American Institute of Aeronautics and Astronautics

1
Design of actively cooled panels for scramjets

Lorenzo Valdevit
1
, Natasha Vermaak
2
, Kathryn Hsu
3
, Frank W. Zok
4
and Anthony G. Evans
5

University of California, Santa Barbara, CA 93106-5050
The operating conditions of scramjet engines demand designs that include active cooling
by the fuel and the use of lightweight materials that withstand extreme heat fluxes under
oxidizing conditions. The goal of this analysis is to provide an optimization tool that can be
used to direct the development of advanced materials that outperform existing high
temperature alloys and compete with ceramic matrix composites. For this purpose an
actively cooled plate has been optimized for minimum weight under three primary
constraints. (i) Resistance to pressure loads arising from fuel injection and combustion, as
well as thermal loads associated with the combustion temperature. (ii) A temperature
distribution in the structure during operation that does not exceed material limits, subject to
a reasonable pressure drop. (iii) A maximum temperature in the fuel (JP-7) low enough to
prohibit coking. It is shown that all design requirements typical of Mach 5-7 hypersonic
vehicles can be met by a small subset of material systems. Those made using C/SiC
composites are the lightest. Others made using Nb alloys and (thermal barrier coated)
superalloys are somewhat heavier, but might prevail in a design selection because of their
structural robustness, facility of fabrication and cost-effectiveness.

Nomenclature
A = area
a = speed of sound
b = width of the actively cooled panel
p
c = specific heat
D = diameter
E = Youngs modulus
f = friction factor
= fuel/air mass ratio
= generic function
g = generic function
H = thickness of the actively cooled panel
h = enthaply
= heat transfer coefficient
k = thermal conductivity
L = height of the cooling channel
= characteristic length
M = mass of the actively cooled panel
m = mass flow rate
Ma = Mach number

1
Postdoctoral scholar, Materials Department. Member AIAA.
2
Graduate student, Materials Department.
3
Graduate student, Materials Department.
4
Professor and Associate Chair, Materials Department.
5
Professor, Materials Department.

American Institute of Aeronautics and Astronautics

2
p = pressure
Pr = Prandtl number
Q = heat flux
q = specific heat flux
R = universal gas constant
= thermal resistance per unit width
r = thermal resistance per unit area
Re = Reynolds number
St = Stanton number
T = temperature
t = thickness
u = velocity of the coolant
V = velocity of the air
V

= volumetric flow rate of the coolant per unit width of the panel
w = width of the cooling channel
x = non-dimensional geometric variable
x,y,z = spatial coordinate
Z = length of the actively cooled panel

Greek symbols
= thermal expansion coefficient (CTE)
= thermal diffusivity
= inverse length constant
= difference
= mixture richness (actual fuel/air ratio divided by stoichiometric fuel/air ratio)
= dynamic viscosity
v = kinematic viscosity
= Poissons ratio
= non-dimensional objective function
*
= dimensional objective function
= mass density
= areal density of the cross-section of the panel
= normal stress

Subscripts and superscripts
0 = inlet conditions
= free-stream conditions
1,2 = generic symbols
3,4 = stations along the vehicle
air = relative to the air
aw = adiabatic wall conditions
c, core = relative to the core web
c, cool = relative to the cooling channels
coke = coking conditions in the fuel
comb = within the combustion chamber
f, face = relative to the face sheet
f, fuel = relative to the fuel (coolant)
fin = relative to conduction/convection in the core web
G = relative to the hot gases in the combustion chamber
h = hydraulic
= horizontal (x) direction
i = generic location in the cross-section
m = mechanical

American Institute of Aeronautics and Astronautics

3
max = maximum
min = minimum
panel = relative to the whole cross-section of the panel
s = relative to the solid material the panel is made of
st = stoichiometric
T = thermal
= total
t = top face
top = top side
TBC = relative to the thermal barrier coating (TBC)
w = wall exposed to the combustion chamber
= between core webs (i.e. near the middle of the channel)
x,y,z = along the respective directions
Y = yielding
panel
T = due to the temperature drop across the panel
tf
T = due to the temperature drop across the top face sheet
* = Eckert reference
= non-dimensional
= generic symbol

I. Introduction
The high specific impulse enabled by air-breathing hypersonic vehicles has motivated continued development of
key technologies. The aerothermodynamics of scramjet engines has been extensively analyzed, and the potential to
achieve positive thrust successfully demonstrated. However, the integration with vehicle design remains a challenge.
Selecting materials and generating designs that resist the associated thermal loads for the duration of the mission is
especially daunting. Active cooling by the fuel is a natural choice, because the proximity of the injection points
facilitates its circulation just before injection. Aspects of the design and performance of actively cooled combustion
systems have already been explored
1-4
, including geometry optimization
5-8
. But, to the best of our knowledge, a
comprehensive treatment that accounts for the thermo-mechanical constraints has yet to be devised. In this paper,
optimization results are presented for a prototypical planar combustor with rectangular channels (Fig. 1), for a
variety of materials. All the work refers to a hydrocarbon-powered scramjet operating at Mach 7. The goal is to
direct the development of advanced materials that outperform existing high temperature alloys and compete with
ceramic matrix composites (CMCs). For options based on high temperature alloys, the possible benefits of
superposing a thermal barrier coating (TBC), such as yttria-stabilized zirconia (YSZ), are explored.
For optimization purposes, it is convenient to have analytic expressions for the temperatures and stresses in the
structure. The temperatures are derived using a two-dimensional resistance network, building upon a fin analogy
9,10

and verified using finite elements. These temperatures are used to provide a constraint on the maximum allowable
material temperature. They are also used as input for analytic calculation of the thermal stresses, again checked
using finite elements. These stresses are superimposed on those induced by the pressure loads inside the combustion
chamber and within the cooling channels, to permit formulation of the structural constraints. By requiring that the
total stresses (thermal plus pressure) remain below the failure strength of each material at its maximum operating
temperature, the optimization results allow the ranking of candidate materials and designs in terms of their weight-
efficiency. While it is preferred to operate scramjets at (or near) stoichiometric fuel/air ratios, in some instances
extra fuel is needed to provide adequate cooling and/or to overcome combustion inefficiencies. To allow for this
possibility, results are also presented for a range of fuel/air mixture richness.
The structure of the paper is as follows. In section II the primary aerothermodynamics concepts that provide the
boundary conditions for the thermo-mechanical problem are summarized. Section III presents the thermal network
used for temperature and heat flux calculations. The details of the stress analysis are presented in Section IV.
Section V illustrates the formulation of the optimization scheme in non-dimensional form and section VI presents its
results. Discussion and conclusions follow.




