AERONAUTICAL SCIENCE
Jerome C. Hunsaker, Consulting Editor
THERMAL STRESSES
McGRAW-HILL PUBLICATIONS IN
AERONAUTICAL SCIENCE
Jerome C. Hunsaker, Consulting Editor
1957
THERMAL STRESSES
The advent of supersonic speeds for airplanes and missiles since World
War II hasconfronted the aircraft structural designer with new problems
involving temperatures, thermal stresses, and material properties. The
engine designer has done considerable work on these problems, but the
airframe designer is just beginning the task of trying to solve these
problems for sheet-metal-type structure. Also, the harnessing of atomic
energy has introducedcomplicated heat-transfer and thermal-stress
problems in the design of nuclear reactors. This book represents an
attempt to give basic information for attacking these problems and to
indicate possible procedures for solving the various problems associated
with elevated temperatures in airplane and missile structures, jet engines,
and nuclear reactors.
The book is not restricted to the classical thermal-stress problem of
finding the elastic thermal stresses for a given temperature distribution
in a structure with no buckling but touches on all phases of the structural-
design problem — the temperature distribution, the elastic and inelastic
thermal stresses in various structures, the combined elastic and inelastic
applied and thermal stresses, the allowable stresses for various materials
and loading conditions, the buckling, deflection, stiffness, fatigue, shock,
and flutter effects of elevated temperatures.
The method of attack on these problems is to set up the complete
problem and then simplify it by making assumptions based on the physical
situation or on experimental data. An analytical solution is obtained
for the approximated problem, which demonstrates the basic parameters
in the problem and allows charts to be constructed, showing how the
stresses vary with these parameters. Refinements in the solution are
then made by investigating the various simplifying assumptions and
obtaining correction factors. In most cases this procedure yields results
with sufficient accuracy for use in the design of the structure. Actually
the thermal-stress calculation is just one of an entire sequence of approxi
mate calculations necessary in the design of a structure. The thermal-
stress-calculation procedures as developed herein are considered to be as
accurate as some of the other steps in the design sequence, and more
accurate than the temperature based on the convective-heat-transfer
vi PREFACE
PREFACE v
SYMBOLS xiii
2- 1. Introduction 21
2-2. Heat Conduction 22
2-3. Heat Radiation 23
2-4. Heat Convection . 26
2-5. Aerodynamic Heating 30
2-6. Heat-balance Equations for Idealized Structure 35
2-7. Numerical Integration for Transient Temperatures 37
3- 1. Introduction 43
3-2. Calculation of Stringer Temperature for Assumed Skin Temperature. 43
3-3. Radiation from Skin to Stringer 47
3-4. Effect of Stringer upon Skin Temperature 49
3-5. Effect of Thermal Resistance of Joints on Stringer Temperatures . . 50
3-6. Comparison of Calculated and Test Temperatures
3-7. Stringer and Skin Temperatures for Constant Heat Source
3-8. Skin and Stringer Temperatures for Missile Structures
... 52
54
58
3-9. Thermal-stress Formulas for Unrestrained Idealized Structure . . 62-v
ix
X CONTENTS
4-1. Introduction 65
4-2. Plates or Beams with Linear Transient Temperature on One Surface or-
Edge 66
4-3. Plates with Constant Heat Source on One Surface 73
4-4. Plates with Variable Heat Source on One Surface 7?.
4-5. Beams with Constant Heat Source on Skin 79-
4-6. Shear Lag in Beams 85
4- 7. Thermal Stresses in Turbine Blades 88
5- 1. Introduction
5-2.
5-3.
Beams with Two-dimensional Temperature Distribution
Thermal Stresses in a Wedge-shaped Wing
.... 94
94
97
5-4. Circular Plates with Temperature Function of Radius 100
5-5. Finite-difference Method for Thermal Stresses in a Turbine Disk . . 102
5- 6. Thermal Stresses in Nuclear-reactor Components 105
7- 1. Introduction
....
153
7-2. Inelastic Thermal Stresses for Uniform Temperature 154
7-3. Inelastic Thermal Stresses for Nonuniform Temperature 158
7-4. Combined Applied and Transient Thermal Stresses 162
7-5. Effective Areas with Thermal Stresses 165
7-6. Combined Applied and Steady-state Thermal Stresses 166
7-7. Design Procedures at Elevated Temperatures 168
8- 1. Introduction 171
8-2. Bowing of Plates and Beams 173
CONTENTS xi
....
178
8-5. Deflection of Plates under Normal Loads 179
8-6. Deflections Due to Cross-section Warping and Rotation 183
8-7. Shear Deflections 187
8- 8. Effect of Thermal Loads on Flutter 187
9- 1. Introduction 194
9-2. Variational Methods 195
9-3. Thermal Stresses in Rectangular Plates 197
9-4. Buckling of Rectangular Plates 199
....
9-8. Composite Body with Cross Section as Two Eccentric Circles 211
INDEX 227
SYMBOLS
General notation
Operator (d^/dx1) + (d'/dy')
e Normal strain
1 Normal stress, psi
F Allowable stress, psi (see Table 6-3 for various allowable
stresses) ; function
9 Acceleration due to gravity, ft/sec2, in./secs
y Shear strain
i (-D*
Stress function; analytic function
Analytic function
9 Heat transfer, Btu/hr
r, 0 Polar coordinates
r, e, z Cylindrical coordinates
8 Arc length; complex variable in unit circle
i Time, hr, min, sec; complex variable on boundary of unit
circle
T Temperature, °F; temperature, °R; temperature change
from datum temperature To
T Shear stress, psi
U, V, w Displacements in x, y, z directions or r, 9, z directions
U Biharmonic function
V Velocity, ft/sec
V Temperature function
w(s) Mapping function
x, y, z Rectangular coordinates
z Complex variable
xiii
xiv SYMBOLS
Material properties
a Coefficient of linear thermal expansion, in./(in.)(°F)
c„ Specific heat, Btu/(lb)(°F)
5 Per cent elongation
E Modulus of elasticity, psi
e Emissivity coefficient
F Allowable stress, psi (see Table 6-3 for various allowable
stresses)
G Shear modulus of elasticity, psi
Convective heat-transfer coefficient, Btu/(hr)(ft2)(°F)
-
h
k Thermal conductivity, Btu/(hr)(ft)(°F); Ea; Ea/(l v)
m Reciprocal of Poisson's ratio
H Coefficient of viscosity
v Poisson's ratio
p Density, lb /in.3
a Radiation constant, Btu/(hr)(ft2)(°F4), 0.1742 X 10~8 for
black bodies
Geometric properties
A, S Area, in.2
o, 6, c, d, LLength, or height, or width, or thickness, in.
c Distance from neutral axis to extreme fiber, in. ; column-end
fixity
e Gap, in.; eccentricity, in.
ec Allowable rivet deflection, in.
h„ t Skin or plate thickness, in.
K, Web thickness
/ Moment of inertia, in.4
/Torsional moment of inertia, in.4
K Buckling coefficient, Eq. (6-2)
L' L/c\ effective column length
P Radius of gyration, in.
s Rivet spacing, in.
W Plate deflection, in.
Loads
M Bending moment, in.-lb
N Torsional moment, in.-lb
p Pressure, force, psi
P Thermal load, lb
Pc Allowable rivet load, lb
P, Q Applied load, lb
q Shear flow, lb /in.
X, Y, Z Components of force, x, y, z directions
Parameters
B Thermal stress parameter, Eq. (3-5); allowable stress
parameter, Eqs. (6-1), (6-2), (6-4); optimum stress
parameter, Eqs. (6-24), (6-27), (6-29)
D Radiation parameter, Eqs. (3-11), (3-15); deflection parame
ter, Eqs. (8-11), (8-12), (8-14), (8-16), (8-18), (8-19),
(8-41), (8-44), (8-45), (8-55), (9-26); stiffness parameter,
Eq. (9-21)
jj Buckling stress parameter, Eq. (6-3)
SYMBOLS XV
THERMAL-STRESS EQUATIONS
AL =
aL(T - To) (1-1)
where L = length of bar
T — = temperature change
To
a = coefficient of linear thermal expansion
To = initial temperature
T = final temperature
This coefficient a is the change in length of the bar per unit length due
to a change in temperature of 1°. The units for a are usually inches
per inch per degree Fahrenheit.
If all the fibers of the body are free to expand or contract, no stress
is produced by the temperature change. However, such an expansion
or contraction generally cannot proceed freely in a continuous body, so
that stresses are produced. External restraints on the body that pre
vent expansion or contraction also produce stresses in the body. These
stresses are known as thermal stresses.
The elongation of a bar of length L due to a uniform stress is /
AL =
§ (1-2)
Example 1-1. A steel rod 20 in. long is fastened between two walls without load
at 90°F. (a) What is the stress in the rod at -30°F? (6) If the walls deflect
0.01 in. under the pull of the rod, what is the stress at -30°F? Use a = 7(10-»)/°F
and E = 29(10") psi from Table B-3.
Solution. For part a Eq. (1-3) gives
*L(T - To) +
^ + 0.01 = 0
or / = -Ea(T - T0)
- 0.01
f = 24,400 - WgXlO') ^^
The value of K in Eq. (1-4) for this case is
X =
1+S#^)=1-°-59=0-41
PROBLEMS
1-1. An aluminum-alloy rod with a cross-section area of 0.50 in.2 is fastened between
two walls at 90°F with a preload of 4,000 lb tension, (a) What will the stress be
at — 10°F? (6) For what temperature will the stress be zero? See Table B-3 for
a and E.
1-2. In Example 1-1 what total deflection of the walls would make the stress
zero? What would this deflection have to be if the rod were 100 in. long?
1-3. A steel wire is stretched between two rigid walls and carries a tensile load
of 4,000 lb at'100°F. If the allowable stress is not to exceed 50,000 psi, draw a curve
of minimum wire diameter against temperature. For what temperature does the
diameter tend toward infinity?
1-4. A 5-ft aluminum-alloy tube 2 in. in diameter with \ in. wall thickness is
supported between rigid walls at 80°F. At what temperature will the Euler long-
column buckling load of the tube be reached? Assume pinned ends.
1-5. Solve Prob. 1-4 if there is a gap of 0.05 in. between one wall and the end of the
tube at 80°F.
THERMAL-STRESS EQUATIONS 3
aMTi - - +^
f-£
To) + = «2L(T2 To)
(1-5)
fiAi = —f2A2
where subscript refers to bar No. and subscript to bar No. and
2
1
1
the cross-section area of the bar. Equation (1-5) can be solved for
is
A
/i= - aii?i[(Ti —
T0)
+ (AiEi/A2E2)
(0:2/0:l) (7*2
—
To)]
-KaJSi{Ti
.
- To) .
1
= Ai.
f2
(1-6)
K
- [a2(T2
- T0)/ai(Ti - To)]
1
=
+ (A1E1/A2E2)
1
is
/i
0,
is
0,
is
/2
there gap present so that the two bars may move some before they
is
a
restrain each other (see Fig. 1-1), then K in Eq. (1-6) becomes
K
- MT2 - To)/ai(Ti To)]
-
[e/aMTi - - To)]
1
=
(1-7)
(AiEi/A2E2)
+
1
where the gap in inches, To determine the value of the gap for zero
is
e
= LMTi - To)
- a2(T2 - T0)] (1-8)
e
0
a
aluminum-alloy block at Ti —
T0 = 400° on the Fahrenheit scale, then
= 0.0048L. Thus for short lengths small gaps may practically elimi
e
4 THERMAL STRESSES
nate thermal stresses. However, for large values of L, small gaps will
have little effect on the stresses.
For the case in which the two bars are of the same material so that
ai = «2 = a and Ei = E2 = E, the stresses
as given by Eqs. (1-6) and (1-7) can be writ
Aluminum
1-2 ft ten in the form
-
(6.5)(10-«)(1)(100) + 12(10-') (2) (100)
+^
- 2/,
3(10')
or ifi = 17,500
Also 2/!(0.80) +/2(1.60) = 20,000
so that = 18,200 psi
/2 = -5,700 psi
Note that the aluminum rod is in compression and suitable provision may have to be
made to prevent buckling.
PROBLEMS
Aluminum
£=3 ft
Steel
1 = 2 ft
|— 3 ft- 4 ft -3 ft —I
5000 lb
Prob. 1-12
1-13. In Prob. 1-12 what temperature change in the aluminum rod will relieve
the steel rod of all load? What temperature change in both rods will relieve the
steel rod of all load?
1-14. Solve Prob. 1-12 if the beam is steel with a moment of inertia oi = 8.0 in.4 I
1-15. As shown, the aluminum-alloy skin is attached continuously to an aluminum-
alloy stringer at 70°F. If the skin is raised to 450°F and the stringer is raised to
120°F, what are the stresses in the skin and stringer? Neglect buckling and bowing.
- 6.0 in. -
Pbob. 1-15
1-3. Thermal Loads on Joints. If the two bars discussed in Sec. 1-2
are fastened together with rivets or bolts which do not deflect, then the
end rivet or bolt must take all the thermal load produced by a tempera
ture change. This can be seen by an examination of Fig. 1-3. Let the
holes for the bolts be drilled in the two bars at temperature T0. For
one bar at temperature Ti and the other at temperature T2 the holes
will not be in line. However, if the end holes are forced into alignment
III II Zfc— 3i
IL_>r2
1 ^J, U u — ^3 — i
'
1 II II II II h-
1 II II II ll
and the end bolts inserted, then all the other holes will be in line. Thus
the other bolts may be inserted, but they will carry no load. On the
other hand, if the inner holes are forced into alignment and the bolts
inserted, the outer holes will not line up (Fig. 1-3).
Now if the end bolts should deflect or there should be a gap because
of oversize holes, then the other holes will not line up so that some load
will be carried by the other bolts. Also, by Eq. (1-7), this deflection
will reduce the total load to be carried. From Eq. (1-5) the magnitude
of the total load is
N 2
P- * a^iAi?!
P- 02^2^2, Tz
U- s -»j«-S-»J
Fig. 1-4. Notation for joints. [By permission of the Journal of the Aeronautical
Sciences, 21(9): 645-646 (1954).]
has been considered by the author (Ref. l).f This deflection may be
due to bearing on the bar, shear in the rivet, bending of rivet, etc. The
deflection e„ of rivet number n can be expressed as
—r P n = 1, 2, . . . , N (1-11)
-
n
-^V
n
= + A2E,
«i(7\ To) (r<) «2(r2 T0)
X (r<) (1-12)
+
ecP
sPc
(r„
- rn+i) n = 1, 2, . . . , N - 1
fa) = 1
rN a N
Z_/ rN
t= i
— =
(2 + H) r-^
Pat
- ^=±?
r.v
n = 1, 2, . . . , AT - 2 (1-13)
ecA i^i
fl + ^)
From Eq. (1-12) P can be written as
P = PoJ J =
1
+ (srN/aH)
- Mr, -
1
Po = A1E1ai{T1 - T0)
i r,)/ai(ri To)]
(1-14)
1 + (A^/A2E2)
where P0 is the load without any rivet deflection. Further the load
on each rivet becomes
Since the end rivet No. 1 carries more load than rivet No. 2, etc., the
primary value needed in analysis and design is qi for different values of
N, s/a, and H, where H may be regarded as the joint-stiffness parameter.
Figure 1-5 shows a graph of qi against H with s/a and N as parameters.
The results shown by Fig. 1-5 indicate that, for the usual values of H,
0 0.05 0.10 0.15 0.20 0.25 1.0 2.0 3.0 4.0 5.0
ecAiEi\ A2E2j
Fig. 1-5. Load Pi on end rivet or bolt. [By permission of the Journal of the Aero
nautical Sciences, 21(9): 645-646 (1954).]
8 THERMAL STRESSES
the end rivet does not take all the load Po: the deflection of the rivets
reduces the total load and divides it among all the rivets. For example,
take aluminum-alloy bars with Ei = E2 = 107 psi, Ai = A2 = 0.50 in.2,
and s = 1.00 in., Pc = 1,500 lb, ec = 0.01 in., whence H = 0.06. Hence,
for s/a = 0.01 and N = 2, Pi = 0.46F0. For this case Eq. (1-13) gives
P2 = 0.43Fo- If ec approaches zero, then H becomes large and the curves
show that the end rivet carries nearly all the load. It should be noted
from the curves that, for four or more rivets, the load on the end rivet
is nearly independent of s/a and N, some small differences occurring in
the region 0 < H < 0.10.
PROBLEMS
3.00 in.
Rivets
Pbob. 1-23
THERMAL-STRESS EQUATIONS 9
If the plate is free of external forces at the ends (no restraints), then to
get the thermal stresses in the plate it will be necessary to superpose on
the stresses in Eq. (1-16) the stresses produced in the plate by tensile
Px = jc_caET{y)bdy (1-17)
and at a sufficient distance from the ends they will produce an approxi
mately uniformly distributed tensile stress of magnitude
Px = Fx = 1
fe
(/*) ' = v
ET(y)
a^ dy (1-18)
A 2cb 2c J_c
Further, if the temperature distribution is unsymmetrical, these tensile
stresses have the resultant moment
M,
jc_c
= aET(y)by dy (1-19)
will apply so that this resultant moment will produce bending stress
a
magnitude
of
M,y _ 3M,y 3y
J [c
(/*)»
= aET(y)y dy (1-20)
2c3b 2c' _
Thus the total stress in an unrestrained thin plate away from the ends
is
-aET(y) + aET(y)y
g
= «ET(y) dy +
£
dy (1-21)
/.
the plate
is
if
For a beam with the temperature varying through the depth of beam,
Eq. (1-21) has the form, from Eqs. (1-18) and (1-20),
/, = -
To =
aE(T0
Tm- Kix - K2y T)
IxyJTy dA - IxJTx dA
II I — 2
dA - IyjTy dA (1-23)
= hJTx
K2
7 7 — I 2
-i-
Tm = TdA
|
axis
y
large and equal to the width of the plate and 2c taken as the plate
is
thickness. Since the plate wide in the direction and the fibers in the
is
fx
and
0
0
aETM
—
1
From the discussion above for the unrestrained thin plate follows that
it
[Eq. (1-21)] for the unrestrained thick plate (away from edges)
aEny)y *
/.'.
+
(1-24)
2c-(rh>
THERMAL-STRESS EQUATIONS 11
r-r.(i-g)
From Eq. (1-21) the stress in the plate is
fx = %aETa
- aETo -£
^1
in plate
iaET0 in stringer
U
3(fc + 2)
If an average temperature of %T0 is used in the plate, then Eq. (1-6) can
be used to give
/. - 2k(aETo)
m+y
.
m plate
. .
iaETo . , .
/. m Strmger
W+2)
Thus the assumption of an average temperature in the plate gives the
correct stringer stress but gives only an average plate stress, which may
be much smaller than the maximum plate stress.
It mustbe remembered that the above formulas do not give the correct
stresses near the ends of the unrestrained beam or the edges of the
unrestrained plate. The stresses are two-dimensional near the ends and
edges so that either the methods of Chap. 9 or the approximate shear-lag
procedure of Sec. 4-6 must be used to get the stresses.
PROBLEMS
T = To + on -c < y < 0
(l
T = To
(l
- on 0 < y < c
|)
find the stress distributions in the plate.
12 THERMAL STRESSES
3.0
(6) T =
To^
»*-?(' +9"
Prob. 1-25 Find the stresses in the beam in each case.
1-26. Change the beam in Prob. 1-25 to a Z section, which is unsymmetrical, and
solve Prob. 1-25 (a) and (6).
1-27. If the temperature is uniform in the cross section of a bar but varies along
the length, then for complete restraint
(a) 7/ - 3". |
(6) r = rm
(j)'
(C) r = r, + (Tm
- t„)
1-28. Derive Eq. (1-6) when the temperatures Tt — T0 and T2 — To in the bars
vary along the length.
1-29. What happens in Prob. 1-27 when the cross-section area of the bar varies
along its length? Solve Prob. 1-27 for
A(x) = A0 + (Am
—
A0) j
1-5. General Thermal Stress-Strain Equations. It was only a few
years after the publication in 1829 of Poisson's famous memoir (Ref. 2)
on the theory of elasticity that Duhamel (Refs. 3-5) investigated the
modifications of elasticity theory to allow for change of temperature.
At about the same time Neumann (Ref. 6) deduced the thermoelastic
stress equations in a form nearly identical to those of Duhamel. In
1873 Borchardt (Ref. 7) discussed the solutions of the Duhamel-Neumann
elasticity equations in two and three dimensions and used potential
theory to solve the circular plate and sphere for general temperature
distribution. In 1879 Hopkinson (Ref. 8) and in 1900 Alibandi (Ref. 9)
gave the thermoelastic equations essentially in the form used today.
THERMAL-STRESS EQUATIONS 13
Today almost all books on elasticity theory give the thermoelastic equa
tions. Reference is made to Timoshenko and Goodier (Ref. 10) for the
equations as given below.
To determine the relations between strains and displacements in an
elastic body, consider a small element dx dy dz of an elastic body (Fig.
A-l). Let u, v, w be the components of displacement in the x, y, z
A'
dy
v+^dx
dx
(du)x
= ~
du
ex
= (du)x
dx dx
Similarly ey = dv/dy,
and e, = dw/dz. To determine the distortion of
the angle between the elements OA and OB, consider Fig. 1-7. The
displacement of point A in the y direction is
»A = v+-dx
while the displacement of point B in the x direction is
uB = u + 3- dy
14 THERMAL STRESSES
Thus the right angle AOB changes to the angle A'O'B', and the change
in the angle is
'*xy
_ vA
— v
Ub — u _ dv du
dx dy dx dy
which is the shearing strain between the planes xz and yz. Thus the six
strains in terms of the three displacements are
ex - du die
6y ~~
dv
dy
e. .. - dw
~dz
"Yxy
_ du
dy
Jli
dx
Ta - du +
dz
dw
dx
(1-25)
dv dw
dz dy
Since there are six strains expressed by Eq. (1-25) in terms of three
displacements, there must be compatibility relations between the strains.
Repeated differentiation and elimination of the displacements gives the
following equations of compatibility:
, d2ey d27*y
=
dy2 dx2 dx dy
d2ey d2e, d2yyi
=
dz2 dy2 dy dz
d2e, d2ex _ d2yx,
-j-
dx2 dz2 dx dz
(1-26)
d2ex dyyx dy,x dy
d
\ (
dy dz dx dx dy dz
)
d% _ -l dyxy dyy,\
A
= ^.Ti5
i
2
dx dz dy dy dz dx
\
d2e,
_ J.
^
dx dy dz dz dx dy
)
Hooke's law and the elongation due to temperature [Eq. (1-1)] can be
used to obtain the thermal stress-strain equations. If stress
is
fx
a
/-
/*
= e, =
-VE -VE
y
Thus,
f,
if
v
THERMAL-STRESS EQUATIONS 15
= vUy+m + «T
ex
^[/x-
ey =
g[/„
- *(/.+/.)] +aT (1-27)
* = ±[f, - »(/.+/,)] + «r
Since free thermal expansion does not produce angular distortion in an
isotropic material, the relations between the shearing stresses and shear
ing strains are unaffected by temperature. They are
= fy, — fx, —
(1-28)
q~ ~q~ ~q
G =
mh) (1-29)
dx dy dz
dtxy dfy dry, _
y
^ (1-30)
dx dy dz
drx, dry, d/, _
,
dx dy dz
(1-31)
t„ cos (n,z) = Z,
f,
From the above equations it is seen that the solution for the stresses
in the general thermoelastic problem has the same procedure as for the
general elasticity problem. That solve Eq. (1-30) subject to Eqs.
is,
(1-31) and (1-26), making use of the relations in Eqs. (1-27) and (1-28).
The relations in Eq. (1-27) introduce the temperature into the problem
through Eq. (1-26).
Consider the effects of different restraints on the body when subjected
to temperature change. If the body completely restrained in every
is
a
[p
v(j>
^
This pressure distribution can by applying certain body forces
be realized
and surface pressures to the body. These forces must satisfy the equi
librium equations (1-30) and boundary conditions (1-31). Put
^
^2v
\
v =
A, —
aET , .
_ 2v
cos {n,x)
1
aE dT
Y
Y. = - _ 2v
cos (n,y)
1
aE dT
-
Z
2v dz
1
Z. = - _ 2v
cos (n,z)
i
dx dy dz — 2v dx
1
dx dy dz dy
aE dT
dr„ dry, df. ^
-
,
i~ ~*~
dx dy dz 2v dz
1
THERMAL-STRESS EQUATIONS 17
fx cos (n,x) + rxy cos (n,y) + tx1 cos (n,z) = cos (n,x)
ctE T
cos (n,x) + /„ cos (n,y) + ry, cos (n,z) = _ ^
cos (n,?/) (1-36)
PROBLEMS
yxy -_ du + , dv _- 1
dy~
rxu
G
b~x
by2 dx2 dx ay
da; 3i/
(1-39)
dx dy
cos (n,x) + tx„ cos (n,2/) = X,
fx
If
<t>
_ d. _- _- ^ ^
,
Jy
J*-dy'' dx2'
rxy
dxdy U-4AJ
then Eq. (1-39) satisfied. Substitution into Eq. (1-37) and then into
is
V«* + aEVT = V = +
£
(1-42)
^
0
18 THERMAL STRESSES
Using Eq. (1-41), the boundary conditions in Eq. (1-40) can be written
in the form
ex
_ du
dx
_ [/. - w(f. + /„)] - ve. + (1 + v)aT
dv _
fly
dy
- Grxy
1
7«*
= E{e, - ,(f.
f,
ccT) + + fv)
(1-46)
ry, = rxx =
0
ez
Jj.dA (1-47)
0
is
shenko and Goodier (Ref. 10), or Sokolnikoff (Ref. 13), or Love (Ref. 14),
THERMAL-STRESS EQUATIONS 19
d/r , 1
~•~
drre
^
. fr —
fe _ n
U
dr r 66 r
"+" + _
r dr r
where /r and /« are the normal stresses in the radial and tangential direc
tions, respectively, and rr$ is the shear stress. This equation (1-48) is
<t>
satisfied by the stress function when
J' _
, d<t>
1
1
d2<t>
dr r2 dd2
r
dH /l dA (1-49)
j9
f
dr\rdd)
<t>
~*~
rdr^~
(
dr2
J
dx2 dy2 r2 dd2
The above formulas give the thermal stresses throughout the entire
body for either one- or two-dimensional temperature distribution.
is a
Now, the body a plate or beam as discussed in Sec. 1-4, then may be
it
if
possible to approximate the stresses away from the ends or edges for
is a
two-dimensional temperature distribution. If the temperature T{x,y)
sufficiently smooth in x, then Eqs. (1-21), (1-22), and (1-24) can be used
to give the normal stress in the beam or plate. This result follows
fx
if
plane sections remain plane under thermal stress. Also the moment M,
in Eq. (1-19) can vary with x, and the conventional bending formula
M,c/I will give good results. The accuracy of this approximation and
the addition of higher-order terms to the Eqs. (1-21), (1-22), and (1-24)
to give more precise results for/x have been considered by Boley (Ref. 12)
(see Sec. 5-2).
PROBLEMS
1-33. Derive Eq. (1-42).
1-34. Derive Eq. (1-43).
1-35. Derive Eqs. (1-45) and (1-46).
1-36. For a steady-state temperature distribution in two dimensions, V2T —
0.
Prove that the thermal stresses in an unrestrained solid plate are zero for this temper
ature distribution. Would the same result hold for plate with hole in it?
a
1-37. Set up equations for the two-dimensional plate similar to Eqs. (1-35) and
(1-36).
20 THERMAL STRESSES
1-38. Write the compatibility equation (1-38) in terms of stresses and temperature.
1-39. Using the relations x = r cos 0 and y = r sin 8 for polar coordinates, derive
the polar form for the operator V2 in Eq. (1-50).
