls of
shuts down two of the four rockei chambers,
0-95 and still increasing'
the XS-1 as the Mach meter in the cockpit registers
over the top surface of the
Small shock waves are now dancing back and forth
;;;g, At an altirude of 40,000 ft, the XS-1 finally srarrs ro ie'el off, and Yeager
firesoneofthetwoshutdownrocketchambers.TheMachmetermovessnrootlriy
through0.98,0.99,to1.02'Here,themeterhesitates'thenjumpstol'06'A
the air ahead ol the needlelike nose o[
stronger borv shock wave is now formed in
theXS-lasYeager.",chesavelocityofT00mi/h.Maclr1.06'at43'000ft.Tlre no L:ss of
flight is smooth; there is no violeni buffeting of the
airplane
-and bccomes
Cilgq\.Yeager
controi as $,as feared by some engineers. At this moment'
of sound. anrl the small but
the li-rst pilot to succesJfully fly fister than the speed
lirst successlul super'oi.ti"
beautiful Bell XS-1, sho*n in nig. 1.2, becoues the

FIGURE I.2
The Ilell XS-1, first manned supersonic aircraft. (Courtes.t'oi the
l'ationul Air unl spate 14usettnt\
4 l.roognxcoN"lPRESSlBL

airplaneinthehistoryofflight.(Formoredetails'seeRefs.land2listedatthe
back of this book.)
Today,bothde!3val's10-hpturbinefromtheWorldColumbianExhibi.
tion and the orange Bell XS-1 are part of the collection of
the smithsonian
Institution of washington, D.C., the former on drspiay in the History of
rvith distinction from the roof of the
Technolog-v Building uia ,1" iatter hanging
NationalAirandSpaceMuseum.Whatthesetwomachineshaveincommonis in
high-water marks
that, separated by more than half a centur]. they represent
in" ..,gi,.,".ring applicatio* of the principles of compressible flow-where the
density"oftheflowisnotconstant.Inbothcasestheyrepresentmarked
J.pu.ir.., from previous fluicl dynamic practice and experience'
Tlreengineeringfluiddynamicproblemsoftheeighteenth.nineteenth.arrd
earl,v trventieth centuries almtst alu,a,vs invol'ed
either the flo""' of liquids or the
lorr-speed fiou, of gases: for both cases the assumption of cclnstant densitr is
quite valici. Hence. rhe farniliar Bernoulli's equation

p + ipl'): corlst (l 1)

rr,asinr,ariabll. emploved vu'ith success. Hou'ever" rl'ith the advetrt

of high-speed
nozzle desisn and the
flows. exemplified u! ,le Laval's convergenr-divergent
,rp"rron;.fiightoftheBellXS-1'thedensitlcannolongerbeassumedconstant
-lndeed.
throughour tie flori,lield. for such flou's the densitv c:in someiinles vary
longer holds' In this light'
Uy ori"r, of magnitude' Consequentl)" Eq' (f i) no
*e.e inde.d a marked departure from pre'ious experience in fluid
,r1ch
"rents
dynamics.
i'e'' it deals with
This book deals exclusi\,elv,'i|i1h tltiit "nlarked rJeparture."
compress.blefows,inrvlrrchthedensityis,lolConstant.Inmodernengineering
A few important
uppii.utiorr. such florvs are the rule rather than the exception.
high-speed
;;l;;i., the internal flo*,s through rocket and gas.turbine engines,
^re h1'p...onic lvind tutrnels. the external flow
subsonic. tran.sonlc, supersonic. and
oi of sound'
or,er modern airplanes'de.signed to..rii. faster than 0.3 the speed
engine' The
and the flo*. inside the corimon internal combustion reciprocatin-g
p,,.po,.ofthisbookistodevelopthcfundamentalcorrceptsofcompress,ible
flow, and to illustrate their use'

1.2 DEFINITION OF COMPRESSIBLE FLOW

this is in contrast
compressible flow is routinely defined as uariable densitl'fiow:
the density is assumed to be constant throughout'
to i.rcomp."ssible florv, ivhere
obviousl.v.inry4tlifeeveryflowofevelyfluidiscgmpressib-letoson1egreateror
lesser exrenr; hence, u t.,rly constani'tensity (iniompressible)
flo*' is a myth'
for almost a1l liquid flows as weil as for the
Hou,ever. as previously mentioned,
are so small
flows of some gases under certain conditions, the density-changes
that the assumption of constant-densitv can be made u'ith reasonabie accuracY'
equation, Eq (1'1), can be applied with confidcnce'
In such cases, Bernoulli's (1.1) does not
Ho*,ever. ior the sub.ject of this book-compressible florv-Eq.
ho,d, and ror our ;:J:"' ;'ffi;J ***"' ;;''*'n"]
thinking.
The simple definition of compressible flow as one in which
the {ensity is
a small element of fluid of volume
variable requires more elaboration. Consider
by the neighboring fluid is p'
u. The pressure exerted on the sides of the element
ar.u*" the pressure is now increased by an infinitesimal eIIsg9j..fu'-The volume
compregfed by the dr'' Since the
of the element will be correspondingly 1mlunt
quat'tity' the to*ptessibility of the fluid' r'
uotrr*" is reduced, 4q. i-1., negative
is defined as
7du (t 2)
L" ap

phy_sically, the compressibility is the fractional change iIr volur.re of the ftuid
not sullicientll'precise'
eteile.rt per rrrit change i.r pressrr". Howe'er, Eq. (1.27 is
We know from experiJn.. th*t when a gas is compressed (say in a bicycle pump)'
its temperature tends to increase, depending on the amount ofl1eat transferred
intooroutofthegasthroughtheboundariesoftheSystem.Therefore.ifthe
i.*p.1uru." of the fluid eleirent is held co-.nstant (due to some heat transfer
*"&u"it*), then the isothermal contpressibilil-t is defined as

1 r'dt r
(i 3)
u\Apl,
I
- -l
-

fluid element (if

On the other hand, if no heat is added to or taken au'ar from the
;he ;"-p..rsion is adiabatic). and if no other dissipative tran:lort mechanisms
,""it ur uiscosiiy and diffusion are important (if the compressiop- rerersible
is
and the
)'

then the compiessio., of the fluid element takes place isentropically.

isentropic compressibility is def,ned as
'l i dl i
r's * _t_t (1.4)
u\ 0p I,
where the subscript s denotes that the partial derir-ative is taken at
constant
q!tropy.
' to-pressibility is a ploperty of the fluid' Liquids ha'e ver-v lou' vaiues of
gases ha'e
compressibility (r. ior *uii. lt 5 x 10-10 m27N at 1 atm) u'hereas
high comprer.iUitiii"t (r, for air is 10-5 nt'7X at 1 atm, more than
four orders
of"magniiude larger than water). If the fluid element is assumed to have unit
density is p : 17'"
*urr, i is the specific-volume (volume per unit mass)' and the
In termq of density, Eq. (i.2) bec-omes
ldp (1 5)
pap
Therefore, whenever the fluid,experiences a change in pressure. .dp, the
corre-
(1'5)
sponding change in-density wili bg dp, where from Eq'
dp: prdp (1 6)
 6 uonEax co\rPRESSrnt-E FLow

To this point, we have considered just the fluid itself. with compressibiltty
being a property of the fluid. Now assume that the fluid is in motion. Such florvs
are initiated and maintained by forces on the fluid, usually created by. or at least
accompanied by, changes in the pressure. In particular. u'e shall see that !_[h-
speed. flows generally involve large pressure gradients- For a given change in
pressure, dp, due to the flow, Eq. (1.6) demonstrates that the resulting change in
dgpsily wiil be small for liquids (u,hich have low values of z). and large for gases
(ri,hich have high values of r), Therefore, for the florv of liquids, relativell' large
pressure gradients can create high velocities rvithout much change in density.
Hence. such florvs are usually assumed to be incompressibie. rvhere p is constant.
{)n tire other hand. for the flow of gases with their attendant large l'nlues ol r-
moderate to strong pressure gradients lead to substantial changes in the dcnsit-v
via Eq. (1.6). At the same time, such pressure gradients create large velocit-v
charrges in the gas. Such florvs are defined as (olnpressible .flotts. where p is a
variable.
We shall prove later that for gas velocities les-. than about ().3 ol the speed
ol sound. the associated pressure changes are small. and even though r is large
frrr gases. dp in Eq. (1.6) may -still be smail enough to dictate a small r1p. For this
reason. !1,g !9*,W.e-d llorv of gas95 can be assumed to be incompressible. For
erample. the flight veiocities of most airplanes from the time of the \\/right
brothers in 1903 to the beginning cf lVorld War II in 1939'\r'ere generallr'iess
than 250 mi/h (11.2 m/s). which is less than 0.3 of the speed of sound. As a
result. the bulk of early aerodynamic literature treats incompressible liou'. C)n the
other hand. flow velocities higher than 0.3 of the speed ol sound ar: associated
with relatively large pressure changes. accompanied br. correspondingly' large
changes in densit-v, Hence. compressibility effects on airplane aerodynamics have
been important since the advent of high-performance aircraft in the i940s.
Indeed. for the nrodern high-speed subsonic and supersonic aircraft of todat,. thc
older incompressible theories are wholly inadequate, and conrpressible lliru
anal1,'ses must be used.
In summary, in this book a compressible flou' rvili be considered as one
wlrere the change in pressure, dp, over a characteristic length of the llo\\'.
multiplied by the compressibility via Eq. (1.6). results i:r a fractional change
in clerrsity. dp/p,wltch is too large to be ignored. For most practi-cal problems.
if the density- qhqqges by 5 percent or more, the flor.r' is considered lo be cor.t-t-
p:essible.