American Institute of Aeronautics and Astronautics

4
II. Thermo-mechanical loads in a scramjet combustion chamber
For tractability of presentation, the ensuing analysis focuses on a specific hydrocarbon-powered hypersonic
vehicle operating at Mach 7 (near the ramjet to scramjet transition). However the procedure is general and can be
readily extended to other flight situations. The preferred fuel at this velocity is JP-7, which facilitates compact and
robust designs
11
. Furthermore, the design of vehicles in this velocity range is most mature and in urgent need of
technological advance. The thermo-mechanical loads on the combustor depend on nearly every aspect of the vehicle
design, including the size and shape of the compression ramp, the size of the combustion chamber, the details of the
injection system, the combustion efficiency, etc. Nevertheless, reasonable (albeit inevitably approximate) estimates
can be derived based on a small set of initial assumptions, namely: (a) the vehicle speed is Mach 7; (b) the vehicle
is 10 m long and 2.5 m tall; and (c) the compression ramp is bi-linear, generating three oblique shocks. The
aerothermodynamic considerations underlying these derivations are clearly exposed in Ref. 11; details governing the
application of these concepts to the present design are provided in Appendix I. The main results are summarized
below.
Three thermo-mechanical loads act on the actively cooled wall of the combustion chamber:
(i) An external pressure load, evenly distributed on the face exposed to the combusting gases,
0.24 MPa
comb
p = .
(ii) A pressure load inside the cooling channels, 10 MPa
cool
p = .
(iii) Thermal loads derived from exposure of one face to the combusting gases, with adiabatic wall
temperature, 3050 K
aw
T = and heat transfer coefficient,
2
535 W/m K
G
h = .
The incoming heat is carried away by the fuel (Table 2) flowing in the cooling channels at a rate per unit width
of the vehicle,
2
0.0082 m /s V =

( represents the fuel/air mixture richness, defined in equation (47)). The fuel
enters the cooling channels at a temperature,
0
400 K
f
T = .
III. Temperature distributions
To obtain analytic estimates of the temperatures, three simplifications are invoked:
(i) The top face of the structure is exposed to hot gases at specified temperature, whereas the bottom face is
thermally insulated. Consequently, all the heat received through the top face is carried away by the
cooling fluid.
(ii) No heat is conducted along the length of the panel (z-direction), either in the structure or in the cooling
fluid. This assumption results in slightly conservative temperature estimates.


Figure 1. (a) Artist rendition of a hypersonic air-breathing vehicle. The proposed multifunctional actively
cooled plate is shown in the inset. (b) Schematic of the actively cooled plate with thermo-structural loads,
variables definition and coordinate system.


American Institute of Aeronautics and Astronautics

5
(iii) The fuel temperature is uniform within the channels at any given cross-section: ( )
fuel fuel
T T z = only.
Namely,
fuel
T is the mixing-cup temperature.
The temperature at every cross-section z is calculated using an electrical analogy
12
(Fig. 2a). The nine conductive
resistances (per unit width in the z-direction) needed to characterize the problem can be grouped into six categories
(see Fig. 1b for definition of the geometric variables):
On the hot side: 1/
G G c
R h t = , 1/
w
G G
R h w =
Across a TBC (when present): /
TBC TBC TBC c
R t k t = , /
w
TBC TBC TBC
R t k w =
Across the face (y-direction): /
face f s c
R t k t = , /
w
face f s
R t k w =


Figure 2. (a) Thermal resistance network used for the temperature calculations, together with the
expression for all the relevant thermal resistances. (b) Effective network.


American Institute of Aeronautics and Astronautics

6
Along the face (x-direction): ( ) / 2 / 2
h c s f
R w t k t = +
Convective resistance on the coolant side (for the portion of the face in direct contact with the coolant):
1/
cool c
R h w =
Combined conductive and convective resistance of the core web (modeled as a one-dimensional thermal
fin
9,10
):
1
1 2
2 2 1
tanh
c c
fin s
s c c s c
h h
R L k
k t t k t

| | (

= | ( `
|
( \ )
.
Here,
G
h is the heat transfer coefficient on the hot gas side,
s
k and
TBC
k are the thermal conductivities of the solid
and the TBC, respectively, and
c
h is the heat transfer coefficient on the coolant side. Flow in the cooling ducts is
assumed to be turbulent and fully developed. Under these conditions,
c
h is given by the correlation
13,14
:

( )
2/ 3
(Re 1000) Pr
2
1 12.7 / 2 Pr 1
f f
c
h h
f
k k
h Nu
D D f

= =
+
(1)
where
f
k is the thermal conductivity of the coolant, Pr is the Prandtl number, 2 /( )
h
D wL w L = + is the hydraulic
diameter of the ducts, and Re is the Reynolds number in the ducts, defined as:

( )
Re
1
h h
f f
u D D V
H
=

(2)
with
f
being the kinematic viscosity of the fluid, V

the prescribed volumetric flow rate of the fuel per unit width
(section II), the areal density of the cross-section (Eq. (29)) and f the friction factor. For fully developed turbulent
flow, the following correlation can be used for the friction factor
13,15
:

1/ 4
0.0791Re f

= (3)
Eqs. (1)-(3) allow calculation of the thermal resistances
cool
R and
fin
R as a function of the geometric variables and
the fuel properties and flow rate.
The temperatures of the structure and the fluid increase continuously along the z-direction, becoming maximum
at the outlet. For convective boundary conditions (and uniform hot gas temperature) the solid and fluid temperature
are interrelated. A three-step approach is adopted:
(i) The thermal network (fig. 2a) is solved as a function of the unknown temperature ( )
fuel
T z .
(ii) The principle of energy conservation is used to relate the fuel and the solid temperatures, thus closing the
system.
(iii) The resulting differential equation is solved for ( )
fuel
T z and subsequently all the other temperatures of
interest are derived.
For step (i), the model can be simplified into the effective network of Fig. 2b, characterized by the five
resistances
1 1 2 2
, , , ,
w c w c
h
R R R R R , where:

1
1
2
2
2
2
2
2
w w w w
G TBC face
c
G TBC face
w w
face cool
c
face fin
R R R R
R R R R
R R R
R R R
= + +
= + +
= +
= +
(4)

American Institute of Aeronautics and Astronautics

7
The pertinent temperatures are the averages of the top face, both over (
c
tf
T ) and between (
w
tf
T ) the core members.
Since the three fluxes , ,
w c
h
Q Q Q are also unknown, five equations are needed to close the system:

( )
( )
( )
1 2
1 2
1
1
( 2 )
( 2 )
/
/
/
w w
aw w w h fuel
c c
aw c c h fuel
w w
w aw tf
c c
c aw tf
c w
h tf tf h
T Q R Q Q R T
T Q R Q Q R T
Q T T R
Q T T R
Q T T R

+ =

(5)
By defining the heat fluxes per unit area as:

/
/
/
w w
c c c
h h f
q Q w
q Q t
q Q t
=
=
=
(6)
Eqs. (5) can be rewritten as:

1 2
1 2
1
1
( 2 / )
( 2 / )
w
aw fuel w w h f
c
aw fuel c c h f c
w
aw tf w
c
aw tf c
c w
tf tf h h
T T q r q q t w r
T T q r q q t t r
T T q r
T T q r
T T q r
= + +

= +

(7)
where the thermal resistances per unit area have been defined as:

( )
1
2
1
1 2
2
1 1
2
1 1
2
2 2 1
tanh
2
2 / 2
f
TBC
G TBC s
f w
s c
f c c c
s
s s c s c
c
h
s
t
t
r
h k k
t
r
k h
t
h h
r L k
k k t k t
w t
r
k

= + +
= +
| | (

= + | ( `
|
( \ )
+
=
(8)
Solving the linear system in Eq. (7), gives:

*
1
*
1
*
1
aw fuel
c c
aw fuel
w w
aw fuel
h h
T T
q R
r
T T
q R
r
T T
q R
r

=
(9)

American Institute of Aeronautics and Astronautics

8
where
*
w
R and
*
c
R are given by:

1 2 1 2 2
*
2
2 2 1 2 2 1 2 2 2
1 2 1 2 2
*
2
2 2 1 2 2 1 2 2
( 2 / 2 / )
( 2 / 2 / ) { [ 2 ( / / )]}
( 2 / 2 / )
( 2 / 2 / ) {
c c w
h h f c f
w c w c w w c w
h h f c f h h f c f
w c w
h h f c f
c c w c w w c
h h f c f h
r r r r r r t t r t w
R
r r r r r r t t r t w r r r r r r t t t w
r r r r r r t t r t w
R
r r r r r r t t r t w r r r r
( + + +

=
+ + + + + + +
( + + +

=
+ + + + +
( )
2
2
1 2 2
*
2
2 2 1 2 2 1 2 2 2
[ 2 ( / / )]}
( 2 / 2 / ) { [ 2 ( / / )]}
w
h f c f
c w
h c w c w w c w
h h f c f h h f c f
r r t t t w
r r r
R
r r r r r r t t r t w r r r r r r t t t w
+ +