1-40. Write Eq. (1-37) in polar coordinates, and verify Eq. (1-42) in polar form
by using Eq. (1-38).
REFERENCES
1. Gatewood, B. E.: Thermal Loads on Joints, /. Aeronaut. Sci., 21(9) :646 (1954).
2. Poisson, S. D.: Memoire sur l'equilibre et le mouvement des corps elastiques,
Mem. Paris Acad., 8:357-570 (1829).
3. Duhamel, J. M. C. : Memoire sur le calcul des actions moleculaires developpees
par les changements de temperature dans les corps solides, Mim. inst. France,
5:440-498 (1838).
4. Duhamel, J. M. C: Seconde memoire sur les phenomenes thermo-mecaniques,
J. icole polytech. (Paris), 15:1-57 (1837).
5. Duhamel, J. M. C: Memoire sur le mouvement des differents points d'une barre
cylindrique dont la temperature varie, /. tcole polytech. (Paris), 21:1-33 (1856).
6. Neumann, F. E.: Die Gesetze der Doppelbrechung des Lichts in Comprimirten
oder ungleichformig erwarmten unkrystallinischen Korpern, Abhandl. koniglichen
Akad. Wiss. Berlin, Zweiter Teil, pp. 1-254, 1841.
7. Borchardt, C. W.: Uber die Elasticitat fester isotroper Korper unter Beruch-
sichtigung der Warme, Monatsber. Akad. Wiss. Berlin, pp. 9-56, 1873.
8. Hopkinson, J.: On the Stresses Caused in an Elastic SoHd by Inequalities of
Temperature, Messenger Math., 8:168-174 (1879).
9. Alibandi, P.: Sulla Elasticite dei solidi complicata da variazoni di temperatura,
Giorn. math., 38:77-91 (1900).
10. Timoshenko, S., and J. N. Goodier: "Theory of Elasticity," 2d ed., McGraw-Hill
Book Company, Inc., New York, 1951.
11. Gatewood, B. E.: Thermal Stresses in Long Cylindrical Bodies, Phil. Mag., ser. 7,
32:282-301 (1941).
12. Boley, Bruno A.: The Determination of Temperature, Stresses, and Deflections
in Two-dimensional Thermoelastic Problems, /. Aeronaut. Sci., 23(l):67-75
(1956).
13. Sokolnikoff, I. S.: "Mathematical Theory of Elasticity," 2d ed., McGraw-Hill
Book Company, Inc., New York, 1956.
14. Love, A. E. H. : "A Treatise on the Mathematical Theory of Elasticity," 4th ed.,
Dover Publications, New York, 1944.
15. Sechler, E. E.: "Elasticity in Engineering," John Wiley & Sons, Inc., New York,
1952.
16. Peery, D. J.: "Aircraft Structures," sec. 7.2, McGraw-Hill Book Company, Inc.,
New York, 1950.
CHAPTER 2
TEMPERATURE EQUATIONS
q =
M (r, - T2) conduction - -M
q = hS(T, — convection
Tf)
q -
aSTthe radiation
AT
q = pcpV -tt- heat storage
21
22 THERMAL STRESSES
'
where q = heat transfer, Btu/hr • -r
= thermal conductivity,
k Btu/(hr)(ft)(°F)
A = cross-section area, ft2
L = length of conduction path, ft
= temperatures of parallel surfaces, °F
h = convective-heat-transfer coefficient, Btu/(hr)(ft2)(°F)
S = surface area, ft2
T. = temperature of surface
T, = temperature of fluid
a — radiation constant, 0.1742(10-8) Btu/(hr)(ft2)(°F4)
Tab, = temperature, °K
P
= density, lb/ft3
= specific heat,
c„ Btu/(lb)(°F)
V —- volume, ft3
AT = change in temperature
At = change in time, hr
In the following sections, the significance of Eq. (2-1) as applied to
actual bodies and fluids will be examined in more detail, with emphasis"on
the heat balance existing in components of aircraft structures.
2-2. Heat Conduction. The conduction relation in Eq. (2-1) can be
written as
dQ = ■
kA^dt (->) (2-2)
V -
where dQ js the heat conducted in the direction x in the time dt. Con
sider the^element of a solid as shown in Fig. A-l. The heat entering
the left face of the element is
F = (dQx)1 =
kdydz^dt
so that the heat change across the element is [Eq. (A-3)]
Similar expressions can be written for (dF)y and {dF), for the y and z
directions. The heat stored in the, element [Eq. (2-1)] is
dT
dQ, = pcp dx dy dz — dt
ot
dQd = qa dx dy dz dt
whence —-
67
= —
pcp
I
\dxl
ft + JT + ay1 dz2 /J
+
pcp
= — v T+
pcp pcp
(2-3)
Solution of Eq. (2-3) for the temperature for various boundary and
initial conditions can be found in the literature (for example, see Refs. 1-8).
Solutions for the aircraft-structures problems considered in later chapters
will be given at the time the thermal stresses are obtained.
Example 2-1. Solve Eq. (2-3) for the steady-state case in one dimension, with no
heat developed.
Solution. Steady state means independent of time so that the left side of Eq. (2-3)
is zero. Thus the equation reduces to
PROBLEMS
a„ = - / fix) sin ax
a Jo a
The heat flow per unit surface area leaving a body per unit time by
thermal radiation is its emissive power W. If two small bodies of area
A i and A2 are placed in a large evacuated enclosure perfectly insulated
and allowed to come to thermal equilibrium, then the bodies will emit
radiations at the rates AiW1 and A2W2, respectively. Let energy
falling on a unit area of any small body in the enclosure be Wb, where Wb
is due to the radiation from the walls of the enclosure. Let the bodies
have absorptivities at and a2, which is the fraction of incident radiation
absorbed. Then the energy balance on the bodies has the form
WB = cTih. (2-4)
Fu = ^
J cos Pi dwi (2-10)
where /3iis the angle between the normal to dAi and the line joining
dAi to dA2 and dwi is the solid angle under which dA2 is seen by dAi
[see, for example, McAdams (Ref. 6) or Eckert (Ref. 8)]. McAdams
(Ref. 6) shows curves for Fi2 for various surfaces.
If the surfaces are not black, it is still more difficult to get an expression
for ^l2 in Eq. (2-9) or for «i2 in Eq. (2-8). For two coaxial spheres or
cylinders whose surfaces radiate diffusely el2 has been calculated as
! i+4i(l_l)/
ei2
=
ei Az \e2
(2-11)
where A i is the surface of the smaller body.For two surfaces whose size
is small compared with their distance apart, the fraction of reflected
radiation which falls back to the radiating surface becomes so small that
it can be neglected. In this case
tU = «l«2 (2-12)
Also, if Ai is small compared with A2, Eq. (2-11) shows that approxi
mately
el2 = ti (2-13)
PROBLEMS
2-6. The inner wall of a thermos bottle is at 212°F and the outer wall at 68°F.
The walls on the sides turned toward each otherare silvered so that t = 0.02. If «u
is given by Eq. (2-7), find the heat loss per unit
area of the inner wall.
26 THERMAL STRESSES
2-7. In Eq. (2-8) take T2 at room temperature 460 + 70 = 530°, and graph q/tu
against Ti. What conclusions can be drawn about the effect of radiation up to
1000°F?
2-8. What approximate value of «i2 between skin and stringer could be used for
the structure shown if its length is 20 in.?
I< 1 Din.
1
t
4 in.
Pbob. 2-8
2-4. Heat Convection. From the convection relation in Eq. (2-1) the
equation for heat transfer between a surface at temperature T, and a
fluid in contact with it at the temperature T/ is
dvy dvy\
(dvy
3vy
Gy + Py+ Fy (2-15)
vx, vy, v,
= velocity components
Use Eq. (A-3) with F = p«x on the left face, F = pvy on the aft face, and
F = pv, on the bottom face, and sum to get total change of flow, which
gives the equation of continuity as
. dvy dv,\ dp dp dp dp _
(dvx
Tz)
+ +,
+,
,
aJ+v'£
+
(2-17)
v"4+v'£
0
dT dJ
for compressible fluids, or
*L-
p dJs =
+
+
(2-18)
0
dx dy dz
dT dT =
±v2
dt dx dy dz pcp pcp
dq =
-k^dA (2-20)
and the thermal conductivity of the fluid close to the surface. Equa
is
k
tion (2-20) has the form of Eq. (2-14) in that the coefficient can be
h
between the layers of the fluid. The heat transfer in turbulence mainly
is
and free of friction away from the wall. This means that the viscous
terms and heat-conduction terms in the above equations are negligible
compared with the remaining terms, except for narrow regions near solid
boundaries, where the velocity and temperature gradients are of necessity
so large as not to be negligible.
The principle of similarity as applied to the above system of equations
has yielded much information on the mechanisms of the convective
processes. This principle can be applied through three procedures:
integral, differential, or dimensional. Using dimensional analysis, an
expression for the heat-convection coefficient h in Eq. (2-14) for the case
of forced convection inside a tube of diameter D can be derived as follows :
Assume
h = CvDbnWpmcpn (2-21)
Htr'L-^T-1 = (r1L)aLb(Mr1L-iy{Htr1L-1T-iy(<ML-3)m(HM-iT-iy
and equate exponents for each unit.
j+n
H:
-1
1 =
= -a-f-j
t:
L: -2 = a + b — f -j — 3m
M: 0 = /
+ m — n
T: -1 = —j — n
a = m b = m — 1
f= n — m j = 1 — n
X"CCf)"(T')" (2-22)
Nu =
where
^ (2-24)
Re = (2-25)
Pr =
(2-26)
^2
is the Prandtl number.
TEMPEBATURE EQUATIONS 29
The few theoretical solutions for example, Jakob (Ref. 1, Chap. 22)
[see,
and Van Driest (Ref. 17)] of the convection equations that have been
obtained can be expressed in terms of the above nondimensional numbers
plus in some cases the Mach number
Ma = - (2-27)
a
T, insulated plate
Thermal
boundary T, hot plate
layer
Vf*
Tj, =
Tf + 2gJcp (2-28)
_ _
Tf. - Tf
*7'aw J*
J- f
. Taw
—
Tf
r (2-29)
vf2/2gJcp
rized as follows:
q = hA(Tw - T„) (2-30)
*k = Nu= fi{Re,Pr,Ma) (2-31)
(2-32)
This formula breaks down for large Mach numbers and when dissociation
occurs in the air (Refs. 9, 13). For incompressible laminar flow the
theoretical local heat transfer for a flat plate is
for Eq. (2-31) and with Taw = Tf in Eq. (2-30). For compressible flow
on a flat plate, Eq. (2-34) still applies provided Taw is used in Eq. (2-30).
However, the "constant" in Eq. (2-34) becomes a function of Mach
number for compressible flow. To avoid changing the "constant,"
Eckert (Refs. 9, 10) proposes the use of a reference temperature Tr to
use for the evaluation of the properties in both Eqs. (2-33) and (2-34).
TEMPERATURE EQUATIONS 31
where y is the ratio of specific heats and r is the recovery factor. Since
it is necessary to have Tr to get y and r, an iteration procedure must be
used on Eq. (2-35) to get Tr. Experimental work on the laminar
boundary layer is reviewed by Kaye in Ref. 13.
For the turbulent boundary layer theoretical work, under assumptions
of constant eddy viscosity and constant eddy conductivity in the turbu
lent portion of the boundary layer, gives
r = Pr* turbulent (2-36)
For theoretical work on heat-transfer coefficients for the turbulent
boundary layer see Van Driest (Ref. 17), Shapiro (Ref. 12), and Kaye
(Ref. 13), where curves are given. Experimental data up to Re = 107
give
=
NuL 0.029(PeL)°-8 Pr* turbulent (2-37)
For Re up to 109 Eckert recommends the Schulz-Grunow formula
0.185 Re Pr* _„
Nul
Ar = (2'38)
(logfle)'-'".
and for large Mach numbers uses the reference temperature Tr in Eqs.
(2-36) to (2-38) to obtain the properties.
Theoretical and experimental formulas and experimental curves for
recovery factors and heat-transfer coefficients on wedges, cones, and
cylinders are given in Refs. 9 and 12 to 14. For airfoils the above formulas
can be used if the Reynolds number Re is based on the local velocity.
However, to obtain this velocity requires the solution of the aerodynamic
problem of velocity and pressure distributions on the airfoil. It is also
necessary to consider shock waves in this solution. Tables for these
calculations are given in Refs. 12, 13, 34, and 35.
Eckert (Ref. 9) discusses the effect of pressure and temperature varia
tions along the plate upon the recovery factor and heat-transfer coefficient.
The effect of pressure variation appears to be minor, but the effect of
temperature variation can be large. Eckert (Ref. 9) also considers heat
transfer at low densities such as occurs at high altitudes.
The coefficients in Eqs. (2-34) and (2-37) are point values so that at a
distance x from the leading edge of a flat plate h in Eqs. (2-30) and (2-31)
can be written [Eq. (2-34)]
If the average value of h over the distance x from the leading edge is
defined as
-ifx Jo
hx dx
Kr = (Rex)
0 8 Pri turbulent (2-40)
0.037^
Usually, however, laminar flow exists over the front portion of the
plate and changes to turbulent flow when the critical Reynolds number
is exceeded. Thus
Kv = - / Qlx)lam dx + (hx)turb dx
x J0 Jxcr J
=
1
/
fx
(hx)turb dx - CXcr
/ {hx)tUrb dx +
C Xer
/ (hx)iam dx
Jo Jo Jo
= 0.036 - Pr»
x
[(Rex)0-*
- (Re„)0
i +
18.5(/2ecr)0-5] (2-41)
The difficulty with Eq. (2-41) is the value of Re„. Where does the
transition from laminar to turbulent flow begin? "This point is the
weakest link in the whole procedure to predict wall temperature and
heat transfer coefficients" (Eckert, page 65 in Ref. 9). See Refs. 9,
18 to 26, and 31 for information on the problem. Gazley's paper (Ref. 21)
is a review of available information and indicates effects of various factors
on boundary-layer transition. Stability theory gives the minimum criti
cal Re for incompressible flow over a flat plate as about 60,000, while
transition experiments indicate a transition range of 3 to 4 million for
incompressible flow over a smooth flat plate with negligible free-stream
turbulence. Negative pressure gradients, decrease in surface tempera
ture, and increase in Mach number tend to raise these numbers. Free-
stream turbulence, concave surface, positive pressure gradient, surface
roughness, and increases in surface temperature tend to lower these
numbers.
For latest information on the problems of aerodynamic heating, see
such journals as the Journal of the Aeronautical Sciences and Aeronautical
Engineering Review.
Example 2-2. What are the adiabatic surface temperatures on an insulated flat
plate over which air is flowing parallel to the surface at 500 mph, 1 in. and 1 ft behind
TEMPERATURE EQUATIONS 33
the leading edge of the plate? The free-stream air temperature is 60°F, and the
pressure is 14.7 psia (sea level).
Solution. To determine the type of flow, calculate the Reynolds number, at each
point. From Eq. (2-25)
Re = V-^
A"
so that the flow is probably laminar at 1 in. but turbulent at 1 ft. From Table B-2
the Prandtl number at 60 + 460 = 520°R is Pr = 0.71 so that [Eqs. (2-33), (2-36)]
c„
= 0.24 (Table B-2)
~j~
that there are small changes in Re, and Taw. However, the change in Taw
is
so
r,
less than 0.5° so that the free-stream temperature can be used for the properties
provided the temperature gradient small.
is
Example 2-3. Determine the rate of heat transfer from a plate ft long and ft
5
wide air flowing parallel to its length under the conditions of 1,000 mph velocity at
is
if
temperature 60°F at sea level with a plate temperature of 250°F. Neglect any shock
waves.
Solution. Assume a reference temperature of 200°F, so that properties can be
evaluated at 460 + 200 = 660°K. Then
= 0.002378(520/660) = 0.00187
p
c„ = 0.24 = 4.5(10-') Pr -
0.69 (Table B-2)
/i
Re
(1,470) (5) (0.00187) _ 3-0(1°
>
4.5(10-')
34 THERMAL STRESSES
which is probably turbulent flow. If the flow were all turbulent, then
i„«, - ou -f
0.88(1,470)' _
~ ^W *
^^
(2) (32.2) (778) (0.24)
(3,600) (32.2)
=
fc =
3,60^ (465)(10-^)(0.24)
(0.037) (1 86) (10-')
^ o
[3-0(107)]o.80.69* [Eq. (2-40)]
= 104
q = KVA(Tw
-
Btu/(hr)(ft2)(°F)
Taw) = (104) (10) (250
- 217) = 34,400 Btu/hr
From Eq. (2-35)
which is close enough to the assumed 200°F reference temperature. If the flow were
all laminar, then
T = Pri = 0.83 Tay, = 208°F
K, =
(0-66)(1f6)(1°-2) [3.0(10')]°«0.685. [Eq. (2-39)]
= 11.9 Btu/(hr)(ft2)(°F)
q = 4,000 Btu/hr
which indicates the large difference between laminar and turbulent heat transfer.
If it is assumed that transition occurs at Beer = 4,000,000, then the distance along
the plate for transition is
, - XXR*«
Xcr - 50-4(107)
5 "
°Q67ft b7
Rex 3.0(10')
Let raT =
^
[x„Pri + (L - x„) Pri] i[(0.67)(0.83) = + (4.33)(0.88)] = 0.87
PROBLEMS
_ (Pay™
,(f)+1
/^V _
*.®r
/ Tg\
W "
Ta 4.94
~ +
Tf \Pf) Mf> V T,)
where P = pressure
subscript a = a point on the body
subscript /
= free stream
AP/q = pressure ratio at body point
and q = p/t//2/2
TEMPERATURE EQUATIONS 35
For given AP/q from the airfoil pressure distribution without shock waves the equa
tions give local pressure, density, temperature, and velocity at the edge of the bound
ary layer. If AP/q = —0.95 at a point on the upper surface 3 ft behind the leading
edge of an airfoil, M/ = 0.90, altitude = 30,000 ft, and the temperature of the skin
is200°F, find the local unit heat-transfer rate.
2-14. Solve Prob. 2-13 for the equivalent point on the lower surface where
AP/q = 0.1.
a
b
A, = skin thickness
Ai,=stringer area
another skin opposite and heat loss to the air inside neglected,
is
if
36 THERMAL STRESSES
then the heat balances for the skin and stringer can be written as follows
[see Eqs. (2-1), (2-8), and (2-30)]:
P.(cp).h,ba
^ = hba(Ta, - T.) - el2<r&a(7V
- 7V) - e.&a(<r7V
- G)
- 8k,h, ^
(T. - Tb) (2-42)
Pb(cp)bAba
^ = wSba(T,*
- TV) + SkX g (T, - Tb)
h. = skin thickness, ft
b, a = dimensions of skin, ft
t = time, hr
Sb
= surface area, ft2, of a unit length of stringer viewed by skin in
radiation transfer
T„ skin temperature, °F [Eq. (2-29)]
= absolute adiabatic
T absolute temperature, °F
Theoretically, Eq. (2-42) can be solved for the temperatures T, and Th.
However, from Sec. 2-5, h is an involved function of temperature T,
and time so that there is no direct solution for the equations. One
procedure is to use a step-by-step solution or iteration solution for small
increments of time. Necessarily the solution can be made only for a
particular problem with numerical values. For example, Tendeland
and Schlaff (Ref. 27) used this procedure on Eq. (2-42) with radiation
omitted; Johnson (Ref. 14) used it but with the stringer omitted so
that Eq. (2-42) reduced to
g -/<*,»)
= + f(x,y0) dx
2/n 2/o /
Jxo
= + [X,f(x,yn) dx
2/i2 2/o
Jxo
= + dx
2/m 2/o /
Jxo
2/l,n+l = =
2/ln 2/l
Repeat the operation to get (£2,2/2), etc The integrations can be per
formed by rectangle rule, trapezoid rule, or Simpson's rule; the more
accurate the integration, the fewer the steps in each iteration.
A method such as the Picard method, although accurate, requires
considerable calculation at each step. It might be used in regions where
y is changing rapidly, but a shorter method can be used elsewhere. If
the intervals are sufficiently small, then
= -
- xi),
?/i 2/0 +f(xo,yo)(x1 x0)
= +/(a;i,2/i)(a;2 etc.
2/2 2/l
is shorter and gives sufficiently accurate results for many problems. For
Eq. (2-43) this procedure gives
r.i - - -
.A\
where T.i the skin temperature at the end of the time interval At, T,o
is
the temperature at the beginning of the time interval, and the proper
is
show the large effects of the emissivity coefficient at high altitude and
38 THERMAL STRESSES
800
1
600
400
6
200 *io
pcpt h=0. 0986 Btu/1 t2'R
800
j-pc 0.197 Btu/
f
><
— e = D.l
600 —<
6
400 1
— jr 0
200 /
//
/■ \ -pcp) is=0 986 Eitu/ft -€=C .1, 0. 5, 1.0
*5 r
50' 100 150 200 250 300 350 400
Time since firing, sec
Fig. 2-36. Average skin temperature of a 1-ft 32.5° cone for transient trajectory.
of the skin thickness or skin heat capacity throughout the entire flight.
These are calculated curves given in Ref. 14 for the trajectory of A-4 at
White Sands, N. Mex., Oct. 10, 1946 (NRL R-3030) ; the cone is assumed
to have infinite conductivity and to be insulated from the rest of the
body.
TEMPERATUBE EQUATIONS 39
Example 2-4. A point 1 ft behind the leading edge of a flat steel plate 0.10 in.
thick, which is traveling at 3,000 ft /sec at 50,000 ft, has a temperature of 700°R.
If the plate reaches 80,000 ft 10 sec later, what is the temperature of the point at
that time? Neglect shock waves, and assume no heat loss on the inside surface of
the plate.
Solution. In Eq. (2-44) p. = 484 lb/ft3 (Table B-3), (c„). = 0.16 Btu/(lb)(°F),
= 0.0083 ft, a = 0.174(10-8) assume e, = 0.5; assume = 20 Btu/(hr)(ft2). Use
ft,
G
;
time interval of sec in the calculations, and evaluate density at 60,000 ft. Assume
5
a
67
= 0.0002240 = 0.000125 (Table B-l)
p
46°700
2~7)
cp = 0.24 = = 4.7(10-') Pr = 0.68 (Table B-2)
M
15132
Re = = 800,000
(3'°00ig()1(0°l7125)
= Pr* = 0.88
r
4b0 - 67 +
- 46n -_
0-88(3,000)2
la. 1050 K
1050°R
T
= 26.5
Btu/(hr)(ft2)(°F)
T 1l _ 7nft
5(26.5(1,050 -
700)
-
0.5[0.174(7)«
- 20] "_ K
,
393) 393)
For the second step assume Tr = 700° so that c„ n, Pr, and Ta, remain the
k,
would give 739°R based on density at 60,000 ft. Thus the accuracy of the result
depends on the choice of density to be used as well as the number of increments
in
time.
Example 2-5. If the plate in Example 2-4 maintains a speed of 3,000 ft/sec at
a
constant altitude of 50,000 ft, what the equilibrium temperature of the plate?
is
Assume T„ and solve by trial and error for T.. Assume T, — 1000°R, and take the
reference temperature at 1000°R. Then for turbulent flow p = 0.0003622(393/1,000)
= 0.000142, cp = 0.25, n = 6.0(10-'), Pr = 0.66, Re = 700,000, r = 0.87, Ta. = 1020°,
k = 0.0262, h = 31.1, so that
and Tr = 843°. This checks close enough so that the equilibrium temperature is
1010°R.
PROBLEMS
2-15. Continue the solution of Example 2-4 for 10 sec more, assuming the velocity
of 3,000 ft/sec is maintained straight up.
2-16. Find the equilibrium temperature of a plate moving at Mach number 3.0 at
30,000 ft. Take the plate as 10 ft long, and get the average temperature. Assume
(a) 6S = 0.1, (6) e, = 1.0.
2-17. Solve Example 2-4 if the plate is aluminum alloy instead of steel.
2-18. Solve Example 2-4 if the plate maintains its velocity at a constant altitude
of 50,000 ft.
REFERENCES
1. Jakob, Max.: "Heat Transfer," vols. I, II, John Wiley & Sons, Inc., New York,
1949.
2. Carslaw, H. S., and J. C. Jaeger: "Conduction of Heat in Solids," Oxford Uni
versity Press, New York, 1947.
3. American Society of Mechanical Engineers and Institution of Mechanical Engi
neers (London): "Proceedings of the General Discussion of Heat Transfer,"
Institution of Mechanical Engineers, London, 1951.
4. Dunsinberre, G. M.: "Numerical Analysis of Heat Flow," McGraw-Hill Book
Company, Inc., New York, 1949.
5. Boelter, L. M. K., V. H. Cherry, H. A. Johnson, and R. C. Martinelli: "Heat
Transfer Notes," University of California Press, Berkeley, Calif., 1946.
6. McAdams, W. H. : "Heat Transmission," 3d ed., McGraw-Hill Book Company,
Inc., New York, 1954.
7. Brown, A. I., and S. M. Marco: "Introduction to Heat Transfer," 2d ed., McGraw-
Hill Book Company, Inc., New York, 1951.
8. Eckert, E. R. G.: "Introduction to the Transfer of Heat and Mass," McGraw-Hill
Book Company, Inc., New York, 1950.
9. Eckert, E. R. G.: Survey on Heat Transfer at High Speeds, WADC TR 54-70,
Wright-Patterson Air Force Base, Ohio, 1954.
10. Eckert, E. R. G.: Engineering Relations for Friction and Heat Transfer to
Surfaces in High Velocity Flow, J. Aeronaut. Sci., 22(8):585-587 (1955).
11. Goldstein, S. (ed.): "Modern Developments in Fluid Dynamics," vols. I, II,
Oxford University Press, New York, 1938.
TEMPERATURE EQUATIONS 41
33. Woolf, J. R., and L. R. Scott: Influence of Through Metal on Heat Transfer
through Aircraft Structural Sandwich Panels, ASME Paper 56-SA-23, 1956.
34. Keenan, J. H., and Joseph Kaye: "Gas Tables," John Wiley & Sons, Inc., New
York, 1948.
35. Ames Laboratory Staff: Equations, Tables, and Charts for Compressible Flow,
NACA Rept. 1135, 1953.
36. Korkegi, R. H.: Transition Studies and Skin Friction Measurements on an
Insulated Flat Plate at a Mach number of 5.8, J. Aeronaut. Sci., 23(2):97-107
(1956).
37. Sibulkin, Merwin: Heat Transfer to an Incompressible Turbulent Boundary
Layer and Estimation of Heat Transfer Coefficients at Supersonic Nozzle Throats,
J. Aeronaut. Sci., 23(2) :162-172 (1956).
38. Snodgrass, R. B.: Rocket Research Report No. XX, Flight Measurements of
Aerodynamic Heating and Boundary Layer Transition in the Viking 10 Nose
Cone, Naval Research Lab. Rept. 4531, June, 1955 (ASTIA AD No. 66606).
39. Goodman, S. : Radiant Heat Transfer between Infinite Parallel Plates, Natl. Bur.
Standards Rept. 4239, August, 1955.
40. Weiner, J. H. : A Method for the Approximate Solution of the Heat Equation,
WADC TR 54-427, Wright-Patterson Air Force Base, Ohio, 1955.
CHAPTER 3
Ph(cP)bAb
^ =
e2i<r&(7Y
- TV) + 8fc'W - r») (3-2)
Even with T, known, Eq. (3-2) cannot be solved by simple means because
of the radiation term with Tb4. As a first approximation neglect radia
tion effects (the examples in Sec. 2-7 show a small effect of the radiation
terms under the temperature range considered), and write Eq. (3-2) as
From Figs. 2-3a and 2-36 and from similar skin-temperature curves in
Refs. 1 and 2 it appears that a simple assumption for T, is
T. = Tm±
l
0<t<h
(3-4)
T. = Tm t>h
where h may be related to time to reach peak velocity, time to reach a
certain altitude, or equivalent time to reach equilibrium temperature.
