Film Cooling

.
RICHARD J GOLDSTEIN

Department of Mechanical Engineering. University of Minnesota. Minneapolis. Minnesota

I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . 321
I1. Adiabatic Wall Temperature and Film Cooling Effectiveness . . .
. 326
A. Incompressible Flow . . . . . . . . . . . . . . . . . . . 326
B. High-speed Flow . . . . . . . . . . . . . . . . . . . . 327
C . Impermeable Wall Concentration . . . . . . . . . . . . . 329
111. Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 330
A . General Remarks . . . . . . . . . . . . . . . . . . . . 330
B. Two-Dimensional Incompressible Flow Film Cooling-Heat Sink
Model . . . . . . . . . . . . . . . . . . . . . . . . . 330
C . Energy Balance in the Boundary Layer . . . . . . . . . . . 331
D . Two-Dimensional Incompressible Flow Film Cooling-Other
Models . . . . . . . . . . . . . . . . . . . . . . . . 338
E. Two-Dimensional Film Cooling in a High-speed Flow . . . . 340
F. Injection through Discrete Holes-Three-Dimensional Film
Cooling . . . . . . . . . . . . . . . . . . . . . . . . 341
IV . Experimental Studies . . . . . . . . . . . . . . . . . . . . 342
A . General Remarks . . . . . . . . . . . . . . . . . . . . 342
B. Two-Dimensional Film Cooling-Incompressible Flow . . . . 351
C . Two-Dimensional Film Cooling-Compressible Flow . . . . . 361
D . Three-Dimensional Film Cooling . . . . . . . . . . . . . 369
V. Concluding Remarks . . . . . . . . . . . . . . . . . . . . . 315
Nomenclature . . . . . . . . . . . . . . . . . . . . . . . 316
References . . . . . . . . . . . . . . . . . . . . . . . . 371

.
I Introduction

The need to protect solid surfaces exposed to high-temperature

environments is an old one. In general the high-temperature environ-
ment is gaseous. and it may be highly ionized as in the stream surrounding
a vehicle reentering the atmosphere or in the constrictor of an electric arc
or plasma jet . During the last twenty-five years. relatively sophisticated
321
322 RICHARDJ. GOLDSTEIN

cooling methods have been used in rockets, reentering space vehicles,

high performance gas turbines, and plasma jets.
One method is to introduce a secondary fluid into the boundary
layer on the surface to be protected. There are different means of
introducing this secondary (or injected or coolant) fluid into the boundary
layer including ablation, transpiration (or sweat or mass transfer),
and film cooling. In ablation cooling, an added coating or “heat shield”
decomposes, and by sublimation and other highly endothermic processes
a significant quantity of gas enters the boundary layer. In transpiration
cooling, the surface is usually a porous material, and the secondary
fluid enters the boundary layer through this permeable surface.
Both ablation and transpiration cooling are primarily designed to
protect the region where the secondary fluid enters the boundary layer.
They are highly effective in this regard as a considerable portion of
the heat transferred toward the wall can be taken up by the injected
coolant right where the heat transfer load is highest. In addition the
gas entering the boundary layer effectively thickens it, decreasing the
heat transfer rate. These two methods do, however, suffer from serious
disadvantages which preclude their use in many applications. The
ablating material is not in general renewable and so ablation cooling
has been restricted to systems with high heat fluxes of short duration,
such as reentering vehicles. This restriction does not apply to transpira-
tion cooling since a coolant can be continually introduced through
the porous surface. However, porous materials to date have not had
the high strength required for certain applications (e.g., turbine rotor
blades) and small pore size often leads to clogging and a resulting
maldistribution of coolant flow. In addition, variation in the external
pressure distribution can result in a nonoptimum secondary flow
distribution through the permeable surface.
Although a secondary fluid is also added to the boundary layer in
film cooling there are considerable differences in operation and even
in goals as compared with ablation and transpiration cooling. A key
difference is that film cooling is not primarily intended as protection
of the surface just at the location of coolant addition, but rather the
protection of the region downstream of the injection location. Film
cooling is thus the introduction of a secondary fluid (coolant or injected
fluid) at one or more discrete locations along a surface exposed to a
high temperature environment to protect that surface not only in the
immediate region of injection but also in the downstream region.
Eckert and Livingood ( I ) examined transpiration and film cooling
(as well as internal convective cooling) to see how a given amount of
fluid could be used most effectively. I n their comparison, however,
 FILMCOOLING 323

the maximum rather than the average temperature of a film cooled

wall downstream of the injection slot was considered. As might be
expected, they found transpiration cooling more efficient in use of
coolant. If a real conducting wall were considered, the average film
cooled wall temperature would be more appropriate, and the difference
in the effectiveness of the two methods would be greatly reduced.
The geometry and flow field at the point of injection are significant
variables in film cooling. In two-dimensional (including axisymmetric)
film cooling not only is the external flow two-dimensional, but the
secondary fluid is also introduced uniformly across the span as in Fig. I .
Secondary fluid can enter through a porous region (Fig. la) or through
a continuous slot at some angle to the wall surface and the main-
stream (Figs. l b and lc). T h e flow downstream of a transpiration

( 0 )

4
ZFR
MAINSTREAM

( C )

FIG. 1. Representative two-dimensional film cooling geometries: (a) porous slot,

(b) tangential injection-step down slot, ( c ) slot angled to mainstream.
324 RICHARDJ. GOLDSTEIN

cooled or ablation cooled region is similar to film cooling, the porous

wall of the transpiration cooled section or coated wall section of ablation
region acting as the slot or injection region. Similarly with liquid film
cooling the cooling effect in the region downstream of the point where
all the liquid is evaporated can be considered similar to gas-to-gas film
cooling.
Although the injection geometry can influence the film cooling
performance in two-dimensional flow, the effect is usually of second
order compared to geometrical effects in three-dimensional film cooling.
In this latter flow (Fig. 2) the injection of secondary fluid is not uniform
across the span, but rather occurs at isolated locations often through
discrete holes in the surface. This can lead to the jets of secondary
fluid being blown off the surface and the mainstream flow coming
between and/or under the coolant jets decreasing the effectiveness of
the film cooling process. Even so, for structural reasons it is usually
impossible to have a truly continuous two-dimensional injection slot,

COOLANT /-

FIG.2. Film cooling with injection through inclined tubes: (a) injection through
single tube inclined at angle a to mainstream, (b) injection through single row of discrete
tubes inclined at angle a to mainstream.
 FILMCOOLING 325

and so interrupted slots and even rows of multiple holes have been
used.
Although film cooling has primarily been used to reduce the convective
heat transfer rate from a hot gas stream to an exposed wall, it could
also be used to shield a surface from thermal radiation if the radiation
absorbtivity of the injectant is high. This can be effectively accomplished
with gas particle suspensions or a liquid coolant. In this review, however,
all the fluids are considered transparent to radiation, and the convective
and radiation heat transfer are then independent and can be treated
separately. Only the effect of film cooling on convective heat transfer
will be considered.
T h e introduction of secondary fluid into the boundary layer with
film cooling may be considered to produce an insulating layer (film)
between the wall to be protected and a gas stream flowing over it.
Alternatively the injected fluid can be considered as a heat sink that
effectively lowers the mean temperature in the boundary layer. As will
be discussed, the secondary fluid usually serves both functions. T h e
introduction of the secondary fluid into the boundary layer at a tempera-
ture lower than the mainstream and its resultant mixing with the fluid
in the boundary layer reduces the temperature in the region downstream
of injection. Note that there is usually considerable mixing of the
injected fluid and the mainstream flow downstream of injection. T h u s
the concept of a film of secondary fluid maintaining its structure for
some distance downstream and isolating the solid surface from the hot
mainstream is not strictly valid, especially with a gas coolant. Although
a separate discrete insulating film is not produced, injection of the
secondary gas can increase the boundary layer thickness and the mass
of fluid entrained into the boundary layer from the free stream. T h e
increased boundary layer thickness tends to decrease the heat transfer
to the wall. However, the increased mainstream flow entrained in the
boundary layer causes increased dilution of the secondary fluid with
a resulting decrease in its effectiveness as a heat sink. T h e significance
and relative importance of these two opposing effects will be discussed
subsequently.
This review is restricted to film cooling with both the mainstream
fluid and the secondary fluid being gases, although not necessarily
the same gas, and with a turbulent boundary layer downstream of
injection. Both two-dimensional and three-dimensional secondary flow
geometries are considered. I n addition, film cooling in compressible
flows as well as in incompressible flows is examined. Although the
emphasis is on adiabatic wall temperatures and uniform mainstreams,
the effects of heat transfer and variable free stream velocity are discussed.
326 RICHARD J. GOLDSTEIN

11. Adiabatic Wall Temperature and Film Cooling Effectiveness

FLOW
A. INCOMPRESSIBLE
In most film cooling applications the heat transfer from the hot gas
to the surface to be protected is not zero. There is usually some type
of internal cooling, or, in a transient problem, the heat capacity of the
wall material itself is used to take up the heat transferred. T h e general
problem in film cooling is to predict or measure for a given geometry,
mainstream, and secondary flows the relationship between the wall
temperature distribution and heat transfer. Conversely, for a given
mainstream and allowable wall heat transfer the requirement may be
to predict the secondary flow needed to maintain the surface temperature
below some critical value.
With constant property flows the velocity distribution is independent
of the temperature field and it is convenient to use the concept of a
heat transfer coefficient. Thus
= h AT = h(Tw - Taat) (1)
where T , is the local wall temperature. A question arises as to the
datum (i.e., base or reference) temperature Tdat to use in Eq. (1).
I n the limiting case of a perfectly insulated (i.e., adiabatic) surface
the heat flux would be zero and the resulting surface temperature
(distribution) is called the adiabatic wall temperature Taw. Thus the
adiabatic wall temperature could be used as the datum temperature.
T h e heat flux with film cooling would then be
q = h(Tw - Taw) (2)
Use of Eq. (2) yields a heat transfer coefficient that is independent of
the temperature difference for a constant property flow. Note that in
the absence of blowing, Tawwould be equivalent to the free stream
temperature or in the case of high speed flow the recovery temperature.
Most film cooling studies have treated the determination of the
heat transfer coefficient and adiabatic wall temperature distribution
separately with primary emphasis on the latter. Often the heat transfer
coefficient is found to be relatively close to the value without secondary
flow, i.e., dependent primarily on the mainstream boundary layer flow.
On the other hand, the adiabatic wall temperature distribution can vary
considerably and is thus harder (and more important) to predict.
I n addition the adiabatic wall temperature is significant in that it is
the limiting value of wall temperature that can be obtained without
internal wall cooling.
Primary emphasis is given to prediction and measurement of the
 FILMCOOLING 327
adiabatic wall temperature distribution even when the assumption of a
constant property flow is invalid, for example in high speed compressible
flow. For this case the use of a reference temperature or reference
enthalpy as with normal boundary layer flow in the absence of injection
will be found useful in predicting heat transfer and will permit application
of Eq. (2).
T h e adiabatic wall temperature is not only a function of the geometry
and the primary and secondary flow fields but also the temperatures
of the two gas streams. T o eliminate this temperature dependence a
dimensionless adiabatic wall temperature, q , called the film cooling
effectiveness is used. For low speed, constant property flow the film
cooling effectiveness is given by

’=
m m
law - 1,
T, - T, (3)

where the temperatures of the secondary fluid T , and the mainstream

fluid T , are assumed constant. Note that in general Taw< T , and
T, < T , in a film cooling application. Since the constant property
energy equation is linear in temperature the film cooling effectiveness
is dependent only on the primary and secondary flows and the position
on the surface. Note that the film cooling effectiveness usually varies
from unity at the point of injection (where Taw= T,) to zero far down-
stream where, because of dilution of the secondary flow, the adiabatic
wall temperature approaches the free stream temperature.

FLOW
B. HIGH-SPEED
For high-speed flow the film cooling effectiveness must be defined
somewhat differently. At the point of injection the wall temperature
T,, would be expected to be the recovery temperature of the secondary
flow or possibly the total temperature of the secondary flow. Far down-
stream the wall temperature might be expected to approach the main-
stream recovery temperature T , (evaluated in the absence of secondary
flow). An expression often used for compressible flow film cooling is

Note that the film cooling effectiveness T,I~ reduces to Eq. (3) when
compressibility effects can be neglected.
An alternate convenient definition of effectiveness for high speed
flows employs the isoenergetic-injection wall temperature distribution,
328 RICHARDJ. GOLDSTEIN

The isoenergetic temperatures (Taw, and TW2,) are the temperatures

on the adiabatic wall for the same mainstream conditions, and the same
secondary flow rate, but with secondary flow stagnation temperature
equal to the stagnation temperature of the mainstream (see Fig. 3).
Note that TaWiis a function of position while the recovery temperature
used in Eq. (4) can often be considered constant. The definition of
effectiveness (Eq. 5) as in Eqs. (3) and (4) gives an effectiveness of
unity at the point of injection and a zero value of effectiveness far
downstream.