=
+ + + + + + +
(10)
Additionally:

*
*
w
aw tf
w
aw fuel
c
aw tf
c
aw fuel
T T
R
T T
T T
R
T T

(11)
For step (b), the temperature in the fluid is obtained via an energy balance:
( ) ( )
,
fuel
f p f c
dT
c V w t Q z
dz
+ =

(12)
where
, f p f
c is the specific heat at constant pressure per unit volume of the fluid and
( ) ( ) ( )
1 1
w c
aw tf aw tf
w c c
T T T T
Q z Q z Q z w t
r r

= + = + (13)
is the total heat flux impinging on the top face over a unit cell of width
c
w t + (Fig. 2a). By using Eq. (11) and Eq.
(13), Eq. (12) can be rewritten as:

( )
( )
* *
,
1
1
0
aw fuel
c
f p f w c aw fuel
c c
d T T
t w
c V R R T T
dz r w t w t

| |
+ + =
|
+ +
\

(14)
With the boundary condition ( )
0
0
fuel f
T z T = = , the solution for the spatial variation of the fluid temperature is:
( )
0
exp
aw fuel
aw f
T T
z
T T

(15)
where

* * 1
,
1/
c
w c
c c f p f
t r w
R R
w t w t c V

| |
= +
|
+ +
\

(16)
The temperatures at the middle of the top face can be expressed in similar form via eqs. (15) and (11):

American Institute of Aeronautics and Astronautics

9

( )
( )
*
0
*
0
exp
exp
w
aw tf
w
aw f
c
aw tf
c
aw f
T T
R z
T T
T T
R z
T T

(17)
All temperatures achieve their maximum at the fuel outlet, z Z = . The relative magnitudes of the convective
resistance at the inner wall and the fin resistance dictate whether the hot spot appears on the top face over or
between core webs; namely:

1
1 2
2 2
2 2 1
tanh
c w c w c c
tf tf s
s c s c c
h h
T T r r L k
k t k t h

| | (

> > > | ( `
|
( \ )
(18)
The temperature at the hot spot (at the face sheet/TBC interface) is then:

max
, *
min
0 1
1
1
2
aw tf top f
aw f
T T r
R
T T r
| |
=
|

\
(19)
where
* * *
min
min{ , }
c w
R R R = .
For the calculation of the thermal stresses, two temperature differences are relevant:
I. The temperature drop across the top face (in the y-direction), both over and between the core webs:

( )
( )
* *
0 1
* *
0 1
exp
exp
c
tf f f
c h
aw f c
w
tf f f
w h
aw f
T t r
R R z
T T t r
T t r
R R z
T T w r

| |
=
|

\
| |
= +
|

\
(20)

II. The temperature drop across the entire panel cross section, measured from the middle of the top face
to the bottom face, both over and between the core webs:

( )
( )
* * 2
0 1
* * 2
0 1
2 exp
2 exp
w
w
panel f
w h
aw f
c
c
panel f
c h
aw f c
T t
r
R R z
T T w r
T t
r
R R z
T T t r

| |
= +
|

\
| |
=
|

\
(21)
These temperature differences are greatest at the inlet ( 0 z = ).
The accuracy of this model has been verified with select finite element calculations. With the exception of
structures made using highly conductive alloys (such as copper), the predicted temperatures were consistent and
remarkably accurate.

IV. Stress analysis
A. Problem statement and boundary conditions
The thermo-mechanical stresses depend on the constraint exerted on the plate by the surrounding components of
the vehicle. Because the actual vehicle design is not known, and for the sake of generality, two idealized sets of
boundary conditions are considered (Fig. 3a-c):

American Institute of Aeronautics and Astronautics

10
Case I: Linear frictionless supports along the edges in the z-direction (Fig. 3a). This constraint prevents
bending in the z-direction, while allowing it in the x-direction (albeit with no rotation at the ends).
Uniform thermal expansion is allowed along all directions, but panel-level thermal bending is
forbidden.
Case II: Two-dimensional continuous bed of rollers (Fig. 3c). Uniform thermal expansion (but no panel-level
bending) is permitted along either direction. The external pressure does not cause the panel to bend
globally, but the internal pressure can bend individual top face segments.

Note that in case I the distance b between the clamps
need not equal the width of the entire combustion chamber.
The same boundary conditions can be used to model
multiple linear, frictionless supports at the back of the plate
(Fig. 3b). In this case, the dimension b represents the
distance between two consecutive supports. Moreover,
case II can be interpreted as the limit of case I as the
spacing between consecutive supports approaches zero. In
both cases, the use of rollers instead of frictional supports
allows uniform thermal expansion of the panel (with no
bending). While the practical implementation can be
challenging, attaining these conditions is essential to viable
solutions. As a simple demonstration, consider the thermal
stress in a uniform plate clamped on all sides (with
frictional supports) subject to a uniform temperature
increase T
16
: /(1 )
x y
E T = = . The maximum
temperature increase that can be sustained by the plate
without yielding is,
max
(1 ) /
Y
T E = . This quantity
is remarkably small for most structural materials of interest
(Fig. 3d). Note that, with the exception of C-SiC, most
materials would yield well below their limit temperature.
To further simplify the problem, both the pressure drop
and the temperature variation along the panel length are
neglected. This assumption, combined with the imposed
boundary conditions, ensure that generalized plane strain
conditions are obtained along the z direction
6
. Furthermore,
it requires that the calculations be performed on one cross-
section only. The inlet cross-section ( 0 z = ) is selected for
this calculation, since the temperature differences in Eqs.
(20)-(21) and thus the thermal stresses are maximum at
that location.
The next step is the identification of the most failure-
susceptible locations in the structure at this cross-section.
The stresses due to both pressure and thermal loads vary
along the member lengths and thicknesses; those due to the
pressure in the combustion chamber depend on the location
within the entire plate (in the xy plane). Because the
thermal and pressure loads often induce stresses of
opposite sign, it is not straightforward to establish a priori
the location of first yield. For this reason, a set of 18
critical points has been identified, clustered around two
failure-susceptible cells: one at the periphery closest to the
supports and the other at the center (Fig. 4). The presence

6
Generalized plane strain refers to conditions wherein the cross-sections z = 0 and z = Z are not allowed to rotate
relative to each other (although they are free to translate).



Figure 3. Possible mechanical boundary
conditions. (a-b) Case I: linear rollers on two sides
(in (b), a mechanically equivalent situation is
shown, where multiple supports are used). (c) Case
II: uniform two-dimensional bed of rollers, with
impeded rotation at the ends. (d) Benchmark
boundary condition: plate sitting on rigid
foundation. The chart compares the temperature
increase from room temperature that would induce
melting (light grey) to the temperature increase that
would yield the structure (dark grey); several
materials of interest are compared. Note that under
this boundary conditions, the full high-temperature
potential of the material would not be exploited.


American Institute of Aeronautics and Astronautics

11


Figure 4. Cross-sections that are likely to fail by yielding: the 18 critical points are depicted. The
two unit cells considered (close to the support, points 1-9, and around the middle of the panel, points
10-18) are loaded differently under boundary condition case I (due to the effect of the pressure in the
combustion chamber,
comb
p ), whereas they are thermo-mechanically identical under case II.

of internal pressure in the core channels (which imposes significant tensile stresses on all members) combined with
the relatively stubby shape of the optimized members (section VI and Fig. 6) makes it unnecessary to design against
buckling
17-25
. Consequently, failure of the structure is averted provided that the Mises stress in the most highly
stressed location remains in the elastic range.
The analytic methods used to determine the stresses in sandwich structures consider the face and core members
in each unit cell as independent beams; the connection between members is modeled using translational (and some
degree of rotational) constraints
17-25
. The accuracy of this approach is dependent upon the aspect ratio of each
element, and decreases as the elements become stubbier. When this approach is used, finite element analyses are
needed to ensure that predictions at the optimal geometries are sufficiently accurate.

