You are on page 1of 78

by both laminar and turbulent boundary layers that form on a flat

surface.
To detennane the norm aI and drag forces on a body caused by
the pressure of a Ruid stream.
• To discuss the e,rfect of Ii~ and flow separation 011 bodies of
vanou s shc.1rpos.

11.1 he Cone pt of the Boundary Laye


\Vhcn a flutd flo,v~ O\' ·r a fla l urfacc. th Iaver of fluid particles adjac
tn the surl,•cc wiU have :t:ro l'elo ity. and larithcr from the surface~
vcloci'ly of each la) ~1 \\ill incre~ s unti~ the.' reach th c free-stre 1

e~ocit) a hol\in ia fig. 11- L Thi I chtl\ ior ~~ callScd by tb.e sh


slre..·s ~hat ac•s h~tw c:rn the lay\!rs of tht: nuid. and ~or a e\\~tonian O
chis str.: ·~is d'ircc'Liy proportional to chc \'c.!locit~ gm atliernt~ r = ,µ (,d u / "
otic~ that thi~ grac.li-.:nl and mhc ~hear 's* f 1Css. r~ l1trgesr at 1lze .'iurface~
both taper orr. tu1U1I fanher from lht: urfacc, the gradient and the sh
strc s app1oach zero. Here the Dtn bcha\'~~ as ir it 1 nonviscous,, sinc
is r111ifar1n* resulting in Hnl\? or no shear or sliding between adjao
layers f nuid. In l 04. Uc.l\\ ig Pr nt.hl r
1
gnizcd this difference in t1
beba j r. an lc1mcd the l caliz d r n \ h u~ the velocity is varia
"'
the h111111dan· . ..la\•er.
Unders~anding boundary
becon1es neoessary to dete;rm
a fluid. The design of pro
m echanica1 and structural e
1

depends on an ana1ys]s of th
this chapter \Ve vv·ill study o
]ay,er,~ vthich occws when the
the surface is relatively .fast. A
those having a high viscosi
understanding its effect on th
anaJysis or numerical modeU
Ba undary Layer Descri
boundary ]ayer can best be il
flo\v of a fluid ,over a ]ong flat
the huU of a ship. a flat section
basic featur,e s of the bow1da
These are. shown in Fig. 11-3.
y

u u

801.mda
(g read
of the boundary layer to grow rapidly, ieve.ntuaUy forming a nirbul
bounda1)1 layer+In spite of this transfitn·mation from laminar to turbul
flo\v, there ahvays

re.rnains, below the turbulent bounda:rv "
]aver,
~
av
thin la111inar or ~'isco11S s,llib,layer of fluid that is ~'s]o\v moving" since t
fluid must cling along the surface of the plate.

Boundary Layer Thickness. ,A t each location along the plate~ t


velocity profile within the boundary laye1»s thickness \Vil] asyrnptotica
approach the fr ee-stream velocity. Since th~s, thickness is not vveU defin
1

engineers use 1lhre'e methods for specifying its value.


1

Disturbance Thickness. Tb.e sin1ple.st vay to report the b>unda


layer's thickness at each location x is to de.fine it as the height l3 \.Vhe.re t
rnaxin1um velocity reached is e qual to a certain pe.reientage of the :f r
1 1 1

str1ea.1n ve]odty. TI1e agreed-upon value is u = 0.99U, as sho,vn in Fig. 11

u u

Fig~ 11-4
(a)

Fig. 11- 5

Displacement Thickness.
specified as a displace,nent t
actua~. surface n1ust be. displa
m.a ss flow \\rith tbis ue\v boun
for the real fluid. Fig.. ll-Sa.
tuu,ne]s and tbe intake of a je
To determine. the distance o
the 1nass floH1 deficit for each
then in the c.ase of the real flu
differential area dA = bdy i
were pre:Sent, then viscous ,ef
m,ass tlo'w at y \Vould then be
the viscosity is therefor·e d1
boundary layer. in'tegra·tion o
total deficit, sho\vn iu dark b
the .same for the idea] fluid in

pU (bo*)

Loss in
uruform n

Since p , U, and b arc constant

a·=
Therefore.., to det,e rmine the
11= u(y) of the boundary ~ay
can be evalua't ed., eitber ana
x a]ong the. plate.
(a)

f ig. 11-6
Momentum rn;ckness. Another 'to treat the veloc
\Vay
d:i sturbance. brought about by the boundary ]ayer is to consider ho\v t
actua] surface should be displaced. so that the rate of 1n<:Jnl.en.turn of t
flow " 'ou]d be the same as if the fluid were ideal. This change in heigh
the surface is c:aUed the rnomenturn thickness 0, Fig. 1l-6b. It r eprese 1

the loss of n1omentum in the boundary laye.r compared to having ide


flol-v. To find it \ve. n1ust detic!nnine the .rate of lnon1,cntu:tn tlo\v deficit
each case. Ff lhe plate has a \Vidth b, then for the real fluid, shown
Fig.. ll-6a, at height y the Ouid passing th.rough the area dA bas
rate of mome.ntum of din u = p(ilQ)u = p(udA)H·. Since dA = b dy~ th
1

driz u = p(ub dy)u. However} if the mass flow dm had a velocity U',, th
the rate of n1on11e ntum flo,v deficit \Vouk1 be p[ubc~i1 ](U - ii). For
cas e of the ~deal fluid., F~g. 1'.1-itlb, the rate of ·momentum flow de.fici
1

1JdQ,l f = p(U0b)U. Therefore] 've require:

p( U@b)U :=;; . f.· ~ pu (U - u)b tf.\'