F Tozi
TWF-
" """""'T~ Tawi

FIG. 3. Temperatures with isoenergetic injection in high speed flow, injection

equal to the mainstream stagnation temperature Tom.
stagnation temperature To,,

The isoenergetic film cooling effectiveness is found to correlate

experimental results better than an effectiveness based on the recovery
temperature without secondary flow. The isoenergetic injection tempera-
ture distribution Tawiis obtained at the same blowing rate and free
stream conditions as with film cooling. Assuming a constant property
fluid, the flow fields are the same in the two cases and independent
of the temperature of the secondary fluid. The viscous dissipation
terms are also the same in the isoenergetic and normal film cooling
runs and subtraction of the temperature distributions from the two
runs eliminates the viscous dissipation effect. Thus the difference
solution vis for the high-speed flow should be the same as for low-speed
flows if the pertinent dimensionless variables describing the flow remain
the same. For low-speed flow qis reduces to the usual incompressible
flow effectiveness (Eq. 3). For these reasons it may be possible to predict
the isoenergetic film cooling effectiveness using the results (theoretical
or experimental) from low-speed flows.
When using compressible-flow film cooling, results in a design
problem not only qis but also the adiabatic wall temperature distribution
for isoenergetic injection TaWimust be known. However, for many
applications the difference between the recovery temperature without
 FILMCOOLING 329

secondary flow, T,, and Tawiis much smaller than the difference between
either of these and the expected temperature of the secondary fluid.
Then the recovery temperature T , may be used as a reasonable approxi-
mation for TaWi.
T h e enthalpy can also be used in defining the film cooling effectiveness
in high-speed flows. This would be useful at large temperature differ-
ences. T h e proper enthalpies could then be substituted (for the tempera-
tures) directly into the above expressions for effectiveness.

WALLCONCENTRATION
C. IMPERMEABLE
I t is often difficult to design test systems with walls that sufficiently
approximate adiabatic surfaces. This is particularly apparent when the
adiabatic wall temperature distribution has large gradients and with a
very high temperature mainstream. I n such cases a mass transfer
process can be used as an analogue to film cooling. Thus instead of
injecting a gas at a different temperature from the mainstream, a gas
of different composition would be injected isothermally. This might
apply particularly in a study of the effects of three-dimensional film
cooling with large density differences (between the primary and secondary
flows). T h e mass transfer analogy is also useful for two-dimensional
film cooling with large temperature (density) differences. T h e injected
gas can be completely different in composition from the free stream,
or only a tracer gas might be used in the secondary flow. If the secondary
fluid is otherwise the same as the mainstream, the use of a tracer gives
results comparable to low density (temperature) differences.
T h e mass transfer process is analogous to the heat transfer process
(neglecting thermal diffusion phenomena) if the equivalent dimensionless
parameters of the flow are the same in the two cases and if the Lewis
number is unity. (The Lewis number is the ratio of the Schmidt number
for the mass transfer process to the Prandtl number for the corresponding
heat transfer process.) T h e turbulent Lewis number as well as the
molecular Lewis number should be unity for the analogy to hold.
If the flow is sufficiently turbulent, variations in the molecular Lewis
number from unity may not play an important role, but in all cases
studied to date the value of the turbulent Lewis number should be
considered.
When using the mass transfer analogy with foreign gas injection,
the quantity analogous to the adiabatic wall temperature is the concentra-
tion of the injected gas at an impermeable wall. Although there is
some question as to the proper concentration to use, the mass fraction C
is the most widely used. Equivalent to the film coolii g effectiveness
330 RICHARDJ. GOLDSTEIN

based on adiabatic wall temperature is an effectiveness based on the

impermeable wall concentration

If the secondary fluid contains a single constituent not contained in the

mainstream, then C, = 0, C, = 1 and Eq. (6) becomes

v c = Ciw (7)

In. Analysis

A. GENERAL
REMARKS
A number of theoretical correlations and predictions have been
developed for the film cooling effectiveness. Since interest is chiefly
in turbulent film cooling, the analyses are at least partly empirical yet
often suggestive of the significant features of the flow. Much of the
interest has centered on relatively simple heat sink models in which
the added secondary flow is considered as a sink of heat at the point
of injection reducing the temperature in the downstream boundary
layer and thus the temperature of the wall. The models have been
applied to two-dimensional incompressible and compressible flows and
lately to three-dimensional film cooling. Other analyses use some of
the recent numerical techniques for predicting two-dimensional turbulent
boundary layers and separated flows to obtain predictions of film
cooling effectiveness.
In this section some of the theoretical analyses will be developed
and differences between them discussed. Comparison with experimental
results will be deferred till the next section in which the various experi-
mental studies are described.

B. TWO-DIMENSIONAL FLOWFILM COOLING-HEAT

INCOMPRESSIBLE
SINKMODEL
The first heat sink model for film cooling was given by Tribus and
Klein (2) in a work primarily concerned with developing kernels to
predict the heat transfer and temperature distribution along non-
isothermal surfaces. At the suggestion of Eckert, they used Rubesin’s
kernel for the wall temperature distribution with a turbulent boundary
layer to predict the temperature on an adiabatic surface downstream
from a line source of heat. The strength of the line source is determined
 FILMCOOLING 33 1

by the net enthalpy flow of the secondary fluid I p2U2CP2(T2 - T,)I

and the calculation ignores any effect of the injection on the mainstream
flow. Integration of the kernel yields
77 = 5.76 Pr2/3Re~.2(p2/pm)o.z
(Cp2/Cpm)(~/M~)-o.s (8)
= 5.76 Pr2/3(C,2/C,,) ,!-O.*

where
,t = ( X / ~ S ) [ ( P ~ / CRe21-0.25
L~) (9)
T h e dimensionless blowing rate parameter M is the ratio of the mass
velocity of the injected fluid to the mainstream mass velocity. T h e
distance x is measured downstream from the point of injection. Tribus
and Klein compare their analysis to results for air injected into an air
mainstream. For p2 = pm, C,, = C,, , and Pr M 0.72,

77 = 4.62 Rei.2(x/Ms)-o.s (10)

T h e parameters in Eq. (8) appear in essentially all of the heat sink
models and they are very useful for correlating data, particularly at low
blowing rates.
Equations (8) and (lo), however, predict higher values of effectiveness
than have usually been found experimentally. This is apparently due to
the assumption that the injected gas does not affect the velocity boundary
layer. Figure 4, however, indicates that the boundary layer is considerably
thickened by injection. This particular figure was obtained for secondary
flow through a porous section ( 3 ) .
Later heat sink analyses were made by Librizzi and Cresci (4 ),
Kutateladze and Leont'ev (9, Stollery and El-Ehwany (6, 7), and
Goldstein and Haji-Sheikh (8). These analyses have a great many
similarities, and all proceed from an initial energy balance on the
boundary layer.

C. ENERGYBALANCE
IN THE LAYER
BOUNDARY
T h e mass flowing within the boundary layer is considered to be
composed of two different fluids from two different streams-the
injected gas (ni2) and the mass which enters (is entrained into) the
boundary layer from the mainstream (&). These gases are assumed
to be well mixed in the boundary layer. At any position downstream
of injection the mass flowing per unit time (ni) in the boundary layer
is given by (see Fig. 5a)
332 RICHARDJ. GOLDSTEIN

c
E
c

m
.- x - 1

u)

! I
u)
W
z
Y
0
r
I-

I, 0.I
5
L
0.08
z
0
0.051 I I I I I I I I
2 4 6 810 20 40 60 80100

DISTANCE FROM TRIP WIRE (cml

FIG. 4. Effect of injection on boundary-layer momentum thickness for injection

through a porous slot.
~~ ~~ ~ ~

Symbol M &(gm/sec) Um(m/sec)

X 0.0389 22.55 56.1
0 0.0263 15.41 55.2
0 0.01 62 10.08 54.6
+ 0.00776 4.90 56.5

[R. J. Goldstein, G . Shavit, and T. S. Chen, J. Heat Transfer 87, 353 (1965).]

The mean temperature T in the boundary layer is given by

Assume constant property ideal gases. The average specific heat Ca

for the boundary layer is given by,
Cu =(hzC92 + kmCDm)/(niz + k m ) (13)
If the wall over which the fluid flows is adiabatic, application of the
steady flow energy equation (see Fig. 5b) at any downstream position
yields
+
(ni2 ni,) CUT= k2Cu2T2 ?ilmCumT, +
(14)
Rearranging terms in Eq. (14) and using Eq. (13) yields
T - T- 1
(15)
 FILMCOOLING 333

, A
CONTROL m
VOLUME
--1
--
-:A=
--
ENTRAINED MASS

-----
I I
I --t"m= m,+m,

FLOW RATE

ENTRAINED ENTHALPY
CONTRF VOLUME /kioocpmT~ Fu)w RATE

(b)

FIG. 5. Control volume when performing (a) mass and (b) energy balances.

I n References (4-6) the mean temperature T in the boundary layer is

assumed equal to the adiabatic wall temperature. (Librizzi and Cresci (4)
did consider the temperature variation across the boundary layer in a
compressible flow model,) Equation (15) reduces to Eq. (3), and the
film cooling effectiveness becomes

These analyses all use essentially the same method to predict m m . They
assume a +th power turbulent velocity profile and a boundary layer
thickness given by
SIX' = 0.376 Re;?'5 (17)
to predict the entrained flow rate & . The parameter x' is the distance
from the point at which the boundary layer starts. Some of the original
derivations use a slightly different value of the constant (0.376), but for
334 RICHARDJ. GOLDSTEIN

comparison purposes they have all been recalculated using Eq. (17).
Equation (17) is valid in the absence of injection and the analyses
assume that it is still valid with injection. T h e primary difference
between the three analyses is the assumed location where the total
mass flow in the boundary layer starts (i.e., where x' = 0). For a 3th
power velocity profile starting at x' = 0, the mass in the boundary layer
from the mainstream at some distance downstream is
mm = pp,U,6 = 0.329pmUmx'
Re;?'5 (18)
Librizzi and Cresci ( 4 ) assume that the boundary layer starts at the
point of injection (x' = x) and at injection (x = 0)
m = m, + mm = m, (19)
Using fi, = pzUzs
mm/& = 0.329(~/Ms)O.~
[Rez(p2/pm)]-0.2 (20)
Putting this into Eq. (16) and using Eq. (9)'

For C,, = C,,

' 1
1
+ 0.329(C,,/c,,) 50.'

'
1
= I + 0.32950.'
Kutateladze and Leont'ev (5) assume the boundary layer downstream
of injection grows as if it had started upstream of injection at some
distance x". I n order to calculate the distance, x", they assume that the
upstream (fictitious) boundary layer grows as a turbulent boundary
layer which started sufficiently far upstream to have a net mass flow
in it at the point of injection equal to the secondary mass flow rate.
Calculation of the mass flow rate from Eqs. ( 1 1 ) and (18) using the
above assumptions gives
m,/& = 0.329(4.01 + [)"" -1 (23)
Inserting this into Eq. (16) gives

' = 1
1
+ (C,,/C,,)[O.329(4.0I + no."- 11 (24)

and for C,, = C,,

' = 1
1
+ 0.249[O.' (25)

Leont'ev (9) extended the model to include the effect of film cooling
 FILMCOOLING 335

on heat transfer and also the effect of rough surfaces on film cooling
performance.
Stollery and El-Ehwany (6, 7) assume the boundary layer starts at
injection (x = x') and also that the total mass flow in the boundary
layer is zero at the point of injection. Thus at x = 0,
m=0
and for x >0
m = gp,u,s
Thus using Eq. (1 1)
m, = ipP,U,6 - m2

Inserting this into Eq. (16) (assuming C,, = C P m )and noting m 2 =

. - - yields
pzU2s, -
1 --
- 8 _P Z_
UZ_
S
rl = 7
-~p,U,6 7 p,U,6
8 m2
Using Eq. (17)
77 = 3.03(r/Ms)-0.8Re,(p2/pm)o.z

or introducing Eq. (9)

77 = 3.03t-O.'

They also indicate in their analysis the effect of foreign gas injection
(when C,, # CPm)and suggest an approach for determining the film
cooling effectiveness with a variable free stream velocity.
Since the above heat sink models assume complete mixing of the
secondary fluid in the mainstream boundary layer, their validity would
be expected to suffer when applied near the point of injection. Both
the Tribus and Klein prediction (Eq. 10) and the Stollery and El-Ehwany
prediction (Eq. 31) essentially assume no mass flow in the boundary
layer at injection yet a finite heat source. Consequently they predict
a value of infinity for the effectiveness at x = 0. T h e Librizzi and Cresci
and the Kutateladze and Leont'ev correlations, by their assumption
of rh = riz, at the point of injection yield an effectiveness of unity at x = 0.
This is a convenience which should not be overlooked. Note that three
of the correlations (Eqs. (22), (25), and (31)) approach the same predic-
tion far downstream where t is large.
It is of interest that these last three correlations (Eq. (22), (25),
and (31)) are in better agreement with experimental data, as will be
shown below, than the Tribus and Klein correlation (Eq. (8) or (lo)),
even though they use the additional major assumption that the tempera-
ture in the boundary layer is constant. That this latter assumption is
336 RICHARDJ. GOLDSTEIN

not true was known from the earliest test results of Wieghardt (10) as
is shown in Fig. 6. The temperature profile is observed to be similar
for different positions downstream of injection.
The reason the correlations (Eq. (22), (25), and (31)) work so well
follows from the unwritten law that sometimes two invalid assumptions
are better than one. Thus the assumption that the boundary layer is
unaffected by the secondary flow would indicate less flow into the
boundary layer from the mainstream (i.e., less dilution) than actually
occurs, reducing m, in Eqs. (1 1) and (16), and thus predict a larger
effectiveness than occurs. However, the mean temperature in reality is
significantly different from the adiabatic wall temperature; T being
between Tawand T, . Thus the assumption used to get Eq. (16) that

’ =
T - T,
T,- T ,
gives a lower effectiveness than the true value

since I Taw - T , 1 > 1 T - T, I. The success of the correlations is

apparently due to these two effects counterbalancing each other.