B. Stresses due to coolant pressure
Case I. The pressure
cool
p inside the cooling channels subjects the core members to uniform tension and induces
a combination of tension and bending on the faces. Using the notation of Figs. 1b and 4, the resulting stresses in the
core members are:

,
, ,
at pts 9,18
cool
cool cool
p
core y
cool c
p p
core y core z
cool cool
w
p t
p p

=
=
(22)
and in the face segments:

( )
( )
( )
( )
2
2
,
2
2
, ,
/ 2 / / 2 at pts 2, 3,11,12
/ 2 / / 2 at pts 1, 4,10,13
/ 2 / / 4 at pts 5,8,14,17
/ 2 / / 4 at pts 6, 7,15,16
cool
cool cool
f f
p
f f
face x
cool
f f
f f
p p
face z face x
cool cool
L t w t
L t w t
p
L t w t
L t w t
p p

=

+

=
(23)

American Institute of Aeronautics and Astronautics

12
Consistent with beam analysis, the through-thickness stresses are neglected.
Case II. The stresses in the core members and the face segments exposed to the combustion chamber are the
same as those in Eq. (22) and (23), respectively, whereas those in the bottom face segments lack the bending
component. Consequently:

( )
( )
( )
( )
2
2
2
,
2
, ,
/ 2 / / 2 at pts 2,11
/ 2 / / 2 at pts 1,10
/ 2 / / 4 at pts 5,14
/ 2 / / 4 at pts 6,15
/ 2 at pts 3, 4, 7,8,12,13,16,17
cool
cool cool
f f
f f
p
face x
f f
cool
f f
f
p p
face z face x
cool cool
L t w t
L t w t
L t w t
p
L t w t
L t
p p

= +

=
(24)
C. Stresses due to combustion chamber pressure
Case I. The entire panel behaves as a clamped-clamped beam under uniform pressure,
comb
p . With the usual
assumption that the shear force is supported by the core and the moment by the face sheets
18
, the stresses in the faces
are:

( )
( )
( )
( )
2
2
,
2
2
,
1
at pts 1,2,5,6
12
1
at pts 3,4,7,8
12
1
at pts 12,13,16,17
24
1
at pts 10,11,14,15
24
comb
f f
p
f f
x face
comb
f f
f f
z fa
b
H t t
b
H t t
p b
H t t
b
H t t

,
comb comb
p p
ce x face
comb comb
p p

=
(25)
The stress exerted on the core members are of the order of the pressure itself, and can be safely neglected relative to
those induced by the pressure in the coolant (Eq. (22)).
Case II. The combustion pressure does not cause global panel bending. The pressure on the top face segments
counteracts that from the coolant; thus since
comb cool
p p << , this contribution can be safely neglected. The stress in
the core is exactly
comb
p , and can again be neglected.
D. Thermal stresses
Case I. At time zero the entire structure is at the same temperature as the fuel at the inlet,
0 f
T , and no thermal
stresses are present. Upon exposure to the combusting gases, the temperature distribution reaches a steady state
(section 3.A). Because the boundary conditions allow uniform expansion of the panel, the problem can be modeled
by superimposing the stresses due to two temperature differences. Contribution 1. By ignoring x-variations, the
average temperature difference between the top and bottom faces,
( )
2
w c
panel panel panel
T T T = + causes the panel to
deform uniformly in both the x and z directions, inducing compression in the top face and tension in the bottom
face
16
. Note that it is not legitimate to ignore the core stretching stiffness in the z-direction, where the core behaves
as a continuous plate. When the core and the bottom face are at the same temperature, the resulting stresses are:

American Institute of Aeronautics and Astronautics

13

( )
( )
( )
( )( )
,
,
at pts 1,2,5,6,10,11,14,15
2 1
at pts 3,4,7,8,12,13,16,17
2 1
at pts 1
1 2
panel
panel
panel
T
face x
panel
panel f c
f c T
face z
E T
E T
E T A A
A A

+
=
( )( )
,2,5,6,10,11,14,15
at pts 3,4,7,8,12,13,16,17
1 2
panel f
f c
E T A
A A

(26)
where ( )
f f c
A t w t = + and
( )
2
c f c
A H t t = are the cross-sectional areas of the face and the core in a unit cell,
respectively. Note that neglecting the core stiffness (letting 0
c
A in Eq. (26)) results in a perfectly bi-axial state
of stress in both faces. Contribution 2. By ignoring x-variations, the temperature difference across the top face
thickness,
w
tf tf
T T = , causes the upper surface of the top face to experience compression and the lower surface to
be in tension, giving:

( )
( )
, ,
at pts 1,5,10,14
2 1
at pts 2,6,11,15
2 1
tf tf
tf
T T
face x face z
tf
E T
E T

= =

(27)
Case II. The distributed support plays no role in defining the thermal stresses, so the solution is identical to that
for case I.
The accuracy of the stress distributions have been tested using selected finite element analyses for near-optimal
geometries. For most materials the agreement is within 10% . The exception was for structures made of highly
conductive material (such as Cu alloys) because the thermal network fails to capture the flow of heat into the bottom
face through the core. Consequently, the model gives overly conservative estimates of yielding. An extension of the
thermal network is underway to accurately model highly conductive materials.

V. Optimization
The objective is to optimize the geometry for minimum weight, subject to the constraints associated with the
structural and functional requirements. In order to fully exploit their potential, air-breathing hypersonic vehicles
should operate at a minimum mass flow rate of the fuel,
st
V V

(or equivalently 1 ). In practice, richer mixtures
might be necessary to compensate for combustion inefficiencies. In order to capture a spectrum of conditions of
interest, optimizations are performed for a range of mixture richness, 1 4 = . Several materials have been
explored (Table 1) and the resulting structures separately optimized. This allows exploration of the feasible alloys as
well as ranking according to weight efficiency. The properties of JP-7 (Table 2) have been used for all calculations.
For generality, and in order to minimize the number of input parameters, the optimization is posed in non-
dimensional terms. Table 3 contains the numerical values of all such parameters. In the following sub-sections, the
optimization variables, the objective function and all constraints are defined.
A. Geometric variables
The geometric variables depicted in Fig. 1b have the non-dimensional forms:

American Institute of Aeronautics and Astronautics

14

1
2
3
4
5
[0.5 0.99]
[0.2 10]
2 [0.01 100]
[0.005 0.02]
[0 0.001]
c f
TBC
x L H
x W L
x t t
x H Z
x t Z
=
=
=
=
=
(28)
with reasonable bounds reported within square brackets. For boundary condition 1, 0.3 b Z = .
B. Objective function
The areal density of the cross section depicted in Fig. 2 is given by:

( )
2
1 2
1 2 3 1
1
1
x x
x x x x
=
+
(29)
and the mass of the panel:

s TBC TBC
M H b Z t b Z = + (30)
The non-dimensional objective function (for each material) is:

2
TBC TBC
s s
t M H
Z Z b Z



= = + (31)
In order to compare the performance of different materials, the results will be presented in dimensional version,
where the merit function is
*
s
= . The mass of the thermal barrier layer is included in Eq. (30) to assure that a
layer of finite thickness will only emerge if it actually reduces the overall weight of the structure.
C. Constraints
Failure by yielding. The stress along the length of a member at any of the critical points (1-18) due to
mechanical loading can be expressed in non-dimensional form as:
Table 1. Candidate materials and their properties ( T
f 0
= 400 K ).