Example 1.1. Consider the 1ow-spced flow of air over an drpla:re u,ing at standard
sea level conditions; the free-stream velocity far ahead of the wing is 100 mi/h. The
fiow accelerates over the wing, reaching a maximum veloci8'of 150 mi/h at some
point on the rving. What is the percentage pressure change between this point and
the lree stream?

Solution. Since the airspeeds are relatively 1orr'. let us (for the llrst and on/.t, timc in
this book) assume incompressibie flow. and use Berntulli's equation for this
 AND INTRODUC''TORY THOUGHTS 7
COMPRESSIBLE FLOW-SOME HISTORY

problem.(SeeRef.lforanelementarydiscussionofBernoulli'sequation,aswellas
role of this equation in the solution
Rei. 104 ro. u *o.Ji.-tuit"a presentation of the
ofincompressiutenow.g"re,'".*,,*.thatthereadelisfamiliarwithBernoulli's
equation-its *. *d its limitations. If not, examine
carelully the appropriate
the frg:- s{1qam and lying
discussions in Refs' i and pcicl$
104) Let '1 aqd 2-'denote
points, respectively' Then, from Bernoulli's equation'
Pr+*Pl'?:Pr+lPv]
or t-Pt:\P(vj'v")
p :- 0'002377 slg&/ft3' Also' using the handy conversion that
At standard sea level,.*{ffi
60 mi/h: 88 ftls, v, : r00- milh : t4'7 tt/s and v': fia mt/h:
22aft/s'lNotethat,asalwalsinthi.sbook,rverr.illuseconsislentunits;for
-English
probiem' or
a"*pf., we will use eithtr the Engineenng Sy'sr9m'-as in this
the Interoatiooa System. See the fJotnote in Sec. 1.4 of this book, as rve11 as Chap'
2ofRef,l.Byusingconsistentunits,noreofotrrbasicequationswillevercontain
"rr.h with this
colversion fr"torr, as q.. and -I. as is found in some references.)
information. we have
Pt - Pz: iPVj - t'i)
: I$.0023711[1zzo;'
* (147)r] : 3r.8lb/rtr
plessure, u'hich at
The fractional chaage in pressure rcferenced to the frce-strealn
srandard. -spa level is pt : 2116lb/ftr' is obtained as

P-L--I,: :];! :0.01s

Pr 2116

Therefore, the pircentage change in pressure is 1,5 pgrlent'

In expanding over the
percent. This is acase where, in Eq.
wing surface, the,oressrire changes by only 1.5
dp is small' .The purpose of this example is to
O 6l,, dp is small, and hence
demonstratethat.iniow.speedflorr'problems,thepercefilagechangeinpressureis
always sma1l, and this, through Eq. (1.6), Justilies lhe assumptirtri
of incompressible
f,ow(.dp:0)forsuchflows.Holvever,athighflorvvelocities,thechangein
This is the regime
pressure is not small, an<l the density must be treated as variable.
floq,-the subject of this book. No/e.' Bernouili's equation used in
of compressible
this exampie is good only fot incompressible flow, therefore it will not appear again
Experience has sholvn that' because it is one
in any of our sribseq*ent discussions.
by students in the study of fluid dynamics,
of the irst equations usually encountered
-Bernoulli's
is a tendeqc5 to use equation for situations where it is not valid'
there
discussions
Compressible nooiii one such situation. Therefore, for our subsequent
in this book, remember neuer to invoke Bernoulii's equation'

1.3 FLOW REGIMES

rvhen orviile
The age of successful manned flight began on December 17, 1903,
and 1tr7ilbrrr Wright rook to the air in their historic Flyer I, and soared over the
windswept ,und dun., of Kili Devil Hills in North Carolina.
This age has
and supersonic
continued to the present with modern, high-performance subsonic
airplanes, as rueli as the hypersonic atmospheric entry of space vehicles' In the
 8 s.ror;r,nn (-oNIpr{Esslr}r-Lr Fr,ow

twentieth century. manned flight has been a major impetus for the advancement
of tluid dynamics in general. and compiessible florv in particular. Hence. al-
though the fundamentals of compressible florv are applied to a whole spectrum of
modern engineering problems, their application to aerodynamics and propulsion
geared to airplanes and missiles is frequently encountered.
In this vein. it is useful to illustrate different regimes of compressible flow
by considering an aerodynamic body in a flowing gas, as sketched in Fig. i.3.
First. consider some definitions. Far upstream of the body, the florv is uniform
u,ith a free-stream uelocit.y of V*. A streamline is a curve in the flowfieid which is
tangent to the local velocity vector V at every point along the curve. Figure 1.3
illustrates only a ferv of the infinite number of streamlines around a body.
Consider an arbitrary point in the flowlieid, rvhere p. T, p, and V are thc local
p{-qssure. temperature, density. and vector velocity at that point. All of the ab6ve
quantities are point properties, and vary from one point to another in the florv. In
Chap. 3, we will shorv the speed oi sound a Io be a thermodynamic property of
tire gas; hence a also varies from point to point in the florv, if a_ is the speed of
sound irr the utriforn-r free stream. then the ratro V_/a_ def,nes the free-stream
.\{aclr rrumber fuf-. Simiiarly. the local Mach nunrber M is defined as M : V/a.
and varie"s.-from point to point in the flowfield. Further physical significance of
\{ach number irill be discussed in Chap. 3. In the presenr section, ,L1 simply r.vill
be used to de{ine four different florv regimes in fluid dynamics, as foilorvs:

liubsonic Flon'
Corrsrricr tire llori over an airfoil section as sketched in Fig, 1.3a. Here. the local
\{ach number is et,er_vrvhere less than unity. Such a flou,. u,here M < 1 at every
p{in-L and hence the flow i'eiocity is ever1,1f i1s1e less than the speed of sound. is
delined as .subsc.tttic fio*,. This flou, is characterized by smooth streamiines and
cor.ir.inuousil' r,arving properties. Note that the initially straight and paraliel
strcanrlirres in the free stream begin to dellect far upstream of the body. i.e.. the
llori' is forewarned of the presence of the body. This is an important property of
subsonic flow. and rvill be discussed further iir Chap. 4. Also. as the flou, passes
or,er the airfoil. the locaj velocity, and Mach number on the top surface increase
above their free-stream vaiues. However, it Mn is suflrciently iess than 1, the
local Mach number everyrvhere wili remain subsonic. For airfoils in common use.
il t\'I*. < 0.8, the flowfield is generally completely subsonic. Therefore. to the
airpiane aerodl,namicist. the subsonic regime is loosell, identified rvith a free
stream where M- < 0.8.

Transonic Florv
lf l'f - remains subsonic. but is sufficiently near 1, the flor,,. expansion over the
top surface of the airfoil may result in iocally supersonic regions. as sketched in
Fig. 1.3fi. Sucir a ryiXgd rqgi.on fiow is defined as transonic fon,. In Fig. 7.3b, M_
is less than 1 but high enough to produce a pocket of locally supersonic flow.
 CO|!,IPRESSIsLE FLOW-SOI\{E HISTORY AND INTRODUCTORY
THOUGT{TS 9

(a)

-b*
M_ < 0.8

4
0.8 <,rr- < I.0

r
I0<n1-<1.2

-- r, \ I 1

FIGURE 1,3
Illustration of different regimes of flow.
 l0 r!{oDERN coMPRESSTBLE FLow

In most cases, as sketched in Fig. 1.30, this pocket terminates rvith a shock rvave
across which there is a discontinuous and sometimes rather severe-change i.n flow
properties. Shock waves will be discussed in Chap. 4. lt M* is increased to
slightly above unity, this shock pattern wiil move to the traiiing edge of the
airfoil, and a second shock wave appears upstream of the leading edge. This
second shock wave is called lhe bow shock, and is sketched i+"__{S,- 1.1-..
(Referring to Sec. 1.1, this is the type of flot'pattern exrsting around the rving of
the Bell XS-1 at the moment it was "breaking the sound barrier" at l4*,: 1.06.)
In front of the borv shock, the streamlines are straight and para1lel, with a
unifolm supersorric free-stream Mach number. l$ passing through that part of
the borv shock which is nearly normai to the free stream. the flow becomes
subsonic. However, an extensive supersonic region again fonns as the IIow
expands over the airfoil surface, and again terminates u,ith a trailing-edge shock.
Both florv patterns sketched in Fig. 1.30 and c are characterized bl, mixed regions
of locally subsonic and supersonic flow. Such mixed flo*'s are delined as trunsonic
fo*,s. and 0.8 < M- < 1.2 is looseiy defined as the transonic reginte. Transonic
flow is discussed at lengtil in Qhap. 14.