Y/8,
FIG. 6. Dimensionless boundary layer temperature profiles at various positions down-
stream of injection: M = 0.74, s = 10 mm, Ta- T, = 28T,(-) e~p[-O.768(y/6~)’~/~],
(- - -) e~p[-O.785(y/6~)~]. Distance from injection in meters: ( 0 )0.05, (+) 0.2, (a) 1.0,
(v) 2.0, ( 0 ) 4.0. [K. Wieghardt, AAF Translation No. F-TS-919-RE (1946).]
 FILMCOOLING 337

Goldstein and Haji-Sheikh (8) use a modified heat sink analysis as an

attempt to correct the two assumptions mentioned above. An overall
energy balance is performed yielding Eq. (15). T h e temperature variation
through the boundary layer is considered, as is the effect of the injection
on increasing the size of the boundary layer, and thus the mass flow
entering the boundary layer from the mainstream.
Assuming a power law velocity profile and a similar temperature
profile (cf. Fig. 6 ) the mean temperature in the boundary layer is
calculated from
-
I' - T , = X(Ta, - T,) (34)

Combining Eqs. (34) and (15),

where h depends on the temperature and velocity profiles. Using the

profiles of Wieghardt (10) and extrapolating the results to zero blowing
rate so they can be compared to Tribus and Klein results gives

l / X = 1.9 Pr2/3 (36)

though the variation with Prandtl number would not be expected to

hold over a large range.
T h e ratio of the mass flow in the boundary layer with blowing m, ,
to the mass flow without blowing, mmo, is determined from experi-
mental results of previous investigations. Figure 7 shows that the mass
added to the boundary layer from the free stream increases with
secondary flow rate and angle of injection (from the mainstream direc-
tion). This figure specifically refers to the flow when the secondary
and primary gases are the same. For different gases

km/km0
=1 + 1.5 x lo-* Re2(p2W,/p,W2) sin a (37)

is obtained where a: is the angle of injection (measured relative to the

wall). Combining Eq. (37) with Eqs. (35) and (36) and using Eq. (18)
to predict mmo(not i , directly),
1.9 Pr2I3
' 1 + O.329(CDm/C,,)['.'fi
where
B = 1 + 1.5 x lo-* R e , ( ~ W m / p , W 2sin
) a (39)
338 RICHARDJ. GOLDSTEIN

5 -
-

4- -
ma0
0 3-
-
8
.€
\

8
.€
2 -

I -
0.9 I I
I 2 3 4 5 6 7 8 910

FIG. 7. Ratio of boundary layer entrained mass flow rate with secondary injection kirm
to entrained mass flow rate for zero injection +zmo as a function of secondary flow injection
angle and flow rate.
Data: (0, . , r , D ) , R . J.Goldstein,G.Shavit,andT.S.Chen,J.Heat Transfer87,353(1965);
( 0 , A), K. Wieghardt, AAF Translation No. F-TS-919-RE (1946);
(v), J.P.Harnett, R. C. Birkebak, and E. R. G. Eckert, J.Heat Transfer 83,293 (1 961);
(D), R. A. Seban and L. H. Back, J. Heat Transfer 84, 45 (1962).
[R. J. Goldstein and A. Haji-Sheikh, in Japan SOC.Mech. Engr. 1967 Semi-Intern. Symp.,
213-218, Tokyo (1967).]

D. TWO-DIMENSIONAL
INCOMPRESSIBLE
FLOWFILMCOOLING-
OTHERMODELS
Hatch and Papell (ZZ) use a theoretical model for tangential injection,
in which they envision that the injected gas remains in a separate film
apart from the free stream and try to calculate the heat exchange between
this film and a turbulent boundary layer atop it. I t would be expected
that such a correlation would fit best very close to the region of injection.
Saarlas (12) uses a boundary layer model to predict film cooling
effectiveness. T h e analysis also permits an approximate calculation of
the heat transfer with film cooling and the effect of a variable mainstream
velocity.
 FILMCOOLING 339

Seban and Back (13, 14, 25) use the similarity of the temperature
profiles to predict film cooling effectiveness based on a uniform eddy
diffusivity across the boundary layer. They divide the flow (tangential
injection) into three regions: a wall-jet-like flow near the slot where
the secondary flow is preserved, a mixing region, and finally a normal
turbulent boundary layer region. They observe (for M < 0.8) that the
initial region of the flow is defined by XIS < 56M1.5 (14). T h e model
uses a linearized form of the energy equation and an upstream effective
starting point of the boundary layer and predicts reasonable values of
effectiveness for tangential secondary flow injection.
T h e wall jet region is particularly significant at a velocity ratio,
U , / U , , greater than unity. Note that it is apparently the velocity ratio
that plays the key role in determining the approach to wall jet behavior
rather than the mass velocity ratio or blowing rate M . T h e significance
of the velocity ratio might be expected since it indicates (for tangential
injection) whether the secondary fluid will tend to accelerate the main-
stream ( U , > U,) or be accelerated by the mainstream ( U , < U,).
Spalding (16) proposes relations for film cooling with a tangential
slot through which a fluid similar to the mainstream fluid is injected.
Although basically empirical the model reduces to a relation similar
to the heat sink models at low blowing rates and behaves similar to
what would be expected for a wall jet at large blowing rates. He predicts,
for 6' < 7,
7)=l (40)
and for 8' 3 7,
?1 = 715' (404
where
6' = O . ~ ~ ( X / I M SRe,0.2
)~.~ + 1.41{[1 - ( U J U J ] x/s}O.~ (40b)

All of the analyses described so far use models that employ considerable
empirical input. I n an attempt to use a more analytical approach
Whitelaw and co-workers (17, 18, 19) have tried to solve the turbulent
flow boundary layer equations for film cooling with tangential injection.
They use the Patankar and Spalding (20) approach in which a mixing
length and effective turbulent Prandtl number distribution are assumed.
Although this method has many difficulties and there are questions
about its validity and accuracy, it does offer the hope of future solutions
valid for the region close to injection point as well as predicting the
heat transfer coefficient. Other references using this approach include
(21)-(23).
340 RICHARDJ. GOLDSTEIN

E. TWO-DIMENSIONAL
FILMCOOLING
IN A HIGH-SPEEDFLOW
Film cooling in a two-dimensional high-speed flow has been
analyzed (24, where the reference temperature (enthalpy) method was
combined with some of the incompressible flow analyses to obtain the
film cooling effectiveness. As mentioned above, an effectiveness using
the isoenergetic wall temperature distribution as a reference appears
to work best for compressible flows. T h e reference temperature used is
T* = T, + 0.12(Tr - Tm) (441)
and all properties in the boundary layer are evaluated at this temperature.
Thus
t* = (x/Ms)(Rez P ~ ~ P * ) (P*/P,)
- ~ . ~ ~ (42)
T h e local wall temperature T , or Tawwould be used in place of T ,
in Eq. (41) if very large temperature differences are encountered.
Corresponding to the Kutateladze and Leont’ev model (Eq. 24)
71s = {I + (Cgm/C,2)[0.33(4.00 + E*)o’8 - 1]}-’.’ (43)
When the injected fluid is the same as the mainstream fluid the
relations derived for high-speed flow are:
Kutateladze and Leont’ev model;
qis = (1 + 0.25&)-0*8 (44)
Librizzi and Cresci model;
q,, = (1 + 0.3340;8)-1 (45)
Stollery and El-Ehwany model;
qi, = 3.03(;0.8 (46)
and Goldstein and Haji-Sheikh model;
qi, = 1.9 Pr2I3(1 + 0.33[0;8p)-1 (47)
where
/3 = 1 + 1.5 x Re,(pz/p*) sin 01 (48)
I n deriving these expressions for high-speed film cooling the constant
used in the boundary layer growth equation comes from a best fit to
experimental skin friction data proposed by Schlichting (25).
Laganelli (26) uses a similar analysis based on reference properties
to predict film cooling performances in supersonic flow. His results
are similar to those given above. He also extends his results to an
 FILMCOOLING 34 1
axisymmetric coordinate system. Librizzi and Cresci (4) have also
considered film cooling in an axisymmetric supersonic flow.

F. INJECTIONTHROUGH DISCRETE
HOLES-THREE-DIMENSIONAL
FILM
COOLIN G

The heat sink concept has been applied to film cooling following
injection through discrete holes (27). With such a geometry there is
little hope of getting a relatively exact analytic description of the velocity
and temperature distributions.
At relatively low mass injection rates the mass addition through a
single hole can be considered to act as a localized heat sink on the
film-cooled surface. T h e transfer process in the boundary layer is
approximated by the conduction equation, the problem being equivalent
to determining the temperature distribution in a semi-infinite solid
medium along whose surface a point heat source is moving in a straight
line with constant velocity. The medium is the mainstream gas, the
strength of the source is determined from the net enthalpy flow added
through the hole and the velocity of the source in that of the free stream,
though in the reverse direction. A major difficulty (and approximation)
is to evaluate an effective thermal conductivity or thermal diffusivity
of the mainstream.
T h e resulting temperature distribution in the mainstream is,

(49)
Along the adiabatic surface Y = 0,

rl(x' )' = +
IMUmD
8r(X/D 0.5)
exp [--0.693 (T)2 ' ]
112

The reference coordinates used in Eq. (49)are shown in Fig. 2; Y is the

vertical distance from the surface, 2 is the lateral distance from the
center plane of the injection hole, and X is the distance downstream
from injection. The distances Y1l2and 21/2 are distances at which the
temperature difference drops to half its value along the centerline on
the tunnel surface. Note that the effectiveness is now a function of
lateral position from the hole centerline as well as of downstream
location. T h e form of this relation is found to approximate the experi-
mental data at low blowing rates and thus it has proven useful. Extension
to higher injection rates faces formidable obstacles, particularly as the
jet appears to leave the surface at large blowing rates.
342 RICHARDJ. GOLDSTEIN

Either Eq. (50) or the direct experimental results for single hole
injection can be used to predict the film cooling performance of a row
of holes using the principle of superposition. As long as the flows from
the individual jets do not interact greatly, superposition appears to
work reasonably well (28). At large blowing rates and far downstream
the jets come together and superposition of single hole results to predict
film cooling from a number of holes cannot be used.

IV. Experimental Studies

A. GENERAL
REMARKS
A summary of some of the experimental studies of film cooling is
presented in Table I. A brief description of the geometry of secondary
injection as well as the range of pertinent experimental variables is
included. Discussion of the results of the individual investigations is
given below.
In many of the tests, film “heating” rather than film cooling is
employed for reasons of convenience. With small temperature differ-
ences the flow can be considered constant property. Then, if there
is no radiant energy input, the dimensionless temperature distribution
in the boundary layer (and thus the film cooling effectiveness) will be
independent of whether the secondary gas is hotter or colder than the
mainstream. This has been discussed previously in the section on
film cooling effectiveness and is also applicable to the determination
of the heat transfer coefficient when a heated or cooled wall is used.
In some applications, however, there can be a considerable temperature
difference between the mainstream and the coolant. Then a key parameter
may be the ratio of the densities of the two fluids. This can, of course,
be studied using real temperature differences, but attainment of adiabatic
wall conditions is very difficult with large temperature differences,
so several investigations have utilized the mass transfer analogy. The
equivalent of the film cooling effectiveness is then the impermeable
wall effectiveness as discussed above. This method is convenient even
at small temperature (density) differences as there are always errors
introduced due to thermal conduction in the wall when finite temperature
differences are employed. This is particularly true in three-dimensional
film cooling though the analogy has not been applied there as yet.
One difficulty in using the mass transfer analogy is the simulation of a
(heat) transfer process at the wall-gas interface. So far only impermeable
surfaces have been used. There is also the lingering question of the
direct equivalence of the analogy; Burns and Stollery (29),in particular,
question whether the turbulent Lewis number is unity.
 FILMCOOLING 343

FILM COOLING-INCOMPRESSIBLE
B. TWO-DIMENSIONAL FLOW

1 . Injection of Air into Air-Constant Mainstream Velocity

T h e first well-known study, Wieghardt ( l o ) ,on film cooling not only
used a heated secondary gas, but was specifically applied to a film-heating
problem, namely, the de-icing of airplane wings. Air was ejected through
a slot (see Table I) inclined at an angle of about 30" to the wall surface.
Note that as in most studies in which injection occurs through a slot
the secondary flow was not fully developed nor was the temperature
profile completely uniform at the slot exit. As previously noted,
Wieghardt found that the temperature profiles were approximately
similar and could be expressed (Fig. 6) in the form
T,
Taw
-

-
T,
T,
- e~p[-0.768(y/6,)~~/~]

Some distance downstream of injection, temperature distributions similar

to this have been found in other film cooling studies indicating the
relative insensitivity of the temperature profile to the secondary injection
geometry. At high blowing rates and near the injection location, however,
similarity of the temperature profile may not be assumed.
Wieghardt found a maximum in film cooling effectiveness at a blowing
rate of about unity (Fig. 8). He was able to correlate his adiabatic wall
temperature distribution at low blowing rates ( M < 1) and some
distance downstream (x/s > 100) with a simple relation
71 = 2 1 . 8 [ ~ / M ~ ] p O . ~ (52)
which, since his range of slot Reynolds number Re, was small, is not too
different from some of the predicted results of the heat sink model.
This simple equation, even today, is widely used to obtain initial
estimates of film cooling performance.
Other workers (30), using the same geometry as Wieghardt, found a
relation similar to Eq. (52) although the numerical value of the constant
was lower (16.9 instead of 21.8). Eckert and Birkebak (31) using the
same geometry were able to correlate their results with Eq. (52).
Others (32) studied film cooling with injection through both a normal
and a tangential slot. For normal injection they found
7 = 2.2(x/Ms)-O.5 (53)
Seban (33) studying the film cooling effectiveness downstream of a
stepdown slot (Fig. lb) correlated his data at low blowing rate ( M < 1)
with the relation
71 = 2 5 M 0 . 4 ( ~ / M ~ ) - 0 . 8 (54)
 TAR1,E I
Density Temp. Blowing Free
Injection ratio, ratio,rate, M = stream
Ref. Geometry gas" pJpm T2/Tm pzUzlpmUm Mach No.