Material T
*
[K]

Y
[MPa] E [GPa] CTE [10
-6
/ K]
k
s
[W/mK]
s
[kg/m
3
]
E T
*
T
f 0 ( )

Y

Nb/Si 1470 370 140 10.5 50.0 6900 4.24
MAR-M246 1090 400 161 16.7 26.0 8440 4.63
Inconel 625 1100 265 164 14.0 20.0 8190 6.06
SiC-SiC 1640 200 240 4.1 17.3 2900 6.18
C-SiC 1810 200 100 2.0 8.7 2000 1.41
TBC (ZrO
2
) 1.0 3000

Table 2. Thermo-physical properties of fuel at 450 K.

Fuel
k
f
[W/m K]
f
[m
2
/s]
f
[m
2
/s] Pr
f

f
[kg/m
3
] T
coke
[K]
JP-7 0.11 2.4810
7
5.34 10
8
4.64 800 600


American Institute of Aeronautics and Astronautics

15

( )
, ( ) ( )
( ) ( )
, ,
( ) ( )
i
m x i i cool comb
m m
Y Y Y
i i
m z m x
Y Y
p p
f geom g geom

= +
=
(32)
where the superscript (i) identifies the location (Fig. 4). The functions
( ) i
m
f and
( ) i
m
g only depend on the geometric
variables
1 5
,..., x x . Similarly, the thermal stresses can be expressed as:

( )
( )
( )
( )
*
( )
0 0 , ( )
, *
0
*
( )
0 0 , ( )
, *
0
1
( , )
2 1
1
( , )
2 1
i
f aw f T x i
T x tf panel
Y Y f
i
f aw f T z i
T z tf panel
Y Y f
E T T T T
f T T
T T
E T T T T
f T T
T T

=


=

(33)
The functions
( ) i
m
f ,
( ) i
m
g ,
( )
,
i
T x
f and
( )
,
i
T z
f are summarized in Appendix II
7
. The Mises yield criterion for mechanical
loads only can be expressed as:

2 2 2
, , , ,
1 18
max 2
m x m z m x m z
i
Y Y Y Y


=

| | | | | |

+ +
` | | |
\ \ \
)
(34)
The corresponding criterion for thermal loads only is:

2 2 2
, , , ,
1 18
max 2
T x T z T x T z
i
Y Y Y Y


=

| | | | | |

+ +
` | | |
\ \ \
)
(35)
Table 3. Non-dimensional parameters used in the optimization.

Nb/Si MAR-M246 Inconel 625 SiC-SiC C-SiC
cool Y
p
0.027 0.025 0.038 0.05 0.05
f
V

4
15 10
comb Y
p
0.00065 0.0006 0.0009 0.0012 0.0012
( )
2
2
crit
crit
f
p Z
p
V

7
1.9 10
*
0
aw
aw f
T T
T T

0.59 0.74 0.73 0.53 0.47
Pr
f

4.6
( )
*
0 f
Y
E T T

4.24 4.63 6.06 6.18 1.41
0
aw coke
aw f
T T
T T

0.93
0.3 0.3 0.3 0.3 0.3
f s
k k
0.0022 0.0042 0.0055 0.0064 0.013

TBC s
k k
0.02 0.039 0.05

TBC s

0.43 0.36 0.37

G
G
s
h Z
Bi
k
=
10.7 20.6 26.7 30.9 61.8



American Institute of Aeronautics and Astronautics

16
and that for combined loads is:

2 2 2
, , , , , , , ,
1 18
max 2
m x T x m z T z m x T x m z T z
i
Y Y Y Y Y Y Y Y


=

| | | | | |

+ + + + +
` | | |
\ \ \
)
(36)
To ensure that the structure does not yield, imposing combined loading only does not suffice since, in some
locations, the stresses due to thermal and mechanical loads have opposite sign.

Maximum material temperature. To ensure that the structure remains below its maximum allowable
temperature
*
T , the maximum material temperature,
max
tf
T (calculated using Eq. (19) at z Z = )), must satisfy the
condition

max
*
0 0
aw tf
aw
aw f aw f
T T
T T
T T T T

(37)

Maximum coolant temperature. Similarly, the maximum fuel temperature (Eq. (15) at z Z = ) must remain
below that for coking; notably:

max
0 0
aw fuel
aw coke
aw f aw f
T T
T T
T T T T

(38)

Maximum pressure drop. The pressure drop is a result of viscous dissipation and other losses in the cooling
channels. To minimize requirements on pumping power, the maximum allowable pressure drop over the 1m length
is taken as 1 MPa
crit
p . Pressure losses at the manifold/panel connections are neglected
8
. From the definition of
the friction factor,

2
( / )
2
h
f
p Z D
f
u

= (39)
and the introduction of the non-dimensional parameter

2
2
crit
crit
f
p Z
p
V

(40)
the pressure drop over the length of the structure can be expressed as:

( )
2
2
2
1
h
f Z Z
p
H D

| |
=
|
\
(41)
where the friction factor f is given by Eq. (3). Thus the constraint is:

crit
p p (42)

7
Equations (32) and (33) also encompass the core member, although in that case (i=9,18) the stresses are in the y-
direction rather than the x-direction.
8
This assumption simplifies the calculations, but can easily be relaxed, should there be a need to consider other
components as well.

American Institute of Aeronautics and Astronautics

17
Manufacturing. To ensure that the optimal designs are compatible with manufacturing constraints, the face and
core member thicknesses are required to be at least 0.4 mm. In non-dimensional form, these constraints are
expressed as:

( )
1
4
1 3 4
1
0.0004
2
1 0.0004
f
c
x
t Z x
t Z x x x

=
=
(43)
D. Optimization scheme
The fuel flow rate was discretized into 100
intervals over the range 1 4
st
V V =

. At each
, the objective function (31) was minimized
subject to the constraints (34)-(43) using the
quadratic optimizer MINCON in the MATLAB
Optimization Toolbox. To avoid local minima, a set
of at least 20 randomly generated initial guesses was
combined with the solution at the previous value of
. The best solution among these was assumed to
be the global minimum. In some cases, a manual
optimization scheme was employed to verify the
accuracy of the numerical results.

VI. Principal results
The minimum mass of panels is plotted in Fig. 5
as a function of the mixture richness, : case I in
Fig. 5a and case II in Fig. 5b. When a TBC layer
enables lower weight, the result is plotted as a
dotted curve. The plots convey a wealth of
information, collected below in three highlights.
(i) For boundary condition I, which induces
the more severe stresses (Fig. 5a), some
materials (notably Inconel and SiC/SiC)
are inadmissible. That is, they do not yield
viable solutions and are thus absent from
the figure. Others only give solutions
above a critical fuel flow rate,
crit
. These
include Inconel with a TBC ( 3.6
crit
= )
and uncoated MAR-M246 ( 1.4
crit
= ).
The remaining materials (the Nb alloy,
MAR-M246 with a TBC and C/SiC)
provide solutions for all flow rates. For the
less stringent boundary condition II,
solutions exist for all materials at all flow
rates.
(ii) The two materials (Nb alloy and C/SiC)
that enable the lightest designs for both
boundary conditions have weight
independent of the fuel flow rate. That is,
they can be designed to operate at the
stoichiometric fuel/air ratio, so that extra fuel for cooling is not needed. For the other materials, lighter
designs are obtained at higher flow rates. However, the weight of the extra fuel may obviate the benefit.



Figure 5. Minimum weight of the optimized structures as
a function of the mixture richness, . Different materials
of interest offer solutions. (a) Boundary condition case I; (b)
Boundary condition case II. The dashed curves refer to
designs that include a thermal barrier coating. Properties for
all the materials are reported in Table 1.