Put this T, in Eq. (3-3), use the initial condition Tb — 0 at t = 0, and
integrate to get
Jl-
pb —
JJt
7V
= e—-
1
± e-B«« t > k (3-5)
im D
8A:,fo,<i
B — B lt —
bAbpb(cp)b
)
t_
h B 0<t<h' ,„„,
otbTm \ab
{a \ e»-l (3-6)
abTm
For a, = «(,, which is the usual with skin and stringer being the
case
same material, the maximum value of this expression, and hence the
maximum stress in Eq. (3-1), occurs at =
<i,
(T.
- Tb\ - <r*
_ (X)_ .„
1
= =
0 1 2 34 B
5 6 7 8
Figure 3-1 shows graphs of Eqs. (3-4) to (3-6) for two selected values
of the parameter B. Figure 3-2 shows the maximum thermal stress
against B [Eq. (3-7)].
The basic parameter in these curves is the nondimensional parameter
B [Eq. (3-5)], involving material properties, geometry, and the equivalent
46 THERMAL STRESSES
R _ 8^75/(60)
a (12)] _
~
6<-ybA„
- Ai
hji _
y 0.10(1) _
~ 1U0
M6(0.10)(0.22) 12(0.30)
o < t < h
(3-9)
T, = Tm t>h
whence for a, = ab
rp
' rp I
"
=
o
- 2B + - e-Bt"> >
\6-l\J)
|-3
2)
2]
[eB(B2
h
m t> t
J-
The maximum value of this expression, and hence the maximum stress,
occurs at = h. For the same value of the parameter the maximum
B
t
value given by Eq. (3-10) larger than that in Fig. 3-2. This due
is
to the fact that less heat has been supplied in the same time under the is
h
Tb.
mately equal to the area under the T, curve. The area under the T,
curve up to = in Eq. (3-9) only one-half that in Eq. (3-4). This
is
<i
<
suggests for small the skin temperature of Eq. (3-9), as well as other
B
to Eq. (3-4) but with different based on the area under the straight
h
a
line. Using one-half the in Eq. (3-9) to get to use on Fig. 3-2 gives
B
<i
result which checks very closely the maximum value of Eq. (3-10).
a
The comparison for large not too good, but the stresses are less
is
B
important for large so that less accuracy required. Thus Fig. 3-2
B
is
PROBLEMS
3-1. If
Tm = 450°F above room temperature, what are the maximum thermal
stresses in the skin and stringer of Example 3-1?
3-2. Solve Example 3-1 for steel instead of aluminum. Calculate the thermal
stresses for Tm = 450°, and compare results with those of Prob. 3-1.
3-3. Draw curves similar to those in Fig. 3-1 for the parameter B = 3.
3-4. Take the maximum value in Eq. (3-10), and draw a curve for the maximum
stress against B similar to Fig. 3-2. Take one-half the abscissas to get a new curve
to compare with Fig. 3-2 (this is equivalent to taking h equal to one-half the
original h), and hence check the area approximation discussed in the text.
3-5. Integrate Eq. (3-8), and derive an equation similar to Eq. (3-10) if T, varies
as a parabola up to t = t\.
3-6. In Eq. (3-6) show that for a, > otb the maximum stress occurs at t = h.
3-7. In Eq. (3-6) show that for a, < ai there is a stress reversal at a certain time
depending upon a,/ab and Bi = B/h. Show that the maximum stresses will occur
at equilibrium (t large) and either at t =- <i or at t < h.
3-8. Use the results in Prob. 3-7 to find the time at which the maximum stresses
and the zero stresses occur if B = 1, the skin is steel, and the stringer is aluminum.
where C is the value of -{f.)m,JHTm in Eq. (3-7) and Fig. 3-2 for con
duction alone and R is the correcting term for radiation,
R =
DT 3
K^4
- 4P3 + 12P2 - 24P + 24)
- e~B(Q*
- +4Q3
- 24Q +
12Q2 24)
- Q4(l - e~B)] (3-15)
D = D1hT0' B = BA Q =
^ P =
B(l+ji)
Example 3-2. Find the value of R in Eq. (3-15) for a steel stringer with Ab =
0.20 in.2, Sb = 3.0 in.2, el2 = 0.5, T0 = 500°R, skin temperature = 1040°F, h = 1.0
min, and parameter B = 1.0.
Solution. The value of Tm from Eq. (3-12) is 1050° + 460° 500° = 1000° so -
that Tm/To = 2.0. With this value and the value B -
1, Eq. (3-15) gives R = 9.3Z».
Now
Un-" nfT>-
ll 0
w&hTp' _ (0.50)(1.74)(0.5)'(8)(1) ft ftfM2
Abpb(cp)b (60)(144)(0.20)(0.28)(0.16)
Hence R = 0.04. In Eq. (3-14) the value of C for B = 1 is C = 0.63 from Fig. 3-2.
Thus the radiation effect is small in comparison with the conduction for this case.
PROBLEMS
»-»! £[('+£)'-'] 4^
TRANSIENT TEMPERATURES AND STRESSES 49
3-4. Effect of Stringer upon Skin Temperature. In Secs. 3-2 and 3-3
for the case of assumed skin temperature the effect of the conduction
to the stringer upon the skin temperature has been neglected. The
tests of Ref. 3 to be nonuniform, decreasing
show the skin temperature
at the stringers and depending upon the relative size of stringer and skin.
The purpose of investigating the effect of conduction on the skin tempera
ture is not to try to determine the nonuniform distribution but to deter
is,
mine a corrected average skin temperature. That suppose Eq. (2-42)
solved for the skin temperature without using the conduction terms, but
is
using the convection and radiation terms. This gives Eq. (2-43), the
solution of which has been discussed in Sec. 2-7. Omitting the conduc
tion terms avoids the geometry of the internal structure, which would
require a complete recalculation for each change in geometry. Thus
it
appears desirable to take the skin temperature obtained from Eq. (2-43)
and modify to allow for conduction to the internal structure, at the
it
is
That let To« be placed in all the convection and radiation terms of
is,
Eq. (2-42) but not in the conduction term. Then solve the Eq. (2-42)
for T, and Tb. For the case of no radiation as considered in Sec. 3-2
Eq. (2-42) reduces to
'ft
°K
,
dt dt Whep,(cp),
dTb =
Bi{T. - Tb) Br = 8kX (3-16)
s
dt
'
bAf,pb(cp)b
T,-Tb = e-<-B'+B* dt + d
(^jf) e<B'+B'» (3-18)
J
ti
t
To, = (3-19)
^
Tom
ti
t
50 THERMAL STRESSES
Then
T. - 0 < t < h
T. -
(3-20)
t > h
Tom
B. + B,
JL /£ _ - Tb\
0 < < < «l
Tom Ri VI ^Om / (3-21)
n > ti
\
t
Tom f?l Tom /
T,-Tb Tb
(3-22)
Tom Tom
Note that Eq. (3-20) has the same form as Eq. (3-6) with a, = o%, except
RiB replaces B. Thus, in Fig. 3-1, the curves apply for T, — Tb in
Eq. (3-20) if B is replaced by RiB, for Tb in Eq. (3-21) if multiplied by
l/Ri, and there will be a new curve for T, given by Eq. (3-22). The
maximum stress can be obtained from Eq. (3-7) if B is replaced by RiB.
Since the skin temperature T, will deviate more and more from the
assumed temperature T0, as time goes on, the above results will be
incorrect beyond t = h for large values of Ri. However, the maximum
— good up to that point.
B it ti,
on the skin. With < RiB the stress will be larger for (Fig. 3-2).
PROBLEMS B
to 4.
0
3-15. Solve Prob. 3-1 by using the factor Ri denned in this section.
3-16. By using Eqs. (3-21) and (3-22) derive an expression equivalent to Eq. (3-6)
which involves Ri and allows the skin and stringer to be different materials.
where hj is the joint thermal conductance in Btu per hour per square
foot per degree Fahrenheit and d is the effective width of contact between
skin and stringer. The equivalent heat flow through two resistances in
series is
greater
hj
than about 50, the geometry of the joint may be more important in
determining Rj than the conductance coefficient For aluminum alloy
hj.
8k.h, 7,200A„
= 90
bd bd
(3-26)
for = 0.10 in., = 8.0 in., = 1.0 in. For the tests of Ref. = 26.0
h,
3,
d
b
to 46.
is
in.,
5
d
The primary result that the effect may be small for some joints and
52 THERMAL STRESSES
large for others. A judicious use of the test data of Refs. 4 and 5, and
of any other available data, in Eq. (3-25) will give an idea of the magni
tude of the effect.
Thus, of the three primary items omitted in the discussion of Sec. 3-2,
it appears that the radiation effect is small (Sec. 3-3), that the stringer
conduction effect on the skin can be taken care of by using a factor Ri
(Sec. 3-4) which is greater than 1, and that the thermal joint resistance
hj
Fig. 3-3. Thermal resistance of joints.
PROBLEMS
3-17. Derive Eq. (3-26) for steel instead of aluminum alloy, and calculate Rj for
hj = 50, 100, 800, 1,200.
3-18. Repeat Prob. 3-17 for h, = 0.035 in., b = 4.0 in., d = 0.50 in.
3-19. Draw curves of Rj against hj for several values of hjbd, for aluminum alloy.
Would practical aircraft or missile structures give values of 8k,h,/bd as high as 1,000?
where the radiation surface is 3.0 in.2 for both direct radiation and skin
to stiffener radiation and ei2 = 0.37 (assumed). For the tests take
To = 520° K, Tm = 520° for directr adiation, Tm = T, for skin-stiffener
radiation.
A check of the skin-stiffener radiation [Eq. (3-15)] shows that it
amounts to about 0.01 T, or about 2 to 3°. Hence it will be neglected.
Table 3-1 shows calculations for the difference between skin and stiffener
temperatures using Eq. (3-14) as compared with the test difference. In
Eq. (3-14) Tm — T, is taken as the reading of temperature gage 6 located
6 in. from the stringer, which is regarded as the average skin temperature.
'-"SKi+fe)4-']1^-^1-^ »*>
Equation (3-27) neglects any change in the stiffener temperature and will
give values on the high side for cases in which the stiffener temperature
is large. Also the stiffener radiates to the unheated oven walls in the
tests. For Tb = 200° the net radiation effect on the stiffener is approxi
mately zero. In Table 3-1 R in Eq. (3-23) is reduced as Tb increases
with R = 0 at Tb = 200°. The test temperature for the stiffener is
taken from temperature gage 11 in Ref. 3. Comparison of the calculated
temperature difference and the test difference in Table 3-1 shows reason
able agreement.
54 THERMAL STRESSES
Spec. h„ Ab, h,
B C R
T. T„ T, - T„ T, - Tb
No. in. in.2 min test test cal. test
^ = Bh(Ta.
- T,) - B,{T, - Tb)
b> = 7,21
8k,h, R _- 8k,h,
. r. \
b2h,p,(Cp), bAbpb(cp)b
Tb)
+ (Bh + B, + Br) jt (T.
- Tb)
dKT'd-
+ BtBi{T. - Tb) = 0 (3-29)
^ (T.
at
- Tb)
= BhTa. at t = 0
(3-30)
J a, Pi P2
Ri
r> = ~
B, + Bi „
fl2 =
Bfc
BT- Fi
— Tb occurs at
The maximum value of T,
h = — p2
In ^ (3-33)
pi p2
Ri + - l(Ri + - 4flJ*
v p2
pi Ri +
R2
R2 + [(Bi +
R2)2
ii!2)2
- 4K2]i
(3-34)
M = p2h = In AT = hB^NR')*
Put Eq. (3-35) into Eq. (3-28), and integrate to get Tb.
Trp-
Tj-
T. Tb T\
rp (3-37)
I
a, a,
■*■ a,
is
Tb
h
(/.).
(T't a, /I max
= (XR2W""1-1" = (3-38)
J-
\
1.0
</Ta.
0.8
b/Ta,
0.6
0.4
//
/
0.2
','
// /
/
//
=5/3,
Figure 3-4 shows graphs of Eq. (3-35) to (3-37) for selected values of
the parameters Ri and R2. Note that N the same for both cases
is
thermal stress from Eqs. (3-38) and (3-1) plotted against i?2* with
fb
typical element of the aluminum-alloy wing structure ft behind the leading edge
A
as shown in Fig. 2-2 with = 0.25 in., = 15 in., Ab = 2.0 in.2 If the datum
h,
is
t
TRANSIENT TEMPERATURES AND STRESSES 57
1.0
= 1.0
0.8
1 2
0.6
1.5
2.0
0.4
0.2
0 1234567°° 1*2)
l/2
Fig. 3-5. Maximum thermal stress for constant heat source.
stresses in the structure and the time at which they occur. Assume turbulent flow,
neglect shock waves, and assume the element is a flat plate.
Solution. For T = 460 + 80 = 540°R the Prandtl number in Table B-2 is 0.70
so that the recovery factor is
r = Pri = 0.89
and from Eq. (2-29)
rv/'
Tf + 2gJcp
T" ~ ~67 +
0.89(3,000)' _ 600 F
(64.4) (778) (0.24)
p2 = 0.000101 at TR = 413°
58 THERMAL STRESSES
From Table B-2 (c,), = 0.24, (c,), = 0.245, Mi - 4.2(10-'), M2 = 5.4(10-'), Pri = 0.70,
Pr2 = 0.68. From Eq. (2-25)
Rei = 4,150,000
(3'°00j(24ff°°145)
•ffe2 = 2,240,000
fc,
rT
From Eqs. (2-37) and (2-24)
ht = = 21.3 Btu/(hr)(ft»)(°F)
0.029|#e°-8Prl
=
h2
17.3 Btu/(hr)(fts)(°F)
From Eqs. (3-32) and (3-28), assuming the joint resistance factor R,- = 1.0,
is
Ri - rr = !-53
+
1
on,
Wis _
=
W " (19.3)(15)(2.0) _
2
(12)(8)(75)(0.25)2
N = 026 =
g^F^T) = 25 min
h
PROBLEMS
3-20. Solve Example 3-3 = 0.072 in., = 4.0 in., and Ab = 0.30 in.s
h,
if
3-27. Solve Example 3-3 the thermal conductance of the joint between the skin
if
and stringer = 800 Btu/(hr)(ft2)(°F) and the contact width = 1.5 in.
hj
is
cf
is
a
TRANSIENT TEMPERATURES AND STRESSES 59
T« = T' +
SI
where for constant acceleration If TM is taken above datum, if
v = at.
Ta, = Te I
—
) vm = ato
where vm
— cruise velocity
<o
= time to reach cruise altitude and cruise velocity
T0 = datum temperature
The value of h during boost is an involved function of time due to
velocity changing with time, density changing with altitude and tem
perature, viscosity, Prandtl number, and specific heat changing with
temperature. If a mean value of h is assumed, then, neglecting radia
tion, Eq. (3-28) can be used for both the boost phase and the cruise
phase, provided the value of To, in Eq. (3-40) is used during boost and
Ta, = Te is used during cruise. This assumption of a mean h but
variable TM allows the heat-balance equations to be solved without using
a numerical procedure. Results for skin temperatures using a mean h
compare very well with those obtained by numerical procedures using
the variable h.
The problem thus is reduced to solving Eq. (3-28) for T, and 2\ for
TM and h given as
Ta' = Te h = huO<t<to
\to) (3-41)
Ta, — Te h — h2, t > to
~2{T'aV
~ + {Bh + B' + Bl>
- + BhB^T' -
Jt {T' Tb) T»)
where Bh = Bhi for 0 < < Bh = Bh2 for The initial condi
f0,
t > to.
t
tions are
-
T. Tb = or =
0
0
f
t
(T. - Tb) = for = (3-43)
0
jt
t
=
and at
- Tb)i - r»),
to
f (T. = (r.
dt
{T' - Tb)i - dt
{T>
- Tb)2
(3-44)
where subscript refers to the solution for < < and subscript
t0
2
1
t
for > io-
t
p—pit
where pi and p2 are given by Eq. (3-32) with K2i = Bhi/B\ and p3 and p4
are given by the same equations but with R22 = Bhi/Bi. From the
initial conditions (3-43)
2p2 fli _ B\
/
Ul =
(p2 -Pl)(Bi«.)'V B« Pi/
(3^6)
C2=
(W2(1+£)-Cl
From Eqs. (3-44) and (3-45)
-
(3-47)
_ - rm
(I??
eP*'°
= 1^22
p,]
+
+
Q
[b.r,
4
i)
p4 p3 \R K.^o))
If
<0,
<i
to.
fi
= —
—-
<i
(3-48)
'
P4 P3 P3C3/
\
TRANSIENT TEMPERATURES AND STRESSES 61
t0,
<i
t0, which will indicate the maximum to be at = t0.
<
h
h
is
and the skin temperature
is
t, - rb n
(3-51)
T
T
E E
and (3-51) against t/U for selected values of the parameters R2i, R22,
2?
l,
and Bito. These curves are similar in shape to the test curves given in
Ref.
6.
62 THERMAL STRESSES
Example 3-4. Solve Example 3-3 if the constant acceleration of the airplane is
such that it reaches the given altitude and velocity from take-off in 1.5 min.
Solution. From the calculations for h in Example 3-3, take h2 in Eq. (3-41) as
h2 = 20 Btu/(hr)(ft,)(°F). For hi in Eq. (3-41), assume an average value between
0 and h2 or hi = 10 Btu/(hr)(ft')(°F). From the other values in the example there
result
Ri = 1.53 Rn = 1.34 = 2Rn B,<0 = 0.48
From Eq. (3-32)
From Eq. (3-47) C3 = -2.56, C4 = 1.24, and from Eq. (3-48) h/U = 2.54, or
h = 3.81 min. From Eq. (3-49)
- ^§f-
5|?
(/.W = (2.56) (0.123)
>•«<
(/»W =
+Y§ (10,600) = 19,800 psi
PROBLEMS
Rj = 0.80.
is
and its cruise velocity of 3,500 /sec in min. typical element of the
ft
ft
70,000
2
is
in., = 6.0 in., Ab = 0.20 in.s If the datum temperature 80°F at = draw
0,
is
<
6
curves for the skin temperature T„ the stringer temperature 7», and the difference
in temperatures T, — Tb for the first min of flight. Assume turbulent flow, neglect
8
3-33. Find the maximum thermal stresses in the element of Prob. 3-32 and the
time at which they occur. Assume Rj = 0.90 for contact resistance between skin
and stringer.
is,
is is
Eq. (1-22) with the last term omitted. If the structure unrestrained
and of one material, then Eq. (1-22) applies. To obtain the stresses
for this case, replace the idealized structure
Fig. 2-2 by the structure of Fig. 3-7, in y-Skin
of
Area-A.-^
which the skin of area
1
at the temper-
A
is
,
ature T, and the web and other flange A2 tw—
-|
are at temperature Tb.
From Eq. (1-22) the stress at any point Area -Aw -
on the cross section of the form
is
=
E(a + mx - aT) (3-52)
f
'
0
where a and m are constants. Consider
V^-Area=A2
the skin to be one material and the web FlG 3.7 structural element for
and bottom flange to be different material, bending.
a
equations
ffdA =0 SfxdA =
(3-53)
0
to evaluate the constants a and m, and hence obtain the stress on the
cross section. Substitute Eq. (3-52), with E, and a, for the skin and
Eb and ab for the web and lower flange, into Eq. (3-53), and integrate
to get
Aa Bbwm = E Ba =
+ +
B +
Dbwm
= EbA2 E,A. + EbAw = E,A, \EbAv
E A
(3-54)
D = E3A, = E,A,a,T, + ^A^Tt,
+
$EbAw
AD -B2 (3-55)
a
AD B>
stress
- a.T.) -h(^T.-
Tb)
= E,(a + =
/.
mbw
E,ab[l + (iEbA2/EbAw)] ^
=
rj
+
1
values for the constant H are needed for the no-bowing and the unre
strained cases.
PROBLEMS
3-34. Solve Prob. 3-1 for the unrestrained case, assuming no bottom flange {A2 = 0).
Where does the maximum stress occur in the web? Compare the stresses with the
results in Prob. 3-1.
3-35. If 6w is relatively large compared with skin thickness and if the web does
not buckle, does it make any difference in the stresses what values 6«, and <w have
as long as Aw - bjtw is fixed?
3-36. An aluminum-alloy channel section at 400° above datum is attached to a
steel skin at 700° above datum. Area of the skin is 0.40 in.2, area of channel web
is 0.50 in.s, and area of bottom flange is 0.60 in.2 Find the maximum stresses in the
cross section.
3-37. Take A2 = 0 in Eq. (3-56), and write an expression for the ratio between H
in Eq. (3-1) and in Eq. (3-56). Graph this expression against the area ratio to
obtain an idea of the difference between the restrained-in-bending and unrestrained
cases.
3-38. Show that for one material Eq. (3-1) is the same as Eq. (1-22) with the last
term omitted.
REFERENCES
4-1. Introduction. In Chap. 3 the case of the skin and stringer, each
at a different uniform temperature, was considered. The results obtained
there will apply to structures with
Heated surface
relatively thin skin and small stringers.
If the skin is relatively thick or
]_L
the stringer is deep, or the stringer T
becomes a full-depth web or spar, then Thick plate
(1
- v)fy _ (1
- v)f,
= _ T_
+ Hc +
aETE ocETe Te
Hc = T{X) AX (4-2)
ZTe fo
where Te is the equilibrium temperature. For the beam Eq. (1-22) can
be written as
fy _ T
+ Hc +
aETE Te
Hc = T{x)b{x) dx (4-3)
fo
Ih =
[ T{x)(x
- c)b(x) dx
T = TE on x = L, t > h
so that FiF't =
W^i'F2 Wi = —
pcp
XL = 0 XJ, = (2n = pn
cos + 1)
| (4-6)
oo
T = Te
i+ X an exp
(- cos
^ - ?l(a;) 0 - - fi
f
n=0
r ^ - (4-7)
(-
= > t,
+
^
&« exp
&|p) cos t
IF = ^ L2
= J*i-
pCpL2
If Eq. (4-7) is put into the differential equation in Eq. (4-1), then
-fT*l (x)
= - ~h
91
= - 2WP + AiX + A2
g'2'(x) =0 02
= A3x + At
Now the condition = 0 on x = 0 requires .4i = A3 = 0, and the
dT/dx
condition T = 0 at t — 0 requires
eo
£ancos^
n=0
= ffl =
^(l-g) (4-8)
a» =
- dx -
J-L \l
C0S (4-9)
2WL U) 17
68 THERMAL STRESSES
At t = h
00
TE+^an
n=0
exp (-PnW) cos
X - 21H1 - f)
= r£ + exp (-pSW) cos
^
- A<
X6"
whence by using Eq. (4-8)
b„ = o"[l - exp (pn'T7)] 44 = 0
(4-10)
Thus, Eq. (4-7) becomes
rB
T 217 L
71=0
(4-11)
T
— 2
= 1 4- — cos t > h
n=0
- exp (p„2lF)
fin = 1 17 =
^ p« = (2n + 1)
|
Figures 4-2 to 4-4 show T/TE in Eq. (4-11) plotted against different
variables with several parameters. Figure 4-2 has x/L as parameter
and t/h as abscissa with fixed W; Fig. 4-3 has x/L as parameter with
W as abscissa and t = h; Fig. 4-4 has WU/h as parameter, ta < h,
1.0
X - —
1.00/ ^
T~
j
0.8 /1 f/ J
1
1
/y ,
0.6
j /
1 I
I , ft
'/' 7 •
//
/> '/II
/11
// 1
0.4
1
//
h/ /
0.2
//
/
=0.20
/ V
x/L
Fig. 4-4. Temperature distribution at t = to < h.
T<u
= TEU/h, with x/L as abscissa. These curves show that W is the
basic parameter in this problem just as B was in Sec. 3-2. A large W
gives small differences in temperatures, while a small W gives large
differences in temperatures. Experimental verification of T in Fig. 3-4
for W of order of 0.1 is given in Ref. 6.
70 THERMAL STRESSES
The thermal stresses for the plate are given by Eq. (4-2), in which
TO i+ - (4-12)
£ [<^L" P(„,0
= 0 < I <
12
71=0
A] fa
00
n=0
„, A (-1)" / pn'WA
The thermal stresses for the beam are given by Eq. (4-3). Let Ai
be the area of the effective skin and stringer flange, Aw the area of the
stringer web, and A2 the area of the stringer flange away from the skin;
then in Eq. (4-3)
ae
2
wt__ 2A^+SA? +
WHc =
£ j-(-T^ + A2y{nt)
n =0
0 < t < h
j ^ [^p^"
00
TO = W +
4l]
+ r5nP(»,<) >
tl
«
-
wl
cA]
C-£
=
^TO (A. 2A0 +
g
+
(<L-%)Av
r(L-c)(-p»A. _ (4-13)
y
Ln-0
< <
0
h
<
TFH6 = (A. - cA
+
2AX)
[J
£
(L-c)(-l)"A. LA, -
cA2]
&nP(n,t)
+
2
Pn Pn
n=
0
>
h
t
ONE-DIMENSIONAL TEMPERATURE DISTRIBUTION 71
The primary interest in Eqs. (4-2), (4-3), (4-12), and (4-13) is the loca
tion, time occurring, and magnitude of the maximum stresses. From
Figs. 4-2 and 4-4 and Eq. (4-2) the location of the maximum stresses
are as follows (T increasing) :
Restrained. Maximum compression on x = L, no tension (stress dis
tribution same as temperature in Fig. 4-3).
Restrained only in bending. Maximum compression on x = L, maxi
mum tension on x = 0.
Unrestrained. Maximum compression on x = L and x = 0, maximum
tension near x = L/2.
For W < 0.10 the maximum tension stress occurs near x = L for the
beam case of Eqs. (4-3) and (4-13). This is due to the assumption
that the entire area A1 is at the surface temperature. For small W
this assumption is not very practical, whence the tension stress at
x = L/2 is taken as the more realistic value. Plots of Eqs. (4-2) and
(4-3)using Eqs. (4-12) and (4-13) for selected values of x/L against t/h
indicate these maximum stresses to occur at time t = fi. Figures 4-5
and 4-6 show graphs of these maximum stresses against W. Note that,
once the stresses on x = L are calculated for the restrained-in-bending
case and for the unrestrained case, the entire stress distribution for both
cases can be obtained by graphical means from Fig. 4-4.
72 THERMAL STRESSES
— —
i1l
1.0 i i i i 1 r i i
i
R= Restrained in bending; U= Unrestraine i
i i 1
-Compression -Ter sion
1 1 1
0.8 p u.o i
Aih 3.4, A 2/Au. -0.4 M/ Aw -1.0, A2/. 4.u,-( ).4
0.6 "- n
U.O
\\
**
\"* k
/
/ \\
r \c
^
0.4 V„ -HA I
\
\ ^<< •*y-o
n —\ V- '
>»
N
v» Of R -0.2
v. Of Jl
>\
0.2
R R
~0( U 1.0
U
0.4 0.8 1.2 0.4 0.8 1.2 1.6
W
Fig. 4-6. Maximum thermal stresses in beams.
This indicates that thick skins on airplanes and missiles will have rela
tively low thermal stresses unrestrained) except for very small values
(if
of h.
It should be noted that, the temperature curve drawn at the par
is
if
area and moment of the area under the temperature curve. The shape
of the curve can be approximated as long as the area and moment of the
area remain unaffected.
ONE-DIMENSIONAL TEMPERATURE DISTRIBUTION 73
(4-16)
where T0 = T at x = 0.
4-3. Add a value for titanium alloy to Eq. (4-15).
4-4. Check points in Fig. 4-4, and show that no terms are needed in the series of
Eq. (4-11) for W > 1.00.