Wieghardt (10) 0.78 1.1 0.22 0.046

to to to to
0.91 1.2 1.9 0.092
+ wall jet
Scesa (60) Air 0.81 1.09 0.2 0.025
to to to to
0.91 1.23 1.14 0.065

Seban, Chan Air 0.81 1.09 0.08 0.037

Scesa (32) to to to to
0.92 1.23 0.916 0.108

Air 0.81 1.09 0.19 0.02s

to to to to
0.92 1.23 1.14 0.065

Chin, Shirvin,
Hayes, Silver
(35)
@m
?,; Air 0.83
to
1.17
0.85
to
1.20
0.25
to
2.5
0.056
to
0.145

7
-
Papell, T r o u t 0.52 0.32 0.0 0.1 5
(39) to to to to
3.54 1.8 13.9 0.80

Hatch, Papell
(11)
7 He 2.45
to
0.34
to
0.018
to
0.53
to
6.5 0.67 1.55 0.57
__.c

Papell (53) Air 1.2 0.37 0.05 0.20

to to to to
4.8 0.85 12.0 0.70
AT ANGLES 90:80:45'

Seban (33) /%-//// Air 0.88 1.14 0.17

to
0.0045
to
u,_ 20.8 0.13

Seban (34) /KL//// Air 0.88 1.0

to
0.27
to
0.09

1.14 0.76
u,_
Chin, Skirvin, 1.15 0.87 0.0512 0.0822
Hayes, Burggraf to to to to
(38) 1.13 0.887 1.026 0.152
MULTIPLE SLOTS

" Unless otherwise noted, mainstream is air. Tests also for single and double rows of holes.
to slot of 10.25 mm and 5.65 m m . ' Values assume same as Seban (33).
EXPERIMENTAL I N FILMCOOLING
STUDIES

Velocity Velocity Slot Starting length Measured Slot

ratio, Urn size, Reynolds No., parameters,Effec- Reynolds Range of
U2/U, mlsec mm Re,, &*Is q/Taw/Cw tiveness No.,Re, wls

0.246 15.8 5.0 8.0 \ los 0.16 Taw 1.0 360 80

to to 10.0 to to to to to
2.44 32.0 15.0 .: lo5 0.36 0.05 1200 800

0.286 9.7 3.175 1.2 \ 105 0.22 Q 0.60 550 10

to to to to Taw to to to
1.35 20.4 2.9 Y lo5 0.33 0.095 2500 130

0.097 13.0 3.175 1.3 K lo5 0.205 q 1.o 580 10

to to to to Taw to to to
1.06 37.0 3.8 'i IW 0.216 0.05 2600 130

0.286 9.7 3.175 1.0 x lo5 0.212 q 0.7 550 10

to to to to Taw to to to
1.35 20.4 2.09 x 105 0.222 0.095 2500 130

0.26 18.9 2.7 15.0 x 106 0.94 Taw 1.0 1400 9

to to to to to to to
2.85 54.0 160.0 X 10' 2.36 0.13 8200 233

168.0 1.59 1.o 0 0

to 3.175 to to to
425.0 6.35 0.0 450,000 540
12.7

0.095 256.0 3.175 Taw 1.0 672 3.18

to to 6.35 to to to
3.84 316.0 12.7 0.16 46,802 271

0.036 147.0 6.35 Taw 0.95 2,500 4

to to slot to to to
2.55 395.0 6.35 0.22 260,000 152
holes

0.193 1.52 1.59 0.4 x lo5 0.03 4 0.95 620 5

to to 3.175 to to Taw to to to
23.6 45.8 6.35 1.1 X lo5 0.12 0.04 7950 300

0.31 30.5 1.59 0.76 X 106 0.03 4 0.75 760 ' 2

to 3.175 to to Taw to to to
0.865 6.35 13.4 x lo6 1.2 0.05 4400 300

0.057 30.0 2.92 10.3 Y lo5 Taw 1.00 413 7.9

to to to to to to
1.13 56.0 19.5 x 105 0.174 6100 177.8

Estimated. Also considered injection through multiple rows of discrete holes. Holes equivalent
 TABLE I
Density Temp. Blowing
Free
Injection ratio, ratio, stream
rate, M =
Ref. Geometry gas' PslPm Ta/Tm PSUpIpmUm Mach NO.

Burggraf, Chin, 1.15 0.87 0.0861 0.0671

Hayes (70) to to to
1.13 1.25 0.1712
MULTIPLE LOUVERS

Hartnett, 0.78 1.02 0.265 0.145

Birkebak, to to to
Eckert (30) 0.98 1.27 0.288

Hartnett,

Eckert
Birkebak,
(46) #
7B
. .. .
Air 0.875
to
0.935
1.07
to
1.14
0.28
to
1.23
0.1 185

Seban, /m&///d Air 0.90 1.03 4.95 0.00965

Back (15) to to to to
0.97 1.11 12.6 0.0149

Seban, Air 0.88 1.13 O.Oh 0.04

Back (14) to to
& 0.70 0.095

Nishiwaki, 0.825 1.21 0.0 0.0 17

Hirata. to to
Tsuchida (41) 0.0975 0.086

/M-///L
~ ~ ~

Seban, Air 0.87 1.0 0.2 0.083

Back (13) to to to to

r&7m
1.00 1.15 0.9 0.1 10
&
Eckert, 0.87 1.15 0.19 0.14
to
Birkebak (31) Air 0.93

Air 0.83 1.05 0.012 0.0961

Shavit, to to to to
Chen (3) 0.95 1.207 0.040 0.1615

Samuel,
4% Air 1.1 I 0.78 0.25 0.040
Joubert (37) to to to to
1.28 0.905 3.18 0.085

Goldstein, Air 3.4 0.8 0.0 3.01

Eckert, Tsou to to

Th
to
Haji-Sheikh (63) 2.04 1.25 0.408
He 0.3 0.31 0.01 3.01
to to to
0.4 0.39 0.02

Velocity accelerated downstream of injection slot to 2.5 and 1.6 times initial values, Values
measured velocitv data.
EXPERIMENTAL STUDIES I N FILMC O O L I N G (Costinrted)
Velocity Velocity Slot Starting length Measured Slot
ratio, urn size, Reynolds No., parameters,Effec- Reynolds Range of
U,/Um mbec mm Re,, &*is tiveness
q/Taw/Cw No., Re, r/s

0.099 25.2 1.62 8.62 x 105 Taw 0.999 530 6.6

to to to to to to
1.44 52.0 18.0 '< 105 0.139 7700 390.7

0.294 50.0 3.12 6.1 x 105 0.244 Taw 0.85 1510 4

to 9 to to to
0.333 0.125 2880 140

0.31 42.0 3.11 4.97 lo5 0.2 Taw 0.96 2,200 6

to to to to
1.37 0.19 10,OOO 198

5.1 3.4 1.59 Taw 1.0 3530 17

to to 3.175 Q to to to
14.0 5.2 6.35 0.2 6960 300

0.0 A 15.8 1.59 0.41 < lo5 0.3 * 9 0.95 0 10

to to 3.175 to to Taw to to to
0.795 29.5 6.35 0.76 x lo5 1.2 0.06 6600 150

0.0 6.0 5.0 Taw 0.7 0 1.4

to to 20.0 to to to
0.118 30.0 50.0 0.035 2200 180

0.20 29.0 1.59 3.3 x 105 0.24 Taw 0.95 lo00 5

to to 3.175 to to to to to
1.03 38.0 6.35 16.8 x lo5 1.36 0.06 7000 280

0.218 50.5 3.2 6.5 x. 105 0.2 T., 1.0 1800 2

to to to to
1.07 0.10 8500 275

0.013 30.5 35.6 0.59 ~-.lo5 0.0359 Taw 0.85 850 1.5
to to to to to to to
0.042 55.0 1.06 x lo5 0.0403 0.05 5800 17.44

0.23 15.2 3.175 Taw 1.0 1420 3.6

to to 6.35 to to to
2.48 30.5 9.525 0.2 22,900 275.0

1050 1.63 4.0 Y 105 0.045 Tow 1.0 0 o j

3.12 to to to to
4.62 0.127 0.1 11,100 73

I050 1.63 4.0 x 105 0.127 Taw I .o 0 0 '

to to to
0.1 23 I 13

taken from Seban (33). ' Based on total step height. ' Based on tunnel dimension rather than
 TABLE I
Density Temp. Blowing Free
Injection ratio, ratio, rate, M = stream
Ref. Geometry gas" pJpm T2/Tm p z U z / p m U m Mach No.

Mabuchi (43) Air 0.788 1.14 0.02 0.02

to to to to
0.878 1.27 0.146 0.03 1

yTgr/
I I- Air 0.84 to 1.12 to 0.013 to 0.095 to
Goldstein, Rask, 0.88 1.20 0.052 0.16
Eckert (42) He 0. i 2 1. I2 to 0.0022 to 0.095 to
1.20 0.0076 0.16

' k rm
~~

Nicoll, Air with 1.o 1.o 0.47 0.06

e$j)3
Whitelaw (18) to
He tracer 2.26

e-m
Metzger, Air 0.96 1.05 0.25 0.04
Carper, to to
Swank (54) 1.49 0.07

Kacker, Air with 1 .o 1.o 0.3 0.06

Whitelaw (36) He tracer
2.1
to

k 1.o I .o

T-
Whitelaw (40) Air with 0.47 0.06
to
2.24
He tracer

Carlson, *&& NP
2.76 0.363 0.5
to
1.98
0.1
to
0.5

Talmor (49) NP 2.76 0.363 0.563 0.115

to to
2.66 0.242

Escudier, 1 .O I .o 0.023 0.0702

Whitelaw (47) to
0.074

Goldstein, 0.33 0.77 0.0085 2.9

Eckert, to to to
Wilson (24) 0.48 1.14 0.0223

Goldstein, 0.84 1.2 0.10 0.08

Eckert, Ramsey, to to
(71) and (72) 2.0 0.17

______

* From Whitelaw (40). As temperatures are presented for only one case, assumed same for all
given boundary thickness. Nitrogen used as mainstream gas. Laminar boundary layer ahead of
EXPERIMENTAL
STUDIFSIN FILMCOOLING
(Continued)
Velocity Velocity Slot Starting length Measured Slot
ratio, Urn size, Reynolds No., parameters, Effec- Reynolds Range of
U,/Um m/sec mm Re,, S*/s q/Tax/Cwtiveness No., Re, x/s

0.26 7.0 15.1 8.0 ’ lo5 0.011 Taw 0.95 358 1

to to to to to to to to
0.185 10.0 51.0 22.0 / lo” 0.092 0.06 1320 35

0.015 to 33.2 to 2.54 8.0 / lo5to 0.037 to Taw 0.80 to 700 to 0.78 to
0.0605 55.0 14.0 lo5
/ 0.046 0.04 4000 31.0
0.018 to 33.2 to 2.54 8.0 lo” to 0.037 to Taw 0.80to 150to 0.78 to
0.063 55.0 14.0 lo5
/ 0.046 0.03 300 31.0

0.47 21.4 6.425 2.4 A lo” 0.107 C, 0.95 4,035 4

to to to to
2.26 0.20 19,500 218

0.26 15.2 0.907 Q ” 312 l/s = 35

to to 2.54 Taw to to
1.55 24.4 2420 70

0.3 21.4 1.87 1.8 lo5 0.3 c w 0.7 730 50

to to to to to to
2. I 11.0 . 105 1.28 0. I 5000 200

0.47 21.2 6.425 2.4 f lo5 0.095 Cx 1 .o 4,035 4

to 7.25 to to to to
2.24 0.107 0.26 19,500 218

0.17 1.59 Taw 1.0 10,350

to to to
0.67 0.1 1 565,000

0.19 1.59 Taw 1.0 13,300

to to to
0.90 0.1 1 368,000

0.023 24.4 25.4 cx I .o 958 0.625

to to to to
0.074 0.1 3000 20.6

1025 12.6 6.0 lo5 0.026 Taw 0.9 I200 0.25

to to to
0.12 3700 8.0

0.12 30.5 23.5 6.0 lo5 0.033 Taw 0.85 10,000 2.0
to to to to to to to
2.38 61.0 21.0 lo” 0.058 0.005 100.000 40.0

runs. ’“ Average heat transfer over section from s = 0 to x = I was measured. “ Calculated from
injection. ‘I Hole diameter. ‘ From kk’hitelaw (40).
 TABLE I
Density Temp. Blowing Free
Injection ratio, ratio, rate, M = stream
Ref. Geometry gas" pr/p~ T,/Tm prUp/pmVmMach No.