American Institute of Aeronautics and Astronautics

18
(iii) The merits of using a TBC are evident from the dotted lines. For case I: (a) a TBC is essential if Inconel is
to be used, but only with 3.6
crit
= ; (b) the weight of the MAR-M246 design is reduced by a TBC and,
more importantly, the coating allows a solution for stoichiometric fuel; and (c) no benefit is obtained for
the Nb alloy. For case II, the only benefit of the TBC is the weight reduction for the Inconel design for flow
rates close to stoichiometric.
Examples of the optimal panel dimensions for the MAR-M246 are summarized on Figs. 6 and 7. Under case
I, the coated and uncoated geometries differ significantly (Fig. 6); conversely, under case II they are nearly identical,
and only the uncoated geometry has been reported (Fig. 7). The optimal dimensions of C/SiC are independent on the
mixture richness and expressed in tabular form (Table 4).
A few remarks hold for both materials. For case II, the panel thickness, H/Z, always hits the lower bound (this is
not true for case I, where the bending loads imposed by the combustion pressure force a thicker panel). Similarly,
the manufacturing constraints (Eq. (43)) are active for both face and core members, resulting in / 2 0.5
c f
t t = at all
values of . Finally, / 0.8 L H , implying that the face thickness is roughly 10% of the panel thickness; the
exception is MAR-M246 under case I at low .
For MAR-M246, the aspect ratio of the cooling channels, w/L, increases as the mixture richness increases; for
the coated case, an asymptotical value ( / 0.5 w L ) is soon approached.
Finally notice that, not surprisingly, the thickness of the coating layer decreases as increases; at 4 = almost
no TBC is necessary.


VII. Interpretation and implications



Figure 6. Optimal dimensions for MAR-M246 (boundary condition case I).


American Institute of Aeronautics and Astronautics

19
Designs tend to become lighter as the
mixture richness increases because the cooling
efficiency increases and hence the thermal
stresses decrease. At sufficiently large , the
thermal stresses become insignificant and the
optimal geometry is dictated solely by
mechanical considerations. The transition
between these two regimes occurs at lower
for materials more suited to high temperature
applications; in particular, C/SiC and Nb are
mechanically dominated at all .
Among the materials considered, C-SiC is
the most weight-efficient. The best metallic
system, Nb, is considerably heavier (by a factor
3); the best superalloy (MAR-M246) is even
heavier (by a factor of almost 5). In some
instances, this weight penalty may be obviated
by the structural robustness and ease of
fabrication of metallic systems.
TBCs play a less significant role for
scramjet combustion chambers than for turbine
engines. The main benefit of interposing a
thermally insulating layer between the structure
and the hot gases is an increase in the
temperature of the wall exposed to the gases, which, in turn, reduces the
heat flux transmitted to the structure (Eq. (58)). Since the adiabatic wall
temperature is much higher than the maximum allowable material
temperature, an increase in the latter by 200-300 C does not reduce the
heat flux significantly.
VIII. Conclusions and future work
An analytical model has been developed for the prediction of
temperature distributions and thermo-mechanical stresses in an actively
cooled multifunctional plate to be employed in the combustion chamber
of a Mach 7 hypersonic air-breathing vehicle. The thermo-mechanical
loads have been estimated based on aerothermodynamics considerations.
A thermal network approach has been used to derive the temperature
distribution, also accounting for the possible presence of a thermal barrier coating. A sandwich panel analysis has
been adopted for the thermo-mechanical stress calculations. The analytic model agrees well with finite element
calculations, with the exception of highly conductive materials. Work is under-way to address this deficiency.
Based on the temperature and stress calculations, the geometry of actively cooled plates has been optimized for
minimum weight under constraints representative of service conditions. The mechanical constraints exerted by the
vehicle have a substantial effect on the thermo-structural stresses and, consequently, on the optimal geometries. A
realistic model of the actual boundary conditions in service is necessary for accurate predictions.
Future work will employ computational fluid dynamics (CFD) calculations to verify the convective correlations
used in the thermal network, coupled with a set of active cooling experiments on near-optimal geometries at heat
fluxes representative of hypersonic flight conditions (via a high-power CO
2
laser). Also, alternative core topologies
will be studied within the framework presented in this paper, with the goal of ascertaining the optimal combinations
of materials and structural designs.




Figure 7. Optimal dimensions for MAR-M246 (boundary
condition case II).

Table 4. Optimal dimensions for
C/SiC (the results are independent
on in the range considered).

Boundary conditions
Dimension Case I Case II
/ L H 0.77 0.84
/ W L 0.43 0.56
/ 2
c f
t t 0.16 0.5
/ H Z 0.011 0.005


American Institute of Aeronautics and Astronautics

20
Appendix I
The aerothermodynamic conditions within the combustion chamber are essentially determined by the Mach
number and the shape of the forebody. In the present work, a medium-size vehicle (~10 m long, and ~2.5 m tall) was
assumed, with a compression ramp shaped to generate three oblique shock waves. Although a smooth compression
ramp shaped to offer isoentropic compression would be more efficient, its benefit over a three-wave design rarely
justifies the increased complexity
11
. All calculations assume a free stream Mach number of 7. Using a standard
thermodynamic approach, the transformation of the working fluid from the front of the vehicle to the outlet of the
combustion chamber (Fig. A1) can be followed.
Conditions at the front of the vehicle (station 0)
9

Since both the aerodynamic loads on a vehicle and the mass flow rate of the air entering the engine scale with the
dynamic pressure, optimized hypersonic vehicles are designed to fly within a very narrow range of dynamic
pressures
11
; the dependence of the dynamic pressure on the atmospheric pressure and the flight velocity, and the
strong variation of atmospheric pressure with altitude dictate optimal trajectories where a specific Mach number is
associated with a specific altitude (Fig A2).The optimal altitude for Mach 7 flight is approximately 30 km. At this
altitude, the air properties can be obtained from the US Standard Atmosphere Chart:

0
0
3
0
0
0
4 2
0
1.197 kPa
227 K
0.0184 kg/m
302 m/s
2114 m/s
8.01 10 m /s
p
T
a
V

=
=
=
=
=
=
(44)

9
The station numbers are not consecutive, consistent with the standard engine thermodynamics notation.



Figure A1. Sketch of a hypersonic air-breathing vehicle with a forebody designed to compress the
air through three oblique shocks (geometry not to scale). The calculated properties of the air at three
locations (stations 0, 3 and 4) are reported (see Appendix 1 for details). Based on the conditions at stations
3 and 4, the properties of the air in the combustion chamber were assumed uniform and idealized as shown.


American Institute of Aeronautics and Astronautics

21
Here,
0
p is the static pressure,
0
T the static
temperature,
0
the mass density,
0
a the speed
of sound,
0
V the velocity and
0
the kinematic
viscosity.