Srrpersonic Florv
A florvfield where .D/ > 1 everyri'here is delined as supelsoriir'. Consitier the
supersonic florv over the u'edge-shaped body in Fig. 7)4. A straight. ohliqrre
shock wave is attached to the sharp nose ol the rvedge. Across this shock u'ave.
the streamline direction changes discontinuously Ahead of the slrock' the str:arl.i-
lines are straight, parallei. and horizontalt behind the shock thel'renrail straight
and parallei but in the direction of the wedge surface. Ur:like the subsonic flou'in
Fig. 1.3a, the supersonic uniform free stream is not forewarned of ihe presence oI
the body untii the shock rvave is encountered. The flou is supersonic both
upstream and (usually, but not ahvays) downstream of the oblique sliock ual'e.
There are dramatic physical and mathematical differences betueen subsonic and
supersonic florvs. as rvill be discussed in subsequent chapters.

Hypersonic Florv
The temperature, pressure, and density of the flor" rncrease almost explosively
across the shock rvave shorvn in Fig. i.31. As 1[- ts increased tcr ]iigher
supersonic speeds, these increases become more severe. At the same time. the
oblique shock wave moves closer to the surface, as sketched in Fig. 1.3e. For
values of M* > 5, the shock wave i9 v_egy clcse to the surface, and the flora.'field
betrveen the shock and the body (the shock la1'er) becomes verl' hot-indeed, hot
enough.!o dissociate or even ionize the ga-s- Aspects of such high-temperature
chemically reacting flows are discussed in Chaps. 16 and 17. These effects-tlrin
shc_':k la-r'ers and hot, chemically reacting gases-add complexity to the anaiysis
t-: such Sou's. For this reason, the flor,v regime for M- > 5 is given a special labei
The choice of M*: 5 as a dividing point..bet*,een supersonic
-ii_rpe.'soriir -fox'.
and hype.rso,," r;:: " .r"
il;-l: il ;.;;;';:::1*:
associated with hypersonic flow appear gradually as Mn is increased, and the
Mach number at which they become important depends greatly on the shape ol
the body and the free-stream density. Hypersonic flow is the subject of Chap. 15.
It is interesting to note that incompressible flow is a special q4W of subsonic
flow; namely, it is the limiting case where Mn '0. Since Mn : Vn/t,-, we have
two possibilities:

Mn - 0 because V* -'+ 0

M*t0becauseo*t&

The former corresponds to no flow, and is trivial. The latter slates that the specd
of sound in a truly incompressible flow would have to be infinitely large. This is
compatible with Eq. (1.6), which states that. for a truly incompressible flou' where
dp : 0r.z.must be zero, i.e., z.gro compressibility. we shali see in chap. 3 that the
rp.id of round is inversely proportional to the square root of r: hence r : 0
irnplies an i-nflnite speed of sound.
There are other rvays of classifying flowfields. For example. ilorvs s'here tire
efl-ects of visqosity, thermaJ. _co-nduction. and maqs 4illuslon are importanl arL-

calied urscoru;flox,s. Such phenomena are dissipative effects ivhicir change the
entropy of the flow, and are important in regions ol iargq.gfadients of ','elocitt'.
temperat-ule, and chemical .qomposition. Examples are bounda,ry l41'er flows. tlou'
in long pipes, and the thin shobk layer on high-akitude hypersonic vehicles.
Frictio.4 drag, flovi'field separarion, and heat transfer all involve viscous.-effeL)ts.
Therefore, viscous flows are of major importance in the study of flL-id dYnamics.
ln contrast, flows in which viscosity, thermal conduction, and diffusion are
ignored are called i\!ry!!!-flows. At first glance. the assurnption of inriscid flou'
ma),appear highlyHifr-Ctive; however, therc are a number of importrnt applica-
tions which do not invoive flows with large gradients. and which readrh can be
assumed to be inviscid. Exarnples are the large regions ol flo'r'over rvings arrd
bodies outside the thin boundary laler on the surface. flo*'through uind tunnels
and rocket engine nozzles, and the flow over compressor and turbir-re blades for
jet engines. Surfaqo*- p;essure distributions. as weil as aerqdl'nan-ric lift and
momei*q on some bodies. can be accurately obtained by means of tire assump-
tion of inviscid flow. In this book, viscous effects r,r'il} not be treated except in
regard to their role in forming the internal structure and thickness of shock
waves. That is, this book deals with compr-essible, inuiscidfiort's.
Finally, we will always consider the gas to be a c--o.lilll1ll.ul11. Ciearl-v-'' a gas is
compcscd of a large number of discre-te atpms and/or qlole".Qules. all moving in a
more or less rando-m..fashion, and frequently colliding lr'ith each other. This
microscopic picture of a gas is essential to thc understanding of the thermody-
namic and chemical properties of a high-tempelature gas, as described in Chaprs
16 and 17. However, in deriving the fundamental equations and concepts ['rr
fluid flows, we take advantage of the fact that a gas usuaIl1' contains a I'rrllc
 12 MoDERN coN{PREssnlLE FLow

number of molecuies (over 2 x 101e molecules/cm3 for air at normal room

conditions), and hence on a macroscopic basis the fluid behaves as if rt were a
continuous material. This continuurn assumption is r"iolated only when the mean
distance an atom or molecule moves between collisions (the mean free path) is so
large that it is the same order of magnitude as the characteristic dimension of the
flow. This implies lqg density, or raryfiedflow. Theextreme situation, where the
mean free path is much larger than the characteristic length and where virtualiy
no molecular collisions take place in the flow. is called free-molecular fiow.In this
case, the flow is eissentially a stream of remotely spaced particles. L.ow-density
and free-molecular flows are rather special cases in the rvhole spectrum of fluid
dynamics, occurring in flight only at very high altitudes (abo;eJp-0^"000 ft), and in
special laboratory devices such as eiectron beams and low-pressure gas lasers.
Such rarefied gas effects are beyond the scope of this book.

I.4 A BRIEF REVIEW OF

THERMODYNAMICS
The kinetic energy per unit n'rass. Y2/2. of a high-speed lloil, is large. As the florv
moves over solid*bodies or through ducts such as nozzies and diffusers, the local
velocity. hence iocal.kipetic energy, changes. In contrast to low-speed or incom-
pressibie flow. these energy changes are substantial enough to strongly interact
with other properties of the flow. Because in most cases high_-.gpeed -flow and
compressibl_e flow are synon-ymous, elergy concepts play a rnajor role in the
stud1, and understanding of compressible flow. In turn. the sci.ence of energy (anC
entropy) is therntodynamics, consequently, thermodynamics is an essential ingre-
dient in the study of compressible flow.
This section gives a brief outline of thermodynamic concepts and relations
necessary to our further discussions. This is in no u'a1' an exposition on thermo-
dvnamics: rather it is a review of oniy those fundamental ideas and equations
*,hich ivi1l be of direct use in subsequent chapters.