I .o I .o

@m
Kacker, Air with 0.288 0.06
He to
Whitelaw (50) tracer 2.66

Air 0.07 1.0 0.021 >0.13

Pai H to to
Whitelaw (59) Air with 4.17 6.87
He tracer

k
rm
-
Argor (Refrig. 12)

Kacker, Air with 1.o 1.o 0.2 0.055

Whitelaw (51) He tracer
to
2.4

/,m,
urn_ Arcton 12
(Refrig. 12)
4.17 1.0 2.21
to
16.7
0.017
to
0.050
Burns,
Stollery (29) He 0.14 1.0 0.071 to 0.050
0.236

4.17 to 1.0 0.14 to 0.050

E;ror Air-Arcton 0.14 4.17
~

Metzger, Air 0.96 1.05 0.25 0.04

Fletcher (45) to to
0.75 0.07

Psi,
Whitelaw (48) @% Hydrogen
Arcton 12
0.069
to
4.17
1.0 0.021
to
6.85
0.03
to
0.06

Williams (58) Nitrogen' 2.38 0.33 0.308 0.04 y

to to to to
3.07 0.42 2.99 2.5
~~ ~ ~~

Goldstein,
Ramsey Eriksen,
Eckert, (28) &
& Air
0.85 1.18 0.1
to
2.0
0.088
to
0.176

* Value for S taken form Whitelaw (40). Based on maximum possible velocity. " Average heat
gradients. Air-hydrogen combustion products form mainstream gas. Accelerated flow. y Hole
EXPERIMENTAL
STUDIES
IN FILMCOOLING(Continued)

Velocity Velocity Slot Starting length Measured Slot

ratio, U, size, Reynolds No., parameters, Effec- Reynolds Range of
U,/Um m/sec mm Re,, S*/s q/Taw/Cativeness No., Re, XIS

0.288 21.4 1.88 2.4 Y lo5 0.0542 * Cw 0.95 745 25

to 3.35 to to to to
2.66 6.35, 12.7 0.366 0.15 44,OOO 150

0.55 10.1 2.54 cw I .o 70.8 2.5

to to to to to
2.21 20.8 0.005 14,250 212.5

0.2 20.8 6.26 2.17 x lo5 0.107 Cw 0.97 1,500 12

to to to to to to
2.4 4.35 x 105 0.191 0.20 18,400 210

0.53 6.1 1.59 0.3 Y lo5 0.236 Cw 1.O 2,220 0

to to to to to to to
4.00 17.4 0.85 i lo5 0.28 0.3 17,420 512

0.51 to 16.8 1.59 0.85 i los 0.236 Cw 0.9 to 113 to 0 to

I .68 0.05 368 512

1.o 17.6 1.59 0.85 '~ 10'to 0.236 to Cw 1.0 to 228 to 0 to

3.03 Y lo5 0.66 0.05 11,300 512

0.26 15.2 1.27 4 " 325 11s = 5.0

to to 2.54 Taw to to
0.78 24.4 3500 70.0

0.55 10.0 2.54 cw 1 .O 70.8 0

to to to to to
2.21 20.7 0.01 14,400 212

0.127 38.8 0.635 Taw 1.O 3,100 12.7

to 1.522 to to to
I .09 0.16 24,100 138

0.118 30.5 11.8 9.7 lo5 0.052 Taw 0.85 5200 3

to to to to to to to
2.36 61.0 17.0 \ lo5 0.125 0.0 52000 80

0.588 30.5 11.8 9.7 Y lo5 0.052 Taw 0.4 12900 3

to to to to
2.36 0.0 51800 80

transfer over section from x = 0 to x = I was measured. " Favorable and nonfavorable pressure
diameter.
352 RICHARDJ. GOLDSTEIN

At large blowing rates several different empirical and semiempirical

relations were used to approximate the wall-jet-like effects observed.
Seban also investigated (34) the influence of mainstream boundary
layer thickness at the point of injection through a tangential slot. Only
a very slight decrease in film cooling effectiveness was found with
increased boundary layer thickness at the point of injection [cf. Refer-
ences (29, 35, and 36)].

0.2 0 -4 08 0.8 I 2

FIG. 8. Film cooling effectiveness at I/S = 100 as a function of blowing rate M .

[K. Wieghardt, AAF Translation No. F-TS-919-RE (1946).

Several investigators have used stepdown slots and correlated their

data with (different) empirical relations (35, 37); others have also
studied the effect of multiple slots (38). Papell and Trout (39)
using tangential injection measured the film cooling effectiveness at
very large temperature differences. Papell and Trout correlated their
results with empirical and semiempirical correlations. Whitelaw (40)
measured the impermeable wall concentration for air injection with
helium as a tracer gas. The orders of magnitude of the results were
found to be similar to previous film cooling effectiveness measurements.
The possibility of turbulent Lewis numbers different from unity was
suggested. This would reduce the value of a direct comparison, but
 FILMCOOLING 353
would still permit impermeable wall tests to suggest trends and give
relative results.
Studies have been made of film cooling downstream of a porous
section through which air was injected (3,41-43). Note that with normal
blowing the velocity distribution near the porous section is severely
affected (Fig. 9). T h e results of these different investigations agree
quite well. A comparison with results for other geometries is shown
in Fig. 10.
I

FIG. 9. Effect of injection on boundary-layer velocity profiles with relatively large

blowing rate through 35.6 nim porous section with a trip wire 11.2 cm upstream of
injection. Data: m2 = 22.55 gmlsec, M = 0.0389, U = 56.0 m/sec. Distance from trip
wire (cm): (A) 6.35, (B) 10.16, (C) 15.88, (D) 21.60, (E) 27.95, (F) 53.40, ( G ) 68.50.
[R. J. Goldstein, G. Shavit, and T. S. Chen, /. Heat Transfer 87, 353 (1965).]

T h e film cooling results for porous injection have been found to

agree relatively well with the analyses of Librizzi and Cresci ( 4 ) ,
Kutateladze and Leont’ev (5), and Stollery and El-Ehwany (6).Actually
there is little difference between these three models, and the resulting
equations approach the same value far downstream.
Comparisons of the Kutateladze and Leont’ev relations with some of
the data for tangential injection are shown in Fig. 1 1. Agreement is quite
good. Figure 12 shows a comparison of some porous wall film cooling
experiments with the predictions of Tribus and Klein (Eq. lo), Librizzi
and Cresci (Eq. 22)’ and Goldstein and Haji-Sheikh (Eq. 38). Note the
3 54 RICHARDJ. GOLDSTEIN

relatively good agreement with the latter two analyses. In Fig. 13,
Eq. (38) is observed to compare favorably with the film cooling effec-
tivenesses obtained for tangential injection. The same relation (Eq. 38)
when integrated to predict average values over the length of a wall (44),
gives good agreement with the average film cooling effectiveness
measured for injection through angled slots by Metzger and Fletcher (45).

I I I I I
.02
10 20 40 60 8ODO 200 400 1000
x/Ms

FIG. 10. Comparison of film cooling effectiveness as determined in various in-

vestigations.
Data: (111//////), Goldstein et al., (++) Wieghardt 7 = 2 1 . 8 ( x / M ~ ) - ~ . ~ ,
(-.-) Hartnett et al. 7 = 16.9(x/M~)-~*~,
(----) Nishiwaki et al. 7 = 1.77 (x/Ms)-O.~,
(--..-) Scesa 7 = 2.20(x/M~)-~.~,
(-.-) Seban 7 = 25.0M0.4(~/Ms)-0.8,
(- -) Hatch et al. q = 1.31 exp(--0.229 Re&* (xh-' - l)/M).
[R. J. Goldstein, G. Shavit, and T. S. Chen, J: Heat Transfer 87, 353 (1965).]

2. Variable Free Stream Velocity and Free Stream Turbulence

Several investigators (14, 46) have reported studies of film cooling
on surfaces with variable mainstream velocity (mainly accelerating).
Little change in the boundary-layer temperature profiles was observed
 FILMCOOLING 355
I .o

F 0.8
u)

y
u)
0.6
W
L
k-
0.4
LL
LL
W

I3
z-I 0.2
0
0
0

5
LL 0.01
10 20 40 60 80 100 200 400 600
- X
Ms

FIG. 11. Comparison of effectiveness for tangential slot geometry with analysis of
Kutateladze and Leont’ev. (-) Eq. (25); ( 0 ) R. A. Seban, J. Heat Transfer 82,
303 (1960); ( v ) S. Papell and A. M. Trout, NASA Tech. Note TN D-9 (1959). [S. S.
Kutateladze and A. I. Leont’ev, Thennal physics of high temperatures 1, No.2, 281-290
(1963).]

and the film cooling effectiveness could be found by multiplying the

effectiveness predicted for uniform mainstream flow by a function of
the local velocity (46). T h e relatively small change in effectiveness
was attributed (14) to the thermal boundary layer being considerably
thicker than the velocity boundary layer. Very strong acceleration
( U , increasing by a factor of 24 or 3) caused a slight decrease in
effectiveness.
Escudier and Whitelaw (47) measured the impermeable wall effec-
tiveness for injection through a porous section with strong adverse
pressure gradients. Little influence of pressure gradient on effectiveness
was observed up to separation, agreeing with the earlier studies
in a favorable pressure gradient (45,46). T h e small effect observed
was an increase in effectiveness. Pai and Whitelaw (48) found little
influence of a favorable pressure gradient on impermeable wall effec-
tiveness unless the boundary layer ceases to be fully turbulent.
Carlson and Talmor (49) report a large change (decrease) in film
cooling effectiveness with acceleration of the free stream. I n their
apparatus the test wall along which the secondary gas is injected is not
flat; a substantial bend occurs at the point of injection, which may
produce separated flow.
They also indicate that increasing the free stream turbulence
356 RICHARDJ. GOLDSTEIN

iy -~

P .lo
9 GEOMETRY BOTH REFERENCES

---
__--
TRIBUS AND KLEIN EON. 10
L l B R l t Z l AND CRESCI EQN. 22
GOLDSTIEN AND HAJI-SHEIKH EQN. 38

SOURCE Re2 M
z V REF.42 982 0.0127
4 A4
LL A REF.42 81 6 0.0155

t
0 REF.42 4444 0.0517
o REF.3 4361 0.0400

*02

I 2 4 6 810 20 40 60 80 100

FIG. 12. Film cooling effectiveness with injection of air through a porous section
including comparison with several analyses. [R. J. Goldstein and A. Haji-Sheikh, in
Japan SOC.Mech. Engr. 1967 Semi-Intern. Symp., pp. 213-218, Tokyo (1967).]

at the slot location can significantly reduce film cooling performance

due to the greater mixing of the secondary gas and the free stream.
Going from a free stream turbulence intensity of 3 to 22% almost
halves the effectiveness some distance downstream of injection. Kacker
and Whitelaw (50) changed the turbulence intensity of the secondary
gas in the injection slot from 5.5 to 9.5 yo and found no significant change
in impermeable wall effectiveness.

3. Slot Geometry
For tangential injection the ratio of lip thickness t to slot opening s
can influence the film cooling effectiveness particularly when the velocity
 FILMCOOLING 357

I .o
.8

.6

F
.4
-
GEOMETRY BOTH REFERENCES
I
ln
ln
w .2
2
W
>
F EOUATlON 38
U
w
h.10
w
-

w .oe - RANWMLY SELECTED DATA FROM REF13 ANQ REF33

E
.06 SYMBOL S
0 M Re2
(mm)
I 0 1.6 0.18 620
4 .04
LL a 6.35 0.26 2420
D 1.6 0.39 1360
0 3.175 0.39 2120
3.175 0.58 3970
.o 2 0 3.17s 0.18 6220

.o I
I 2 4 6 8 1 0 20 40 60 80 100

FIG. 13. Film cooling effectiveness with tangential injection including comparison
with analysis of Goldstein and Haji-Sheikh. [R. J. Goldstein and A. Haji-Sheikh, in
Japan SOC.Mech. Engr. 1967 Semi-Intern. Symp., pp. 213-218, Tokyo (1967).]

ratio U,/U, is near unity. This has been demonstrated using the mass
transfer analogy by Kacker and Whitelaw (50,51) and Burns and
Stollery (29). Figure 14 shows the effectiveness for different values of t / s .
For a lip thickness less than about 40% of the slot opening, the effects
are small. T h e influence of lip thickness also diminishes as the velocity
ratio U,l U , is decreased. Similar phenomena are reported by Sivasegaram
and Whitelaw (52).
T h e significant reduction of film cooling effectiveness that occurs
for large lip thicknesses is probably due to the pronounced separated
and reverse flow region at the lip edge. Under those conditions the
simple heat sink models cannot be used directly, though Eq. (25) and
358 RICHARDJ. GOLDSTEIN
I.o
0.8

0.6

’ 0.4

0.2
10 20 30 40 50 100
x /s
FIG. 14. Effect of increasing slot lip thickness on impermeable wall effectiveness for
0.126, ( A ) 0.38, ( v ) 0.63,
tangential injection at p 2 / p m m 1 and U,/U.Z= 1.07: t / s : (0)
( 0 ) 0.89, (+) 1.14, ( 0 ) 1.90. [S. C. Kacker and J. H. Whitelaw, J. Mech. Engr. Sci. 11,
22(1969).]