Jo
or

1
l·€ . H .
e= 00
u 1 1 -udy
( )'
{11-
0

In summary we now have three definitions for the boundary la


'.f

thickness.: 8 r efers to the height to '~hich the boundarr layer disturbs


1

flov,.r, up co \Vhere the ve~odty becomes 0.'99U; and 8 and (~ define t


heights to \Vhich the surface n1ust be-displaced 01 repositioned so tha
the f]uid \Ve.re ideal aud flowing ·with the free-streru:n. velocity U it \Vo
produce the same. rates of mass and momentum fio\v, respec::bve1y; as
the case of tbe. rea,l fluid.
Boundary Layer Classifi
fluid de.ve]ops on the. surface
within the bclundary layer, an
]an1inarflo\v begins ·~o transit
can be used to do tbis; since. b
boundary layer <levehlpmen
the Reynokls ntunbe.r on tbe
is the distance do"vnstream
Therefore
'

_
R

From experirne:nts, [t has bee


1

do~'n a·t about Re.~ = ](] 05 ),


value when this happe.us is r
the p~ate, the uniformity of
changes occurring along the p
e.stabHsh a consistent 11alue
R evnolds number to be
1

"

For e.xa1nple.,for air at a ten11p


at 25 :m/ s, the boundary Jaye
distance of Xcr = (Rex)c,P
= 0.302 m from the front of t
A irflow over t he surface of tnis
t rH!S.e
flow over the flat plate in Fig. ll--8, \¥here the fluid is incompr·essibie a
has ste.ady larninar flow. Expe.riments have sho\vn that as the.fluid n1ov
1

over the plate., the streamHnes for the. flo\v graduany begin to cur
upward. so that a particle locat,e d at (x. y) has velocity components u. a
v. For high R eynolds nun1bers, the boundary ~ayer is very thin, and so
vertical component iJ will be much smaUe·r than the ho1d zontal compon
u. Also~ due to viscosity, 'the changes of u and v in they direction, t11at
au /fly, av/iJy, and iflu/ aj, \Vil[ be m.uch greate.r than chang,e s au/iJx,. av/
and iJ~~i/ilt 2 in the x direction.. Furthern1ore. because the S'lreanili
within the boundary layer 01dy .slightly curve up\vard, the press
variation in the y direction that caus,es this curvatur.e is practica
constant~ so that flp /iJy ~ 0. Finally,~ since the pressure above t
boundary layer is constaut, the.n within the boundary ~ayer, due. to
smaU height, np/ilx = n. 'Vith lhese assumptions, Prand'll was able
reduoe the three Navier--Stokes equations, Eq. 7--75~ to just one in th
direction,. and it along \'\i·ith the continuity equation~ be,comes
:.'I
riU :1 _
uU u:J,i i
u- + v- - ij --:;
ox dy dy-
-au + -ilv_o
- ,
ax av _,

To obtain the \'e[ocity distribution within the boundary lay er it is ne,cess


to so1ve these equations simultaneously for u and V', using the bounda
conditions u = ·v = 0 at v = 0 and u = U at -y = ;c. In 1908 Paul Blasius., w
~

\Vas one of PrandU's graduate students; did this usmng a nQmerica] analy
See Ref. [16]. He presented his results in the: form of a curve, sho\\"Il
F~g. ll-?~ that is plotted on axes ot.iliJnensionle.ss .ve]oc~ty 1J/ U ve.rsus t
dimensionless pa1ameter (y/x)YRex. Here Rex JlS defined by Eq. 11
For convenience, numerical values for this curve a.r e Hsted 1

Table 11- 1. Thus, f;o r a speoeified point (x , y) and free-stream velocity


1

Fig. 11--8. the v,elocity u for a particle \V]thin the boundary laye:r can
determined from the curve or the table.As ,e xpected .. the. so~ution indica
that the velocity of the bo'l.indary layer approaches the free-str,e.
ve1udty asyn1ptoticaUy., so that uj ,u ~ 1 as y--+ .x.
Disturbance Thickness
the velocitv t)f flow u is at
J

u/ U = 0.99. Fron1 B~asius~s s

Thus,

l..a

·u sing this result, it is worth n


actually be. For exainple, if
Reynolds. nu·m ber reaches
.t,r :::;;:;; l 00 mm, lhen the 1am1n
is onJy 0.707 rnm.
Displacement Thick
Fig. 11- 9, is substitute.d in
nlun,e rica[ integration, th
u boundary layers beco1nes

Actual surface
x
Velocity pro.file ror 10nc~ obtained,, th!s ~hickne
actual surface location of the sohd bounda
inviscid or ideal. For exam:p
Fig. ll-lO ch.au1.ber to acco1nm<C,date r8
the chamber vvill be un.ifor1n
- (·.·till)·
To -µ. -
. dv-· .·. y =o·
(ll-

The velocity gradient at y = 0 can he obtained by measuring it .f rom t


graph of Blasius's solution, Fig.11-9. It is sho\vn to be

a(~)
}J ... ..r=--
d ( ..;.."
. )·I .; ; ; Q.332
v Re.rr.. .
..,,. . ~\·-0

For a spe.cific ]ocation _,t and constant value.s of UI and 1!'~ the Reyno~
number Rex = Ur / v wH1 a1so be constant, and so d(u / U) = du / U a
=
d(yYRev/x) dy VRe.'t/x. After rearn1ng en1ent. the deriva'live du/ 1

then becoines

d.r..•c = 0.332(.' .,.,.


.d} -~
u),/Re . -~

Substituting this into Eq. 11- 7 gives tbe result

To = 0.332µ. ; ( U)·' VRex


-
1
r(11

With this equation, we can no\\' caku]ate d1e she.ar stress on the plate
any position x from the p[ate's leading edge. Notice that this stress \
become sn1aller~ as the distance x increases~ Fig. 11-lla.

You might also like