Perfect Gas
,A.gas is a collection of particles (molecules. atoms. ions. electrons. etc.) rvhich are
in more or less random motion. Due to the eiectronic structure of these particles,
a force field pe-rvades the space around them. The force field due to one particle
reaches out and interacts with neighboring particles. and rice versa. Hence, these
lelcis are called interntolecular torces. The intermolecular force varies with
distance betrveen particles; for most atoms and molecules it takes the form of a
.-r.eai' attractive force at iarge distance, changing quickly to a strong repelling
j:'rce *t close distance. In general, these intermolecular forces influence the
:::;r1,i;*n ol the particles; hence they also influence the thermodynamic properties
,:: tnt eas. rr'hich are nothing more than the macroscopic rami{ication of the
- -.::i::a rlilt:on.
 A,,he
"*;;;;ffiffiH;;" ;, ";;ffi;
applications, the gas particles are, on the average, widely separated. The average
dista-nce betyeex paI{iqies is usually more than 1Q--q1-p,.!ecglal-.diame-ters, which
-,:

corresponds t0 a very weak attractive force. As a result, for a large number of

engineering applications, the effect of intermolecular forces on the gas properties
is negligible. By definition, q,p-erLect gps is one iry whigh intermolecular lorces are
ryeglected. By ignoring intermolecular f,orces, the equation of state for a perfect
gas can be derived from the theoretical concepts of modern statistical mechanics
of kinetic. theory. However, historically it rvas first synthesized from laboratory
measurements by \-oFert Boyle in the seventeenth century, Jaggg-e-.9 Challes in the
eighteenth century, and Joseph Gay-Lussac and John Dalton around 1800- The
empirical resuit rvhich unfolded from these obsen'ations was
p7': t{RT (1.7)

n'herep ispressure(N,/m2 orlb/ftz),72'isthevolqlr-9.9f thesvstem(m3orft-').

,1{ is 'lhe mass. of the -system (kg or slug). R is the speq$r-g45 constant
tJ/{Ue.K) or (ft.lb)/(slug'"R)loR).r which is a differe.lit value for different gases.
and f is the temperature (K or This equation of state can be u'ritten in
r-uan), forms, most of u,hich are summarized belou' for the reader's corrvenience.
For example. if Eq. (1.7) is divided by the mass of the s)'stcm.

(t.s)

n,here c is the spe*.ific',,olume (*3,Zkg or ft3lsiug). Since the densit5 p : 1 /,,

Eq. (1.8) becomes'
: PRTl, (1.e)

Along another track .r'hich is particularh' usefui in chemicalil' reacting

systems, the early fundamental empirical obsen'ations also led to a form for the
equation of state as folloivs:
p7": -4'27 (1.10)
rr'Irere thc num.}e1 of moles of gas in the svstem. and @ is the universal gas
.,{.is
coltstant. which is the sameJ-gr all gases. Reca]l that a mole of a substalce is that
amount thich contains a mass numericall;' equal to the rnolecular rveight of the
gas. and rvhich is identifled rvith the particular s-vstem of units being used, i.e.. a
kiiogram-mole (kg' mo1) or a slug-.mole (slug' mol). For example. for pure

iTu.o sets of consistent units uil1 be used througirout this book, tiie International System (SI) and lhe
English Engineering System. In the SI system, the units of force, rnass, length. time, and temperature
are rhe ne*'ron (N). kilogram (kg), meter (m), second (s). and Kelr'in (K). respectivelv: in the English
Engineering S)-'stem they are the pogg;!,(Lb), siug, foot,(ft). second (s), and Ranline ("R). respectivelv.
The respective units of energv are joules (J) and foot-pounds (ft lb).
 14 MoDERN coMPRESsIBLE FLow

diatomic oxygen (Or), }-kg I mol has a mass of 32 kg, whereas 1 slug. . mol has a
mass of !-2 slug. Because the masses of different molecules are in the same ratio
as their molecular weights, 1 mol of different gases always contains the same
numbe'I'_of .molecules, i.e., 1 kg'mol always contains 6.02 x 1026 molecuies,
independent of the species of the gas. Continuing with Eq. (1.10), dividing by the
number of moles of the system yields
p"/., : gT (1.11)
v,here /r' is the molar volumelr*71Xg. mol) or ft3/(slug . mol)1. Of more use in
gasdynamic problems is a form obtained by dividing Eq. (1.10i by thg mass of
the system:

(1 .r 2)

where u is the specific volume as before, and 4 is the mole-mass ratio

t(kg. n,qlkg and (slug ' mol)/slugl. (Note that the kilograms and slugs in tlre
above units do not cancei. because the kilogram-mole and siug-mole are entities
in themselves; the "kilogram" and "slug" are just identifiers on the mole.) Also,
Eq. (1.10) can be diyiC.d -by the systeu volume, yielding
P: CzT (1.13)
where C is the concentration [(kg . mol)/mr or (slug - mol)/ftr].
Finally, the equation of state can be expressed in terms of particles. Let ,\',
be the number of particles in a mole (Avogadro's number, *'hich for a kilograin-
mole is 692 x 1026 particles). Muttiplying and dividing Eq, (i.l-1) by,Y.,.

p: (N^c)(#), (r .i4)

Examining the units. 1V,.{ i.! physically the number density (number of
particles per unit volume). and 9/N, is the gas constant per particle. u'hich is
precisely the Boltzmaln constant k. Hence, Eq. (1.14) becomes
P:nkT (i.1s)
where n denotes number density.
In summary, the reader will frequenily encounter the different forms of the
perfect-gas equation of state listed above. However, do not be confused; they are
all the same thing and it is wise to become familiar with them all. In this book,
particular use will be made of Eqs. (1.8), (1.9), and (1.12). Also, do not be
confused by the variety of gas cor,stants. They are easily sorted out as follows:

1. When the equation deals with moles, use the unive_rsal,gas constant, which is
the "gas constant per mole." It is the same for ali gases, and equal to the
follcwing in the two systems of units:
g: 8314 J/(ke .mol . K)
g: 4.97 x t04 (ft . Ib)/(slug . mol ' 'R)
 An*D INTRODUCTORY TII0UaHTS l5
COMPRESSIBLE FLOW-SOL{E HISTORY

the specific'gas constant R' which is

2. When the equation deals with ma-ss' use for different gases' and is
the "gas consta4t p"1 "*f *uil'-It is different
related to the ,rrri,J^* gas constant ' R
: q/'/'{' where
"//
is the molecuiar

R: 287J/(kg ' K)
R: 1716 (ft ' lb)/(slus' "R)

3.Whentheequationdealswithparricles'usetheBoltzmann-constantk'*'hich
is the "gas constant per particle":
k:1.38 x7A-23 J/K
k : 0.565 x 10-23 (ft'lb)/'R
gas? It has been-experimentaliy
How accurate is the assumpdon of a perfect
1 atm or less) u'-djt^ hig"['-temperatures
determined that, at-.iow prgr.u'it (ne41
(starrdard temperature,,fiiK, and above)' the vaiue py/RT for most pure gases
by l_ess than 1 percent. However, at very cold
temperatures
a"ri"i", from unity gas are more ciosely packed together'
and high pressures, tie molecule' of tht
more *important' Under these
and consequentiY irri.t*ol*fur fo-rces become
such cases. the perfect gas equation
conditions, the gas i, i"n".a as a real gas. In
ofStateraustberep}acedbymoreaccuraterelationssuchasthevan-derWaais
equation
(1.16)
{0.#it,-o):nr
and b _areconstants that depend on the
a type of gas. As a general rule of
where
of state \ ar) approximately as
thumb, deviations from the perfect gas equation
lhe temperatures arrd
p/T3.In the uur, *u;o.iiyir gutaj"amic. applications'
'J."rrur., are :
srch thaii; be applieri with confld'ence. Sucl.r willbe the
pRr-can
case throughout this book'
Intheearlylg50s,aerocynamicistsweresuddenlvconfrontedrvittrhypcr-
26,000 ft,zr it km/s). The shock layers
sonic enrry vehicles at uetocitles as high as
reactions in the airfloiv
about such vehicles were hot enough to cause chemical
in the aerody-
fJir.o"iutio.r,ionization, etc.). At th-at time, it t'ecanre fashionable
"reai gas However' in
namic literature to denote such conditions as "qt:t:'"
classical physical a re.al 8as !q defined ui ot" in rvhich intermolecuiar
"f,"*iri,y,
f-or-ces are importarrt, ut'a if'" a"n"tti"n is completely divorced from the idea of
ho:'.t followed such a
chemical reactions. In the preceding paragraphs' "l:
gut, us will be discussed at length in
classical definition, nor a ctremicalty rJacting
a mixture of perfect gases'
6i"p. io, most problems can be trlated by assumingbecause
where rhe relation i-: p|i still hoids. i1o*"r.r, B 7 9/-,l and (
variesduetoihechemicalreactions,thenRisavariab'lethroqghouttheflow.It
pr,"no*ena as "real gas effects," and
*;"r;il, ;;;;;;;:;iio ta"",iry,ru"n we wili deal with "chemically
this term will not u" us"d in this book. Rather,
mixtures .f p;;;;;'ses," $'hich are the subject
of Chaps' 16 and 17'
reacting
16 MoDERN coMPREsslBLE FLow

Example 1.2. A pressure Yjeqqgl that has a volume of 10 nf is used to store

ai, io. op.ratlng a supersotic wind tunnet. If the air pressure and-
higb-pressure
t"irp.r",,rr. inside the u"sset ate 2.q.ai49-an-
g l-tt1K' respectively' what is the npss of
qlr stored in the vessel?