4-5. A symmetrical aluminum-alloy beam has one flange of area 0.50 in.2 heated
linearly from 100°F to 500°F in 2.0 min. The web is 0.1 in. thick and 3.0 in. deep.
Find the maximum stresses in the restrained-in-bending case.
4-6. A titanium plate is heated linearly from 100°F to 800°F in 1.5 min. If the
stress in the plate is not to exceed 10,000 psi, what is the maximum thickness of the
plate?
4-3. Plates with Constant Heat Source on One Surface. This case
corresponds to Eq. (4-1) with TM = TE and h constant. From Eq. (4-5)
take
T = Ae-»w>< cos \x + B
The condition
on x = L
gives
= X„L = Pn
Pn -y cot pn (4-17)
To satisfy T = 0 at t = 0, take
ao
n =0
74 THERMAL STRESSES
•0
whence TB = ) an cos
~
L
— Af
TE cos (pnx/L)dx 2T
and o„ = = — sin
: -. (4-18)
cos2 (pnx/L) dx Pn + sm pn cos pn
fQL
— - sin P" exP (-Pn2W<i) cos (pnx/L)
Z/V
Thus =
2
1
TB p" + sin p„ cos p„
H =
2
J
(
-T"l^ 57 .
/
^
/-/ + sin p„
;
P« (Pn cos p„)
n=
0
4)
has plotted the stress on the heated surface against the parameter
for various values of hL/k. He also plots the maximum stress on the
heated surface against hL/k. These stresses on the heated surface are
the largest stresses in the plate.
Przemieniecki (Ref. 14) has drawn curves for both the restrained-in-
bending and the unrestrained cases for the stress on the heated surface
against the parameter N in Eq. (4-21). His results are shown in Figs.
4-7 to 4-9. Cheng (Ref. 28) has obtained the maximum stresses for
various hL/k on the surface and at the center for the symmetrical case
and on both surfaces for the unrestrained case.
Figures 4-7 to 4-9 show the maximum surface stress to depend on the
parameter hL/k. For = 19.3 Btu/(hr)(ft2)(°F) (see Example 3-3) and
h
If
in feet. the stress in the plate for
-
restrained-in-bending case
is
L
not to exceed 0.10a£/7V(l ?), then, from Fig. 4-8, hL/k = 0.36 and
< 1.38 ft. For titanium with = 10, < 0.18 ft. Thus, for values
L
L
Table 4-1. The First Four Roots, p„, of p„ = (hL/k) cot p„t
hL/k Po Pi Ps P3
N=kt/pCpL2
Fig. 4-8. Thermal stresses in heated surface of plate restrained in bending. (By
permission of J. S. Przemieniecki, Transient Skin Temperatures and Stresses Attained
in High Speed Flight, Tech. Office Rept. 77, The Bristol Aeroplane Company, Ltd.,
Aircraft Division, June, 1954.)
ONE-DIMENSIONAL TEMPERATURE DISTRIBUTION 77
1-0 1 1 1 1 1 1 1 i 1 1 1 1 1 1 1 1 1
0.8
k/hL
Fig. 4-9. Maximum thermal stresses in plates. (By permission of J. S. Przemieniecki,
Transient Skin Temperatures and Stresses Attained in High Speed Flight, Tech. Office
Rept. 77, The Bristol Aeroplane Company, Ltd., Aircraft Division, June, 1954.)
PROBLEMS
4-7. Verify the firstthree roots pi, p2, ps for hL/k = 1.0 in Table 4-1.
4-8. Use the results of Prob. 4-7 in Eq. (4-19) to draw temperature curves against
Wt/h for x/L = 1.0 and x/L = 0.0.
4-9. A titanium-alloy plate is subjected to constant heat source at Te = 700°
and h — 30 Btu/(hr)(ft2)(°F). If the stress in the unrestrained plate is not to
exceed 5,000 psi, how thick can the plate be?
4-10. Figure 4-7 shows the stress to be near its maximum value for a considerable
time. If the plate is steel with thickness 2 in., use Eq. (4-21) to calculate the interval
of time involved in 0.02 < N < 0.20.
4-11. Derive Eq. (4-20).
4-12. In the theory of boundary-value problems the roots of Eq. (4-17) form a
sequence of eigenvalues which define the set of eigenfunctions cos (p„x/L). This set
of functions is orthogonal; i.e.,
[L
JO
cos
L
cos P=5 dx = 0
L (4-22)
for m n. For m = n the value is not zero. This theory was used to isolate and
evaluate the constants in Eq. (4-18). Prove the theory for this case by making the
integration in Eq. (4-22).
4-4. Plates with Variable Heat Source on One Surface. In Eq. (4-1)
assume h to be constant, and take Ta, in the form »
T_ t
+
(x/LY - - 2(k/hL)
1
TE k 2W
30
q
Li
t
o
t
W sin pn cos p„)
+
Pn2(P«
n=0
(4-24)
- Z/V
P" sin P" exP (-P"2W<o) cos (pnx/L)
t>t
T
_
2
I
0
Te W Pn2(P" + sin p„ cos p„)
\
71=
- exp (pn2W)
0
= W p" =
X cOt P"
'
/3„
1
of
prescribed
4)
Eq. (4-24) for < < for several values of hL/k. The thermal
0
to
t
Wtl*+2j
BO
sin p"
WHe = < <
to
t
hL L-i pn
to
n=0
00
= W + >
2
to
t-l
I
Pn
n=
0
^
to
(P^inPn + 2cosPw)-2
TO ^
V
= < (4-25)
+
12
Q
fc
J
n=0
TO
BO
sin P")
-
2
= 12 g=
^P(n,0 >
<0
<
^
+i22cos
Pn
n=0
pn) exp (— p^Wt/tp)
P(n,0 = (sin
Pn2(Pn + Sin pn coS p„)
The maximum value of the surface stresses will occur at > to. Since
<i
the stress curves are rather flat (Fig. 4-7), the maximum can be deter
mined approximately by calculating the stresses for about three values
of > t0. Values for < can be read from Heisler's curves in
t0
t
t
Ref.
4.
Equation (4-1) can be solved, in manner similar to that for the linear
a
ONE-DIMENSIONAL TEMPERATURE DISTRIBUTION 79
case, for other variations in Ta, such as a parabola or t" variation. How
ever, the coefficients in the series become more involved, and higher-
degree polynomials in x have to be introduced. Also, the Laplace trans
form can be used to solve the system (4-1). Kaye and Yeh (Ref. 5) used
this method with both h and Ta, having linear variations. They give
charts (Ref. 5) for obtaining the temperatures for this case and apply
the charts to a previous problem (Refs. 15, 16) solved by numerical
procedures.
For other work on plates see papers by Schneider (Ref. 17) on various
cases of heat generation, by Ness (Ref. 18) on sweat cooling, by Sprague
and Huang (Ref. 19) on edge heating, and by Ambrosio and Ishimoto
(Ref. 37) on the distinction between thin and thick plates.
PROBLEMS
4-13. Show that Eq. (4-24) reduces to Eq. (4-11) for hL/k = ».
4-14. Derive Eq. (4-25).
4-16. If W = 0.2 and hL/k = 1.0, use the results of Prob. 4-7 to draw temperature
curves against t/U for x/L = 1.0 and x/L = 0.0.
4-16. What are the maximum stresses in Prob. 4-15 for the restrained-in-bending
case?
4-17. Derive Eq. (4-24).
4-5. Beams with Constant Heat Source on Skin. Whereas the tem
perature distributions in in the plate or beam for a
Sec. 4-2 are the same
prescribed surface or edge temperature, they are not necessarily the
same for a constant heat source. Section 4-3 gives results for the plate
•26 W-Zhs
bw=2L -J
-x=<l
f
Fig. 4-10. Structural element for multiweb beam.
for a constant heat source. These results will apply to the beam only
if it is actually a plate heated on one edge, having no flange on the heated
side. For the skin exposed to a constant heat source with webs inside
(Fig. 4-10) conduction through the skin as well as the web must be
considered. The equations are
80 THERMAL STRESSES
h d'Tb
0 < x < L
dt Pb(cp)b dx2
dT. k, d2T,
dt P«(cp)« dx2
+ TT^VlT
p.{cp).h,
h
(Ta.-T.) L<x<L + b
= 0 on x = 0 (4-26)
dx
T. = Tb on x = L
' ox /
on i=L
T. = T6 = 0 at « = 0
Pohle and Oliver (Ref. 7), Schuh (Refs. 8, 9), Parkes(Refs. 10, 33), Hoff
and Torda (Ref. 11), and Goldberg (Ref. 12) have considered this set of
equations (4-26) under various assumptions and by several methods,
including Laplace transform and numerical integration.
Schuh (Ref. 8) has used numerical integration to solve the system (4-26)
and construct charts of temperature against the parameter
i
~[l
r hpb(cp)b
hph{cp)b
, , , , | (4-27)
\_kbp,(cp),h,
for various locations in the skin and web at selected times. The web
thickness equals the skin thickness in the calculations. For the skin
<, <i>
1.0 =.°*
«-20
10
«-5
0.8
0-2
0.6
0.2
3
7
0
5
1
q=Uh/khs-)V2
Fig. web stresses for
Maximum = 7.5. [By permission from H. Schuh,
R
4-11o.
Transient Temperature Distributions and Thermal Stresses in a Skin-shear Web Con
figuration at High Speed Flight for a Wide Range Parameters, Journal
of
the
of
and web the same material, he gives curves (Ref .8) of stress /»,
at the
</>
center of the web against q for various ratios of = 2h,b/hwL and for
values of = 2h,/hw. Figure 4-11 shows his results for the
R
several
maximum stresses. These results are for the symmetrical box beam, or
82 THERMAL STRESSES
0.8
1 o.e
y
£ 0.4
\R- = 2.0, 1.0
0.2
1 2 3 4 5 6 7 8
?=i|*Ms]l/2
Fig. Time at which maximum stresses occur.
4-12. [By permission from H. Schuh,
Transient Temperature Distributions and Thermal Stresses in a Skin-shear Web Con
figuration at High Speed Flight for a Wide Range of Parameters, Journal of the
Aeronautical Sciences, 22(12): 829-836 (1955).]
assumption simplifies Eq. (4-26) into two separate equations for the
skin and for the web. If the effect of the web upon the skin is neglected,
then Eq. (3-28) gives for constant heat source
T,
Tb
= 1 - e-B"' Bh =
Pth,(Cp),
(4-28)
g |[7Vl-e--)-T]
= on x = L (4-29)
ONE-DIMENSIONAL TEMPERATURE DISTRIBUTION 83
The solution is
= +
rB
1
-Ciexp(- <rjcosr 2/anexpV-
n— 0
~tr) cos^
? n J
hP,{cp)X p»(e,)»L' (4-30)
Ci =
—
£ —
: p„ = m,- cot p„
m,- cos <7 9 sin q
2q2 sin p„
a„ = — —
(q2 Pn2) (pn + sin p„ cos p„)
Note that, if
h replaces hs and Bh is a specified constant, then Eq. (4-30)
represents another case in Sec. 4-4 for TM = TE(l — e_B*'), a variable
heat source. If there is no resistance at the joint, or the skin and web
are integral structure, then m, = « and the expressions simplify to
Ci = sec q pn = (2n + 1)
| an =
^^ij (4-31)
,
A.
, f, C, / . , V • 1
v- <^ J H
+ sm
L1 7 v- 2/ 3
.
+ 4„
\L . n
Cl
\L-,
(cos
- + sm
— c — —g] exp /
sin
3
1
1
, • \
q)
c «2SWA
j 2 J ^— ^
00
j. V L - •
- csinpnl exp / Pn2WfM
\- -ir)\
+ a- T 1
p« 1
1 -i-
+ sm n
( p">
2/ -^r~ J
where 4,
= skin area at temperature T,
Al = flange area any) next to skin at temperature T\
(if
0,
respectively. For different materials use Eqs. (3-52), (3-55), and (4-16).
Figure 4-13 for the restrained-in-bending case shows the maximum
thermal stresses against the parameter for several values of the joint
q
age geometry of = based on Fig. 4-11. The curves show that the
5
parameter rrij for the joint may increase the thermal stresses for small
84 THERMAL STRESSES
values of q but may actually decrease the stresses for large values of q.
The time at which the maximum stresses occur is similar to Fig. 4-12,
except for small values of q, for which the time increases as my decreases.
Levy (Ref. 20) gives stresses and curvature of beams for a linear
temperature penetration part way into the beam.
Kotanchik (Ref. 13) gives test results on heating one side of a multiweb
beam. The temperature distribution is similar to that calculated in
Ref. 7 and shows some decrease in skin temperature at the web. The
temperature in the web is low, and a check of the measured stresses is
made by using average skin and web temperatures in the equations for
hjLI fe-0.( )
htb/
Fig.
0 12345678
4-13. Maximum web stresses for various joint thermal resistances (</>= 5).
[By permission of the Journal of the Aeronautical Sciences, 24: 152-153 (1957).]
stress in Sec. 3-9. The stresses in the skin checked very well up to
buckling of the skin. Cobb and Brouns (Ref. 27) give theoretical and
experimental temperatures and stresses for a rapidly heated box beam.
Seide, Taylor, and Harvill (Ref. 36) give curves for temperature and
stress for various geometry and material of the box beam, including
some work on contact resistance between skin and web and on end
effects. See also Quinville and Sneyd (Ref. 38) for a discussion of contact
resistance.
Example 4-1. A multiweb beam similar to Fig. 4-10 is heated on both surfaces
by a constant heat source of h = 40 Btu/(hr)(ft2)(°F) and TE = 500°F above datum.
If the material is aluminum alloy with h, = 0.15 in., hw = 0.08 in., 26 = 15 in.,
2L = 10 in., find the maximum thermal stress in the web.
ONE-DIMENSIONAL TEMPERATURE DISTRIBUTION 86
q
_ 5 f 40(12) li la
12L75(0.15)J
.
* " (2) (0.15) (7.5) _ 5 b *p 2(0.15)
•J'75
0.08(5) -008"
From Fig. 4-11
whence by interpolation
= 40,000 psi
(/»)m»x = (0.62) (13) (10) (500)
PROBLEMS
fx- 0
( (
1
Fig. 4-14. Thermal stresses along beam.
Consider the beam idealized to that shown in Fig. 4-15, where the
webs take only shear loads and the stringers take only axial loads. If
the beam is assumed restrained at the ends by compression load P in
stringers Ai and a tension load 2P in stringer A 2, then this restraint can
*
t
y
O
a
<\ 1
2P- { , 2P
J 0 '■ ( A2
a
( t P OA1
l h
Free end
Fig. 4-15. Beam for shear-lag calculations.
(4-33)
= h,rxy = Gh. +
3
Oh. —
(«2 Mi) (4-34)
D« =
^[(i + l)l«d* + E«^-T.\
where D = d/dx. Differentiate again to get
D2q — p2q =
Gh,a
a
D(T2 - T,) = 0
q = Cie~px + C2epx
(4-36)
ONE-DIMENSIONAL TEMPERATURE DISTRIBUTION 87
From Figs. 4-14 and 4-15 the boundary conditions can be taken as
q = 0 for x = L, L large
P = qdx = - Ai/i at x =
(4-37)
fQL
0
This gives the shear flow as
= pPe~"x
(4-38)
q
Superposition of the loads gives zero loads in the stringers at x = and
0
stringer load variation of
a
pl =
/;gdx = p(i-e-)
P2 = 2P(1 - e~") (4-39)
it
If
stiffness will have to be provided along the beam until the shear becomes
small. In fact the shear flow in thin webs will tend to be constant in the
direction between stiffeners. If the shear flow is assumed constant at
x
_ aAiE
V
1
,
^~*1J
|
\Gh.[\ + (2A./A,)]]
V
the web thick with relatively small flanges, or the beam plate
If
is
is
a
heated on the edges, then, with the assumption of constant shear flow
direction for distance the shear-flow variation in the direc
in
the
d,
x
tion
is
dA
[afxd.
.
(4-42)
Jy
P, =
j*
dy (4-43)
q
is
d
A
3
i
88 THERMAL STRESSES
Example 4-2. In Fig. 4-15 let a = 5.0 in., h. = 0.050 in., Ai = 0.80 in.8, A2 =
0.25 in.2 (5 in. of web), and/i = 10,000 psi due to temperature. Find the maximum
shear stress in the web, the maximum load in the end stiffener, and the minimum
distance over which the end effects can be balanced out.
Solution. With E/G = 2(1 + *) = 2.5, Eq. (4-35) gives p = 0.19, whence
Thus the maximum shear stress in the web is 1,520/0.050 = 30,400 psi. The maxi
mum load in the end stiffener is [Eq. (4-40)]P. = 5(1,520) = 7,600 lb, and Eq. (4-41)
gives the minimum distance d as 5.3 in. Note that, for the web riveted to the
stringers Ai, the load on the rivets will be approximately ^(0) = 1,520 lb/in. over
the end 5.3 in. Note also that d = a approximately, agreeing with the conventional
design practice of using a uniform shear flow over a distance approximately equal
to the width of tfce panel.
PROBLEMS
4-23. Solve Example 4-2 if h, = 0.10 in. and 5 in. of web is used for the area A2.
Compare results with those in the example.
4-24. Consider the beam in Fig. 4-14 to be an aluminum-alloy plate 0.25 in. thick
and 10 in. wide with a symmetrical temperature distribution
Use d = 5 in. in Eq. (4-42), and find the maximum shear stress in the plate. If the
last 2 in. of the plate is assumed to resist the stiffener load given by Eq. (4-43), will
this load buckle the plate?
4-25. Solve Prob. 4-23 for a temperature of A i at 500°F above datum and the rest
of the beam at 0°F. Assume the plate to be steel alloy, and use Eq. (4-44).
4-26. Solve Prob. 3-36 for the maximum shear flow near the ends if the web thick
ness is h, = 0.050 in.
where heat is received from the hot gases and conducted to the blade
root and turbine disk. For T, = 2\ at x = 0 (the root) and dTJdx = 0
at x = L (the tip) Eq. (4-45) gives
cosh X(L —
a;)
Ta, T, — {Ta, —
T\) cosh XL
(4-46)
X2 = h_
h,hg
For large XL, T, — Ta, at the tip. Also, as discussed in Sec. 4-5, in
many cases conduction has little effect on the distribution of the temper
ature so that Ti may be nearly equal to T0>. For the hollow blade with
allowance for cooling-air temperature change due to convection, radiation,
radial conduction, and rotation Eq. (4-45) becomes (Ref. 30)
d2T AT
hjPj (4-47)
= G2 =
F
kPah,
It
Cp (j
hjj + h„P„
G3 =
kPah,
where hi, h0, hn heat-transfer coefficients for inside cooling air, outside
gases, and nozzle radiation
= inside, average, and outside perimeter of blade
P
P0
P
»j a,
1
= angular velocity
TP
Ref. 30.
Equations (4-45) to (4-47) assume the blade temperature to be uniform
at any blade cross section. Actually
there considerable variation across the
is
is
the
a
is
is
is
Fig. 4-18. Stress field for hollow turbine blade.
ONE-DIMENSIONAL TEMPERATURE DISTRIBUTION 91
alent to the case in Eq. (4-47) or to Eq. (4-45) if an average h and TM are
used. Finally there is a temperature drop across the thickness of the
blade wall due to difference of inside-air temperature and outside-gas
temperature. Figure 4-17 shows the temperature field as obtained by
Pollmann (Ref. 31) for the blade in Fig. 4-16.
Since the blade cross section is unsymmetrical, Eq. (1-23) can be
used for the thermal stresses. The temperature T0 as denned in Eq.
(1-23) represents a temperature field of straight lines, which is shown in
Fig. 4-17. If the points where the two temperature fields have equal
temperature differences are connected together, there result lines of
equal temperature difference which represent, by Eq. (1-23), lines of
equal thermal stress. Figure 4-18 shows the thermal-stress field as
determined by Pollmann (Ref. 31) from the temperature fields in Fig.
4-17.
PROBLEMS
4-27. Derive Eq. (4-46).
4-28. What effect does the spanwise temperature variation given by Eq. (4-46)
have upon the thermal stresses given by Eq. (1-23)? See Sec. 5-2.
4-29. Show that the stress in the blade due to centrifugal force is
-
= >p2R2
(I A)
for a constant cross section with R the radius of the turbine disk plus length of the
blade and a measured from the blade tip toward the disk center.
REFERENCES
(1954).
10. Parkes, E. W.: Transient Thermal Stresses in Wings, Aircraft Eng., 26(298) :373-
378 (1953).
11. Hoff, N. J.: Structural Problems of Future Aircraft, 3d Anglo-Am. Aeronaut.
Conf., 1951, Roy. Aeronaut. Soc, pp. 77-114; app. I, Aerodynamic Heating and
Thermal Stresses, by Paul Torda and N. J. Hoff, pp. 103-110.
12. Goldberg, M. A.: Investigation of the Temperature Distribution and Thermal
Stresses in a Hypersonic Wing Structure, J. Aeronaut. Set., 23:981-990 (1956).
13. Kotanchik, J. N., A. E. Johnson, Jr., and R. D. Ross: Rapid Radiant-heating
Tests of Multiweb Beams, NACA TN 3474, September, 1955.
14. Przemieniecki, J. S.: Transient Temperatures and Stresses in Plates Attained in
High Speed Flight, J. Aeronaut. Sci., 22(5) :345-348 (1955); also, Tech. Office
Rept. 77, The Bristol Aeroplane Company, Ltd., Aircraft Division, June, 1954.
15. Massachusetts Institute of Technology Staff: Thermal Stresses and Deflections
in Supersonic Aircraft Wing, Air Force Tech. Rept. 5786, 4 pts., Wright-Patterson
Air Force Base, Ohio, November, 1949.
16. Kaye, Joseph: The Transient Temperature Distribution in a Wing Flying at
Supersonic Speeds, Aeronaut. Sci., 17:787-808 (1950).
17. Schneider, P. J. : Variation of Maximum Thermal Stress in Free Plates, J. Aeronaut.
Sci., 22:892 (1955).
18. Ness, Nathan: On the Temperature Distribution along a Semi-infinite Sweat-
cooled Plate, J. Aeronaut. Sci., 19:760-768 (1952).
19. Sprague, G. H, and P. C. Huang: Analytical and Experimental Investigation of
Stress Distributions in Long, Flat Plates Subjected to Longitudinal Loads and
Transverse Temperature Gradients, WADC TR 55-350, Wright-Patterson Air
Force Base, Ohio, December, 1955.
20. Levy, S. : Thermal Stresses and Deformations in Beams, Aeronaut. Eng. Rev., 16 :
62-70 (October, 1956).
21. Gatewood, B. E.: Effect of Thermal Resistance of Joints upon Thermal Stresses,
J. Aeronaut. Sci., 24:152-153 (1957).
22. Kuhn, Paul, and J. P. Peterson: Shear Lag in Axially Loaded Panels, NACA TN
1728, October, 1948.
23. Peery, D. J.: "Aircraft Structures," sees. 17.12-17.15, McGraw-Hill Book Com
pany, Inc., New York, 1950.
24. Gatewood, B. E. : Shear Distribution in Beams with Variable Webs, J. Aeronaut.
Sci., 16:749-753 (1949).
25. Reissner, E.: Analysis of Shear Lag in Box Beams by the Principle of Minimum
Potential Energy, Quart. Appl. Math., 4:268-278 (1946).
26. Heldenfels, R. R. : The Effect of Nonuniform Temperature Distributions on the
Stresses and Distortions of Stiff ened-shell Structures, NACA TN 2240, Novem
ber, 1950.
27. Cobb, J. O., and R. C. Brouns: Studies of Thermal Effects in Aircraft Wings,
thesis, Massachusetts Institute of Technology, 1952.
28. Cheng, C. M.: Resistance to Thermal Shock, J. Am. Rocket Soc, 21:147-153
(1951).
29. J. Bradshaw: Thermal Stresses in Turbine Blades, Phil.
Lighthill, M. J., and F.
Mag., ser. 7, 40:770-780 (1949).
30. Livingood, J. N. B., and W. B. Brown: Analysis of Spanwise Temperature
Distribution in Three Types of Air-cooled Turbine Blades, NACA TR 994, 1950.
ONE-DIMENSIONAL TEMPERATURE DISTRIBUTION 93
31. Pollmann, E.: Temperatures and Stresses on Hollow Blades for Gas Turbines,
NACA TM 1183, September, 1947.
32. Schmit, L. A., and F. L. Williams: The Effects of Thermal Radiation on Aircraft
Structures, pt. II, The Response of a Simple Structure to Radiant Heating,
WADC TR 54-384, Wright-Patterson Air Force Base, Ohio, 1955.
33. Parkes, E. W.: The Alleviation of Thermal Stresses, Aircraft Eng., 26:51-53
(1953).
34. Fritz, R. J.: Evaluation of Transient Temperatures and Stresses, Trans. ASME,
76:913-921 (1954).
35. Kuhn, Paul: "Stresses in Aircraft Structures," McGraw-Hill Book Company,
Inc., New York, 1956.
36. Seide, P., E. C. Taylor, and V. A. Harvill: Thermal Stresses in Thick Skin
Multicell Wing Structures, Rept. NAl-54-482, Northrop Aircraft, Inc., July, 1954
(ASTIA AD No. 64433).
37. Ambrosio, A., and T. Ishimoto: Analytical Studies of Aircraft Structures Exposed
to Transient External Heating, WADC TR 54-579, vol. I., Thermal Response of
a "Thin" Plate under the Influence of a Constant Temperature Edge, May, 1954
(ASTIA AD No. 73860), vol. II., Thermal Response of a Finite Plate and the
"Thin" Plate Criteria, November, 1954 (ASTIA AD No. 75855).
38. Quinville, J. A., and J. M. Sneyd: Methods of Calculating Heat Conduction for
Transient Aerodynamic Heating of Supersonic Wing Structures, Rept. N Al-54-452,
Northrop Aircraft, Inc., June, 1954 (ASTIA AD No. 78265).
39. Howland, W. E., E. A. Trabant, and S. A. Hawkins: A Mechanical Computing
Device for the Analysis of One-dimensional Transient Heat Conduction Problems,
ASME Paper 56-SA-27, 1956.
CHAPTER 5
(5-1)
- $j (c2
- y2)x
The elementary beam theory gives only the first term in fx, gives /„ = 0,
and gives the correct rxy in this case. However, for a beam long com
pared with its depth, the second term in fx and the value of /„ are small.
Note that the stresses in Eq. (5-1) only apply away from the ends of the
beam.
94
THERMAL-STRESS PROBLEMS IN TWO DIMENSIONS 95
For more general loading on the beam, Seewald (Ref . 2) has shown that
/» has the form
fx = fxi + =
^ (FX2) + Fx r**
Yx
+ (GW + Gx (5-3)
where fx\ is given by Eq. (1-21), Fx involves terms in the fourth- and
higher-order derivatives with respect to x, Gx involves the fifth and higher
derivatives, and
- -h
SI T dy L Ty dy + +
L
y dy
(?)
T
6
aE t1 (?)']
- -
/> ^ dy
(f)
+ 2i)
L
dy
[10
21 10
+
(?)']
i
aE -LTdy+W+^)]LTdy
-iXi-($)]LTydy (5'4)
—Gxi
Tdy-y Tydy +
jV_
Ty2 dy
)_
J_
aE
|*
/lr*-i[1+(?)]/-.IV*
5)+
ft)'
20
-(f)']/!1**
+
21
&[*+
For the in which the temperature independent of y, the
71
is
case
stresses are (Ref.
3)
96 THERMAL STRESSES
73
= -\W- c2)
SF +
ISO
- myV + 0 +
(152/4 7c4)
* . '
7T.=-kW- c2>°
Ti + 360
- °™ - g + *■>
• • •
(5-5)
5 I - - - - a?
ife
= • • •
s? 3y2)
+
c2) (^2 c2) (7c2
=
For example, take To(x/L)2 in Eq. (5-5), whence the maximum stress
T
on = +
is
c
2/
--
(£)
= t„, =
fy
<xETa
/.