(38) could possibly be modified to account for the role of geometry

in the entrainment of mainstream flow into the boundary layer.
Sivasegaram and Whitelaw (52) report the effect of injection angle
on film cooling effectiveness. As expected [cf. Eq. (38)] the larger the
angle the smaller the film cooling effectiveness due to the greater mixing
of the coolant with the mainstream at the point of injection. Papell (53)
and Metzger and co-workers (45,54) find a similar trend.

4. Effect of Large Temperature DajGerences

Few experimental studies have used the extreme temperatures that
might be encountered in film cooling applications. Large temperature
differences can introduce significant errors in assumed boundary condi-
tions and make accurate measurements difficult, particularly in getting
adiabatic wall temperature distributions.
Film cooling studies with large temperature differences include those
by Papell and Trout (39) and Papell (53) on a flat plate. In their tests
the temperature of the hot gas stream was as high as about 800K.
Milford and Spiers (55) examined film cooling in a gas turbine combus-
tion chamber at temperatures to 1950K. Lucas and Golladay (56, 57)
measured film cooling performance in rocket nozzles and combustion
chambers with free stream gas temperatures up to 3000K. Williams (58)
studied film cooling in a rocket nozzle with a free stream temperature
of about 870K. Other studies made at high temperatures are discussed
in the section on compressible flow film cooling.
 FILMCOOLING 359

5. Foreign Gas Injection

There have been few studies of film cooling with the heated or cooled
injection of a foreign gas into an air mainstream. Hatch and Papell (ZZ)
injected helium through a near tangential slot into a hot air mainstream.
Other workers (42) injected heated helium through a porous section.
Burns and Stollery (29) find relatively close agreement between these
data and a correlation similar to Eq. (31) though the constant is consider-
ably larger, having been increased empirically to give the best fit with
experimental data.
T h e adiabatic wall temperature results for helium injection from (42)
were found to be somewhat higher than the prediction of Eq. (21)
and (24). However, Fig. 15 shows that Eq. (38) fits the data relatively
well.
Investigations with isothermal foreign gas injection have been
performed to study the effect of a density difference between the injection
gas and the mainstream using the mass transfer analogy. Nicoll and
Whitelaw (18) and Burns and Stollery (29) used the mass transfer
analogy, injecting foreign gases through tangential slots into an air
mainstream and measuring the impermeable wall concentration.
Figure 16 from Reference (29) shows the variation of impermeable
wall concentration at a velocity ratio U J U , close to unity for different
density ratios, p 2 / p m . I n this study the influence of boundary layer
thickness on effectiveness is found to be small. T h e thickness of the
slot lip plays a significant role near the slot for a relatively light injection
gas. A decrease in effectiveness is found with increasing lip thickness
which is attributed to increased mixing in the separated region imme-
diately downstream of the lip.
With the heaviest coolant, increasing the velocity ratio U,l U ,
increases the effectiveness, though past unity the increase is small.
For helium injection the effectiveness continues to increase considerably
even for velocity ratios U,/U, greater than unity.
Pai and Whitelaw (59) measured the impermeable wall effectiveness
downstream of a tangential slot through which hydrogen, air (with a
helium tracer), argon or Arcton 12 (Refrigerant 12) were injected.
With injection of a relatively dense gas the effectiveness reaches a
plateau at a velocity ratio U,/U, of about unity (Fig. 17). For light gas
injection increase of the velocity ratio above unity continues to yield
further increases in effectiveness. For air injection, the effectiveness at
most locations increases with blowing rate, finally reaching a plateau
at a velocity ratio U,lU, of about unity. From unity to the highest
velocity ratio used (-3.1) the effectiveness stays approximately constant.
360 RICHARDJ. GOLDSTEIN

SYMBOL AT( C ) Tar C) M

Po0
I
4.04- V 44.8 27.2 0.0022 2 I83
-
U
D 45.7 27.2 0.00330 185
0 58.7 29.2 0.00442 360
0 56.0 30.1 0.00683 364
.02 - -
WTA FROM REF. 4 2

.o I 1 I I 1 I I I

FIG. 15. Film cooling effectiveness with injection of He through a porous section
into a mainstream of air including comparison with analysis of Goldstein and Haji-Sheikh.
[R. J. Goldstein and A. Haji-Sheikh, in Japan SOC.Mech. Engr. 1967 Semi-Intern.
Symp., pp. 213-218, Tokyo (1967).]

It should be noted that the lip thickness t was about 60% of the slot
opening in these tests. Their results were in reasonable agreement
with calculations made using the turbulent boundary-layer equations.

6. Heat Transfer
Measurements have been made of the heat transfer with film cooling
on a surface over which a uniform mainstream flowed (30) and on a
surface with a pressure gradient (46). Except at large blowing rates, they
reported that shortly downstream of injection the heat transfer coefficient
 FILMCOOLING 361

reduces to the heat transfer coefficient one would expect with no blowing
(Fig. 18). Near the injection region the blowing usually causes a slight
increase in heat transfer coefficient. T h e driving force in defining the
heat transfer coefficient is the difference between the actual wall tempera-
ture and the adiabatic wall temperature. Scesa (60) and Seban and
co-workers (24,32,33) found similar results in that the heat transfer
coefficient was not significantly altered by blowing, although in these
studies the heat transfer coefficient was sometimes found to be reduced
slightly by the blowing. T h e difference in injection geometry used in (30)
as compared to that used in (33) may account for this different trend.
Metzger and co-workers (45,54) observed a slightly larger effect
(increase) on heat transfer coefficient than the other studies, particularly
at large blowing rates and close to the injection location.

FIG. 16. Effect of density ratio, p a / p m , at a velocity ratio U,/Um 1 on impermeable

wall effectiveness for tangential injection. [W. K. Burns and J. L. Stollery, Intern. /. Heat
Mass Transfer 12, 935 (1969).]

C. TWO-DIMENSIONAL
FILMCOOLING
COMPRESSIBLE
FLOW
Many applications of film cooling occur in high-speed flows. Although
the incompressible flow results can often be used for compressible
flow problems, this transformation must be checked experimentally.
This is particularly true if the wall geometry is such as to produce
shock interactions in the film cooled region.
I n several reports (62-63) measurements have been made of the
adiabatic wall temperature distribution downstream of a step-down
slot in supersonic flow. Either air or helium could be injected tangentially
into an air mainstream which had a Mach number of approximately
362 RICHARDJ. GOLDSTEIN

1.0 2.0 3 .Q
VELOCITY RATIO U2IUm

FIG. 17. Impermeable wall effectiveness for air injection with He tracer as a function
of velocity ratio U , / U m . [S. C. Kacker and J. H. Whitelaw, J. Heat Transfer 90, 469
(1 968).]

three. Both heated and cooled secondary flows were used. Due to the flow
over the edge of the splitter plate (separating the secondary and main-
stream flows) there is an expansion fan, a lip shock, a separated region,
and a reattachment shock, whose magnitudes are dependent on the rate
of secondary mass addition. T h e effect of blowing rate on the flow field
is shown by schlieren photography in Fig. 19a and 19b. At the larger
secondary flow rates choking occurs in the injection slot.
T h e results were correlated, using a film cooling effectiveness based
on the isoenergetic flow conditions as described earlier. I n measuring
the isoenergetic film cooling effectiveness two test runs are required
 FILMCOOLING 363

2.0

I .8 GEOMETRY

I. 2

I. 0

I I I

at each blowing rate-one to obtain the isoenergetic wall temperature

distribution and the other to obtain the film-cooled wall temperature
distribution. Since the total temperature of the mainstream may be
somewhat different in the two runs it is useful to normalize the tempera-
tures in Eq. ( 5 ) by dividing them by the total temperature of the free
stream for the test in which they are obtained. An empirical correlation
of the results for air injection at low blowing rates (Fig. 20) is

771s = 550(&2.0 for M < 0.12 (55)

where
5 =(x/~’)[(i/~)~o.4 (56)
Note that the step height h‘ (slot height plus lip thickness) is used,
indicating the importance of the geometry. These test results are of
the order of magnitude of the Tribus and Klein correlation indicating
higher values of effectiveness than is found in most of the subsonic
studies. At higher values of blowing rate ( M > 0.12) the effectiveness
results are considerably higher than even the Tribus and Klein equation,
the empirical correlation over the range of parameter studied being

vis = 1 6 2 ( ~ / M h ’ ) - l . ~ for 0.12 < M < 0.408 (57)

A comparison of these empirical correlations and some of the other
predictions is shown in Fig. 21. Note that, compared to subsonic correla-
tions, the supersonic results for injection through a tangential slot
 M - 0.136
FIG. 19a. Schlieren photographs for injection through a step down slot in supersonic
flow for mainstream Mach number = 3.01, slot opening s = 4.62mm, step height
h' = 6.07 mm. [R. J. Goldstein, E. R. G. Eckert, F. K. Tsou, and A. Haji-Sheikh, AIAA
(Am.Inst. Aeron. Astronaut.) 1.4, 981 (1966).]
 M 0.412
FIG.19b. Schlieren photographs for injection through a step down slot in super-
sonic flow for mainstream Mach number = 3.01, slot opening s = 4.62 mm, step height
h' = 6.07 mm. [R. J. Goldstein, E. R. G. Eckert, F. K. Tsou, and A. Haji-Sheikh, AZAA
(Am.Inst. Aevon. Astronaut.) 1.4, 981 (1966).]
 FIG.20. Correlation of film cooling effectiveness for supersonic mainstream flow
with heated air injection at small blowing rates. [R.J. Goldstein, E. R. G.Eckert, F. K.
Tsou, and A. Haji-Sheikh, Univ. of Minnesota, Heat Transfer Lab. Rept. H T L T R 60
(1965).]

FIG.21. Film cooling effectiveness as predicted by subsonic and supersonic correla-

tions. [R. J. Goldstein, E. R. G. Eckert, F. K. Tsou, and A. Haji-Sheikh, Univ. of
Minnesota, Heat Transfer Lab. Rept. HTL TR 60 (1965).]
 FILMCOOLING 367
indicate a much more substantial length immediately downstream of
the slot where the effectiveness is close to unity although further down-
stream the effectiveness diminishes more rapidly with distance than
with subsonic film cooling.
For helium injection through the tangential slot, the film cooling
effectiveness based on the recovery temperature (obtained from tests
with no blowing) was used (62,63). T h e results are correlated by the
relation
rlr = 10,000(x/h’)-2~0M0~8for 0.01 < M < 0.02 (58)
Mukerjee and Martin (64) studied film cooling with injection or air
into a Mach 1.5 to 1.7 air mainstream. Their system had approximately
tangential injection though the relative lip thickness was much greater
than in References (62) and (6.3). I n fact, the slot opening was only
from 10 to 30% of the total step height h’. T h e secondary flow was
apparently not increased to the point of choking. They report an
empirical relation and compare their measured isoenergetic film cooling
effectiveness with the results of Reference (63) and the equation of
Tribus and Klein. At low blowing rates relatively good agreement is
found with this latter equation though some of the results indicate
considerable deviations. T h e measured effectiveness values are in
qualitative accord with the earlier study (63). Differences between the
two studies are possibly due to the difference in injection lip geometry.
Interestingly, at high blowing rates a significantly shorter length along
the wall for which the effectiveness is unity and then a more gradual
diminishment of effectiveness with distance is found in Reference (64)
as compared to References (62) and (63).
Parthasarathy and Zakkay (65) conducted an extensive series of tests
for film cooling with an axisymmetric Mach 6 air mainstream. Helium,
hydrogen, argon, and air were employed as coolants with sonic injection
in the upstream, normal and downstream directions. T h e boundary
layer thickness at the injection location was much larger than in most
other studies. For downstream injection they correlate their results
with the relation
rlo = K [ ( ~ ~/- 0~. 8 1)- 0 . 7 (59)
where K = 155, 120, 35, and 30 for injection of hydrogen, helium,
air, and argon, respectively. Normal injection gave less effective cooling
than injection in the downstream direction, while no correlation could
be obtained for injection in the upstream direction. It should be noted
that the definition of effectiveness ?lo uses the free stream stagnation
temperature as a reference rather than the recovery temperature or
isoenergetic temperature.
368 RICHARDJ. GOLDSTEIN