Solution Recall that 1 atnr : 1.01 x 10s N/r/' From Eq' (1'9)

p: #: "?ji;3i#1"' : n a6kg/nt

The total mass stored is &en

M : 7.p : 0A)(23.46)
:
F;t;l
Example 13. Calculate the isothermal compressibility for air at a pressure of
0.5 ator.

Solution From Eq. (1.3)

1/du\
rr: -;\A),
From fu. (1.8)
RT
u- p

Thus
I 0u\ RT
l-:-l
\ dP )r P-

Hence

ar:-:l#),:-(#)(
Weseerhattheisotheruralcompressibilityforaperfe.ctgasissimplythereciprocal
?:)
of the pressule:
11
p 0.5
In terms of the Internatignal System of units, where p : (0'5X1'01 x 105) :
5.05 x 104 N/rrf,
1.98 x i0-s mzA
In terms of lhe English Engineeriog System of units, where p : (0'5X2116) :
10581b/tt2.

,T 9.45 x L0-4 ft? lb

 HrsroRY AND INrRoDUcroRY rHou(iHrs 17

Internal Energy #ff;Lo\\'--soME

particles in random
Returning to our microscopic view of a gas as a collection of
motion, tire indiX[g[ue!]<inetic energy of each- pSrticle contributes to
the ovg1qll
of the g;;'M*ri.orer, if the particle is a molecule, its.rotational and
"r"rgy *otion. (see Chap. 16) also contribute to the gas"e.-nergy' Finally' the
viUra'iional
small
*otio, or _"J_""tron, in both aioms'and molecules is a sourc_e-,9f -e..uergy.This
to a massive portrait in
sketch of atomic and molecular energies will be enlarged
can consist
Cnup. f e ; it is sufficient to note here that the energy--of, 4.Plrticle .of
In turn, energies, summed over all the
,*.*f different forms of motion. these
gas'.Moreover' if the
paiticles of the gas. constitutethe ltternal energ), e, of the
particles of the gas (called the stu/em) are rattling about in their state of
l.mu*im,r* disorder" (see again ihap. t6;. the system of p.rrticles will be in
eQuilibrium'nnow
to the macroscopic vierv of the gas as a continuum. Here,
temperature' and
equ-[ibrium is evidenced by no gradients in velocity' pressure'
the system, i.e., the sYstem has uniform
chemical concentratlons tirroughout
properties. For an equilibrium iyttt- of a real gas where intermg-!9cul1r folces
and alio for an equilibrium chemically reacting mixture of
perfect
,..i-po.,unt, Let e
gur"., ih. internal energy is a function of both temperature and volume'
(internal energy per unit mass). Then. the
i..,o," the specific intJinal energy.
:
enthatpy. /l, is defined. per unit mass' as l-t e '* pu' and we have

e : e(7, t,)

h : h(r. p) (1.17)

lor both a real gas and a chemicaily reacting mixture of perfect gases'
Ifthegasislrolchemicaiiyreacting,arrdifweignoleintermolecularforces,
and enthalpy
the resultin{system is a thermalll' peyfect gas. where internal elergy
are functions of temperature only' and r'r'here the specil&*b9.ats at constant
volume and pressure. c,. and cp. are also functions of temperaiule o$ly:

e: e{T)
h : h(r)
tle : c,, dT

dh: crdT (i.18)

The temperature variation of c,. and c, is associated with the vibra"tional an<i
electronic motion of the particles, as wiil be explained in Chap. 16.
Finally, if the specffic heats "are constant, the system is 4ucalorically perfect
gas, where
^_ Lt,t
e -^.r
h: coT (1.1e)

In Eqs. (1.i9), it has been assumed that h : e :^".0*?! T : 0'

l8 MoDERN coMPRESsIBLE FLow

In many compressible flow applications, the pressures and temperatures are

moderate enough that the gas can be considered to be caloric-ally perfect. Indeed,
there is a large bulk of literature for flows with consunt specific heats. For the
first half of this book, a qglggg44y_p,95fect.ga9 wil! be assumed. This is the case
for atmospheric air at lgmpglgtpls*}glgw.-_10@.K. However, at higher tempera-
tures the vibratiolal...motion of the O, and N, molecules in air becomes
important, and the air becomes tl_r-g5mally ppffecq with spec-ific heats that vary
wiih iemp.rature. Finally, when the temperature exceeds 2500 K, the O, molecules
begin to diss-octte into O atoms, and the air becomes chemicSlly reaqJinS. Above
4000 K, thffi.'molecules begin to dissociate. For these chemically reacting cases,
from fqs. (f .11), S'.{gpegdl on both-I anC. u, and fr depends on both T and p'
(Actually, i" equitiiiiium iiiermodynamics, any state variable is uniquely deter-
mined by any two other state variables. However, it is convenient to associate I
and u with e, and I and p with fr.) Chapters 16 and 17 will discuss the
thermodynamics and gasdynamics of both thermally perfect and chemically
reacting gases.
Consistent rvith Eq. (1.9) and the definition of enthalpy is the relation
t.:' cr-c,:R (1'20)

where the specific heats at.constant pressure and constant volume are defined as

0h t
0T)p
-I

_l
0e't
and ATJ,
respectiveh. Equation (1.20) holds for a caloricaliy p.erfect or a thermally perfect
gas. Il rs, frqf ;3li$ for either a chemicgliy r-eacting or a regJ gas' Two useful forms
of f,q. (1.20.; can be simply obtained as follows. Divide Eq' (1'20) by co:
c.. R
(1.2i)
cp cp

Define y = cr,/ct.For air at standard conditions' y: i'4' Then Eq' (1'21)

becomes

1--:-'l1R ("
Solving for cr,

(1.22)

Similarly, by dividing Eq. (1.20) by c,,, we find that

(1.23)
Equations,r.rr, -;;;;ffi ;:X;'; ffiJ;;;;
subsequent treatment of compressille. flow'
t':;
wiit Ue useful in our
Example 1.4. For the pressure vessel in Example 1.2, calculate the total internal
energy of the gas stored in the vessel.

Solution. From Eq. (1.23)

: R : 281 : 717 's l/kg' K
', 7-1 T;1
From Eq. (1.19)
e : r,T: (717.5)(100) :2.153 x 105 i/kg
From Example 1.2. *e caiculated the mass of air in the vcssel to be 234.6 kg. Thus'
thc totai intcmal cncrgr i.
r------ -_l
r,: .\'te: (234.6)(2.1-<3 x 10') : i5
05 x 10'J I

First Larv of Thermodi'namics

Consider a st'slcr71. ivhicir is a {rxe<l mass tlf gas Separated from the surroundings
by a flexible bounciarv. For the time being. assume the system is stationary, i'e'. it
has no directed kinetic energy. Let 6q be an incremental amount of heat added
to the systenr across the houndary (say bl"direct radiation or thermal conduction)'
Also. let 6rl denote the u'ork done on the svstem by the surroundings (say by a
displacernent of thp boundan. sOg-qezing the volume of the systen to a smalier
value). Due to thc uro.leculei itrutldll ol thr- gas. rhe systcm has an iq19rna1 ele-rgy
e- (This is the specific internal energy if *'e assume a system of unit mass.) The
l1-eapd{ed ald yqlk dglg ori th-e sytl-elll c4use a change in eneqgy, and since the
s)'stem is stationary, this c,lr-ange iq Engfgy is simply de:
:6q :,ie (1.24)
- 6u I

Tlris is the fr.it lav' o-f rhermod)'nanilcs: it is an empirical result confirmed b1'
laboratory an<1 practical experienr:e. In F-q (1.24), e is a state variable- !Je-nce' r/e
is aa exact ,lifferential. ancl its yalue depelds only on the initial a4d fi43i states of
the sl,sren-r. In contrast. 69 and 6N depend on the prg-"cess in going from the
initial and flnal states.
For a given de. there are in generai an infinite number of difi-erent rvays
(processes) by u,hich heat can be added and u'ork done on the system. We will be
prirharily concerned rvith three types of Processes as follow:

l. Adiabatic process- otre in rvhich.go.kg!. is added to or taken away from the

system
2. Reuersible process_ one in which qo dissipative phenomena occur, i.e.. rvhert
the effects of v-iscositl'. thermal conductivit)', and ntass diffusion are absent
3. Isentropic proce.es-one u'hich is both adiahatic and reversible
 20 MoDERN coMPREssrtsLE FLow

For a reversible process, it

can be easily proved (see any good text on
thermodynamics) that dy.:,.:p dutwhere du is an incremental change in specific
volume due to a displacement of the boundary of the system. Hence, Eq. (1.24)
becomes
Eq - pdu: de (1.2s)
If, in addition, this
process is also a.{jqb-a1ic (hence isentropic), Eq. (1.25) leads to
some extremely use${_.fie*gpodynamic formulas. However, before obtaining these
formulas, it is usefui to review the concept of entropy.