Conditions at the inlet of the combustion
chamber (station 3)
The compression system increases the
temperature and the static pressure of the air,
while slowing the flow significantly. Generally
speaking, the efficiency of the combustion
process increases with increasing air
temperature at the inlet of the combustion
chamber. There is, nonetheless, a limit above
which losses due to dissociation and other
effects prevail. These two considerations result
in an optimal temperature at the combustor
inlet, typically
3 0
7 T T = . This condition,
together with Eq. (44) and the shape and size of
the vehicle (which determines the number of
oblique shock waves), fully characterize the
compression process. The software HAP
(GasTables)
11
was used to determine the
conditions of the air at the inlet (for simplicity
assumed to coincide with the end of the
compression stage).
For this design:

3 0
3 0
3 air
3
3
3
3 0
200 0.24 MPa
7 1590 K
1460 kJ/kg
1.5
770 m/s
1150 m/s
0.062
p p
T T
h
Ma
a
V
A A
=
=
=
=
=
=
=
(45)
where A
3
is the cross-sectional area of the combustion chamber and A
0
is the capture area at the inlet of the vehicle
(i.e. the area defined by the streamlines that enter the combustion chamber). If the overall vertical size of the vehicle,
0
2.5 m H = , this results in a combustion chamber height,
3
15.5 cm H = .
Conditions at the outlet of the combustion chamber (station 4)
The chemistry of the fuel dictates the stoichiometric fuel/air ratio,
st
f , i.e. the proportion in which all reagents
are transformed into combustion products (assuming perfect combustion). The complex chemical composition of JP-
7 makes it extremely difficult to calculate this quantity from first principles. Nonetheless, a reasonable estimate can
be obtained by modeling JP-7 with
12.5 26
C H (Ref. 11), which yields:

fuel air
0.0675 kg /kg
fuel
st
air
m
f
m
= =

(46)
In actual service, air-breathing engines often operate at fuel/air ratios that deviate considerably from stoichiometric.
By introducing the fuel/air mixture richness, , the actual fuel/air ratio can be expressed as:




Figure A2. Flight trajectories (altitude VS Mach number).
Adapted from Ref. 11. Contours of uniform dynamic pressure
(
2
0 0 0
2 q V = ) and air mass flow rate entering the engine (
0 0
V )
are depicted. Dynamic pressures of ~ 48 kN/m
2
and air mass flow
rates of ~ 50 kg/m s
2
are reasonable, resulting in the correlation
~ 7 ~ 30 km Ma H .

American Institute of Aeronautics and Astronautics

22

st
f f = (47)
This allows calculation of the enthalpy increase of the working fluid within the combustion chamber. At an injection
temperature of approximately 500 K, the increase associated with complete combustion of a unit mass of the fuel is
approximately:
, 7 f
27500 kJ/kg
PR JP
h (Ref. 11). Neglecting the mass increase of the working fluid due to fuel
injection (the combustion process only adds heat to the working fluid)
10
, and modeling the combustion process as
isobaric, yield the conditions at the exit of the combustion chamber (station 4):

4 3
4 3 air
0.24 MPa
3320 kJ/kg
PR
p p
h h f h
= =
= + =
(48)
All other properties of air at station 4 can be calculated from Eq. (48) via the Mollier diagram
11
; in particular:

4
3
4
4
2800 K
0.295 kg/m
989 m/s
T
a

=
=
=
(49)
Since the velocity of air does not change significantly within the combustion chamber
11
, the Mach number at the exit
of the combustion chamber can be calculated as:

3 4
4
4 4
1.17
V V
Ma
a a
= = (50)
As expected for a Mach 7 vehicle, the flow in the combustion chamber is barely supersonic (consistently with Mach
6-7 denoting the ramjet-scramjet transition).
To be conservative, the temperature in the combustion chamber is assumed to be uniform and equal to the exit
temperature. The chamber will therefore be characterized by the following conditions:

2800 K
3320 kJ/kg
1.2
1150 m/s
0.24 MPa
comb
comb
comb
comb
comb
T
h
Ma
V
p
=
=
=
=
=
(51)
Adiabatic wall temperature and heat transfer coefficient
The heat flux from the combusting gases to the walls of the combustion chamber is typically calculated using the
following correlation
12

11
:
( )
* *
comb aw w
q St V h h = (52)

10
The validity of this assumption was checked against a more elaborate thermodynamic analysis of the combustion
process (via the software HAP(Combustion) (Ref. 11)): no significant differences were observed in the conditions of
air at station 4.
11
As we are in a regime where non-linear effects (such as dissociation) can be dominant, c
p
is not constant,
requiring the use of the Mollier diagram in converting enthalpies to temperatures.
12
This correlation was originally introduced to estimate heat flux to leading edges, but has been largely adopted for
combustion chambers as well.

American Institute of Aeronautics and Astronautics

23
where
*
St and
*
are the Stanton number and the density of the air, both calculated at the Eckert reference
temperature (see below), and
comb
V is the velocity of the air;
aw
h and
w
h are the adiabatic and the actual wall
enthalpies, respectively. For turbulent flow, the following correlation is available for the Stanton number:

*
*2/ 5 *1/ 5
0.0287
Pr Re
x
St = (53)
where
*
Pr is the Prandtl number of the air and

*
* *
Re
comb comb
x
p V L
RT
= (54)
is the Reynolds number. L represents the length-scale of the problem, which in this case can be approximated as the
length of the forebody of the vehicle (~6 m).
The quantities marked with an asterisk need to be evaluated at the reference temperature,
*
T . The Eckert
reference enthalpy is defined as
11
:

2
*
0.22
2 2
comb w comb
h h V
h
+
= + (55)
where a unitary recovery factor was assumed (equivalent to neglecting the thermal conductivity of the gas). The
reference temperature can be obtained from Eq. (55) via the Mollier diagram. Note that the enthalpy at the wall,
w
h ,
in Eq. (55) is unequivocally linked to the wall temperature, and therefore is an unknown. Eqs. (52)-(55) were solved
for various
w
T between 800 and 1000 K, and no appreciable difference was observed in
*
T (and even less on
*
St );
*
2050 K T = was used in all subsequent calculations. This implies:

* 5 2
* 3
* 7
* 4
6.27 10 N s/m
0.4065 kg/m
Re 4.47 10
8.47 10 St

=
=
=
=
(56)
with
*
Pr 1 = and 288 J/kg K R = .
The adiabatic wall enthalpy, or recovery enthalpy, is that attained by the stream-line at the wall if perfectly
insulated. With a unitary recovery factor, the adiabatic wall enthalpy is equal to the total enthalpy:

2
2
comb
aw T comb
V
h h h = = + (57)
Conversion of Eq. (52) into the more customary heat transfer coefficient expression assumes that a constant
specific heat could be used for both the adiabatic and the actual wall temperatures. Although this is not the case, the
error introduced with such an approximation would not be significant, relative to the degree of uncertainty in Eqs.
(52)-(53). For 1.35 kJ/kg K
p
c = , Eq. (52) can be rewritten as:
( )
G aw w
q h T T = (58)
where the heat transfer coefficient is

* * 2
535 W/m K
G comb p
h St V c = = (59)

American Institute of Aeronautics and Astronautics

24
The adiabatic wall temperature is 3050 K
aw
T = .
Static pressure inside the cooling channels,
cool
p
The pressure inside the cooling channels is dictated by optimal combustion considerations, as it greatly affects
the mixing efficiency of air and fuel in the combustion chamber: the higher the Mach number, the lower the
residence time of air molecules in the combustion chamber, which in turn requires higher fuel exit velocities at the
injectors. This is achieved by pressurizing the fuel channels. An analytical expression for
cool
p in terms of the Mach
number of the vehicle would require a detailed knowledge of the vehicle design. In the present work, a
representative value 10
cool
p MPa = was used, which is realistic for hypersonic flight conditions in the Mach 7-9
range
13
.
Volumetric flow rate of the fuel (coolant)
The volumetric flow rate of the fuel (per unit width of the combustion chamber) can be expressed as:

2 0 0 0
0.0082 m /s
f
air st
f
f f f
m
f m f V H
V

= = = =

(60)
where
3
800 kg/m
f
= is the density of JP-7,
0 0
,V are given by Eq. (44) and
0
H is the height of the overall
vehicle, from the leading edge to the bottom of the combustion chamber (
0
2.5 m H = ).
Appendix II
All the stress components in Eqs. (22)-(27) can be conveniently expressed in the form of Eqs. (32)-(33) via the
introduction of four non-dimensional functions (refer to Fig. 4 for the locations of the points):
( ) ( ) ( ) ( )
, ,
, , ,
i i i i
m m T x T z
f g f f .
They assume different values depending on the boundary condition.
Boundary condition 1