Dannenberg (66) injected helium nearly tangentially on a hemisphere

close to the stagnation point over which a Mach 10 airstream with a
stagnation temperature of 4800K was flowing. The peak heating rate
was reduced by a factor as large as 2.5 and immediately downstream
of the injection point almost complete protection (7,rn 1) could be
obtained. T h e most effective film cooling was found when the velocity
of the injected coolant was matched to the velocity of the gas stream
at the point of injection.
Film cooling of a Mach 2.4 axisymmetric nozzle was studied by
Lieu (67). T h e mainstream air was heated to 670K. Injection took
place near the entrance (subsonic region) to the nozzle through a slot
inclined at an angle of 10" to the main flow. Optimum film cooling
performance was obtained when the free stream and coolant velocities
were approximately equal. A modified version of the Hatch and Papell
correlation (11)was used to correlate the results.
Redeker and Miller (68) used film cooling in the stagnation region of
a cylinder exposed to a Mach 16 crossflow. Nitrogen and helium could
be injected either normal or tangential to the surface. Considerable
reduction in heat transfer was found with injection. With normal
injection the aerodynamic heating could be cut in half while with
tangential injection it could be reduced to one-tenth of the non-film-
cooled value.
The downstream film cooling effects of nitrogen, helium, and argon
injected through a transpiring flat plate into a Mach 8 airflow was
studied by Woodruff and Lorenz (69). T h e reduction in the turbulent
heat flux in the downstream region was found to be relatively inde-
pendent of Mach number. Use of the blowing parameter eliminated
the influence of Reynolds number and the nature of the coolant on the
results.
Studies have been made (24) of film cooling with a Mach 3 main-
stream and injection of air through a narrow porous strip. Use of a
reference temperature enabled them to correlate their adiabatic wall
temperatures, in the form of isoenergetic film cooling effectiveness
with modified subsonic incompressible flow relations (see section on
analysis ).
I t should be recalled that other film cooling results with compressible
flow could not be correlated well with modified incompressible flow
correlations, I n those, however, the flow geometry was much more
complicated. Usually a step down slot of some type was used and the
resulting flow pattern, including lip shock, separated region and
reattachment shock could be such as to preclude simple correlations.
 FILMCOOLING 369

FILMCOOLING
D. THREE-DIMENSIONAL
I n many applications of film cooling, design considerations prevent
the use of continuous slots for introduction of the coolant. Discrete
holes may be used for injection, or a slot with discontinuities (due to
structural supports) may be used. If the mainstream is essentially
two-dimensional in the absence of injection, blowing through discrete
openings will result in a nonuniform flow across the span of the film
cooled wall. This is the type of three-dimensional film cooling that will
be discussed and reviewed in this section. T h e film cooling effectiveness
for an adiabatic wall is still of interest, but now the effectiveness is a
function of lateral position as well as downstream distance.
T h e film cooling effectiveness for injection through discrete holes is
usually considerably less than for slot injection at the same rate of
secondary flow per unit span. I n addition, as the blowing rate M is
increased past a relatively low value (perhaps M -. 0.5 for pz m pa),
the effectiveness for injection through discrete holes falls off rapidly.
These phenomena can be understood qualitatively by considering
the interaction of a nontangential jet and a mainstream. There is usually
ample room across the span for mainstream air to flow between the
individual secondary flow entrances. At low blowing rates the jets
entering the flow are quickly turned toward the surface by the main-
stream. As the blowing rate is increased, the jets penetrate into the
mainstream permitting mainstream gas to flow around and under the
entering secondary flow jets. This separates the injected fluid from
the wall and results in relatively low values of film cooling effectiveness.
At still higher blowing rates the jets penetrate further and mix more
with the mainstream.
It should be noted that the dynamic head or dynamic pressure ratio
(p2U22/pmU,2), rather than the blowing rate M is probably the parameter
to use, for a given geometry, in predicting the secondary flow for which
significant penetration of the jet (and reduced effectiveness) occurs.
T h e dynamic head ratio would be important in predicting results for
an application where the densities of the secondary and mainstream
flows are quite different from test results for approximately constant
density studies.
Another important parameter would be the geometry of the hole
through wbich the secondary fluid enters. A geometry which turns
the secondary fluid (and thus the jet momentum) towards the wall as it
enters the mainstream would be desirable in terms of optimizing the
film cooling performance.
370 RICHARDJ. GOLDSTEIN

Wieghardt (10) covered his continuous slot with perforated sheets

to study the effects of both a single row and two rows of holes running
transversely across the wall. He studied only one blowing rate with
this geometry. With two rows of holes the effectiveness was relatively
uniform across the span although less by a factor of two than with the
same air flow through a continuous slot. With injection through a single
row of perforations he found very low values of effectiveness except
directly downstream of the central region of each hole.
Papell (53) measured film cooling with injection through multiple
rows of discrete circular holes. Injection sections with either two rows
or four rows of holes could be inserted in the wall. The holes were
at an angle of 90" to the mainflow. Data were taken over a large range
of injection rates and could be correlated using an empirical modification
of the relation he used for film cooling through a continuous slot.
Use of rows of punched crescent louvers to inject a film coolant has
been reported (70). The louvers apparently turn the individual jets down-
stream so the problem of jet departure from the surface was not

L
TUBES

FIG. 22a. Injection section and coordinate system for a row of inclined jets. Detail
and flow field are shown for only a single jet interacting with a mainstream. [R. J. Gold-
stein, E. R. G. Eckert, V. L. Eriksen, and J. W. Ramsey, Israel J. Technol. 8, 145 (1970).]
 FILMCOOLING 37 1
encountered. The data was correlated using the same parameters as
were used for film cooling through a number of two-dimensional
slots (38). Far downstream the louvers were almost as effective as slots
in protecting the surface.
Several publications (28, 71, 72) have appeared from the University
of Minnesota on measurements of film cooling with injection through
circular tubes (ending flush to the surface) inclined at various angles
to the main flow. Both single tubes and a transverse row of tubes were
used. The general flow configuration is presented in Fig. 22, which
shows qualitatively the flow of the jet entering the mainstream.
Figure 23 shows the film cooling effectiveness downstream of a single
hole through which air enters at an angle 01 of 35" to the main flow.
Even along the hole centerline (2 = 0) the effectiveness is considerably
less than what would be expected for injection through a continuous
slot as shown by the top two curves (Eq. 52). Off centerline (2# 0)

UNNEL FLOOR

FIG.22b. Flow field and coordinate system associated with laterally inclined jet
interacting with a mainstream. [R. J. Goldstein, E. R. G. Eckert, V. L. Eriksen, and
J. W. Rarnsey, Isruel J. Technol. 8, 145 (1970).]
372 RICHARDJ. GOLDSTEIN

0.9 I I I I I I I I
I
0.8 3 =21,8&-io: ASSUMING 30 S W I N G ACROSS SPAN

6 0.7

0.6
2
w
2
6 0.5
w
B 0.4
(3
z
c 0.3

2
LL
0.2

0.I

0
0 5 10 I5 20 25 30 35 40 45
DIMENSIONLESS DISTANCE DOWNSTREAM, X/D

FIG. 23. Axial effectiveness distributions for injection through a single hole at an
injection angle of 35" and M = 0.5. [R. J. Goldstein, E. R. G. Eckert, and J. W. Ramsey,
J . Eng. Power 90, 384 (1968).]

the effectiveness is even less. T h e results shown in this figure are for
M = 0.5, which is approximately the optimum blowing rate to maximize
the film cooling through a single tube at an angle of 35".
Figure 24 shows how the effectiveness varies with blowing rate at
different downstream positions. Data for a single hole and a row of
holes inclined at 35" to the mainstream are presented here. Note that
for a single row of holes the effectiveness reaches a maximum at a
blowing rate M 0.5. This could be interpreted as the blowing rate
(for p z w pm) above which the jet is no longer turned by the mainstream
to hug the wall along which it enters; above that value it increasingly
penetrates into the main flow. At higher blowing rates not only is the
effective protection per unit mass of coolant reduced, but the absolute
value of effectiveness is reduced as well.
At low blowing rates the flows of the individual jets from a row
of holes appear to be independent of one another. T h e two-dimensional
adiabatic wall temperature distribution can then be approximated by
 FILMCOOLING 373

0.7
I I I
GEOMETRY

c0.6 '$zD = 0.22 105

- 0.124
W
z
0.5
-
z _
D -0.0
w
I
5W 0.4 SINGLE ROW0
lL
U
W
0.3
z
J
0
g 0.2
z
-I
C0.l

0
0 .5 1.0 1.5 2 .o

BLOWING R A T E , M
FIG. 24. Comparison of the centerline film cooling effectiveness for single hole and
multiplehole injectionat an injection angle of 35"with the flow for various blowing rates M .
[R. J. Goldstein, E. R. G. Eckert, V. L. Eriksen, and J. W. Ramsey, Israel I. Technol. 8,
145 (1970).]

superposition of the effect of a number of single film cooling jets (28).

At larger blowing rates where the jets tend to blow off the wall the
flows from the individual holes interact. This interaction of the jets
results in effectively blocking part of the region where the mainstream
might flow around the jets. The secondary flow is then more effectively
turned toward the wall, giving considerably higher effectiveness at
large M than would be given by a single jet or by superposition (cf.
Fig. 24). At large blowing rates the effectiveness for injection through
a row of holes increases with position downstream and then remains
approximately constant for a considerable distance.
If the injection tube is inclined laterally (Fig. 22b), the film cooling
effect is spread out further across the span. At a given blowing rate
the average effectiveness across the span can be higher for this geometry
than for normal injection or injection through a tube inclined downstream
(28). Figure 25 shows contours of film cooling effectiveness for injection
through different inclined tubes. Lateral inclination seems to impede
the penetration of the jet into the mainstream at moderate values of M.
374 RICHARDJ. GOLDSTEIN

GEOMETRY AY

-2
' o
.I0
b
.I5
I
c
20
d
25
c
.30
Y'I.0

-
I I

-
~ I
-I - Q=W. U.35' ~

C
0
I - I
I I I

DIMENSIONLESS DISTANCE DOWNSTREAM X/D

FIG.25. Lines of constant film cooling effectiveness for single hole injection at
M = 1.0 for various angles of injection. [R. J. Goldstein, E. R. G. Eckert, V. L. Eriksen,
and J. W. Ramsey, Israel J. Technol. 8, 145 (1970).]
 FILMCOOLING 375

It should be recalled, however, that M is based on the velocity within

the injection tube. As the tube is inclined at a greater angle from the
normal the elliptical exit hole area increases and the component of the
injection velocity normal to the mainstream decreases, tending to
decrease the penetration of the jet into the flow. The effect on penetration
is apparently more significant when the other component of the injection
velocity is in the lateral direction rather than the downstream direction.
Metzger and Fletcher (45) measured the average film cooling effec-
tiveness (in lateral and downstream directions) following injection
through a row of holes inclined downstream. Their trends for the
average film cooling effectiveness are similar to those from other
studies (71).They also measure the average heat transfer downstream
of the holes. Aside from the region close to injection it appears, at least
for moderate blowing rates, that the average heat transfer coefficients
can be approximated by the values determined without blowing, i.e.,
for a normal two-dimensional turbulent boundary layer.

V. Concluding Remarks
Considerable understanding of film cooling processes has developed
in the last twenty-five years. Recent important applications indicate
that there are still significant advances to be made.
Further work on numerical solutions to the equations for turbulent
flows should enhance our ability to predict two-dimensional film
cooling phenomena. Accurate predictions for film cooling injected at
an angle to the mainstream with a relatively thick splitter plate, with
high-speed flow or with large density differences may, however, prove
elusive.
For secondary flow through discrete holes or even interrupted slots,
the difficulties in predicting film cooling performance are even greater.
The resulting three-dimensional flow is not yet accessible to anything
but simplified analysis. Much work must still be done experimentally
to understand the effects of hole geometry, density differences, and the
interaction of individual jets on the adiabatic wall temperature distribu-
tion. In addition, information on the effect of the mass addition on the
local heat transfer is required.