Entropy and the Second Larv of Thermodl'namics

Consider a block.of lce in conlact with a red-hot plate of steel. Experience tells us
that the ice rvill warm up (and probably melt) and the steel plate rvill cooi doun.
Hor*'ever. Eq- (1.2a) does not necessarily say this *'i11 happen. Indeed, the Iirst
lqlz-a-llows that the ice may get cooler and the steel plate hotter-j*st-.a.s long as
e3grgy is conserved ciuring the process. Obviously, this does not happen; instead-
nature imposes another condition on the process. a condition rvhich tells us in
w.ht_gh directior a process will take place. To ascertain the proper direction of a
process, let us define a new state variable. ih-e entropr'. as follorvs:

,
as:
64.".
T
where s is the entropy of the system, 6q.., is an incremental amount of heat
added reversibiy to the system, and I is the system temp€iature. Do not be
confused by the above definition. It defines a change in entropy in terms of a
reversible addition of heat, 6q..,. Horvever, entropy is a s,tqtq variable. and it can
be used in conjunction with alJ type of process, reversible or irrerersible. The
quantity 6q,., is just an artifice; an effective value of 6q,".. can alrvays be assigned
to relate the initial and end points of an irreversible process. u'here the ac'tual
amount of heat added is 6q. Indeed. an alternative ancl probabll' more lucid
relarion is

(1.26)

Equation (1.26) applies in general; it states that the change in entropy during an1'
incremental process is equal to the actual heat added divided by the temperature.
6q/7, plus a contribution from the irreversib-le dissipative phenomena of vi-sc"os-
ity, thermal c*onductivity, and mass.diffusion occurring wirhin the system. dr,-"-,.
These dissipative phenomena always increase the entropy:

(1.21)

The equal sign denotes a reversib-1.e. process. where by definition the atrore
dissipative phenomena are absent. Hence, a combination of Eqs. {7.26) and (1.27)
 coMpRESsIBLE FLow-soML HISToRY AND INTRoDUCToRY TIIouGHTS 2l

yields

(i.28)

Furthermore, if the process is adiabatic, 6q : 0, and Eq' (1'28) becomes

(1.2e)

The
Equations (1.28) and (1.29) are forms of the second lly'.9f .t"b?f!1lgdynanics'
s"cond law'te1ls us in what direction a process will take piace. A process will
proceed in a direction such that the entrpPY of the svstem plus surroundings
alrtay!-increases, or at best stays the same' In our example at the beginning
of
to be both the ice and steel plate combined' The
ttrls seciion, consider the system
simultaneous heating of the ice and cooling of the plate f ields a net inc-rease
irl
eqlfopy fot the system. On the other hand, the impossibie situation of the ice
g"tti.rg cooler and the plate hotter would yield a net de-crease in .entropy'. a
iir,rutL, forbidden by the second Iaw. In summary, the concept of entropy in
combination with ths second law allorvs us to predict the direction that nature
takes.

Calculation of EntroPS'
Consider again the firs_t law in the form of Eq. (1.25). If lr'e assunle that the heat
:
is reversibli, and \\,e use the definition of entropy in the form 64.., IJ'r, then
Eq. (1.25) becomes
Tds-pdu:de
l1f at: de + pdtl (r 30)

Another form can be obtained in terms of enthalpy. For example. h1 definition

h:e*pu
Differentiating, \r'e obtain
dh:de*pdu*cdP (1 31)

Combining Eqs. (1.30) and (1'31), we have

Tds: dh - udp (1.32)

Equations (1.30) and (1.32) are important, and shouid be kept in mind as much
as the original form of the first law, Eq. (1.24)'
For-a ther_mally perfect gas, from Eq. (1 18). rve have dl!: ,odr. Substitu-
tion into Eq. (1.32) gives
dT udo
ds:co, -i (1.33)
22 MoDERN coN{PREssTBLE FLow

Substituting the perfect gas equation of state pu : RT into Eq. (1.33), we have
. c,dT
ds: Rdp (i.34)
, - p
lntegrating Eq. (1.3a) between states 1 and2.

sr-sr: I
uT,
fT:
co {I - *rn!,Pt (1.35)

Equation (1.35) hoids.fgr a thermally perfect gas. It can be evaluated-if" c, is

known as a function ol T.If we further assume a caloricallyper{-ect gas, u'he-r"e-.co
is constant, Eq. (1.35) f ields
.r
:c I' in*-Rln- (1.36)
I1 PT

Similarly, starting with Eq. (1.30). and using de: c,clT. the change in
entropy can also be obtained as

.r--.r,:c' ln2*^rn?
Ir I'i
(1.37)

As an exercise, shou,this yourself. Equations (1.36) and (i.37) allow the calcula-
tion of the ch4gge*in-eltropy betrveen two states of a calori-callv pe-rfect gas in
terms of either the pressure and temperature. or the volume and temperature.
Note that engopy.ls a function of borlt p and T. or {.'ar.rd I. even for the simpiest
case of a caloricalll"perfect gas.

Example 1.5. Consider the air in the pressure vessel in Example 1.2. Let us no*' heat
the gas in the ve-ssel. Enoug! h,eat is added to increase the tenrperature to 600 K.
Calculate the change in entropy of the air inside the vessel.

Solution The vessel has a constant volume: hence as the air temperaturc is
increased, the pressure also increases. Let the subscripts 1 and 2 denote the
conditions before and after heating. respeclivelr-. Then. from Eq. (1.8).
Dt T, 600
pt Tt 300

In Example 1.4. we found thal c,.:7L'1 .5 J/kg'K. Thus, from Eq. (1.20)
tp: c,.+ R:717.5 + 28'l : L0A .5 l/kg'K
From Eq. (1.36)

r: sr : ,r^+ - R,,X
: 1004.5 1n2 - 281 ln? : 491 .3 I /kg ' K
From Example 1.2. the mass of air inside the vessel is 234.6 kg. Thus, the total
 INTRODUCTORY THOUGHTS 23
COMPRESSIBLE FLOW-SOME HISTORY AND

entropy chaage is

: M(sz- sr) : (234'6)(497'3) : i.167 x 105 J/K

S, - S,

Isentropic Relations
An isentropic process was deflned above as adiaba"tiq and revelsible' For an
:
u-aiuuutic piot"is, 6g 0, and for a reversible process, dr;o', : 0' Hence' from
Eq. (f.ZO), an isenti_opic_.process is one in which d{:0, r.e.. the enlroBy is

c-onstallt..
i*portunt relations for an isentropic process can be obtained directly from
fq.s. (1.36) and (1.37), setting rz: rr-For example' from Eq' (1'36)

o:ror"t-Rhb
c-
,"i:
n-
iil"T
T"

, (.-/R
-tz '
Pz [ \' (1.38)
pr:- \T )
Recaiiing Eq. (1.22),
', : 'l
R v-1
and substituting intb Eq. (i-38).
D- ,r.'rt/Q-t)
r! I i I
(1.3e)
n.
rL l7'l
\ r,/

Similarly, from Eq. (1.37)

T. l')
0:c, lnt'+Rlnn
,"i,.- : - i.i,E
c. T"

c,/R
u2 t/ r1z\\
,, : \r,/
(1.40)

From Eq. (1.23)

cr, 1

R:t_r
Substituting into Eq. (1.40), we have

',
t)t
: (L\-""-" (1.41)
\fr/
U N{oDF-RN coMPRESSI}lLE I--Low

Recall that pr/pr: ut/uz.Hence. from Eq. (1.41)

9r: lT, lLtr'--nt

O1
lrl (1.42)

Summarizing Eqs. (1.39) and (1.42).

i ol \'I Trlrttr-t'
\p,/ -\r,t
Pz ( 1.43)
ttn-

F4uation (1.43) is jmportant. It relates pressure. density, and temperature for at.t
i-sentropic process, and is very frequently used in the analYsis of compressihle
flows.
You might legitimately ask the questions r'''/i-r'Eq. (1.43) is so important'
and wfty it is frequently used. Indeed. at first thought the concept of an isentropic
process itself may seem so restrictir,c-adiabatic as u'ell as reversibie-that olre
might expect it to find only Iimited applications. Hou'ever. such is not the case.
For example. consider the florvs over an airfoil and through a rocket engine. In
the regions adjacent to the airfoil surface and the rocket nozzle wails. a boundary
layer is formed u'herein the dissipative mechanisn.rs of viscosit!. thermal conduc-
tion. and diffusion are strong. Hence. the entropv irtcreases uithin these bound-
ary layers. On the other hand, consider the tluid elements outside the boundary
layer. where dissipar:ve effects are negltgibls. \4e1scrver. no heat is being added or
taken away- from the fluid eiements at ti-i.sc pt'ritt: - itctiee. titc flot is adiabatic.
As a resuit. the fluid elements outside the boundary ia1'er arc e.tperiencing
adiabatic and reversible processes-namel-v. isentropic flon. Moreover. the vis-
cous boundary layers are usr,rall1' thin. hence large regiotrs ol the tlowfields are
isentropic. Therefore. a studl of isentropic flous is dircctlr appiicable to manY
types of practical florv problems. In turn. Eq. (1.a3) is a pou'erlul relation ior
such fious. r,alid for a calorically perfect gas.
This ends our brief rcvics'of thermodlnanrics. Its purpose has been to give
a quick summar), of ideas and eqr-iatiotts u,hich u'ill be ernploVed throughoui our
subsequent discussions oi compressihle flou. Aspects of the thermodl'narnics
associated u,ith a high-temperature chemicall-v- reacting gas ri'ill be developed as
necessarv in Chap. 16.