0
If large compared with
I
then the stress very small. Since these
L
is
is
c,
equations do not apply at the ends of the beam, necessary that
is
it
L
be several times before they will give correct results at the center of
c
x
will be small so that the one-dimensional formulas of Sec. 1-4 and Chap.
4
will give good results. That just as elementary beam theory gives
is,
stress theory give good results for variable temperature along the beam.
a
is
large compared with the depth so that the variation of the temperature
over the surface of the wing produces small thermal stresses compared
with the variation through the thickness of the wing.
For work on box beams with spanwise and chordwise temperature
variations see papers by Heldenfels (Refs.
4,
5).
PROBLEMS
6-1. If= TE{y/cy{x/LY, use Eq. (5-3) to find the stresses at the point =
T
c,
y
= L/2 in an aluminum-alloy beam 1.0 in. wide, 4.0 in. deep, and 32.0 in. long
x
(see Fig. 1-6 for coordinate system). Take Te — 500°F above datum.
5- 2. What are the stresses in beam (o) linear in and independent of x,
T
if
is
a
6-4. How much correction to the elementary bending formula does Eq. (5-2) give
for cantilever beam with triangle loading? Take the width constant and the
a
2400 i i i
-Adia batic wall tempe raturt
2000
M ach n umbe r = 6
A :celer ation of lg (32.2 ft/se c2)
1600
/
t5 1200
J
-Bott Dm si rface
800 /
r
400
Top bundle
1
—
—■ ~
/
nitia --Mk Jplanf
-.it- \
ten ipera ture
'
. atM-1.4-1
2 3 4 5.5
Distance from leading edge,(ft]
Bottom surface
Fig. 5-1. Chordwise temperature distribution in wedge-shaped wing. [Massachusetts
Institute of Technology Staff, Thermal Stresses and Deflections in Supersonic Aircraft
Wings, Air Force Tech. Rept. 5786, 4 pts., Wright-Patterson Air Force Base, Ohio,
November, 1949 (ASTIA ATI No. 67934).]
5-2 shows the temperature variation through the thickness at the mid-
chord of the same wing in Fig. 5-1. The results in Figs. 5-1 and 5-2
were obtained by numerical integration in two dimensions. However,
a one-dimensional integration in the thickness direction gave essentially
the same results, indicating that the chordwise conduction of heat had
very little effect. There is some effect near the leading and trailing
edges, where the chordwise temperature is large. Thus the temperature
98 THERMAL STRESSES
800
700
600
g 500
TO
CD
E 400
300
200
100
0
1 3 5 7 9 11 13 15
of the wing. The stresses due to the chordwise variation can be obtained
from Eq. (4-3), where b(x) represents the varying thickness of the wing
and T(x) is the chordwise temperature distribution, obtained by averaging
the temperatures through the thickness at each point. Unless this
temperature can be approximated by an equation, it will be necessary
to evaluate the integrals in Eq. (4-3) numerically.
120
At Ma :h number=6
80
—
FVlidplane
40
/
0
Upper s jrface
-40
-80
120
./&^\ 1
/ 1
^— Lower surface
1
/ If
\\
160
o
200
X
240
\
280
\
320
\
360
\
400
1
440,
a)
00 <U1
480 *o
a>
Ml
-a .
oo OJ 1
520
liling
ladin
560
600
11 13 15
3
7
1
Station numbers
Fig. 5-3. Stress distribution in wedge-shaped wing. [Massachusetts Institute Tech
of
nology Staff, Thermal Stresses and Deflections in Supersonic Aircraft Wings, Air Force
Tech. Rept. 5786, pts., Wright-Patterson Air Force Base, Ohio, November, 1949
4
Example 5-1. Using the temperature distribution in Figs. 5-1 and 5-2, calculate
the approximate thermal stresses at the leading edge and at mid-chord on the surface
and at mid-plane. Use the same assumptions as to material properties as used in
Fig. 5-3, and compare results with Fig. 5-3.
Solution. Approximate the Mach number curve in Fig. 5-2 by symmetrical
a
6
= 25
+
600
T
100 THERMAL STRESSES
(-T + 197)
— 120,000 psi on surface
U
f. 48,000 psi at mid-plane
PROBLEMS
6-6. Assume the wing in Fig. 5-1 to have an average chordwise symmetrical
temperature distribution the same as that for W — 0.08 in Fig. 4-4. Assume the
symmetrical distribution through the thickness at the mid-chord to be the same as
that for W = 0.40 in Fig. 4-4. Approximate these two curves, and calculate the
thermal stresses at the leading edge and on the mid-chord surface and mid-plane by
superposition.
6-7. In Prob. 5-6 use the W — 1.60 curve in Fig. 4-4 at the quarter-chord point,
and calculate the quarter-chord surface and mid-plane stresses by superposition.
6-8. Suppose the chordwise average temperature distribution in Prob. 5-6 is
unsymmetrical, being the case of W = 0.08 in Fig. 4-4 for the aft half of the wing
and W = 0.16 for the forward half. Use Eq. (1-22) to obtain the thermal stresses
due to this chordwise temperature variation.
whence
■
= j j ?y rT{r) dr dr (5-7)
dU dV aE [b _, , ,
/ rT{r) dr = M (5-8)
dr dr b j0
i±(r±)\l±(r™)]-0
r dr\ dr) dr dr
(5-9)
\-r
\
/
whence = dr2 log + + C3 log C4 = C2r2 +
U
+
C2r2 C4 (5-10)
r
=
M =
Mr2
+
u
c>
Tb izr c< (5-11)
d(U - V) d2(U
- V) = d2(U
- V)
J'
1
= +
1"
dr r2 dd2 dr2
-
r
d(U V)
1
dd
r
= rrB =
—
whence rT(r) dr
^r
fr
0
^-
(5-12)
aET M
^
=
T + ^J0 rT{r)dr =
f'rTdr
ft
Jo
b
a
is
instant —
the edge of the plate maintained at the temperature zero,
is
0,
t
(5-13)
102 THERMAL STRESSES
•
2£
(5-14)
n-l
Jaeger (Ref. 23) gives tables for the stresses in Eq. (5-14) for the time
parameter N = kt/b2 varying from 0.005 to 1.0. For small values of the
parameter N the series converges very slowly so that he uses the Laplace
transform and the resulting error functions to obtain the stresses. The
maximum tangential stress, which occurs on the surface r = b and is the
largest numerical stress in the plate, is
/.-aSr.[l-4(^)' + 2f +
|(^)l + ^+...]
for small values of N.
Horvay (Ref. 24) considers the case of a transient temperature applied
to the edge of the disk and to the surface of a cylinder.
PROBLEMS
5-9. Derive Eq. (5-13).
5- 10. Obtain Eq. (5-14).
6- 11. If the plate is steel with T0 = 600°F, what are the stresses in the plate
[Eq. (5-14)] at r = 6/2 and t = 2 min?
5-12. Show that the stresses for the circular plate with a hole of radius a in its
center are
ft = N
(l + +
£ rT(r) dr - aET (5-15)
N =
^]barT(r)dr
Take the radial stress as zero on r = a and r = 6. Note that Ci in Eq. (5-10) must be
taken as zero so that the displacements of the ring will be single-valued (Sees. 9-5, 9-6).
where p is the density of the disk. If the stresses are obtained for this
centrifugal force, then the total stress in the elastic disk can be obtained
by superposition of the thermal stresses in Eq. (5-12) and the stresses
from the centrifugal force.
<t> Equation (5-16) is satisfied by the stress
function denned by
rfr = =
+
Pp2r2
4>
(5-17)
/.
|?
Eliminate u from the strain-displacement-stress equations
€r = = (/r - "fe) 6e
= = Ue
- Vfr) (5-18)
tr
I
I
F
and substitute for the strains to get the following equation for the stress
function </>:
r20 +
r^-* + (3 + v)ppV =
(5-19)
0
The solution of Eq. (5-19)
is
_ +
y-^pVrs
r Cj
= Cir
+
(5-20)
v^
+
(3
C2 = Ci =
0
=
^PP2a>2
- r2)
(5-21)
—+ , +—
3*
3
1
v
»"2
=
,
,
/«t
g— PP2^2 PP
g
For the total stress in rotating elastic disk of constant thickness with
a
the disk material vary with the temperature, and the stress in parts
of
(5-16)
h
becomes
- hf, pV'hr' =
+
jr
(rhfr) (5-22)
0
104 THERMAL STRESSES
Add the temperature strain aT in Eq. (5-18), and eliminate the displace
ment to get
+
+ y)ir/r-/9)=0
^)-(1 (5-23)
Equations (5-22) and (5-23) together with boundary conditions will give
fr and fe. However, p, E, v, a, T, and h may be functions of the radius r
so that the equations cannot be solved directly. The finite-difference
method, which is used in many engineering problems, can be used to
solve Eqs. (5-22) and (5-23).
Choose a number of discrete point stations along the radius, 0, 1, 2,
. . . , n — I, n, . . . . If A is a point halfway between point n — 1
and point n, then rA = (r„-i + rn)/2 and approximately
(rhfr)A = |(r„_lA„_i/r,n_i + rnhnfr.n)
W')a =
dr
+ (Pnhnrn2 + Pn-ikn-iTn-i2) — 0
^
fe,n fe,n—i Vyfr.n ■ Vn— ifr.n— l \ rp rp
iin iin
- /*,*) - /fl.n-l)]
-G<n-l
- ¥(r«
l/„ -
^
+ + Pn-l) (/r.n-l
r-O fC1 "n)(/r.n
_ %
+
, (1
g-^— j
n
= 0
|-
Or Cnfr,n — Dnfe,n = ^n/r.n-l + Gn/s.n-l —
Now the stresses at any station can be written in terms of the stresses
at station 0 (on the boundary of the hole in the disk, fr = 0, fe = fe,o)
where the coefficients Ar,n, Br,n, A$,n, and Be,n can be determined by
substitution of Eq. (5-26) into Eq. (5-25). Since /9,0 is arbitrary, the
coefficients of /»,0 and the constant coefficients must be zero, whence
where the terms Kn, L„, etc., are known in terms of the C„, Dn, etc., in
Eq. (5-25). For a solid disk /r,0 = fe.o, and Eq. (5-26) gives
From these known coefficients the coefficients at all other stations can be
obtained by successive applications of Eq. (5-27). At the rim r = b
= k" 7 Brb
ho (5-30)
whence the stresses at all stations can now be obtained from Eq. (5-26) .
Here /r,& is the known radial rim stress, and Ar,b and Br,b are the coeffi
cients as determined in the last of the successive applications of Eq. (5-27).
The bulk of the computation in the above finite-difference procedure
is contained in the determination of the coefficients Kn, L„, etc., in Eq.
(5-27). However, if the stations are chosen according to an assigned
value of (r„ — r„_i)/r„ and if v is assumed constant as = then the
-J,
v
p (5-31)
Kn =
Tn
-L„ =
nn n-l
-
then 0.834tn 0.015F n 0.104tn + 0.077F„
= L'n = 0.043t'„ + 0.767Fn
K'n 0.343i„ 0.153Fn
Mn = -(0M7inPn-i + 104p„)r„2p2 0.092F.J?; - (5-32)
- 0.920M;
0.
PROBLEMS
6-16. Solve Prob. 5-15 by the finite-difference method, using stations as specified
in Eq. (5-31). Omit centrifugal forces, and assume constant properties so that
Eq. (5-32) simplifies still further. Compare these approximate results with the
exact results of Prob. 5-15.
6-17. Verify Eq. (5-32).
6-18. If the stress due to centrifugal force in steel disk not to exceed 100,000 psi,
is
a
rate at which heat can be removed. This removal of heat requires the
106 THERMAL STRESSES
-kV2T = Q
(5-33)
where Q is the heat generated per unit volume per unit time and k is the
coefficient of thermal conductivity. Consider the case of the temperature
distribution in a moderator structure pierced with regularly spaced
coolant channels. Assume that the region surrounding each circular
coolant channel constitutes an insulated cylinder with a uniform heat
source and an internal convective boundary. Let b be the outer radius
and a the inner radius of the cylinder. Then Eq. (5-33) gives [see
Eq. (1-50)]
d'T
dr1
1
r dr
-r
dT _- 1 d
dr\
( dT\
dr J
_ _ Q
k
(5-34)
^- = 0 on r = 6 T = Ta on r = <
dr
The solution is
The thermal stresses in the hollow cylinder are given by Eq. (5-15)
with fr and fe replaced by (1 — v)fr and (1 — v)fe and from Eq. (1-46)
and (1-47), where
(1
-
(1
-
(1-- v)f. = 2N - aET
(5-37)
(1 =
(1
- v)(fr + fe)
N rT dr
a2 J a
'*)-(i-5D(i-^+*5(S-'+»hd
F,(r) = -4 (4 - ) + 8 -2 In -
/
1
\a2 a2 a
The maximum numerical stresses in Eq. (5-38) are the tangential and
axial tensile stresses on r = o,
40)
a
b
r
r
b
108 THERMAL STRESSES
graph of the parameter P2 against b/a, from which the maximum stresses
in Eqs. (5-39) and (5-41) can be obtained at once. Glasstone (Ref. 21)
shows graphs of Pifr, Prfe, and PJ,
in Eq. (5-38) against r/a with b/a
as parameter.
Thompson (Ref. 22) considers the hollow cylinder with the heat source
a function of the radius and solves in particular the case of Q = constant
f or a < r < c and Q = 0 f or c < r < b. For other work on temperature
distributions and thermal stresses in nuclear-reactor components see, for
example, Glasstone (Ref. 21) and Thompson and Rodgers (Ref. 25).
12
Fig.
3 456789 b/a
5-4. Parameter P2 for thermal stresses in thick-walled cylinders.
PROBLEMS
5- 19. Derive the expressions for/x in Eq. (5-37).
6- 20. Derive Eq. (5-38).
5-21. By using Fig. 5-4 graph r/a in Eq. (5-40) against b/a.
5-22. If the hollow cylinder is uranium-aluminum alloy with Q = 10' Btu/(hr) (ft3),
a — in., and physical properties the same as those for aluminum, how large can 6 be
if the stress in the cylinder is not to exceed 40,000 psi?
5-23. Many reactor designs have plate-type fuel elements which can be idealized
as an infinite slab with a uniformly distributed heat source. Use Eq. (5-33) with
Q constant to find the temperature distribution in the slab. Take T = T0 at x = 0,
the center of the slab, and T = Z\ at x = ±L, the surfaces of the slab.
5-24. Use Eq. (4-2) to get the stresses in the slab of Prob. 5-23.
THERMAL-STRESS PROBLEMS IN TWO DIMENSIONS 109
REFERENCES
with these factors. Table 6-1 shows a possible listing of the independent
factors that can be used to define almost any structural problem, includ
ing fatigue, dynamic, and high-temperature problems. In each of these
problems the physical and mechanical properties, which are functions of
some or all of the factors in Table 6-1, are needed. The primary physical
constants needed in aircraft structures are listed in Table 6-2. The
mechanical constants or, more correctly, the allowable stresses as defined
by type of stress and of the local cross section of the structural component
are listed in Table 6-3.
100
1 1
-Compression
E- -Tens ion~
80
60
Ftu
40
Fey
20 IIlonge tion
1.25
F = 4 0,000 psi
?r
1.00 j '
t ie U Sift —
s—Via
0.75
Strc in (in /in.Jl
0.50
/
/ /
0.25 V
\ \
1.0
V
0.8
1
,v
\ L A \ Vs
\
0.6
\ \ 400
\
\
50
600°
EJ1
w \\\
1
:
3C 0=F-
0.4 2C 0°F-
Ro om te Tip. -
\\
\
V
u\
0.2
\
\
00T
W
inuuin lerrip.
\
300T--
\
\W
\
Ir
10 20 30 10 20 30 40
0
Stress (ksi)
Fig. 6-3. 2024-T3 aluminum-alloy-clad sheet: tangent and secant E in compression
hr soaking time).
(i
ALLOWABLE STRESSES AT ELEVATED TEMPERATURES 115
sheet material; (13) test specimens used; and (14) tested in air. From
the allowable stresses in Table 6-3, Fig. 6-1 shows (5) E for tension and
compression; (6) per cent elongation (5); (7) ultimate tension stress
(Ff„) (8) yield-tension (^j,) and yield-compression (F^) stresses.
; Figure
6-2 shows (1) short-time compressive stress-strain curves. Figure 6-3
shows (2) compression tangent modulus curves and (3) compression
secant modulus curves. Figure 6-4 for physical properties in Table 6-2
shows (2) specific heat cp, (3) coefficient of thermal expansion a, (4)
thermal conductivity k. Table 6-4 gives some approximate data on
emissivity coefficients, item 7 in Table 6-2.
14.0 0.25 r
(i l./in. F|10<
tu/lb° F]
a cp Mean k
k
i y' Btu-in./ft2hr°F]
12.8 0.22 820
ss
0.20
100 200 300 400 500 600 700 800
2TFI
Fig. 6-4. 2024-T3 aluminum-alloy-clad sheet: thermal properties.
Emissivity
Material and condition
coefficient
TT2E ,fPy_ i
t)Fcy \L'J B2
KE
(tV _ i
.
F^ =
T]Fcy T7/
where L' —
L/c* with c the column-end fixity
K = edge-support coefficient for plates
E —
elastic modulus
Fey — yield stress
=
77 function of stress-strain curve in terms of Et and E, (tangent
and secant moduli) and depends on type of cross section
t The discussion on room-temperature procedures in this section and Figs. 6-5, 6-7,
Journal of the Aeronautical Sciences, 18(10) :657-
6-9, 6-12 are used by permission of the
664 (1951).
ALLOWABLE STRESSES AT ELEVATED TEMPERATURES 117
Stowell (Ref . 34) gives expressions for ij for different type sections. How
ever, these values can be approximated by two values (Ref. 35),
E,
for plates
\ +
\ i1 + 3
w) (6-3)
jj = for columns
-gf
For 2024-T3-clad aluminum alloy at given stress and temperature,
ij
can be calculated from values of Et and E, given in Fig. 6-3.
To solve for Fc and F„ in Eqs. (6-1) and (6-2) with given material
a
and geometry, necessary to determine by trial and error from Eq.
it
is
17
(6-3) and tangent and secant moduli curves for the material at the given
temperature. Also curve of Fc/Fcy and Fcr/F^ against can be drawn
B
a
B
r\
eE
where n the stress-strain shape factor (Refs. 36, 37). This gives
is
E E
E' =
(WW,)-l Et = (6-5)
+ + (3n/7)(F/^)-i
1
1
which can be put into Eq. (6-3) for t\. The stress-strain curves of the
aluminum and magnesium alloys at room temperature can be approxi
mated closely (Refs. 35, 38) by n = 10. Further, study of the stress-
a
soaking times indicates that the curves can be approximated by Eq. (6-4)
and that n varies from to 12. Some values are shown in Table 6-5.
7
temperatures.
17
78 10
i
212 2 7
212 2 7
212 10 8
212 100 7
212 1,000 10
400 1 9
400 2 12
400 10, 100, 1,000 7
500 1,2 10
500 10 7
500 110, 1,000 10
600 1 10
600 2, 10 8
ALLOWABLE STRESSES AT ELEVATED TEMPERATURES 119
drawn, which will apply at all temperatures. Figure 6-5 shows these
curves, in which S = FJFcy or Fcr/Fcy. Also shown is Johnson's parabola
for short columns, which differs little from the column curve in Eq. (6-1),
4[1
- (Fc/Fcc)]
1
B1
(6-6)
where B is defined in Eq. (6-1). For sections that do not fail by local
instability, Fcc = Fcy;otherwise, Fcc is the crippling stress of plates,
formed or extruded sections, or stiffened plates. Table 6-6 shows com
parison of test data (Ref. 70) and values calculated from Fig. 6-5 for
75S-T6 columns at various temperatures.
For columns with eccentricity e the maximum extreme fiber stress Fc,
is (Ref. 63)
„ P , Mc PL , eAc A
(6-7)
2 \ElJ 2 \fJ
where B is defined in Eq. (6-1). This gives
(Fc/Fcc)*
-
1 .k
(6-8)
B 2 cos-l {(eAc/I)/[(Fcc/Fc) 1]}
with Z stiff eners (Kef. 44), and plates stiffened with angle stiff eners
(Ref. 45).
The maximum allowable stress or ultimate allowable stress is the allow
able crippling stress for local failure of a section and is the allowable
column stress for column failure in either the long or the short column
13
12
11
10
-
One side free
(compression]
0.385
Fcr=m)2 -\
"6" for shear is the smaller dimension of the plate
1
a/b
Fig. 6-6. K for compression, shear, and bending of flat plates.
range. In most cases the allowable crippling stress, which may be higher
than the buckling stress owing to the fact that the edges and corners
of the section may be able to carry higher stresses than the fiat elements
and which may involve yielding in the corner, must be determined from
empirical formulas or curves obtained from tests. The NACA has done
some theoretical work on calculating maximum stresses (Refs. 56,
57, 60).
ALLOWABLE STRESSES AT ELEVATED TEMPERATURES 121
0.2
6l/62
Fig. 6-7. K and M values for angles in compression. [By permission of the Journal of
the Aeronautical Sciences, 18(10): 657-664 (1951).]
Schuette (Ref. 62) shows that for a certain range of values of F„ the
maximum stress F^ can be expressed in the form
F
*• cc F
* max
Ct
6 = = M MB-
F~
*■cy T~
1 cy \riFj (6-9)
where M
depends upon the geometry of the section. Figure 6-12 shows
a plot of Eq. (6-9) for selected M values. The maximum stress is taken
122 THERMAL STRESSES
as equal to the buckling stress to the left of the intersection of Eqs. (6-9)
and (6-2). M values for various sections as based on test data (Refs. 35,
62, 44) are given in tables on Figs. 6-7 to 6-11. For sections in which K
is not available procedures as in Ref. 52 or 61 can be used. For stiffened
plates effective areas can be used (see Sec. 7-4).
A formula for many materials at room and elevated temperatures to
bw M-extrusions;
JW-formed in
formed in "O"
tw "2" cond.
cond.
bp
\ I 1
t
.*— *
bw
-J-
bF
t
V.— bw -4
K
1 \bwj
0.5
replace Eq. (6-9) for plate elements is given by the NACA in two 1956
reports (Refs. 65, 84). The formula is
•^Compression leg
1.20
Tj '-f
/ .i-
*w bw
', -1.5C
1.00
^
May be different i n
«L» 1 «j» t
u a IU I irun
comp ressic >n leg
0.80
ko
0.60
0.40
0.20 _ ft
r-=K Y
)
0 2 4 6 8 10
f>w /b.
Fig. 6-9. K and M values for channels and Z's in bending. [By permission of the
Journal of the Aeronautical Sciences, 18(10): 657-664 (1951).]
Example 6-1. For a 2024-T3 aluminum-alloy-clad formed channel in Fig. 6-8 with
bF = 1.0in., bw = 2.0 in., tw = tF = 0.072 in. at 400°F find the buckling and crippling
124 THERMAL STRESSES
stresses. If the channel is 30 in. long and fails as a pin-ended column about an axis
parallel to the flanges, find the allowable column stress.
Solution. From Fig. 6-8 with 6*./6w = 0.50 and tw/tF = 1.0. K = 2.62. From
Fig. 6-1 at 400°F, E = 9.3(10») psi, and Fcy = S1,000 psi. Equation (6-2) gives
*-£(&)'
From the plate curve in Fig. 6-5 S = F„IFcy = 0.83, whence
' '
I
I
'
M=0.80 for all -r
t«
bw/b,
Fig. 6-10. K and M values for Z-stiffened panels in compression.
Although M can be obtained from Fig. 6-8, it is not applicable for this section (see
Fig. 6-12 for B = 0.99). If the channel is formed in the T condition, the yield
stress in the corners is increased so that the ultimate strength is larger than for a
channel formed in the O condition and then heat-treated. If the channel is formed
in the O condition and heat-treated, Fig. 6-12 gives Fcc = F„ = 25,700 psi. If it is
formed in the T condition, Fig. 6-12 gives Fcc = 0.90(31,000) = 27,900 psi.
To determine the column stress, calculate
/ = (2)(1.0)(0.072)(1.0)2 +
°-072(2 °)3
ai9 ^4
ALLOWABLE STRESSES AT ELEVATED TEMPERATURES 125
whence p = (//A)* = 0.81 in. and from Eqs. (6-1) and (6-6)
K = 4.8. Table 6-7 shows the comparison of the test stresses (Ref. 129)
with the stresses calculated by Figs. 6-5 and 6-12 or Eqs. (6-9) and (6-6).
An end fixity of c = 3.75 is used in the column-stress calculation, and
the crippling stress is increased 11 per cent to allow for four stiffeners on
three skin panels. Note the large increase in the stresses for the 10-hr
soaking period at 400°F. This result is a propertyof 2024-T3 aluminum
alloy, but not necessarily of other alloys. It emphasizes the wide varia
tions in the same material due to different temperature histories.
1.2
- Formed in /"-condition
24 ST
0.8
-Fee1 MM J, l/2
0.6
0.80=
0.4
060
0.4 8
0.2
E uckli ig str iSS-'
B
Fig. 6-12. Allowable
crippling stresses. [By permission of the Journal of the Aero
nautical Sciences, 18(10): 657-664 (1951).]
PROBLEMS
-aTE, SE,aT dA + P
F = + E.
SE,dA
fE,aT dA +P
e = -aT + (6-H)
JE. dA
Figure 6-13 shows these curves (dotted) for selected values of eap. Note
that these curves cannot actually exist. For the case of Fig. 6-13 they
represent an applied strain with the thermal stresses omitted. The actual
stress distribution depends upon the sequence of application of the applied
load and the temperature change. If the temperature change occurs
first and the external load is applied second, then the solid curves of
Fig. 6-13 are obtained from Eq. (6-11). Although the solid and dotted
curves are drawn for the same eap, the external loads are not quite the
same in the inelastic region. Even with the same external load, it is
apparent that in the inelastic region the thermal stresses decrease and
practically disappear so that the ultimate stress is essentially the average
- SE.y dA _ {Ee(y
- yNA)2 dA
VNA Ieff (6-13)
-JWTdl JWdA
which depends upon the stress level. There is little effect in the elastic
range, but there is a large effect in the inelastic range for large temperature
gradients. The analysis of the strains can be made by adding the bend
ing strain to Eq. (6-11) and plotting the strains in the same way as in
Fig. 6-13, provided the temperature change occurs before the applied
load.
Consider Eq. (6-11) and Fig. 6-13 for the case in which P is a compres
sion load. For no buckling the yield stress is reached first on the edge
of the plate. For the plate as a column about an axis perpendicular to
the plate a column curve may be constructed as follows: Assume eap in
Eq. (6-12), which determines P. Use this eap in Eq. (6-11) to get e and
hence E, from Fig. 6-3. This procedure assumes the temperature
change is applied before the external load. Calculate
A graph of Fc —
P/ A against L'/p gives the column curve for the par
ticular symmetric temperature distribution. Figure 6-14 shows the
column curve for plates with the temperature distribution in Fig. 6-13,
plotted on the basis of the scales in Fig. 6-5 with Fcy the yield stress for
T = 300°F. The column curve from Fig. 6-5, which applies for uniform
temperature, is also shown for comparison.
For the unsymmetrical temperature distribution, strains due to the
bending stresses must be included and allowance made for eccentricity
due to bowing and shift in the neutral axis. Equation (6-13) will give
the neutral-axis shift provided Et is used for the column instead of E,.