ACKNOWLEDGMENT
Several colleagues were of great aid during the preparation of Table I and in reviewing
the manuscript for errors. Particular thanks are due to D. R. Pedersen, who also offered
invaluable assistance in preparing the figures and text for publication.
 RICHARDJ. GOLDSTEIN

NOMENCLATURE

C mass fraction of foreign gas Stanton number with injection

Ctw mass fraction of foreign gas at an Stanton number without secondary
impermeable wall injection
C, mass fraction of foreign gas present injection slot height
in secondary flow lip thickness of slot at injection
C m mass fraction of foreign gas present adiabatic wall temperature
in mainstream adiabatic wall temperature with
C, specific heat isoenergetic injection
c, average specific heat of gas in datum or reference temperature
boundary layer, see Eq. (13) used in defining the heat transfer
C,, specific heat of secondary fluid coefficient
C,, specific heat of mainstream mainstream stagnation temperature
D diameter of injection tube stagnation temperature of secon-
h convective heat transfer coefficient dary stream
h' step height, i.e., sum of slot height isoenergetic stagnation temperature
and lip thickness of secondary stream (should
K empirical constant used in Eq. (59) equal Torn)
Le Lewis number; ratio of Schmidt wall recovery temperature in ab-
number to Prandtl number sence of secondary flow
M blowing rate or blowing parameter wall temperature
pz U a l P m u m wall temperature at point of in-
Ma, injection Mach number jection, high-speed flow
Mam mainstream Mach number wall temperature at point of in-
m mass flow rate per unit span in jection with isoenergetic injection
boundary layer at any point temperature at a distance y from
including both secondary fluid the surface
and fluid entrained from main- temperature difference
stream temperature of secondary fluid at
ti2, secondary fluid mass flow rate per injection
unit span paU,s mainstream temperature
tibo mass flow rate per unit span in property reference temperature
boundary layer of fluid entrained mean temperature in boundary
from mainstream layer, Eq. (12)
m m o mass flow rate per unit span in velocity in boundary layer
boundary layer of entrained fluid velocity of secondary fluid in in-
with no secondary injection jection slot (rica/p,s)
Pr Prandtl number mainstream velocity
q heat flow per unit time and area molecular weight of injection gas
ReHD Reynolds number based on hy- molecular weight of mainstream
draulic diameter of tunnel gas
Re. mainstream Reynolds number distance downstream from point of
based on distance downstream injection through hole (down-
of injection p m U m x / p m stream edge) (Fig. 2)
Re,, mainstream Reynolds number distance from point of injection
based on starting length (Fig. 1)
pco U m x ' l p m distance from starting position of
Re, slot Reynolds number based on turbulent boundary layer
slot height p z U z s / p n starting length of Reference 5.
 FILMCOOLING 377
y distance normal to adiabatic wall impermeable wall effectiveness,
Y distance normal to surface in three- based on concentration, Eq. (6)
dimensional film cooling studies isoenergetic film cooling effective-
(Fig. 2) ness, Eq.( 5 )
Y l / 2 vertical position at which film cooling effectiveness based on
( T W , Y, 0) - T m ) / ( T ( X0,O)
, - Tm) total temperature of gas stream
=B film cooling effectiveness based on
lateral distance from centerline of recovery temperature, Eq. (4)
injection (cf. Fig. 2) dimensionless temperature para-
lateral position at which meter ( - T m ) / ( Taw - Tm)
( T W , 0 , Z ) - Tm ) / ( T( X,0,O)- Tm) viscosity of secondary fluid
=& viscosity at reference temperature
angle of injection in YX-plane T,
(Fig. 2) viscosity of mainstream fluid
injection parameter of Reference 8, dimensionless film cooling para-
Eq. (39) meter defined in Eq. (9)
boundary layer thickness dimensionless film cooling para-
boundary layer momentum thick- meter defined in Eq. (40b)
ness dimensionless film cooling para-
thermal boundary layer thickness meter for high speed flow defined
boundary layer displacement thick- in Eq. (42)
ness density of secondary fluid
turbulent thermal diffusivity density at reference temperature T*
parameter defined in Eq. (56) density of mainstream fluid
film cooling effectiveness, low- angle of injection in XZ-plane
speed flow, Eq. (3) (Fig. 22)

REFERENCES
I. E. R. G. Eckert and J. N. B. Livingood, NACA Rept. 1182 (1954).
2. M. Tribus and J. Klein, Heat Transfer, Symp. Univ. Mich 1952, 21 1 (1953).
3. R. J. Goldstein, G. Shavit and T. S. Chen, J. Heat Transfer 87, 353 (1965).
4. J. Librizzi and R. J. Cresci, AIAA (Am. Inst. Aeron. Astronaut.) J. 2, 617 (1964).
5. S. S. Kutateladze and A. 1. Leont’ev, Thermal physics of high temperutures 1, No. 2,
281-290 (1963).
6. J. L. Stollery and A. A. M. El’Ehwany, Intern. /. Heat Mass Transfer 8, 55 (1965).
7. J. L. Stollery and A. A. M. El-Ehwany, Intern. J. Heat Muss Transfer 10, 101 (1967).
8. R. J. Goldstein and A. Haji-Sheikh, in Japan Soc. Mech. Engr. 1967 Semi-Intern.
Symp., 213-218, Tokyo (1967).
9. A. I. Leont’ev, “Advances in Heat Transfer” (T. F. Irvine, Jr. and J. P. Hartnett,
eds.), Vol. 3, p. 33-100. Academic Press, New York, 1966.
10. K. Wieghardt, AAF Translation No. F-TS-919-RE (1946).
11. J. E. Hatch and S. S. Papell, NASA Tech. Note. 2“-130 (1959).
12. M. Saarlas, Ph.D. Thesis, Univ. of Cincinnati (1967).
13. R. A. Seban and L. H. Back, /. Heat Transfer 84, 45 (1962).
14. R. A. Seban and L. H. Back, J. Heat Trunsfer 84, 235 (1962).
15. R.A. Seban and L. H. Back, Intern. J. Heat Mass Transfer 3, 255 (1961).
16. D. B. Spalding, AIAA (Am. Inst. Aeron. Astronaut.) J. 3, 965 (1965).
378 RICHARDJ. GOLDSTEIN
17. S. C. Kacker, B. R. Pai, and J. H. Whitelaw, “Progress in Heat and Mass Transfer”
(T. F. Irvine, Jr., W. Ibele, J. P. Hartnett, and R. J. Goldstein, eds.), Vol. 2, p.
163-1 86. Macmillan (Pergamon), New York, 1969.
18. W. B. Nicoll and J. H. Whitelaw, Intern. J. Heat Mass Transfer 10, 623 (1967).
19. S. C. Kacker, W. B. Nicoll, and J. H. Whitelaw, Imperial College, Dept. of Mech.
Engr. Rep. TWF/TN/30, London, 1967.
20. S. V. Patankar and D. B. Spalding, “Heat and Mass Transfer in Boundary Layers.”
Morgan-Grampian Press, London, 1967.
21. M. Wolfshtein, Imperial College, Dept. of Mech. Engr. Rep. SF/TN/7, London,
1967.
22. E. H. Cole, D. B. Spalding and J. L. Stollery, Imperial College, Dept. of Mech.
Engr. Rep. EHTITNI11, London, 1968.
23. B. R. Pai, Imperial College, Dept. of Mech. Engr. Rep. EHT/TN/9, London, 1968.
24. R. J. Goldstein, E. R. G. Eckert and D. J. Wilson, J. Eng. Ind. 90, 584 (1968).
25. H. Schlichting, “Boundary Layer Theory,” 6th ed., p. 600. McGraw-Hill, New
York, 1968.
26. A. L. Laganelli, Intern. Heat Transfer Conf., 4th, Versailles/Paris, 1970 Pap. No.
69-IC-191 (to be presented).
27. J. W. Ramsey, R. J. Goldstein, and E. R. G. Eckert, Intern. Heat Transfer Conf.,
4th, VersaillesiParis, 1970 Pap. No. 69-IC-136 (to be presented).
28. R. J. Goldstein, E. R. G. Eckert, V. L. Eriksen, and J. W. Ramsey, Israel J. Technol.
8, 145 (1970) (cf. NASA CR-72612;also Univ. of Minnesota, Heat Transfer Lab.
Rept. H T L T R 91 (1969)).
29. W. K. Burns and J. L. Stollery, Intern. J. Heat Mass Transfer 12, 935 (1969).
30. J. P. Hartnett, R. C. Birkebak, and E. R. G. Eckert, J. Heat Transfer 83, 293 (1961).
3 1. E. R. G. Eckert and R. C. Birkebak, in “Heat Transfer, Thermodynamics and Educa-
tion, Boelter Anniversary Volume” (H. A. Johnson, ed.), p. 150-163. McGraw-Hill,
New York, 1964.
32. R. A. Seban, H. W. Chan and S. Scesa, Am. SOC.Mech. Engrs. Pap. 57-A-36 (1957).
33. R. A. Seban, J. Heat Transfer 82, 303 (1960).
34. R. A. Seban, J. Heat Transfer 82, 392 (1960).
35. J. H. Chin, S. C. Skirvin, L. E. Hayes, and A. H. Silver, Am. SOC.Mech. Engrs.
Pap. 58-A-107 (1958).
36. S. C. Kacker and J. H. Whitelaw, Intern. J. Heat Mass Transfer 10, 1623 (1967).
37. A. E. Samuel and P. N. Joubert, Am. SOC.Mech. Engrs. Pap. 64-WAIHT-48 (1964).
38. J. H. Chin, S. C. Skirvin, L. E. Hayes, and F. Burggraf, J. Heat Transfer 83, 281
(1961).
39. S. Papell and A. M. Trout, Nasa Tech. Note TN D-9(1959).
40. J. H. WhiteIaw, Aeron. Research Council, London, Current Pap. No. 942, 1967.
41. N. Nishiwaki, M. Hirata, and A. Tsuchida, in “International Developments in Heat
Transfer,” part IV,p. 675. ASME, New York (1961).
42. R. J. Goldstein, R. B. Rask, and E. R. G. Eckert, Intern. J. Heat Mass Transfer 9,
1341 (1966).
43. I. Mabuchi, JSME (Bulletin of Japanese Soc. of Mech. Engr.) 8, 406 (1965).
44. E. R. G. Eckert, R. J. Goldstein, and D. R. Pedersen, A Discussion of AIAA Pap.
69-523 by D. E. Metzger and D. D. Fletcher. (cf. Reference 45).
45. D. E. Metzger and D. D. Fletcher, AIAA ( A m . Inst. Aeron. Astronaut.) Paper 69-
523, to published in J. Aircraft (1969).
46. J. P. Hartnett, R. C. Birkebak, and E. R. G. Eckert, in “International Developments
in Heat Transfer,” Part IV, p. 682. ASME, New York, 1961.
 FILMCOOLING 379
47. M. P. Escudier and J. H. Whitelaw, Intern. J. Heat Mass Transfer 1 1 , 1289 (1968).
48. B. R. Pai and J. H. Whitelaw, Imperial College, Dept of Mech. Engr. Rep. E H T
TN/A/l5, London, 1969.
49. L. W. Carlson and E. Talmor, Intern. J. Heat Mass Transfer 11, 1695 (1969).
50. S. C. Kacker and J. H. Whitelaw, J. Heat Transfer 90, 469 (1968).
51. S . C. Kacker and J. H. Whitelaw, Intern. J . Heat Mass Transfer 12, 1196 (1969).
52. S. Sivasegaram and J. H. Whitelaw, /. Mech. Engr. Sci. 11, 22 (1969).
53. S. S. Papell, N A S A Tech. Note TN D-299 (1960).
54. D. E. Metzger, H. J. Carper, and L. R. Swank, (1.Engr. Power) 90, 157 (1968).
55. C. M. Milford and D. M. Spiers, in “International Developments in Heat Transfer,”
Part IV, p. 669. ASME, New York, 1961.
56. J. G. Lucas and R. L. Golladay, Nasa Tech. Note TN N-1988 (1963).
57. J. G. Lucas and R. L. Golladay, N A S A Tech. Note TN D-3836 (1967).
58. J. J. Williams, Ph. D. Thesis, Univ. of California, Davis, California, 1969.
59. B. R. Pai and J. H. Whitelaw, Aero. Research Council, London, Paper 29928, H.M.T.
182, 1967. Also Imperial College Dept. of Mech. Engr. EHT/TN/8, London, 1967.
60. S. Scesa, Ph.D. Thesis, Univ. of California (1954).
61. R. J. Goldstein, F. K. Tsou and E. R. G. Eckert, Univ. of Minnesota, Heat Transfer
Lab. Rep., H T L T R 54, 1963.
62. R. J. Goldstein, E. R. G. Eckert, F. K. Tsou, and A. Haji-Sheikh, Univ. of Minnesota,
Heat Transfer Lab. Rept. H T L T R 60 (1965).
63. R. J. Goldstein, E. R. G. Eckert, F. K. Tsou, and A. Haji-Sheikh, A I A A ( A m .
Inst. Aeron. Astronaut.) J. 4, 981 (1966).
64. T. Mukerjee and B. W. Martin, in “Proceedings of the 1968 Heat Transfer and Fluid
Mechanics Institute” (A. F. Emery and C. A. Depew, eds.), p. 221. Stanford Univ.
Press, Stanford California, 1968.
65. K. Parthasarathy and V. Zakkay, Aerospace Research Lab. Tech. Rep., Contract
F33615-68-C-1184 Project 7064, Wright Patterson Air Force Base, Ohio, 1968.
66. R. E. Dannenberg, N A S A Tech. Note TN D-1550 (1962).
67. B. H. Lieu, U. S . Naval Ordance Lab. NOLTR No. 224, White Oak, Maryland,
1964.
68. E. Redeker and D.S. Miller, in “Proceedings of the 1966 Heat Transfer and Fluid
Mechanics Institute” (M. A. Saad and J. A. Miller, eds.), p. 387. Stanford Univ.
Press, Stanford, California, 1966.
69. L. W. Woodruff and G. C. Lorenz, A I A A ( A m . Inst. Aeron. Astronaut) J . 4, 969
(1966).
70. F. Burggraf, J. H. Chin, and L. E. Hayes, J. Heat Transfer 83, 286 (1961).
71. R. J. Goldstein, E. R. G. Eckert, and J. W. Ramsey, J. Eng. Power 90,384 (1968).
72. R. J. Goldstein, E. R. G. Eckert, and J. W. Ramsey, N A S A CR-54604; Also Univ.
of Minnesota, Heat Transfer Lab. Rep. H T L T R 82, 1968.