Example 1.6. Consider the florv through a rocket engine nozzle. Assume that the gas
flow through the nozzle is an isentropic e.xpansion of a caioricalJ,v.pelfept gas. In th':
combustion chamber, the gas which results from the combustion of the rocket fuel
and oxidizer is at a pressure and temoerature of 15 atm arrd 2500 K, respectively:
the moiecular r#eight and spec.ific heat at constant pressure of the combustion gas
are L2 and 4151 t /kg ' K, respectively. The gas expands to super.sonic' speed
through the nozzle, with a temperature of 1350 K at the nozzle exit. Calculate thc
pressure at the exit'
 AND INTROI)UCTORY THOU(iHTS 25
COMPRESSIBLE FI-OW-SOME HISTORY

SolutionFromourearlierdiscussionontheequationofstate,

R: 3:: i# :6e2 8 l/ks K

From Eq. (1,20) '

c., : cp - R : 4L5'l - 692'8 : 3464 l/kg' K

Thus
cr 4757 _11
1: q:1464-',"
From Eq. (1.43), u'e trave

P: I Tr\""-" :-\zsoo/
1ll4lr:'rr " :o.rro*
P':\T/
p, :0'025Pt: (0.024s)(15 atm) : i 0 372 atm
I

1.5 AERODYNAMIC FORCES ON A BODY

Thehistcryoffluiddynamicsisdominatedbythequesttopredictforcesona
uo.iy*oulngthroughuRuia-'upsmovinglhroughwater.andinthenineteentl.I
exatlples'
to name.lust a ferv
and twentieth centuries aircraft morir"rg through air.
in his Principiu (1687) u'as oriented in
Indeed, Newton's treatment of fluid flow
on an inclined surface The calculation of
puri ,o*rras the prediction of forces
remains a L-entral thrust of nrodern
I"roayna*ic and-hydrodl'namic iorcer-stil1
llori..ri'hich governs the
nuia iyr."*ics. Thii is eipecially true for_compressible
supersonic. a.d
a..oayrra*ic lift and d.a! on trigrr-speea srrbsonic. rransonic.
Therefore. in se'eral seclions of this book' t]re
hyperionic airplanes, and fiissiles.
rvili be applied to the practical calculation of
fundamentals of compressible flow
aerodynamic forces or high-speed bodies'
force to a
The mechanism b1: which nature transmits an aerodlnamic
from on11.trr'o basic sources: surface
surface is straightfo*,u,d. This force Stems
consider. for example, the arrfoil of unit span
pressure and surface shear stress.
sketchedinFig.l'4.Letsbethedistancemeasuredalongthesurfaceolthe
are functions
airfoil from the nose. In general, the pressure p and shear stress-r
otfs;p:p(s)andr:-"1,;.Th.'.pressureandshearstressdistributionsare force to the
the only *.ur. that nature has to communicate an aerody'namic
consider an elemental surface area.dS on rvhich is
airfoil. To be more specific,
exertedapressurepactingnormaltodSandashearStressTactingtangentialto
as, ,, sketched in r.ig. i.+. l.t n and m be unit vectors perpendicular
and

purattet, respectively, Io tire elemint dS' as shorvn in Fig'

1'4' For future
a vector dS n dS; hence dS is a vector
discussion, it is convenient to define =
1'4' force
normal to the surface with a magnitude dS' From Fig'
the elemental
dF acting on dS is then
dF: -PndS + rmcls: -PdS -' rmdS (i'44)
26 MoDERN coN'IPRESsIBLE FLow

'f

FIGL'RE I.4
Sources of aerodynamic force; resultant force and its resolution into lift and drag.

Note from Fig. 1.4 tltat p acts toward the surface, whereas lS
: n dS is directed
away from the surface. This is the reason for the minus sign in Eq. (1.44)- The
uai aerodynamic force F acting on the complete body is simply the sum of all
the element forces acting on all the elemental areas. This can be expressed as a
surface integral, using Eq. (1.44), as follows:

r : Sar : -fitds + fftm ds (1.45)

On the right-hand side of Eq. (1.45), the first integral is the pressure force on the
body, anj the second is the ihear, or friction force. The integrals are taken over
the complete surface of the bodY.
consider x, y, z orthogonal coordinates as shown in Fig. 7.4.Let x and y
be parallel and perpendicular, respectively, to v*. lf F is the net aerodynamic
force from eq. 1i.+S;, then the yft L and drag D are defined as the components
of F in the 1 and x directions, respectively. In aerodynamics, z- is called .the
relatiue wini, and lift and drag are always defined as perpendicular and parailel,
respectively, to the relative *ird. Fot most practical aerodynamic shapes, I is
generated mainly by the surface pressure distribution; the shear stress distribu-
t-ion g.n".ully makes only a small contribution. Hence from Eq. (1.45) and
Fig. 1.4, the aerodynamic lift can be approximated by

t = y componenr of l- #r rr1 (1.46)

With regard to drag, from Eq. (1.45) and Fig. 1.4,

D: xcomponent ,rf-fuas] + , component "r\ff,^atl 0.47)

pressure drag skin-friction drag
 rn this r*-, ;;;;*,;;il, ;;:#;#ffi.:T r:
1.3. For many bodies, the inviscid flow accurately determines
the surface pressure
with Eq'
distribution. For such bodies, the results of this book in conjunction
is due both to
(1.46) allow a reasonable prediction of lift. on the other hand, drag
pressure and shear strer" distributions via W' 0'47)' Since
.we
will not be
fonsidering viscous flows, we will not be able to calcrtlate skin friction drag'
Moreover, the pressure drag in Eq. (1.47) is often influenced by flow separation
from the body-also u uir*u, effect. Hence, the fundamentals of inviscid
compressibl" ffo* do not lead to an accurate prediction of drag for many
situations. However, for pressure drag on slender supersonic shapes due to shock
we
waves, so-called waue drag, inviscid techniques are usually quite adequate, as
shall see in subsequent chaPters'

Example 1.7' A flat plate with a chord length of 3 ft and an iofinite span

lperpendicular to the page ir Fig. 1.5) is immersed

in a Mach 2 flow at standard sea
plate
level conditions at an anlle of attack of 10o. The pressure distribution over the
:
is as follou,s: upper surface, pr const : ll32 1b/ft2; lower surface, 1: : const :
3568tb/ff . ffr" l""a shearstress is given by r- : l3/to'', where_r*, is in pounds
per square feet and { is the distance in feet alonB the plate from the leading edge'
Assume that the distribution of r,. over the top and bottom surfaces is the same'
Both the pressure and shear stress distributions are sketched qualitatively in Fig.
1.5. Calculate the lift and drag per unit span on the plate'

Solution Considering a unit sPan,

- ffo os : l,' o,
at + tt u], : [ - (1132)(3) + (:s0r)(l)ln: 7308n
l* lo3

From Eq. (1.46)

: )' component or : :
#r rt1
L - 7308 cos 10o per unit span
I @
From Eq. (1.47)

drag: wave drag = D. : x .omponenr .t rt]

Pressure
[- ffr
Hence

4 :73o8sin1oo: perunitspan
@
Also from Eq. (1.46)

Skin-friction drag = Dt: xcomponent

"t [ff"r"
rt]

d{]* : 16zst4/sl:om: 3e.13m

fitmds: [rrft-"
 ,L
28 MoDERN coMPRESSIBLE lrt-ow

Pt: 1132 tb711z

10' / -
a
_ L
Ivt,

p* : 7176 tb/tt1

p3 - 3568 tb/ft:

Pressure distribution

7=
-/ |
I < /C'.t! t,
to/tt:

Sherr strers dlstrlbutlon

FIGI-RE 1..5
(;eometn for ExamPle 1.7.