Consider the case of a plate supported by stringers along the edges
and loaded by a compression load P on the ends. Let the edge stringers
be at a uniform temperature Tb and the plate at a higher uniform temper
ature T,. Equation (6-11) gives the strain in the plate and in the string
ers. When the strain in the plate reaches ecr = K(t/b)2, the plate will
buckle. The buckling stress in this case is given by Fig. 6-5. Note
that, when e = e„ in Eq. (6-11), the load P may be tension or compression
130 THERMAL STRESSES
;
Inifor m ten iperal ure o 300 F
0.4
^onu liforrr tern; >erature (F g6-l
^\
0.2
B
Fig. 6-14. Column stresses for nonuniform temperature.
PROBLEMS
o.oio r
/r=41.0 ks i
R = 0.975
0.008 400
38.0 1
J
.952
.9031 i. 6.0
356
0.006
34.0
,
—
.808
0,004 f~
0.002
%
f.-'- >hort time st ress
0 1 2 3 4 5 6
Time (hr]
Fig. 6-15. Compressive creep curves for 2024-T3 aluminum-alloy plates at 400°F;
b/t = 20; 10-hr exposure at 400°P prior to loading.
132 THERMAL STRESSES
That
is,
necessary to reduce the applied stresses on the structure.
it
is
necessary that an allowable stress be used which permits only a certain
specified amount of deformation in a specified time.
The usual stress-strain curve at the given temperature cannot be
obtained in this case because the strain depends upon the time as well
as the stress. However, for given load or stress at given temperature,
a
a
strain-time curve can be obtained by test. Typical strain-time curves
a
for several stresses at given temperature are shown in Fig. 6-15 (Ref.
a
family of curves drawn for different stresses, then cross
If
is
129).
a
a
plot of stress against strain can be made for any fixed time. This gives
family of apparent stress-strain curves with time as parameter.
a
a
= Ae-<*/RT
(6-16)
r
where = a constant
A
= gas constant
T R
and = temperature, °R
If the creep rate assumed to be inversely proportional to time, then
is
(6-17)
t)
e
|
is
C
a
t
Q(F) indicates that the activation energy depends on stress. This tem
perature-time parameter T(C + log designated as the Larson-Miller
t),
parameter (Refs. 85, 77), checks test data over a wide range of times and
temperatures for many steels, high-temperature alloys, and aluminum
alloys. Although a universal value of = 20 checks the test data
C
reasonably well, Heimerl (Ref. 77) shows that better correlation can be
obtained by using different values of for different alloys. Heimerl
C
(Ref. 77) also discusses several other parameters that have been proposed
ALLOWABLE STRESSES AT ELEVATED TEMPERATURES 133
70
—r~r
Temp., "F
o 90
60 - 212
"9
o 300
o- 300
» 300
50 v 400
V *- 450
-oa>- * 450
' 600
40
30
20
Creep rate
1% strain •
10
Rupture
i
10 12 14 16 18 20 22 24 26xl03
Rupture, r(17 + log t) or creep rate, TOO-log r]
12 14 16 18 20 22 24 26 28x103
1% strain, T(17 + log t|
Fio. 6-17. Master rupture and creep curves for 24S-T3 aluminum alloy.
e = AeBFtK + J (6-18)
|T+460|(20 + log10t)/103
Fig. 6-18. 2024-T3 aluminum-alloy-clad sheet: master creep curves.
is
Fcy
and an n value [Eq. (6-4)] can be determined, then the procedure of Sec.
6-3 can be used. However, the iso-stress-strain curves tend to have the
form of Eq. (6-18) rather than Eq. (6-4) so that may not be possible
it
to use an n value in Eq. (6-4). On the other hand, Figs. 6-16 and 6-18
ALLOWABLE STRESSES AT ELEVATED TEMPERATURES 135
indicate that n < 10 may give a fair representation of the curves. Based
on the 0.2 per cent offset yield stress in each case, Fig. 6-16 shows graphs
of Eq. (6-4) for n = 10 and n = 5. Thus curves based on n values
can be used to approximate the iso-stress-strain curves, and in particular
the curves of Figs. 6-5 and 6-12 based on n = 10 can be used provided
it is recognized that they are conservative for stresses greater than F^
and slightly unconservative for stresses from about 0.5Fcy to F^.
Mathauser and Deveikis (Ref. 129) have made creep tests on 2024-T3
aluminum-alloy Z-stiff ened panels (see Sec. 6-3 and Table 6-7 for descrip
tion of the panels and short-time tests on the panels) for creep times up to
about 6 hr at 400°F after soaking 10 hr at 400°F. Table 6-8 shows com
parison of test stresses with calculated stresses. The stresses are calcu
lated by using the test creep time to obtain Fcy (0.2 per cent offset) from
Fig. 6-16, whence Eqs. (6-9) and (6-6) are used for Fcc and Fc. See dis
cussion on Table 6-7 for geometry of the panels.
Table 6-8 shows that, for creep times under 10 hr, the procedures of
Sec. 6-3 are applicable when the iso-stress-strain curves of Fig. 6-16 are
used. For longer creep times Fig. 6-18 is applicable for approximate
calculations. King (Ref. 133) has made creep tests up to 256 hr on
2024-T4 and 7075-T6 machined, integrally stiffened panels. His column
is so short that essentially the maximum stress is the crippling stress Fcc.
Buckling occurs beyond the yield stress so that the buckling and crippling
stresses are on the cutoff in Fig. 6-5 or 6-12 (assuming the cutoff to apply
in creep and using n = 5 to represent the creep curve). If the Larson-
Miller parameter is calculated for Bang's tests (Ref. 133) and the yield
stress read from Fig. 6-18 for his room-temperature yield of 50,000 psi
for 2024-T4, then the resulting Fce = l.2Fcy checks his test data (the
tests are 0 to 15 per cent higher than the calculated values). Also the
strain curve of e = 0.010 in Fig. 6-18, based on {F^rt = 50,000 psi,
checks the test data to about the same accuracy as the 1.2Fc„ stresses.
Mathauser and Deveikis (Ref. 84) show that Eq. (6-10) checks the test
data closely for plates when E, and from iso-stress-strain
Fcy are taken
curves as in Fig. 6-16. They also show that the Larson-Miller parameter
correlates the plate strength for each b/t ratio tested, with the curve for
b/t = 20 being nearly the same as the tensile rupture curve.
For cases in which Eq. (6-18) applies, Fcy can be calculated. Put
e = Fcy/O.IE in Eq. (6-18) to get (here Fcy is defined in terms of a slope
(6-19)
which can be solved for Fcy for particular case by trial and error or by
a
(6-20)
1
hit
*2E/(u/Py
F= + ABEtK exp (BFc)
K° ZL>
6
Example 6-2. Solve Example 6-1 the channel section to have life of 100 hr.
if
is
TM -
~~ (400 +460) _
(20 + log 100) = 1Qn
ltfU
1,000
ALLOWABLE STRESSES AT ELEVATED TEMPERATURES 137
PROBLEMS
kFu 3.25(1 - v)
Ea hL
(6-22)
for column stress at one load may not be the most efficient at a different
load. It is possible to analyze this problem of geometry and load effects
by the combination of geometry and load into one parameter, designated
as a structural index. For the case of a column of given length L to take
a given load P equate the column buckling stress to the applied stress
[Eq. (6-1)],
(jj
P
Fc = r,Ear2 = =
/
A
Fe _ E_ PY J_
£
so that =
(
s
(6-23)
FcyVi AFcy2Ly B2
V
where P/L2, involving both the given load and the given length, may be
designated as the structural index.
800
-Stainless W
SAE 4340
RC-130B
400 Inconel X
V ^-XA78S-T6
75S-T6
°F
Fig. 6-19. Variation of compressive-yield-stress-density ratio with temperature.
is
son's parabola in Fig. 6-5, then Eq. (6-23) by the use of Eq. (6-6)
becomes
±
_Fc = B2 >
2
Fcv
Fc
4
B2 < (6-24)
2
+ B4
4
B
VCX AFcy2L2)
Against the parameter Eq. (6-24) graphs the same as Euler's curve in
B
Fig. 6-5 and slightly above Johnson's parabola, but having the end points
the same as the parabola. In the parameter are terms for the column
B
142 THERMAL STRESSES
end fixity, the cross-section geometry, the material properties, and the
structural index. For a given structural index Eq. (6-24) can be used to
compare different materials provided the end fixity c and cross-section
parameter p2/ A remain invariant for the different materials. The com
parison can be made at different temperatures and for creep conditions
provided the proper E and Fcy are selected from short-time stress-strain
curves, or from iso-stress-strain curves (Figs. 6-3 and 6-18). Figure 6-20
shows comparisons made for some titanium alloys, steel alloys, and alumi
num alloys against short-time temperature for c(p2/A)(P/L2) = 50 psi
(Ref. 107). For the parameter = 200 psi Ref. 107 shows considerable
change in the efficiencies, RC-130B is best up to 350°, stainless W from
350 to 750°, RC-130B from 750 to 950°, Inconel X from 950° up.
800 r
where D is the average tube diameter and t is the wall thickness. How
ever, there is a limit to the D/t ratio because of local buckling of the walls.
If the local buckling stress is equated to the column stress, then the opti
mum value of D/t and hence p2/ A can be obtained. For this value of
ALLOWABLE STRESSES AT ELEVATED TEMPERATURES 143
p2/ A the highest possible stress for the tube is obtained. Shanley (Ref.
108) gives
)OPt
(6-26)
2
(t
0
P/L2/
where = 0.40, based on tests, and independent of material. How
is
fc2
ever, the optimum p2/ depends upon material (E) and upon the struc
A
tural index. This means that the materials will compare differently for
this case from that shown in Fig. The parameter in Eq. (6-24)
B
6-20.
for the optimum tube becomes
With this in Eq. (6-24) the most efficient tube for each material can
B
E
is
then for the best tube in the other materials (constant and P/L2)
c
[Eq. (6-26)]
(£)-»(£)'
0
(?)
= 71 for steel
h.
A9-r
c
= 57 for titanium
(5)
0r Inconel X
C
f
Z
These values give smaller value for in Eq. (6-24) and hence larger
B
a
in Fig. 6-20 for optimum tubes, all the curves being closer to 75S-T6 at
room temperature.
formed or extruded section restrained to act as
If
the column
is
a
column about one axis, the problem of comparison becomes more difficult.
The section has local buckling and crippling, which determines one
arrangement of the area to get the highest stress while the column requires
a different arrangement of the area. As indicated in Sec. 6-3, for the
short column range not permissible to equate the crippling and
it
is
of Sec. 6-3, best cross section can be obtained for a given material and
a
for given structural index P/L2. Just as for the tube, the p2/A of this
a
144 THERMAL STRESSES
(bj\ 1
[UPIO FJ 1 V
W/.„( & P/L2)
fA
2 V k1CTT2
, + Q(tF/tw)(bF/bw) (6-28)
, K
[I + 2(tF/tw){bF/bw)}2 Kitw
where K is given in Fig. 6-8 as a function of tF/tw and bF/bw. The term
hiK*, which is independent of material, can be optimized by selecting
tF/tw and bF/bw and using Fig. 6-8. This optimum value may be limited
by practical dimensions and requirements of standard thicknesses; the
load and the material will influence any modifications in the best value.
The optimum column stress for the b/t ratio in Eq. (6-28) is
F\ = /27M2cr2kiKi E* -P\*
F \ 64 FJV)
-
tc _ O
4
Fcc
which applies for crippling stresses below the cutoff (Fig. 6-12). If the
crippling stress has the cutoff value, then the optimum b/t ratio is the
largest value on the cutoff.
The optimum arrangement for plates and stiffened plates is still more
involved than for columns. Several cross-section parameters similar to
p2/ A may be present. The structural index takes the form Pi/b, where
P, is the load intensity in pounds per inch and b is the plate width.
ALLOWABLE STRESSES AT ELEVATED TEMPERATURES 145
Considerable work has been done on these plate problems (Refs. 105 to
109, 115 to but in many cases it is assumed that the cross-section
120),
parameters remain invariant when different materials are compared.
Although in many cases fair comparisons can be obtained with this
assumption, in other cases it is possible that unfair comparisons may be
made in the sense that the heavier materials may appear worse than they
actually are in comparison with the lighter alloys. A procedure similar
to that used above for the channel-section columns can be used for
stiffened plates. When p2/ A is expressed in terms of b,/t, (Fig. 6-10),
several geometry constants involving ratio of area in skin to area in
stringers, centroid of stringers, etc., result (Ref. 115). Optimum values
on these geometric constants as a group may or may not carry over to
other materials, depending upon load and practical arrangement. If
they carry over exactly and if the column stress equals the buckling
stress for all stresses, then the optimum curves for various materials and
temperatures in Ref. 117 can be used to compare aircraft materials.
PROBLEMS
6-17. Plot Eq. (6-24) on Fig. 6-5, and compare with Euler's curve and Johnson's
parabola.
6-18. A pin-ended column of 24S-T3 aluminum alloy 30 in. long is to take 100,000 lb
at 400°F. What is the minimum cross-section area of a solid circular column (a) for
short-time loading? (6) for a life of 100 hr?
6-19. Solve Prob. 6-18 for a circular tube. What are the tube diameter and wall
thickness?
6-20. Solve Prob. 6-18 for a channel section. What are the dimensions of the
channel? Can a ^-in. bolt be put through the flange?
6-21. Solve Prob. 6-20 for titanium with Fcy = 100,000 psi and no creep.
6-22. Derive Eq. (6-28).
6-23. Derive Eq. (6-29).
6-24. Equations (6-28) and (6-29) were derived on the assumption that optimum
column stress would occur in the short column or inelastic range. Show that this
assumption is correct by showing that the long column or elastic formula gives a
lower stress.
REFERENCES
46. Bijlaard, P. P.: Theory and Tests on the Plastic Stability of Plates and Shells,
J. Aeronaut. Sci., 16(9) :529-541 (1949).
47. Bijlaard, P. P.: Analysis of the Elastic and Plastic Stability of Sandwich Plates
by the Method of Split Rigidities, I, /. Aeronaut. Set., 18(5) :339-349 (1951), II,
J. Aeronaut. Set., 18(12) :790-796 (1951).
48. Bijlaard, P. P.: Determination of the Effective Width of Plates with Small
Deviations from Flatness by the Method of Split Rigidities, Proc. 1st U.S. Congr.
Appl. Mech., Chicago, 1951, pp. 359-362.
49. Krivetsky, A.: Plasticity Coefficients for the Plastic Buckling of Plates and
Shells, J.Aeronaut. Sci., 22(6) :432-435 (1955).
50. Van Der Maas, C. J.: Charts for the Calculation of the Critical Compressive
Stress for Local Instability of Columns with Hat Sections, J. Aeronaut. Sci.,
21(6):399-403 (1954).
51. Johnson, J. H., Jr.: Critical Buckling Stresses of Simply Supported Flat Rec
tangular Plates under Combined Longitudinal Compression, Transverse Com
pression, and Shear, J.Aeronaut. Sci., 21(6):411-416 (1954).
52. Needham, R. A.: The Ultimate Strength of Multiweb Box Beams in Pure Bend
ing, /. Aeronaut. Sci., 22(11) :781-786 (1955).
53. Needham, R. A.: The Ultimate Strength of Aluminum Alloy Formed Structural
Shapes in Compression, /. Aeronaut. Sci., 21(4) :217-229 (1954).
54. Abramson, H. N., H. A. Williams, and B. C. Woolpert: An Investigation of the
Bending of Angle Beams in the Plastic Range, /. Aeronaut. Sci., 22(12) :818-828
(1955).
55. Lo, Hsu, H. Crate, and E. B. Schwartz: Buckling of Thin-walled Cylinder under
Axial Compression and Internal Pressure, NACA TN 2021, January, 1950.
56. Bijlaard, P. P., and G. P. Fisher: Column Strength of H-sections and Square
Tubes in Postbuckling Range of Component Plates, NACA TN 2994, August,
1953.
57. Mayers, J., and B. Budiansky: Analysis of Behavior of Simply Supported Flat
Plates Compressed beyond the Buckling Load into the Plastic Range, NACA TN
3368, February, 1955.
58. Semonian, J. W., and J. P. Peterson: An Analysis of the Stability and Ultimate
Compressive Strength of Short Sheet-Stringer Panels with Special Reference to
the Influence of Riveted Connections between Sheet and Stringer, NACA TN
3431, March, 1955.
59. Johnson, J. H., Jr., and R. G. Noel: Critical Bending Stress for Flat Rectangular
Plates Supported along All Edges and Elastically Restrained against Rotation
along the Unloaded Compression Edge, J. Aeronaut. Sci., 20(8):535-540 (1953).
60. Stowell, E. Z.: Compressive Strength of Flanges, NACA TN 2020, January,
1950.
61. Cozzone, F. P., and M. A. Melcon: Nondimensional Buckling Curves — Their
Development and Application, J.
Aeronaut. Sci., 13(10) :511-517 (1946).
62. Schuette, E. H.: Observations on the Maximum Average Stress of Flat Plates
Buckled by Edge Compression, NACA TN 1625, February, 1949.
63. Peery, D. J.: "Aircraft Structures," chap. 14, McGraw-Hill Book Company,
Inc., New York, 1950.
64. Anderson, M. S.: Compressive Crippling of Structural Sections, NACA TN 3553,
January, 1956.
65. Anderson, R. A., and M. S. Anderson: Correlation of Crippling Strength of Plate
Structures with Material Properties, NACA TN 3600, January, 1956.
66. Heimerl, G. J., and W. M. Roberts: Determination of Plate Compressive
Strengths at Elevated Temperatures, NACA TN 1806, February. 1949.
r
ALLOWABLE STRESSES AT ELEVATED TEMPERATURES 149
91. Sherby, O. D., and J. E. Dorn: An Analysis of the Phenomenon of High Tem
perature Creep, Proc. Soc. Exptl. Stress Anal., 12(1):139-154 (1954).
92. Pao, Yoh-Han, and J. Marin: The Creep Deflections of Beams and Columns,
presented at the Institute of Aeronautical Sciences annual meeting, January,
1953, Preprint 406.
93. Pao, Yoh-Han, and J. Marin: An Analytical Theory of the Creep Deformation
of Materials, J. Appl. Mech., 20:245-252 (1953).
94. Hult, J. A. H.: Critical Time in Creep Buckling, J. Appl Mech., 22 :432 (1955).
95. Wahl, A. M., G. O. Sankey, M. J. Manjoine, and E. Shoemaker: Creep Tests
of Rotating Disks at Elevated Temperature and Comparison with Theory,
J. Appl. Mech., 21:225-235 (1954).
96. Pao, Yoh-Han, and J. Marin: Deflection and Stresses in Beams Subjected to
Bending and Creep, /. Appl. Mech., 19:478-484 (1952).
97. Symposium on the Thermal Barrier, November, 1954, American Society of
Mechanical Engineers, New York.
98. Carlson, R. L., and A. D. Schwope: A Method of Estimating Allowable Load
Capacities of Columns Subject to Creep, Proc. 2d U.S. Natl. Congr. Appl. Mech.,
1954, pp. 563-568, American Society of Mechanical Engineers.
99. Coffin, L. F., Jr.: Study of the Effects of Cyclic Thermal Stresses on a Ductile
Metal, Trans. ASME, 76:931-950 (1954).
100. Freudenthal, A. M., and J. H. Weiner: On the Thermal Aspects of Fatigue,
J. Appl. Phys., 27(1) :44-50 (1956).
101. Parkes, E. W.: Wings under Repeated Thermal Stress, Aircraft Eng., 26 :402-406
(1954).
102. Cooperative Investigation of Relationship between Static and Fatigue Proper
ties of Wrought N-155 Alloy at Elevated Temperatures, NACA TN 3216, April,
1955.
103. Gerard, G.: Life Expectancy of Aircraft under Thermal Flight Conditions,
J. Aeronaut. Sci., 21 :675-680 (1954).
104. Symposium on Thermal Fracture, compiled by W. D. Kingery, J. Am. Ceram.
Soc, 38:1-54 (1955).
105. Gerard, G.: Buckling Efficiencies of Plate Materials at Elevated Temperatures,
J. Aeronaut. Sci., 22:194-196 (1955).
106. Clark, J. W.: Buckling Efficiencies of Plate Materials at Elevated Temperatures,
J. Aeronaut. Sci., 22:659 (1955).
107. Heimerl, G. J., and P. J. Hughes: Structural Efficiencies of Various Aluminum,
Titanium, and Steel Alloys at Elevated Temperatures, NACA TN 2975, July,
1953.
108. Shanley, F. R.: "Weight-Strength Analysis of Aircraft Structures," McGraw-
Hill Book Company, Inc., New York, 1952.
109. Micks, W. R.: A Method for Determining the Effects of Elevated Temperatures
on Structural Design and Weight, Rand Rept. P-498, March, 1954.
110. Ferguson, R. L.: A Further Investigation of the Effect of Surface Finish on
Fatigue Properties at Elevated Temperatures, NACA TN 3142, March, 1954.
111. Predvoditelev, A. A., and B. A. Smirnov: Theory of Dynamic Creep, NACA TM
1330, September, 1955.
112. Smith, G. V.: "Properties of Metals at Elevated Temperatures," McGraw-Hill
Book Company, Inc., New York, 1950.
113. Baer, H. W.: Prediction of Very Short Time Creep Buckling from Very Short
Time Tensile Creep Properties, Proc. 2d U.S. Natl. Cong. Appl. Mech., 1955,
pp. 569-576, American Society of Mechanical Engineers.
ALLOWABLE STRESSES AT ELEVATED TEMPERATURES 151
It should be noted that elastic and inelastic stresses as used here are
not associated with the conventional yield stress defined by the 0.002
strain offset line. The secant modulus at the yield stress may be any
where from 80 to 40 per cent of the elastic modulus, depending upon
material and temperature. With elastic thermal stresses defined
as those
calculated for a constant E
(usually room-temperature elastic modulus)
in Chaps. 1 to 5, and with inelastic thermal stresses defined as those
calculated with a variable E (whether due to temperature or due to the
proportional limit being exceeded), it follows that a considerable inelastic
effect can occur before the yield point is reached. In other words, a
permanent strain of 0.002 due to inelastic action is permissible within the
yield-stress definition. As Fig. 6-2 shows, the major inelastic effect arises
from the decrease of stress with temperature for a given strain. Some
consequences of this inelastic effect upon design are considered in
Sec. 7-7.
PROBLEMS
7-1. If the load equation in Eq. (1-5) is replaced by/iAi +/2A2 = Q, where Q ia
an applied load, show that for the elastic case
/, = -M,(r, - ro) +
AiEiEfA2E2 a -4; (£-/.) (7-D
7-2. Apply a load Q to the ends of bar No. 1 in Fig. 1-4. If the rivets deflect
according to Eq. (1-11) and Tl = T2, find the loads on the rivets. Does the end
rivet have the highest load in this case?
7-3. Solve Prob. 7-2 with the bars at different temperatures. Does superposition
apply in this case?
7-4. Find the value of the applied load Q needed in Prob. 3-1 to make the maxi
mum stress equal to the yield stress for 2024-T3 aluminum alloy.
values of K for a
temperature increase to 600°F and return to room tem
perature. The recovery on cooling from 600°F is approximated as being
equivalent to the 500°F stress-strain curve in Fig. 6-2 for all temperatures
below 500°F. The tension curves are assumed to be the same as the
compressive curves in Fig. 6-2, and unloading is assumed to be elastic.
In Fig. 7-1 the stress cycle is essentially elastic for the K =
^ case,
assuming E to recover. There is an inelastic effect for K \, \, and 1 =
so that there is a residual tension stress at the end of the cycle, though
the yield stress is not reached in the K = and cases. \
In the K = | \
and ^ cases there is no inelastic effect in tension so that on subsequent
-30
-20
K = 1/2
1/4
-10
K=\ /8
/
—'
/
- '1/4 1/
/'
f y 1/2 /
+ 10
/
/ ——
— Te mp. ncrea sing
75
Temperature, °F
Fig. 7-1. Thermal stresses in restrained bar for temperature cycle to 600°F.
LM = U < <
10-5(2-^)(1+A1°4)
t t3
LM = 10.5
(l + A log t > t3
where <i
= 2 min
<2 = 60 min
t3 min
= 70
Figure 7-2 shows thermal stresses against the parameter for K = 1 in
Eq. (7-2). It is assumed that on cooling the material recovers to the
LM — 18 curve in Fig. 6-18. The tension iso-stress-strain curves are
1.0,
assumed to be the same as for compression, and in Fig. 7-2 the tension
stress is shown on the compression scale as a reflection in the = 0 axis. /
Creep in tension at room temperature due to the residual stress is obtained
from Fig. 6-18 by starting from the LM = 18 curve and holding constant
strain with time increasing.
If a second cycle is started after 100 hr at room temperature, then by
STRESS ANALYSIS AT ELEVATED TEMPERATURES 157
Fig. 7-2 the second cycle essentially coincides with the first cycle at 600°F
so that the remainder of the second cycle follows the first cycle.
In the above discussion of inelastic thermal stresses K has been assumed
to be constant. However, for actual structure such as two bars restrain
ing each other or a stringer restraining the skin, Eqs. (1-6) and (3-1) show
K to depend upon the E of each bar so that K depends upon the stress-
strain curve and temperature through E. Thus, as the temperature varies
through the cycle, K will also vary. From Eq. (3-1) write the strain as
e, = -K(a.T. - abTb) K =
; + pj^rj (7-4)
where E,and Eb are the secant moduli for the skin and stringer, respec
tively. The variation of E with temperature has little effect on K, but
the variation due to exceeding the proportional limit can have large
effects. For example, let Tb = 0 and bh, — Ab so that in the elastic
region K = % for the same materials in the skin and stringer. But at
600°F in Fig. 7-1 E, = 3.5(10«) and K = . f
However, this change occurs
in the range from 400 to 600°F in Fig. 7-1, and the figure shows that in
this range the K = 1 curve is only slightly above the K — curve. \
Thus the variation in K in the inelastic region affects the strain con
siderably but affects the stress very little so that the variation can be
neglected in many cases.
In Chap. 3 expressions for a,T, — abTb in Eq. (7-4) were obtained under
various heating conditions as functions of time. For these functions the
procedures of Figs. 7-1 and 7-2 can be used to find the stresses, whether
elastic or inelastic, short-time or long-time. The maximum values of
— abTb given in Chap. 3 do not necessarily give the maximum stresses
a.T.
in the inelastic case (see Figs. 7-1 and 7-2, where maximum T — To is
525°F but the maximum stress for K = 1 occurs about T - T0 = 325°F).
The maximum stresses depend not only upon a,T, — abTb but also upon
T, and the corresponding stress-strain curves.
-
Example 7-1. Consider the skin-stringer thermal-stress problem discussed in
Sec. 3-2. In Eqs. (3-4) to (3-6) assume Tm = 600 75 = 525°F, B = 1.0, a, =
m = a for aluminum alloy, 2bh, = Ab so that K = in J Eq. (7-4) (see Fig. 3-1 for the
variation of T„ Tb, and T, — Tb with time).
Solution. If the time is short (ordinarily it must be in order for B to be 1.0), the
procedure of Fig. 7-1 can be used to obtain the thermal stresses against the time.
Figure 7-3 shows both the elastic stress for E = 107 psi, which varies with T, — Tb
of Fig. 3-1, and the inelastic stress for 2024-T3 aluminum alloy, based on the stress-
strain curves of Fig. 6-2. For the inelastic calculation the peak stress of 21,000 psi
occurs at t/h = 0.70 as compared with 29,000 psi at t/h = 1.00 for the elastic case.
The residual tension stress will decrease toward zero with time by creeping (see Fig.
7-2). The stringer stress is one-half the skin stress, but of opposite sign. Note that
the yield stress is never reached in this example, all the inelastic action occurring at
less than the yield stress.
158 THERMAL STRESSES
PROBLEMS
7-5. Solve Example 7-1 for the case of bh. = 2^4.6.