2
2
2
( )
2
2
1
at pts 1,4,10,13
2 2
2
1
at pts 2,3,11,12
2 2
2
1
at pts 6,7,15,16
2 4
2
1
at pts 5,8,14,
2 4
f
f f
f
f f
f i
m
f f
f
f f
H t
w
t t
H t
w
t t
H t
w
f
t t
H t
w
t t
| |
|
|
\
| |
+ |
|
\
| |
= |
|
\
| |
+ |
|
\
17
at pts 9,18
c
w
t

(61)

13
D. Marshall and T. A. Jackson, personal communication.

American Institute of Aeronautics and Astronautics

25

( )
( )
( )
( )
2
2
2
( )
2
at pts 1,2,5,6
12
at pts 3,4,7,8
12
at pts 12,13,16,17
24
at pts 10,11,14,15
24
0 at pt
f f
f f
i
m
f f
f f
b
t H t
b
t H t
b
g
t H t
b
t H t

s 9,18

(62)

0 0
( )
0 0 ,
0
at pts 1,5,10,14
at pts 2,6,11,15
at pts 3,4,7,8,12,13,16,17
0
panel tf
aw f aw f
panel tf
i
aw f aw f T x
panel
aw f
T T
T T T T
T T
T T T T f
T
T T




+
=

at pts 9,18

(63)

( )
( )
0 0
( )
,
0 0
0
2
at pts 1,5,10,14
2
2
at pts 2,6,11,15
2
2
at pt
2
f c panel tf
f c aw f aw f
f c panel tf
i
T z
f c aw f aw f
f panel
f c aw f
A A T T
A A T T T T
A A T T
f
A A T T T T
A T
A A T T
+

+
+
+
=
+

+
s 3,4,7,8,12,13,16,17
0 at pts 9,18

(64)

Boundary condition II
The thermal functions
( )
,
i
T x
f and
( )
,
i
T z
f are the same as for case I. The mechanical functions are:

American Institute of Aeronautics and Astronautics

26

2
2
2
( )
2
2
1
at pts 1,10
2 2
2
1
at pts 2,11
2 2
2
1
at pts 6,15
2 4
2
1
at pts 5,14
2 4
2

2
f
f f
f
f f
f
i
f f
m
f
f f
f
f
H t
w
t t
H t
w
t t
H t
w
t t
f
H t
w
t t
H t
t
| |
|
|
\
| |
+ |
|
\
| |
|
|
= \
| |
+ |
|
\

at pts 3,4,7,8,12,13,16,17
at pts 9,18
c
w
t

(65)

( )
0 1:18
i
m
g i = = (66)

Acknowledgments
This work was supported by the ONR through a MURI program on Revolutionary Materials for Hypersonic
Flight (Contract No. N00014-05-1-0439).

References
1
Buchmann, O. A., "Thermal-structural design study of an airframe-integrated scramjet", NASA 3141, 1979.
2
Rakow, J. F. and Waas, A. M., "Thermal buckling of metal foam sandwich panels for convective thermal protection
systems", Journal of Spacecraft and Rockets, Vol. 42, No. 5, 2005, pp. 832-844.
3
Rakow, J. F. and Waas, A. M., "Thermomechanical response of actively cooled metal foam sandwich panels for thermal
protection systems", 46th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics & Materials Conference, AIAA, 2005,
pp. 1-23.
4
Song, K. D., Choi, S. H. and Scotti, S. J., "Transpiration cooling experiment for scramjet engine combustion chamber by
high heat fluxes", Journal of Propulsion and Power, Vol. 22, No. 1, 2006, pp. 96-102.
5
Scotti, S. J., Martin, C. J. and Lucas, S. H., "Active cooling design for scramjet engines using optimization methods", NASA
100581, 1988.
6
Youn, B. and Mills, A. F., "Cooling Panel Optimization for the Active Cooling System of a Hypersonic Aircraft", Journal of
Thermophysics and Heat Transfer, Vol. 9, No. 1, 1995, pp. 136-143.
7
Flieder, W. C., Richard, C. E., Buchmann, O. A. and Walters, F. M., "An Analytical study of hydrogen cooled panels for
application to hypersonic aircraft", NASA CR-1650, 1971.
8
Walters, F. M. and Buchmann, O. A., "Heat Transfer and Fluid Flow Analysis of Hydrogen-Cooled Panels and Manifold
Systems", NASA CR-66925, 1971.
9
Lu, T. J., Valdevit, L. and Evans, A. G., "Active cooling by metallic sandwich structures with periodic cores", Progress in
Materials Science, Vol. 50, No. 7, 2005, pp. 789-815.
10
Valdevit, L., Pantano, A., Stone, H. A. and Evans, A. G., "Optimal active cooling performance of metallic sandwich panels
with prismatic cores", International Journal of Heat and Mass Transfer, Vol. 49, No. 21-22, 2006, pp. 3819-3830.
11
Heiser, W. H. and Pratt, D. T. Hypersonic air-breathing propulsion, AIAA education series, AIAA, Washington, D.C.,
1994.
12
Holman, J. P. Heat transfer, McGraw-Hill series in mechanical engineering, McGraw-Hill, New York, 2002.
13
Bejan, A. Convection heat transfer, Wiley, Hoboken, N.J., 2004.
14
Gnielinski, V., "New Equations for Heat and Mass-Transfer in Turbulent Pipe and Channel Flow", International Chemical
Engineering, Vol. 16, No. 2, 1976, pp. 359-368.
15
Blasius, H., Forshungsarb. Ver. Deutsch. Ing., Vol. 131, 1913.
16
Boley, B. A. and Weiner, J. H. Theory of thermal stresses, Wiley, New York, 1960.

American Institute of Aeronautics and Astronautics

27
17
Rathbun, H. J., Zok, F. W. and Evans, A. G., "Strength optimization of metallic sandwich panels subject to bending",
International Journal of Solids and Structures, Vol. 42, No. 26, 2005, pp. 6643-6661.
18
Valdevit, L., Hutchinson, J. W. and Evans, A. G., "Structurally optimized sandwich panels with prismatic cores",
International Journal of Solids and Structures, Vol. 41, No. 18-19, 2004, pp. 5105-5124.
19
Valdevit, L., Wei, Z., Mercer, C., Zok, F. W. and Evans, A. G., "Structural performance of near-optimal sandwich panels
with corrugated cores", International Journal of Solids and Structures, Vol. 43, No. 16, 2006, pp. 4888-4905.
20
Wei, Z., Zok, F. W. and Evans, A. G., "Design of sandwich panels with prismatic cores", Journal of Engineering Materials
and Technology - Trans. of the ASME, Vol. 128, No. 2, 2006, pp. 186-192.
21
Wicks, N. and Hutchinson, J. W., "Optimal truss plates", International Journal of Solids and Structures, Vol. 38, No. 30-
31, 2001, pp. 5165-5183.
22
Wicks, N. and Hutchinson, J. W., "Performance of sandwich plates with truss cores", Mechanics of Materials, Vol. 36, No.
8, 2004, pp. 739-751.
23
Zok, F. W. et al., "Structural performance of metallic sandwich panels with square honeycomb cores", Philosophical
Magazine, Vol. 85, No. 26-27, 2005, pp. 3207-3234.
24
Zok, F. W., Rathbun, H. J., Wei, Z. and Evans, A. G., "Design of metallic textile core sandwich panels", International
Journal of Solids and Structures, Vol. 40, No. 21, 2003, pp. 5707-5722.
25
Zok, F. W., Waltner, S. A., Wei, Z., Rathbun, H. J., McMeeking, R. M. and Evans, A. G., "A protocol for characterizing the
structural performance of metallic sandwich panels: application to pyramidal truss cores", International Journal of Solids and
Structures, Vol. 41, No. 22-23, 2004, pp. 6249-6271.