Hencc. recalling that shear stress acts on both sides,

Dt : 2(39.13)cos 10o : -l lb unit span

177 lper
The totai drag is
D:D,,+D]
D : t t691b - 11 .11b : ,ft46 lb-l
i-l
drag accounts
Note: For this example. the drag is mainly wave drag; skin-friciion
for oniy 5.7 percent ol the total drag. This illustrates an irnportant point. For
 TIIOU(]IITS 29
CONIPRESSIBI-E, F'LO\\' ' SOME IIISTORY AND iNTRODUCI'ORY

wave drag is
at a reasonable algle of attack' the
supersonic flow over slender bodies drag' For
leve1' far exceeding the skin-friction
the primary d'"g ;;"-;;;io"t ou sufltce, because the
discussed in this book
such applicatrorrr, r-n.-iJ."ld nrethods we see here also
wave drag (p.t**;;;;;;"t* tt c,o'"lned
from such methods'
fo"""a on the reductioo of wave-drag-because it is
why so.u"t uttt'itlo"i' jtug-to*po"tt't' At smaller angles of attack' the relative
frequeotly tf,. pti*tty-
higher altitudes, u'here viscous effects
proporrio! of o, i.o b io.i"ur.r. Also, at
is lorveil' the relative proportion of Dt to
D
become stronger (;; Rd;;l;t number
increases.

1.6 MODERN COI\{PRESSIBLE FLOW

steam-nozzies ol de Laval
In Sec. 1.1. \r'e saw hon'the con'r'ergent-<iivergent
the uorlJof p'1::i::l-::,1]'i"t'nt applica-
helped to usher compressible flot'into major-attention unti.l
tio,s. Ho*,er,.r..orrpr-.*ibl" floru did not begin to receive
theadverrtorj"tp.op,,t,ionandhiglr.speedfl*ightcluringWorldWarII'Indeed.
of compressible flo$'
ber*.een 1g45 and 1g60. the fundamentals andlpplications
b1'
i....-. essentiaih "classic." generaill' charaglgdgg4
i t'" constant speci{ic }reats'
1. Treatment oi a caloricalll pelfect glls'
bur usually' approximate solulioils
- l:xaclsolutions of no"'ln nng diqension'
2.-
three-dime*sional florvs' These
i;'r]-*'; ri,r.uri.J equation;) foi two- and
solurior-rsrr.ereclosedlc.,rrn.rieldingequationsorfclrnrulasfortlredesired
inlorrn:rtion.e,...p,lnn,\\.erethemerhodolCharacteristics.arteractnumeri-
flows (see Chap' 11)'
of compressible
cal aoproach applicable to cr'rtain classes
soluticrt: to the flC$,o\.er a sharp' right.circular
irrrd the eract far.lor-\'laccoll
10)' Both of these exceptions required
colte at zero angle of attack lsee Cirap'
endeavors before the advent ol the
numerical solutions $ hich \\'e re iaboiious
irroderir high-speed digitai coilputer'

gotl.1 te\ib(roks orl chssical compressible

florr. have been $ritteu since
\larrr
through 17 at tl-re errd of this book' Tjre
1945. So,re ol rhem are lisred as Refs.3
readerisstronglr.".",,,g.otostudl'thesereferer]Ces.becauseatliolough
flo'n' is esser.rtial to modern applications'
u;rcierstar.r6ing or .ru*.;i .-o*pr.t.iule
Sirrceapproximareh,.lg60.compressibleflorr'lrasentereda..modern,.
period. characterized b1'

l. The necessit,r' of dealing rvith high--temperature- chemicalll''

':u:ti*g qi:::
aid rocket engines' hence requiring a tna-1or
associated u'ith h,vf ersonlp- -fllght a calorically
extension ,na *oiin.ution oi the classical liter;rrr-,re based on
perfect gas.
rvhich is a nerv third dimension in
2. The rise of computationS! fl-gi{pynamics'
pre'ious existing dime*sions of pr.rre
fluid dynami.r, ;;;;1.;";ir[' ihe.
adve,t of modern high-speed digital
experimenr una pur.',lreory. tiith thi fluid dynamics
of computational
computers. ona tn" *U'equent development
30 MoI)ERN coMPREsstBLE FLow

as a distinct discipline, the practical solution of the exact governing equations

for a myriad of complex compressible flow problems is now at hand. In brief,
computational fluid dynamics is the art of replacing the governing partial
differential equations of fluid flow with numbers. and advancing these num-
bers in space and/or time to obtain a final numerical description ol the
complete flowfield of interest. The end product of computational fluid dynam-
ics is indeed a collection of numbers, in contrast to a closed-form analytical
solution. However, in the long run the objective of most engineering analyses.
closed-form or otherwise, is a qtrantitative description of the problem, i-e-,
numbers. (See, for example, Ref. 18.)

The rypdgl-go-n-tpres5;bie flow of today is a mutually supportive mixture of

classical analyses along with computationai techniques, uitir the treatment of
nongaloricaily perfect gases as almost routine. The purpose of this book is to
provide an understanding of compressible flow from this point of vieu'. Its intent
is to b-lend the important aspects of classical compressible flou'*'ith the recent
techniques of computational fluid dynamics- Moreover, th-q*SIIIpart of the book
rvill deal almost exclusively rvith a qloricaliy perfect gas. In turn. the secpnd part
rvili contain a logical extension to re_alms of high-temperature gases, and the
results u'ill be contrasted with those frorn classical analyses. In addition. various
historical aspects of the development of compressible flo*'..both classical anJ
modern. will be included along with the technical material. In this fashion. it is
hoped that the reader rvill gain an appreciation of the heritage of the discipline.
The author feels strongiy that a knowledge of such historical traditions and
events is important for a truly fundamental understanding of the discipline.

1.7 SUMMARY
The compressibility is generically defined as

ldt
- -uap
--;- (1.2)

hence
dp: prdp (i 6)

From Eq. (1.6), a flow must be treated as compressibie $'hen the pressure
gradients in the flowfield are large enough such that. in combination rvith a lirrge
Qg,gggh value of the compressibility, r, the resulting
densin changes are too large
to ignore. For gases, this occurs when the flow Mach number is greater than
C&gggg1. In short, for high-speed flows, the density becomes
a va,Liable; such
variabie-density flo*'s are called compressible flows.
High-speed, compressible flow is also' high-energy flou'. Thermodynamics is
the science of egelgy and entropy; hence a study and applcation of compressible
flow involves a coupling of purely fluid dynamic fundamentais u'ith the results of
thermodynamics.
 compressib,';;Jil;" r* ffi;; "*'"^.,#;:;"I*'l'
regimes' each with its
In turn, this range or Mach number is subdivided into four
o*r, aistinguishing physical characteristics and different analytical methods'
flow' Each of
supersonic, and hypersonic
These regimes ur" ,ulrorri", transonic,
these regimes is discussed at length in this book'

PROBLEMS
are 5'6 atm and 850'R'
l:t ti et the nose of a missile ia flight, the pressure and temperature
;;;;rl""jv.?rr""rate the dinsity and specific volunte. (Note: I atm : 2116 ta7* -1
"--^'
/ 1-lt n the reservoir of a supersonic wind tunnel, the pressure and temperature of air are
' 10 atm aod 320 K, respectively. Caictriate the density, tle number density,
ald the
:
--- mole-mass ratio. (Note: I atm 1'01 x 105 N/rf')
!"{5. r.lo, a calorically perfect gas, derive the relation cp - cu: R' Repeat tlre derivation
for a thermallY Perfect gas.
wave in air
1.4. The pressure and temiperature ratios across a given portion of a shock
of and
"." ir/p, - 4.5 a.,d rr/rr:1.6g7, where 1 andthe change in entropy in units of
2 denote conditons ahead
behind the shock *urf, resp"ctively. Calculate
(a) (ft ' ib)/(slug ' "R) and (D) J/(kg' K)'
1.5. Assume that the flow of air through a given duct is isentropic. A: one
point in the
duct, the pressure and temperature are pt : 1800 lb/ft] and 4 : 500'R' r€spec-
tively. At a second point, the tempersture is 400'R. Calculate the pressure and
density at this second Point.
1.6. Consider a tooln that is 20 ft long, 15 ft wide, and 8 ft high' For standard sea
ievel
conditons, calculate the mass of air in the room in slugs. Calculate the weight 'n
pounds. {Note: If you do not know what standard sea 1evel conditions are, consult
any aerodynamics iext, such as Reference lM, for these values. Also. rhel can
be

obtained from any standard atmosphere table')