7-6. Figure 7-1 indicates that variation of K due to change of E, in Eq. (7-4) has
little effect on the stress. What is the effect if the stringer instead of the skin yields
so that K varies because of What happens to K when T, — Tb starts to decrease?
7-7. Solve Example 7-1 for the case in which the skin is aluminum alloy and the
stringer is steel.
7-8. Find the inelastic thermal stresses for the temperature variation in Fig. 3-6.
Use Te = 600 — 75 = 525°F for aluminum-alloy skin and stringer.
t/h
Fig. 7-5. Inelastic-thermal-stress distribution as function of time.
[ExdA
c
SE dA
where E is the secant modulus. Since E depends upon the strain, Eq.
(7-5) must be solved by trial and error unless some simple expression is
/
assumed for E. Assume ec, eb, and c, calculate e, read from stress-strain
curves, and check for equilibrium.
Figure 7-6 shows the thermal stresses for the restrained-in-bending case
= in Eq. (7-5)] for the two temperature distributions used in Fig. 7-4.
[eb
The tension stresses are reduced small amount from the elastic values,
a
TE = 600 - 75 = 525°F
Using the stress-strain curves in Fig. 6-2 and finding ec by trial at various
times, curves for the stresses at various positions in the plate can be
obtained and plotted against time. Figure 7-7 shows the results for the
elastic and inelastic stresses. As in the restrained case of Fig. 7-5, the
largest reduction in the compression stresses on the heated surface.
is
The maximum compression starts on the surface and shifts into approxi-
STRESS ANALYSIS AT ELEVATED TEMPERATURES 161
10 0 10 -20 30 -40
/Iksil
Fig. 7-6. Inelastic thermal stresses in plates restrained in bending
PROBLEMS
7-9. By using Fig. 6-18 construct a 1,000-hr creep curve for the problem shown in
Fig. 7-4.
7-10. Construct curves similar to those in Fig. 7-4 for TB = 450 - 75 = 375°F.
7-11. Figure 7-6 shows the elastic and inelastic thermal stresses for a plate restrained
in bending. Construct a similar figure for the unrestrained case.
7-12. As time goes on in Fig. 7-5, creep occurs and the stresses approach zero.
What will be the stress in the plate after 10 hr at the equilibrium temperature of
600°F?
7-13. What are the maximum elastic and inelastic unrestrained stresses for the
problem given in Figs. 7-5 and 7-7?
7-14. What are the stresses in Fig. 7-7 after 10 hr at 600°F equilibrium temperature?
_ (Q/AbEb) -(a.T. -
abTb)
e'
1 + (bh,E,/AhEb) (7-6)
eb = e, + (a,T, — abTb)
Consider the skin-stringer problem in the example of Sec. 7-2 for which
the thermal stresses are given by Fig. 7-3. Assume an applied tension
stress of 8,000 psi to be acting on both the skin and the stringer from
t/t1
Fig. 7-8. Combined applied and thermal stresses in skin and stringer combination.
the inelastic region but has little effect on the stresses. Figure 7-8 shows
the resulting stresses. Although the yield stress never reached at any
is
time, the stresses are not the same as would be obtained by addition of
the elastic thermal stresses of Fig. 7-3 to the applied tension stress.
Let the external loads be applied at a time later than = At any
0.
t
time up to t/h = 0.62 the thermal and applied stresses add directly (Fig.
Beyond t/h = 0.62 the
application of the external load will pro
7-8).
duce a different stress distribution. Figure 7-8 shows the case when the
load applied at t/h = 1.0. For application of the load at t/h > 1.0
is
is
164 THERMAL STRESSES
thermal and applied stresses shows a different result from that for tension
(Fig. 7-8). Addition can be made up to about t/h = 0.62, then from
t/t\ = 0.75 to 2.0, the skin stress is unchanged by the applied load, but
beyond t/h = 2.0 the skin stress changes owing to inelastic effects in the
stringer. If the compression load is applied at t/h = 1.0, the skin stress
will not change and the same curves result as in the previous case. How
ever, if the external load is applied at t/h = 2.0, the stresses change as
shown in Fig. 7-8. Note that, in this compression case, the yield stress
in the skin is reached at t/h =1.0 and in the stringer at t/h = 4.0.
Creep for t/h > 4.0 will tend to equalize the applied stress in the skin
and stringer. There is some permanent strain, which is less than the
yield strain.
Suppose the skin-stringer structure with the stresses shown in Fig. 7-8
has been through one cycle and returned to equilibrium at room temper
ature. Assume no residual stress to be present, and assume no recovery
in strength properties from the 600°F stress-strain curve in Fig. 6-2. Let
the structure be subjected to a second loading similar to that used in
Fig. 7-8. The maximum stresses occurring at t/h < 1.0 will disappear,
and the maximum will be at t/h = 1.0 (+11,500 psi depending on the
loading). For t/h > 1.0 all the cases will be essentially the same as in
Fig. 7-8, except no stress in either skin or stringer will exceed ± 11,500 psi.
The yield stress is not exceeded at any time.
If the +8,000 psi in the above discussion is regarded as a limit load,
then, since the yield stress is not exceeded under the combined thermal
and applied stresses, the structure is satisfactory for the limit-load design
requirement. If the ultimate applied stress is taken as +12,000 psi,
then the critical condition will occur at 600°F, for which the ultimate
allowable stress is about +13,000 psi (no buckling or column action).
Failure will not occur because for t/h > 1.0 the strains will adjust to
keep the stress in the 600°F skin below +12,000 psi until the stringer
heats up. Then, at t/h « 4.0, the combined stresses will be approxi
mately + 12,000 psi in both the skin and the stringer. Failure occurs at
a strain much larger than the thermal strain so that beyond the yield
point the thermal effects disappear. If the 12,000 psi stress is held for
some time, failure will occur owing to creep. Thus for short-time
strength the applied ultimate loads determine failure, while combined
thermal and limit applied loads determine the yield condition. Just as
in the above example of Fig. 7-8, in many cases the thermal stresses may
be small enough to have no effect on meeting the yield condition. If the
thermal stresses cause yield when acting alone, then yield will usually
occur in the combined case.
For the thick plate heated on one surface, thin plate heated on one
edge, or beam heated on one edge with resulting nonuniform temperature
STRESS ANALYSIS AT ELEVATED TEMPERATURES 165
-aT -
eb(^
+ ee + = Ee
f
-
I J
_ faET dA _ M
+ JaET[(x/c)
- dA
+
Q
1]
e" (WrlA e"
eb ETC- /VI _ 112^4 \'-')
JE dA JE[(x/c) dA
r
l]2
Q
c
The procedure for solving Eq. (7-7) the same as discussed in Sec. 7-3
is
and demonstrated in Fig. 7-7. All the possible combinations discussed in
Sec. 7-4 for the uniform-temperature case will be present in this non
uniform problem. The addition of applied stresses in Fig. 7-7 gives
effects similar to those shown in Fig. 7-8 for the uniform case. In
a
similar way the thermal stress has no effect on the ultimate strength
condition but may in some cases affect the yield condition for limit
applied load.
For some test work on heated box beams with and without applied
loads, see Refs.
9,
10.
PROBLEMS
7-15. Verify the curves in Fig. 7-8 for the applied stress of —8,000 psi.
7-16. Change the areas in the problem of Fig. 7-8 to bh, = 2Ab, and construct the
curve for an applied stress of —8,000 psi.
7-17. Add axial stresses of + 10,000 psi to the linear-temperature distribution case
in Fig. 7-6.
7-18. Change the maximum temperature in Prob. 7-17 from 600°F to 500°F, and
solve for an applied stress of — 15,000 psi. Is the yield stress exceeded anywhere on
the cross section?
for the variable stress distribution in the skin after buckling. The usual
procedure of calculating the effective skin area on the basis of the stringer
stress not applicable, since under thermal-stress conditions the stringer
is
Eq. (6-2) as
-*(*)' (7-8)
166 THERMAL STRESSES
and define an effective width b, from the skin strain e. adjacent to the
stringer by
(7-9)
e'--
H
1 + B
e' -
<r
e"
e. = H +
^ +
[(H
+
B£)B>ecr]
e, > e„
(7-H)
eb
= e, + (a,T, —
abTb)
is
and assume the buckling strain to be e„ = —0.001. For Ab = 2h,b
it
follows that Q/Ab = 12,000 psi and
±
-
(^r)
= > (7-12)
/,
12,000
±
e.
ft
ecr
i
In the case of the applied tension load buckle appears in the skin at
a
t/h ~ 0.50 and disappears at t/h ~ 1.40. The skin stress adjacent to
the stringer larger in compression for 0.50 < t/h < 1.00 than that on
is
Fig. but then reverses more rapidly than on Fig. 7-8, becoming
it
7-8,
tension at t/h = 1.25. The stringer stress smaller for t/h > 0.50.
is
In the case of the applied compression load the stresses are similar to
Fig. 7-8, but with the maximum compression in the skin up to —28,000
psi and the maximum tension 11,000 psi. The stringer has more com
pression throughout the time interval. The buckle appears at once and
still present at t/h = 4.0. There permanent compression strain in
is
is
both skin and stringer at t/h = 4.0, but its magnitude less than the
is
yield strain. Figure 7-9 shows the skin and stringer strains. Note that
these strains give the stresses from Fig. 6-2 up to t/h = 1-0, but when
elastic unloading starts in the skin strain increments must be used to
get the stresses.
7-6. Combined Applied and Steady-state Thermal Stresses. Ordi
STRESS ANALYSIS AT ELEVATED TEMPERATURES 167
narily the time during which transient thermal stresses act is sufficiently
short so that they have little effect on creep of the structure under applied
load. However, as Figs. 7-2 and 7-4 show, creep occurs in restrained
structure under steady-state temperature distributions. Thus, if two
different materials are restraining each other under a steady-state tem
perature or if for some reason such as cold fuel or refrigeration there is
a steady-state temperature distribution on a restrained structure, then
thermal stresses will be present to affect the creep of the structure. Also,
if residual thermal stresses are present after a transient temperature has
reached equilibrium, creep must occur to remove these stresses.
Yield strain
Skin strair
-a IT.-
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
t/h
Fig. 7-9. Combined applied and thermal strains with buckling.
read on Fig. 6-18 LM = 20.5 = 1.06(20 + log whence = 0.25 hr. Thus takes
t),
it
t
that the applied stress actually increases somewhat, owing to the decrease in effective
area as the strain increases.
Example 7-3. In Fig. 7-9 at t/h = 1.0 the temperature in the skin is 600°F and
in the stringer 270°F. The stresses are —12,500 psi in the skin and —9,300 psi in the
stringer. If the temperatures are maintained at 600 and 270°F for t/h > 1.0, find
the creep stresses and strains.
Solution. Since the stringer stress is elastic and with all the applied load in the
stringer producing only —12,000 psi, it follows that the skin will creep along an
approximate constant-strain line in Fig. 6-18. Calculate LM = 1.06(20 + log t) for
various times, and read the skin stress on the —5.5 strain line on Fig. 6-18. For
1.0 hr, LM = 21.2, /. = -8,000 psi, fb = -10,200 psi; for 10 hr, /, = -5,000 psi,
= -11,000 psi.
fb
PROBLEMS
7-19. Verify Fig. 7-9, and calculate the stresses corresponding to the strains in
Fig. 7-9.
7-20. In Example 7-3 what are the stresses after 10 min?
7-21. Figure 7-8 shows residual stresses at t/h = 4.0 for the no-applied-load case.
What will the stresses be after 10 hr?
flight history of speed, altitude, and time for the aircraft. From this
predicted history the worst combinations of applied, thermal, and allow
able stresses must be selected to use in the design. Some work has been
done on this problem, but much more must be done in order that the
lightest possible structure can be designed for the given airplane or
missile. See, for example, Goldin (Ref. 11), Goldman (Ref. 12), Behr-
bohm (Ref. 13), and Bisplinghoff (Ref. 15).
Essentially at elevated temperatures the designer faced with two
is
factors which tend to increase the weight of the structure. (1) The
material properties are much lower than at room temperature, and (2)
thermal stresses may add to the applied stresses. To obtain the lightest
possible structure, consideration must be given to designing to avoid
either one or both of these problems. For flights of short duration,
insulation may help to avoid the first problem and reduce the effects of
the second problem. The weight of the structure plus insulation may be
considerably less than the weight of the structure alone when designed
for the reduced material properties. For longer nights insulation plus
cooling can avoid problem and the total weight may be less than the
1,
REFERENCES
11. Goldin, R. : Design Criteria for Heated Aircraft Structures, ASME Paper 56-AV-14,
March, 1956.
12. Goldman, G. M.: Discussion of Safety Factor Requirements for Supersonic Air
craft Structures, ASME Paper 56-AV-18, March, 1956.
13. Behrbohm, H. : On a Minimum Time Flight Path with Regard to Stress and Heat
Limitation, SAAB TN 26, Saab Aircraft Co., Linkoping, Sweden, 1954 (ASTIA
AD No. 76721).
14. Manson, S. S.: Behavior of Materials under Conditions of Thermal Stress, NACA
TN 2933, July, 1953.
15. Bisplinghoff , R. L. : Some Structural and Aeroelastic Considerations of High Speed
Flight, J. Aeronaut. Sci., 23(4) :289-329 (1956).
16. Hilton, H. H., H. A. Hassan, and H. G. Russell: Analytical Studies of Thermal
Stresses in Media Possessing Temperature-dependent Viscoelastic Properties,
WADC TR 53-322,Wright-Patterson Air Force Base, Ohio, 1953.
17. Weiner, J.: An Elastoplastic Thermal Stress Analysis of a Free Plate, ASME
Paper 56-APM-6, June, 1956.
18. Horton, W. H.: The Influence of Kinetic Heating on the Design and Testing of
Aircraft Structures, Proc. Conf. High-speed Aeronaut., 1955, Polytechnic Institute
of Brooklyn.
19. Weiner, J. H., and H. Mechanic: Thermal Stresses in Free Plates under Heat
Pulse Inputs, WADC TR 54-428, Wright-Patterson Air Force Base, Ohio, 1955.
CHAPTER 8
8-1 to 8-5 for typical shell-type aircraft structure. The more general
problem of deflections and buckling in two dimensions is considered in
Sec. 9-4.
Since the deflection of a structure is a measure of its stiffness through
the stiffness parameters EI for bending and GJ for torsion, it is apparent
that the thermal stresses may change the apparent stiffness of the struc
ture. This change in the parameters EI
and GJ can influence the
behavior of the structure, not only in its deflections, but also in its
stability and dynamic responses, for these parameters appear in the
formulas for column buckling and for the natural vibration frequencies
J
of the structure.
Ordinarily Eand G are considered as material properties and / and
as geometric properties of the structure. However, in an actual struc
ture, these properties are affected by several factors. The most impor
tant factors, based on the conventional methods of structural analysis,
may be listed as follows (Ref . 8) :
M= jAaET{y)ybdy (8-1)
where y is measured from the centroid of the section. Now the radius of
curvature of the plate or beam is
(8-2)
(8-3)
r= *!
aTxHbQ.
- v) °
(8-4)
_ c
Wm = —s beams
8c
174 THERMAL STRESSES
Formulas for Hb are given in Chap. 4 for the various temperature dis
tributions considered there. Levy (Ref. 1) shows curves of l/aT^r
against depth of penetration of a ramp-type temperature distribution
in beams.
For unsymmetrical beams Eq. (1-23) gives the two components of
deflection as
- 1
tw \ - aJ^lfl2
r. = -L
ocK2
1
(IT.). =
a*'a (8-5)
PROBLEMS
+
To A,.
T = To o < y < d
T -0 d < y < h«
where y is measured from one edge of the beam. If the flange areas are each A\ in.2
<«,
and the web is in. thick, derive the formula for the maximum deflection due to
bowing of the beam. If allowed to vary, what value of gives the largest deflec
is
d
d
?
8-6. Let the top flanges and top skin of the unsymmetrical 75S-T6 beam shown be
-Ai-3.0 in.2
-A4=1.0in.2 A3=3.0in.2-i
Prob. 8-6
at 500°F, and let the webs, bottom flanges, and bottom skin be at 100°F. If the beam
200 in. long and cantilevered, find the maximum thermal stresses and the tip deflec
is
tions. Neglect the skins and webs in calculating moments of inertia, and neglect
shear deflections. The skins and webs are 0.050 in. thick.
THERMAL EFFECTS ON DEFLECTIONS AND STIFFNESS 175
AL
where L
is the original length before deflection and W = Wm sin (tx/L)
with Wm the maximum deflection. Approximately
4t-ijT(S)'*-Tr <m>
If
the initial deflection of the column or plate is approximated by
Wi = Wmi sin (irx/L), then, from Timoshenko (Ref. 2), approximately
w- -
r=w=5 (8-7)
where Pcr is the critical Euler column load and P is the applied load.
In the case of temperature change the load P is determined by strain
so that Eq. (8-7) can be written as
W
-
.
Wm = -. (8-8)
1 (eo/ecr)
where is the middle plane or middle plate strain and ecr is the buckling
e0
al = e0 + -j- = eB +
Ui
T -_ <
aT
T\
~ e0
1
.
'
mt
1 —
D<
(e0/e„)_J- (A)
(8-11)
2L(ecr)» 2p
176 THERMAL STRESSES
2.0
1.0 )•**
1.8
—' ^'ield c urves for
i\ V
1.4
1.2
0.40
D .1.0
1
^ &
^0.1 ]
).20
N
0.8
n.-n
'I w"00
^
0.6
0.4 0
0.4
0 20^
0.2
0.10
0.02 ).00
Figure 8-2 also shows the case of zero initial deflection in which the
deflection can be taken as zero up to e0 = ecr, whence Eq. (8-9) gives
W
= (RT - l)1 (8-12)
(aT)cr
Note that the temperature Tcr defined in Eq. (8-10) is essentially a refer
ence temperature. Although it may be called a buckling temperature,
it does not cause failure of the column and actually indicates the start of
deflection only for the ideal case of no initial deflection. However, when
the temperature exceeds Tcr byvarying amounts, yielding of the column
in bending may occur. This yielding can result in residual stress when
the column is cooled (see Sec. 7-2).
To determine when yielding may occur, consider the strain em in the
outer fiber. The strain em consists of the axial strain and bending strain
that
-
so
= Ae0(Wm Wmi)c
eo +
or 2=
e0
- 1 = (Wm
2
Wmi)c
(8-13)
feS-1) (8-14)
For selected values of p/2c and e„/ecr Eq. (8-14) can be plotted on Fig. 8-2
with Di as parameter. The intersection of the corresponding Dt curves
gives the temperature at which yield occurs. The case of em/e„ = 3 and
p/2c = 0.2 is shown in Fig. 8-2.
If the temperature on the cross section of the column is nonuniform,
then the column may bow as well as deflect owing to the column load.
It may also rotate (see Sec. 8-6). If the column is simply supported and
held at the ends, then the deflections can be obtained by using the bow
ing deflection in Eq. (8-4) as the initial deflection in Fig. 8-2 and using
the average temperature on the cross section as T in Fig. 8-2. Note
that, for a large temperature gradient and a small average temperature,
the stress exerted by the supports is tension instead of compression and
the deflection is less than the bowing deflection. The occurrence of
yielding will depend upon the thermal-stress distribution on the cross
section.
If
an external applied load is acting on the column, then the end sup
ports may be simply supported but they cannot be held. Under this
condition there will be no thermal load for the uniform-temperature case,
and the analysis for the applied load proceeds as for the room-temper
178 THERMAL STRESSES
ature case, with the allowable column load given in Chap. 6. For the
nonuniform temperature case Eq. (8-7) can be used to approximate the
deflection, with Wmi taken as the bowing deflection in Eq. (8-4).
Boley and Mechanic (Ref. 3) have considered column deflections when
the temperature varies along the column length as well as on the cross
section.
PROBLEMS
- rfe
mz
R* = Di (8-16)
where the length of the buckle L is taken as the width b. Thus Figs. 8-1
and 8-2 apply for this case provided D = Wm/l.2lh„ Di = Wmi/\.2\h,.
In many cases the heated plate is simply supported and held on all
four sides by the cold stringers and ribs of the structure. Assuming
complete restraint in both x and y directions, the stresses in each direc
tion will be equal so that the buckling strain is (page 334, Ref. 2)
Ui - W
'
W
0.60M1 + (Va)2P
V '
0.60/41 + (a/6)2]*
If the strain e0 in Eq. (8-8) is taken as the average of e0x and eoy,
THERMAL EFFECTS ON DEFLECTIONS AND STIFFNESS 179
then Fig. 8-2 gives the maximum deflection provided the expressions
D = Wm/0.S5h. and D{ = Wmi/0Mh. are used.
For the case in which the plate is restrained by stringers of finite area
and in which an applied load is acting, the deflection of the plate can be
obtained from Fig. 8-2 by using the strain e, in the plate adjacent to the
stringer. This strain e, is given in Eq. (7-11) in terms of temperature
change, applied load, and relative effective areas of plate and stringer.
Take ecr as in Eq. (8-15), and replace aT by e, in Eq. (8-16). Thus, in
Fig. 8-2, take RT = e,/ecr, D = WJl.2lh„ D, = W 1.21h,. J/
For yield
ing in extreme fiber from bending Eq. (8-14) becomes
(8-19)
Example 8-1. Find the deflections of the buckled skin in Fig. 7-9, (o) at t/ti = 1.0,
(6) at t/h = 4.0, and (c) after 1 hr. Assume initial deflection to be W„i/1.21h, ' 0.2C.
Solution, (a) At t/h = 1.0 Fig. 7-9 gives e,/e„ = 5.3, whence Eq. (8-12) gives
Wm/1.21h, = 2.1. Although Eq. (8-12) applies for zero initial deflection, Fig. 8-2
shows that for large e,/e„ or aT /(aT)„ the initial deflection has little effect.
(6) At t/h = 4.0, eje„ = 2.3, and Fig. (8-2) gives WJ1.21K = 0.79.
(c) From Example 7-2, it takes 13 min for the residual stresses to disappear, at which
time the applied stress is -9,600 psi. Take LM = 1.06(20 + log 1) = 21.2. From
Fig. 6-18 at / = -9,600 psi and LM = 21.2, e. = -5.5, whence, by Eq. (8-12),
PF„/1.21A. = 2.1.
PROBLEMS
8-13. In Example 8-1 the applied stress in part c increases with creep because the
effective skin decreases. Use Eq. (7-12) to correct the stress, and solve part c for the
changing stress.
8-14. Remove the applied stress in Prob. 8-13 when the 1 hr is up, and cool the
structure slowly to room temperature. How much permanent deflection remains?
8-15. Use Figs. 6-1 and 6-18 to calculate the creep life of the structure in Prob. 8-13.
8-16. Does any yield in bending occur in the structure of Example 8-1?
(8-20)
180 THERMAL STRESSES
fc3
0.
ki
fci
at edges, = 0.028.
fc,
To find the deflection and stress in
lb/in.2
p
flat plate acting as a membrane,
a
LULU -N
U
consider Fig. 8-3 for a l-in.-wide strip
J
lb/in.
Tw
1
1
W
pP d6 = 2N sin = N
^
d&
N
for small deflections. Employing
the usual approximation for p, the
— — differential equation for W
is
dx
\*
V
(8-21)
dx2 N
whence, with W = at = and =
b,
x
x
0
W =
^(b-x) (8-22)
The tension stress in the plate is/( = N fh,. To find N, equate the elon
gation of the plate due to stress to the elongation from deflection, whence
by using Eq. (8-22) in Eq. (8-6)
=
Eh, 2ilP [afir^J (8-23)
Wm =
Zi = 0.36P» 0.35P* (8-24)
h,
ducing deflection and the plate has an initial deflection of Wmi at the
if
of the form of Eq. (8-22), then the total deflection and stress can be
expressed as
Wn,
= 0.35(P PT Pi)*
+
+
h,
(8-25)
Pi =
ft.-6.88(|)'(«T). 23.5(^)3
THERMAL EFFECTS ON DEFLECTIONS AND STIFFNESS 181
2t =
(pVtW+p? (8-26)
^ =
0.35[(Pr + P,)» - (P,)»]
Note that Eqs. (8-25) to (8-27) apply to a curved plate provided the
shape of the plate can be approximated by the parabola of Eq. (8-22)
and Wmi is the center deflection of the curved plate relative to the flat
position.
When the simply supported and held plate takes the load in both
tension and bending, the differential equation for the deflection is (see
Timoshenko, Ref. 2)
dW _ NW pbx . px2 EI
dx2 r>
D on
2D 'or*
1
2D i (8-28)
_ pb4
|
cosh U[l - (2x/b)] _ + <?>-*) (8-29)
16DU* \ cosh U
From ft = N/h„ fb
= Mc/I, and Eq. (8-29) the center stresses and
deflection are
W
2, = 0.366C/2 S„ = 0.75PFi(t/) ^ = 0.142PP2((7)
_ - _ - U (8-30)
KM 2(1 sech
ip
U)
F2LU)
(gVg) 1 + sech
Although Eq. (8-6) will give the change in length of the deflection curve
of Eq. (8-29), the resulting expression when equated to the elongation
due to stress is a ninth-degree transcendental equation to evaluate U.
Instead of obtaining approximate solutions of this involved equation, it is
simpler to approximate the deflection curve itself by an equation which
will give essentially the same change in length as Eq. (8-29). From
_ _
Timoshenko (Ref. 2) an approximate deflection curve for this case is
tx W2
w
W-^rsmT
W0 .
7-1 =
Nb2
= ,Q
(8-31)
where W0 is given by Eq. (8-20). With this deflection curve, Eq. (8-23)
becomes
F" - Y' = 0.061P2 (8-32)
182 THERMAL STRESSES
s = si + (si2 -
0.00137)5 (8-33)
si = 0.031P2 + 0.037
~=
(?)
St = 0.055 S„ = 0.75 0.142 (8-34)
the plate has initial deflection or curved so that its shape can be
If
is
approximated by Eq. (8-31) and subjected to temperature change,
is
a
then Eq. (8-34) becomes
(8-35)
Yi
Pt
negligible).
All the stresses and deflections of this section are expressed in terms
of parameter P, which involves E. If the stresses are inelastic, then
a
trial procedure.
a
PROBLEMS
8-17. If P > 100, show that Eq. (8-34) reduces to Eq. (8-24) plus a bending stress
of Xb = 1.55P*.
8-18. long 0.064-in. aluminum plate 10 in. wide loaded by a uniform load of
A
is
deflections.
8-21. Solve Prob. 8-19 with the condition in Prob. 8-20.
8-22. Derive Eq. (8-23).
THERMAL EFFECTS ON DEFLECTIONS AND STIFFNESS 183
while compression
loads will tend to increase
it,
each (y-axis]
producing normal component tend Fig. 8-4. Plate warping due to thermal
a
5).
From Fig. 8-4 there are bending moments and torsional moments
tending to deflect and twist the plate cross section. These moments at
= L/2 are
x
-
4
L
J
W, = WdA (8-36)
1/ /1
'
No = - 6r2fdA = 6dA
J
(8-37)
184 THERMAL STRESSES
Wm =
1 - (pJs/SEI) °m =
1 - WZ/4GJ)
(8-38)
If Wmi and $mi are the initial deflections due to either initial warp or to
applied load, then by Eqs. (8-38), depending upon the signs of P& and No,
the thermal loads will either decrease or increase these deflections. On
short plates these effects will be reduced because of the wall restraint
on the deflection of the cross section.
As in Secs. 8-3 and 8-4, the thermal loads and moments in Eq. (8-38)
depend upon the strains so that Eq. (8-38) can be written as
Wm
1 - Wmi
(e0/ecr)
&m
1 —
dmi
(e0c/ecrc)
^8 ^
where eo = median plane strain at center of rotation of plate
For a moment acting in the plane of the plate the critical buckling
moment in the lateral instability case (Ref. 2) is
M„ =
I (EIJGJ)*
,
whence = — —
0.63r07r / IjV ...
ecrc I
-p- J (8-40)
^ =
-AIT
=
—±L~
(«D.
Rt —
ecrc LI - (eoc/ecrc)} 2L(ecrc)»
Just as for the column, there is no failure of the plate in torsion, although
yielding may occur (Fig. 8-2).
In writing Eq. (8-39) it is assumed that the strain e0c on the hot edge
of the plate determines the rotation regardless of the temperature dis
tribution. This is true for a linear temperature distribution because
both the temperature strain and the moment strain for buckling are linear.
However, the assumption is an approximation for other temperature dis
tributions unless the correct strain ecrc for the stress distribution is
obtained. It is possible to evaluate e„c = (aT)crc from Eq. (8-38) by
taking
- 4GJ (8^2)
(aT)crc
whence
Suppose the plate in Fig. 8-4 is allowed to rotate at the ends about
the axis of rotation for 6 but is otherwise held and simply supported.
If dm is the angle of one end relative to the other and the hot edge of
the plate remains straight, then
AL = \U + (r0ft.)*]»
-L= Mm)2
2L
Hence Eq. (8-41) applies to this case with
If the plate has local buckling on theedge, then AL is larger so that from
Fig. 8-1 the strainis smaller.
e0c
2Wm2 „ TWm
for vibration in bending and in torsion; for static torsion use Eq. (8-44).
For the cantilever plate with applied load Q, RT in Eq. (8-41) becomes
(aT)oe = jJ aTdA
(8-46)
frequencies in this case may be considerably lower than in the first case
(Fig. 8-5). Further, if the amplitude is sufficiently large to produce con
siderable yielding, then the thermal stresses may be reduced and the fre
quency increased. It appears that both the soft-spring and hard-spring
(Ref. 13) effects may be present, in which the frequency decreases with
increasing amplitude for the soft spring and the frequency increases with
increasing amplitude for the hard spring.
Test data by Vosteen and Fuller (Ref. 6) on a cantilever plate heated
on the edges show increases in torsional deflections with and without
applied torque and decreases in both the first-bending and first-torsion
vibration frequencies. See Sec. 8-8 for comments on calculated and test
data for this plate.
PROBLEMS
8-26. A circular cantilever rod has a temperature of 0°F for 0 < r < 0.8a and a
temperature T°F for 0.8a < r < a. If the initial deflections are W„
that Eq. (8-38) gives
Wm = Wmi I
1 - 0.58ar
8-27. Let the plate in Fig. 8-4 be 40 in. long, 10 in. wide, and | in. thick. The tem
perature varies linearly from 80°F on one edge to 380°F on the other. Assume the
cross section to be unwarped but to have initial deflections at the center of 0.1 in. and
1.0°. Find the deflections produced by the thermal stresses for an aluminum-alloy
plate.
8-28. Solve Prob. 8-27 for a steel plate.
8-29. Solve Prob. 8-27 for a 20-in.-wide plate with 5 in. on each edge at 380°F and
the middle 10 in. at 80°F.
8-30. Solve Prob. 8-28 for a cantilever plate.
THERMAL EFFECTS ON DEFLECTIONS AND STIFFNESS 187
7 = 2ed
- eb (8-48)
where ed is the strain in the diagonal and eb is the strain in the chord.
Similarly the shear deflections due to thermal stresses in a buckled skin-
stringer panel can be approximated as
yT = 2(a,T, - abTh)
- eb (8-49)
where eb is the strain in the stringers. For combined thermal and applied
axial stress is given by Eq. (7-11).
eb If 100A; is the per cent of the shear
flow q (in pounds per inches) carried in tension, then the total shear
deflection is
2.5q ,
7 = -Tpr f or q < qcr
tE
(8-50)
7 = kyr +
m [2-5(1 -k) + 2fc(1 + k)] for q > Qcr
7 - L ~
G
= 2j^1
tE
it follows that the effective skin thickness for the diagonal-tension case
with temperature is
7
=
1 - k + 0.80fc(l + k) + OAOktEyr/q
^8-51')
=
33.64 (ElV 15
(GJ\ (a
Uh
-L^\m) Wa==L\Tj (8-52)
J* =
(i - isY «- =
fi _ (8-53)
who \ ecr/ £oao \ ecrc/
1.00
0.40
0.20
0.10
0.05
0.2
0.02
0.00
shows a graph of Eq. (8-53) against RT [Eqs. (8-41) and (8-46)] with
Dt [Eqs. (8-44) and (8-45)] as the parameter. It is apparent from Fig.
8-5 that large reductions in the frequencies can occur owing to thermal
stresses. Ordinarily the applied axial load Q [Eq. (8-46)] is zero in the
vibration components of the airplane or missile.
Example 8-2. Determine the deflections and change in natural frequencies of a
cantilever aluminum-alloy plate 40 in. long, 20 in. wide, and 0.25 in. thick. The
THERMAL EFFECTS ON DEFLECTIONS AND STIFFNESS 189
center 16 in. of the plate is at 0°F, and the edge 2 in. are both at 300°F. The initial
deflections are Wmi = 0.10 in. and 9m< = 1.0°.
Solution. From Eq. (8-46)
(aT)oc = %
0
(300)(1) - 60a
For the torsion calculation Eq. (8-43) gives
(aT)eOJ ot(0.40)(300)(20)(l)73
fr'fdA _
^JT8
J™
2(i)a dr 24Qr,
dr^
60).2
= 0.86(10-')
, .
n
Di =(10)(1.0)(100) " nil
011
. ..
f0r defleCtl0n
(57.3)(40)(17.4)i
n = x(10)(1.0)(100) ftlo , ., ..
Di = 012 f°r Vlbratl0n
(57.3)(4)(40)(8.6)*
whence Fig. 8-2 gives D = 1.60 and 6m = 14.5°. With large deflections and 0ml- as
equilibrium point Fig. 8-5 gives ua/a>ao = 0.28; with 8m as equilibrium point wa/wao =
0.80.
For the bending deflection Eq. (8-36) giveszero for Pb, and hence Wm = Wmi,
unless the plate warped. Let the given = 0.10 in. be an average of 0.08 in. at
is
Wmi
the center and 0.12 in. deflection at the edges. Then approximately-
j
j-
= x(0.10)(100)
Ui U-U44
(4) (40) (20.5)4
whence Fig. 8-2 gives
D = 0.73 Wm - 1.7 in.
the warping reverses in the bending-vibration case when passing through zero
If
deflection (the presence of torsion deflection tends to produce such reversal), then
a
the bending frequencies will be affected. Figure 8-5 gives wj/o>*o = 0.25 for Rt = 1.5
and Di = 0.04. For small amplitudes about Wm there essentially no effect, and
is
w»/w*o = 1.0.
Example 8-2 indicates that the bending deflection and vibration are
quite sensitive to any warping of the cross section when large thermal
190 THERMAL STRESSES
stresses are present. Also it appears that both the bending and torsion
are sensitive to the distribution of the thermal stresses. The expression
for ecrc in Eq. (8-43) may be large or small, depending upon the stress /
and the stiffness GJ. It is possible for the stress distribution to make
the denominator in Eq. (8-43) negative, zero, or positive, so that the
thermal-stress effect is destabilizing, neutral, or stabilizing. In the latter
case the frequencies increase, and the deflections decrease. During a tem
perature and thermal-stresscycle in which the edges of a plate are heated
rapidly and then allowed to cool, the torsion frequency may decrease;
then recover some when heating stops on the edges, then decrease again
as the stress distribution changes in the plate, and finally recover as the
entire plate cools. It is also possible for some recovery to arise if the
edges buckle locally, decreasing the destabilizing load. If the stress in
any part of the plate exceeds the proportional limit, the deflection
recovery and the frequency recovery may not be the same.
Test data by Vosteen and Fuller (Ref. 6) on a cantilever plate heated
on the edges from 80 to 300°F in 16.5 sec and then allowed to cool show
a decrease in torsional and bending frequencies similar to Fig. 8-5, some
recovery in torsional frequency when heating stops, then decrease and
finally recovery as the plate cools. Torsional deflections occur with and
without applied torque, following Fig. 8-2. Recovery of the deflection is
not the same as in the torsional frequency, owing to some buckling in
the edges of the plate. If initial warping and deflections are estimated
and some allowance made for the effects of buckling, increased GJ due to
short length, and shear lag (the plate is only 20 in. long for a width of
20 in.) upon the thermal stress, then the deflections and frequencies
obtained in the test can be checked approximately by the procedures of
Example 8-2.
It must be emphasized that these effects on the deflections and fre
quencies arise because of the thermal stresses and their distribution. If
the structure has no thermal stresses or the stresses are distributed
properly, there are no effects, except because of change in elastic E and G
with temperature or change in /
and J
due to effective areas. However,
thermal are present in airplane and missile structures under
stresses
transient conditions and in many cases are distributed so that the above
destabilizing effects are produced. In a solid double-wedge wing, with
the temperature distribution shown in Fig. 5-1, these effects may be large.
In a box-structure wing with thermal stresses in the skin and stringers,
as discussed in Chaps. 3 and 4, there is ordinarily very little destabilizing
effect [the tension loads are adjacent to the balancing compression loads
so that the denominator of Eq. (8-43) is approximately zero]. In a
multiweb wing with tension in the webs and compression in the skin
there can be small destabilizing effects. See Ref. 8 for examples of
various types of structure.
THERMAL EFFECTS ON DEFLECTIONS AND STIFFNESS 191
Budiansky and Mayers (Ref. 10) have considered the effective torsional
stiffness of thin wings for various flight conditions. Since they use the
equivalent of Eq. (8-38) rather than Eq. (8-39), they obtain large reduc
tions in GJ, with (GJ)e becoming zero in many cases. This would give
infinite deflection and zero torsional frequency. Such results are possible
if the load is an applied load, but thermal loads are internal self-balancing
loads produced by strain. And this strain is a finite amount, determined
by the temperature distribution.
As noted at the beginning of this section, the flutter speed of the wing
or tail of an airplane or missile depends upon the bending and torsional
frequencies d>h and coa. For a dense wing with coh/oja small Theodorsen
and Garrick (Ref. 11) give the following approximate formula for the
flutter speed of a wing in bending-torsion flutter,
bwa
- r-'jndhiT (8-54)
where = flutter
Vf speed
b = semichord
o)„ = torsional frequency
a = ratio of distance from elastic axis to mid-chord b
e- - iHrao +
C»T
-
D~-mbr> <*«>
PROBLEMS
8-31. Solve Example 8-2 for a triangular temperature variation on the edge 2 in. of
the plate, starting at 0°F 2 in. from the edge and going to 300°F at the edge.
8-32. In Example 8-2 assume the symmetrical temperature distribution is 0°F for
a distance of a in. on the edge, then 300°F for 2 in., and 0°F the rest of the way to the
center of the plate. For what value of o will the thermal stresses produce no effect
on the torsional stiffness? Is it possible to select a so that
the stiffness is increased?
8-33. Assuming Eq. (8-54) to apply, what are the nutter speeds of the plate in
Example 8-2?
8-34. Use Eq. (8-54) to plot a curve of flutter speed against the number a in Prob.
8-32.
8-35. Use Eq. (8-55) to determine what value of applied twist angle must be used
in Example 8-2 to remove all thermal stress at the edge of the plate. Does yielding in
the plate occur in this case?
8-36. In a skin-stringer panel the O.lO-in.-thick skin is at 400°F, and the stringer is
at 150°F. If the material is aluminum alloy, the stringer strain is zero, the shear flow
is 1,000 lb /in., and all the shear is carried in diagonal tension, find the effective skin
thickness.
REFERENCES
1. Levy, S.: Thermal Stresses and Deformations in Beams, Aero. Eng. Rev., 15(10) :
62-70 (1956).
2. Timoshenko, S.: "Theory of Elastic Stability," McGraw-Hill Book Company,
Inc., New York, 1936.
3. Boley, B. A., and Harold Mechanic: Thermo-elastic Stresses and Deflections in
Beam Columns, WADC TR 54-425, March, 1954 (ASTIA AD No. 75595).
4. Boley, B. A., and I. S. Tolins: Thermo-elastic Stresses and Deflections in Thin-
walled Beams, WADC TR 54-426, December, 1954.
THERMAL EFFECTS ON DEFLECTIONS AND STIFFNESS 193
Fourier-series analysis, and some of the sets of functions that arise are
the trigonometric functions and the Bessel functions. This procedure
is
V =
/// V0 dx dy dz (9-3)
Now several theorems and principles can be derived from the strain
energy and the forces acting on the body: principle of virtual work,
theorem of minimum potential energy, Castigliano's theorem, principle
of least work, theorem of minimum complementary energy. See Sokol-
nikoff (Ref. 2) and Timoshenko and Goodier (Ref. 3) for derivations of
these theorems. The minimum-potential-energy theorem may be used to
demonstrate the variational methods. It may be stated as follows:
Those displacements which satisfy both the boundary conditions and the
equilibrium conditions will make the potential energy a minimum. This
problem of determining the displacements to make the potential energy
a minimum can be attacked by the calculus of variations. The calculus
of variations can be used to convert the problem into an equivalent prob
196 THERMAL STRESSES
p =
nnx
where s =
s(x,y), and sx = ds/dx, etc. Now replace s(x,y) by s(x,y) +
en(x,y), where n(x,y) is any function that vanishes on the boundary of
the body and e is an arbitrary small parameter. If s{x,y) makes P a
minimum, then this minimum occurs when e = 0 so that dP/dt = 0 at
e = 0.As shown in books on calculus of variations and in Ref. 2, this
condition gives
IJL(8)n(x,y) dxdy = 0 (9-5)
where L(s) = — -— —- —— + — —
ds dx dy dsy dx2 dsxx
il
dsx
,
2 dF_ dF_ = 0
(Q-6)
dx dy dSxy dy2 dsyy
<t>
by Eq. (1-41) for the two-dimensional case, then Eq. (9-6) gives Eq. (1-42)
and the minimum-energy theorem leads to the same solutions as the equi
librium and compatibility conditions do. However, the minimum-energy
theorem can be used for obtaining approximate solutions by several
schemes.
The Rayleigh-Ritz method (Ref. selects certain functions with
2)
N N
uN = anRn(x,y) vN = bnQ,>(x,y) (9-7)
^
n=l n=l
„-l,2
g
= = N (9-8,
0
0
SOLVING THE THERMAL-STRESS EQUATIONS IN TWO DIMENSIONS 197
where fn(x,y) satisfy the boundary conditions on s and Eq. (9-5) becomes
//L(Sjv)/„ dx dy = 0 n = 1, 2, . . . , N (9-10)
= Kx)g(y) (9-12)
so that from Eq. (1-41)
= fg" = f"9 = -f'g'
U
(9-13)
U
rxu
theorem to obtain a function that gives the best approximation for the
/
Since has been shown in Sec. 5-2 that in long plates the thermal stress
it
the most logical selection for the function the one-dimensional solu
is
g
198 THERMAL STRESSES
- 2vAiff"
ja_a
2EV =
{A^ + A2(/")2 + 2(1 + k)A4(/T
+ 2Ea[(Ab AtX)f (A7 + AJOf"] dx
+
(9-15)
)
where Ai =
j^{g")2 dy A2 = J^g2dy
= = \gg'\\ - A,
A
jb_bgg"dy
3
A< =
/!>')* dy
=
Wit*
A 5
A,
jb_b
Ae = A7 = T„ = Yg dy
f'b
Yg" dy dy
g
If in Eq. (9-4) = and given by Eq. (9-15), then Euler's equa
is
F
f
s
tion (9-6) gives the following ordinary differential equation for deter
<t>
mining the function for the best approximation for the function iQ
/
Eq. (9-12):
A4"" - + v(A3 AW +=
+
A
J
2[A<
-Ea(A, + A6X + AsX") (9-16)
The boundary conditions are
= fg" = rxy = -f'g' = on x = +a
fy fx
0
0
=
f"9 =
t„ = -f'g' = on = ±b
0
0
/
(9-18)
Equation (9-18) can be solved for any given X in Eq. (9-11), whence the
stresses are given by Eq. (9-13).
Heldenfels and Roberts (Ref. have solved the above equations for
4)
= T0 - < <
+
(l
T
|j
T1
0
b
y
(9-19)
- To + rl + -b<y<0
T
(l
Jf
SOLVING THE THERMAL-STRESS EQUATIONS IN TWO DIMENSIONS 199
+
C3 C4
+21 + (1,365)*
1
fci2
(9-20)
\
1
fc2s 1362
J
C2 = C3 =
0
fc2
DC = — fci cosh fcja sin fc2a — sinh kia cos fc2a
fc2
4
A;2a
With this and the stresses in Eq. (9-13) checked the test data (Ref.
4)
/
±
a
PROBLEMS
9-1. Derive Eq. (9-15).
9-2. Derive Eq. (9-16).
9-3. Write out the expressions for the thermal stresses for the functions and in
g
/
Eq. (9-20).
9-4. Draw curves for the stress distributions in Prob. 9-3 for the case of a = 18 in.,
= 12 in.,Ti = 150°F, E = 10.4(10«) psi, a = 12.7(10-') /°F.
b
9-5. Solve Prob. 9-4 for the case of a — 48 in. and determine how far the shear-lag
effects reach from the ends of the plates (see Sec. 4-6).
differential T\„- If
the deflections of the plate are assumed to be small,
then the stresses are given by Eqs. (9-13) and (9-20). The deflections
of the buckled plate are given by the equation (Ref.
6)
dW Eh,3
-
d2W
/
xy (9-21)
dx dy 12(1 v2)
mirx
amn cos
2a
and the Rayleigh-Ritz method (see Sec. 9-2) used to obtain set of
is
simultaneous equations for the amn. For buckling the determinant of the
200 THERMAL STRESSES
coefficients must equal zero, which gives (Ref. 5) for the case of a/b = 1.57
= 5.39
(9-22)
After the plate buckles, the deflections may be large so that the stresses
will change owing to stretching of the plate mid-plane surface. In
this case, Eq. (1-42) becomes (Ref. 6)
and W is given by Eq. (9-21) provided the in Eq. (9-21) are based
stresses
on Eq. (9-23) by using Eq. (1-41). This means that Eqs. (9-21) and
(9-23) must be solved simultaneously. In Ref. 5 the solution is made
by using the Galerkin method (see Sec. 9-2) to obtain the center deflec
tion Wm of the plate as
b2E£pih'
= 5-39 + 1.12(1 - v2)
^4 (9-24)
°
™lD
= 5-39
(
1 - W^j + 1-12(1
- "2) h, (9-25)
— i _ .
'
9-2
(W
,
w n
aT^ Wm L2aTUr
Except for a different value of D,-, Eq. (9-26) is the same as Eq. (8-11)
[see Eq. (8-16) for Dt for restrained plates at uniform temperature], and
its graph will be the same as Fig. 8-2. This indicates that it may not
be necessary to solve the large-deflection equations (9-21) and (9-23) for
the thermal-buckling and -deflection problem. It is necessary only to
solve for the buckling temperature in the small-deflection equation and
use that result in Eq. (8-11). However, Eq. (8-11) is based on the
average net compressive strain (see Sec. 9-4, where RT = e,/eer for the
skin-stringer problem) so that an average net aT\ for the compression
SOLVING THE THERMAL-STRESS EQUATIONS IN TWO DIMENSIONS 201
strain with the corresponding aTicr should be used in Eq. (8-11) with
Di in Eq. (8-16). For the temperature distribution of Eq. (9-19)
(aTi)„ = 0.25aTi, whence A
in Eq. (9-26) becomes Wmi/l.lQh,. This
compares satisfactorily with the restrained plate value of Wmi/1.21h, in
Eq. (8-16).
Test data in Ref. 5 for deflections of a 24- by 36- by 0.25-in.-thick plate
check Eq. (9-25) closely for increasing temperature, and the resulting
curve follows the corresponding curve in Fig. 8-2 closely.
9-5. Methods Using Complex Variables. It was shown in Sec. 1-6
that the thermal-stress problem in two dimensions could be reduced to
solving the system of equations [Eqs. (1-43) and (1-44)],
V2F = kT VAU = 0
'
dU , .dU .
f , .v. , , dV ,
1
dV (9-27)
dx dy J„ dy dx
with Q(x,y) the function conjugate to the harmonic function V2U. Define
^i(z) = ^2(2)) an(i differentiate Eq. (9-28) to get
1ST
+
l%
=
*l(z) + z*[{2) + Ui) (9-30)
*u *u ,w *M
+ i = + + (9.31)
dx dy w (s)
L
H + i^ = g{6) + ih(6) on (9-32)
^+'7i =
^)+^K0+*(0 (9-34)
Let p(s) be any polynomial which has no zeros inside L and which is such
that p(s)w(s) has no poles outside L except perhaps at infinity. Then
fit will be assumed that w' (s) is not zero either in the unit circle S' or on its circum
ference L. For the region S finite let w(0) = 0.
JV. Smirnoff, Math. Ann., 107:313-323 (1933). In fact Smirnoff proves these
results under less restrictive conditions than those given here.
SOLVING THE THERMAL-STRESS EQUATIONS IN TWO DIMENSIONS 203
V = A Tw'{t)w'{t) In p
A* / 8' J dm (9-36)
s
it\
£i
£
t
= — t); —
here has the arguments
p2
t)(s T
(s
dV _ dV = _
Jo f1
f2* Tw'(t)w'{t)R dR
k
d<t>
—
Jo
dx dy 2xw/(s)
s
t
appearing in Eq. (9-37) in form to which the formula of Cauchy and the
a
dition of the form given in Eq. (9-33), then Eq. (9-37) can be written in
the following form,
*r _ _ TW'(twmdRdtt
-
(('
=
+
Jr f
/')
\Jo
f
dx dy 2«w'(s) J\t\ =
s)
<(<
R
<
|s|
\t\
\t\
The function <t>i(z) defined in Eq. (9-29) and used in connection with
the biharmonic function in Eq. (9-28) will be examined next. The
U
definition of 0i(z) given in Eq. (9-29) does not determine <t>i{z) com
pletely from V2U. The function Q(x,y) conjugate to V2U determined
is
204 THERMAL STRESSES
except for a real constant, whence <pi(z) is determined except for a term
(a + ib) + Biz, where a, b, and B are real constants. Since for the
calculation of the stresses it is necessary to know only the expression
V2£7 = /,+/„ + V27 [see Eq. (1-41)], it follows that <t>i(z) may be sub
jected to the conditions
0i(O) = 0 Re = 0
4>[(0)
j (9-39)
The expressions for the stresses can be readily obtained in terms of the
functions </>i, and V. They are
U
,
+
.
W =
3m — 4
-2rtT
, , .
- ecz
m
- 1
2G
■«»+*<» +
(9-41)
where m = 1/v. The expressions for the stresses and displacements in
terms of the polar coordinates r and 6 of the s plane can be obtained by
using the following transformations:
. . sw'(s) , . . , . = . .
Ur + W" = ^M + ^ /'+/» /»+ /„
rite7 (a) I
/.-/r + 2trrf = £^
S2w/(s)
(/, - /, + 2irw) (9-42)
r2tD'(s)
Now consider the form of the functions <£l(2) and ^i(2) in the expres
sions for the stresses and displacements when the cross section of the body
is multiply connected. Let the region be bounded by n + 1 simple closed
curves Co, C\, . . . , C„, which have no common point and with C0
enclosing all the other curves. It is permissible for Co to recede to
infinity. Since the physical considerations demand the stresses, the dis
placements, and the temperature to be single-valued in S, it is necessary
to require that the right-hand members of Eqs. (9-40) and (9-41) be
single-valued. Hence from Eq. (9-40) it follows that Re [<t>[(z)] = |V2(7
must be single-valued. However, in this case, the function Q(x,y) conju
gate to V2C may not be single- valued and, in fact, may increase by a
constant value on going once around a simple closed curve C'k enclosing
C*(fc= 1,2, . . . , n). Designate this increase by iAk (Ak real), whence
n
<S,
where <t>f(z) and analytic in
is single-valued except perhaps at infinity,
and Ot, are real constants. Integration of the terms of Eqs. (9-40),
bk
(9-41), and (9-27) around the closed curve C'k enclosing Ck shows that
ak
jpj
= - - iY*
)C4,
+ CX. (9-43)
f <
<
X* + iYk = - (X + iY) ds
(
*iC0 =
- (X* iYk) ln - *) *f (z)
+
+
(z
£
1}
8^(mm_
n (9-44)
8^141}
and ^* (z) are single-valued and analytic in
S,
(9-45)
9-6. The Hollow Circular Cylinder. Let be the region between two
S
region on a ring formed by two concentric circles about the origin in the
plane; let C0 go into circle of radius C\ into circle of radius
B
A
b,
a
s
< b), and region into the circular ring between circles and B.
A
(a
Now the boundary condition in Eq. (9-27) must hold on C0 and Ci, whence
the condition (9-34) must hold on the circles and B. For this case
A
=
** for on = l>2
(f
*'
+
+
wwm
*
♦«>
i
A<>
)
(9-46)
Since the region multiply connected, the functions and \p(s) will
is
4>(s)
206 THERMAL STRESSES
where <t>* and \p* indicate the terms of the Laurent series which have
nonnegative exponents, while </>* and \p* have negative exponents. In a
way analogous to that for Eq. (9-34) the Cauchy integral formula can be
applied to Eq. (9-46) to obtain the functions 4>*, yp*,
HlW = _ y
Zy 4(n
—
1)
4.
z"-1
0lW = y a- vf;
Zy 4(n + 1)
(9-49)
n=2 n=l
Differentiation gives
ay
A0);
= lnr +
da;
+
^(Ci + fDi) |(Alnr-^+
(Ci - tDO + ffiW + zG[(z) + Gi(z)]
+
(9-50)
j
zero, and make use of the mapping function = to give the following
z
\dV .dV =
,
1,
(9-51)
2
i
J—.
the mapping function w(s) = s, and multiply Eq. (9-46) by
Then integrate the first equation = of Eq. (9-46) and its conjugate
(i
1)
over circle for exterior to A, and integrate the second equation
A =
2)
(i
s
of Eq. (9-46) and its conjugate over for interior to B. This gives
s
four equations for determining the functions 0P, <t>d,ipp, and \pd defined in
Eq. (9-47). The results are
- -
A(b2\nb
- In
A
a)
a2
<t>P(s)
= kG(s) + "FT + ^0
-A
- iD1)s2
b2 a2
2
. k(d
+
l
4(a2 b2)
(in
Ms) (Ci
b
I
s> 1
US) =
[
,
,
J
where G(s) = Gi[w(s)], H(s) = Hi[w(s)]. Use = w(s) = in these
s
z
functions to give <t>i(z) and ^i(z), and then determine from Eq. (9-28).
Put the stress function U
= — in Eq. (1-49) to obtain the stresses
V
U
0
, kA a2 b2 - r2 . b\
(.
b
=
-2ynr-T2¥^2lna)
fr
- fl2)
(Cl c0S Di Sin
+
2^
d)
d
a2^ b2
- (h2 — rJVr! — n2\
-
1c
T">
6
275 6,
A
+ r2,
/.
JbA a2b2
6
6
,
/.
4a2r2 - (3r2
- + _ „,
fc
62)(r2 a2) n
Di sin
.
,
(Ci cos
+
0)
0
2r-3 a2 + 62
PROBLEMS
9-6. Compare the results of this section with those in Prob. 1-36. Discuss reasons
why only a few terms in steady-state temperature distribution in Eq. (9-48) contribute
to the stresses in Eq. (9-53).
9-7. Use Eqs. (9-41) and (9-42) to determine the displacements for the hollow
circular cylinder of this section.
9-8. Use Eqs. (1-46) and (1-47) to find the axial stress and axial strain in the hollow
circular cylinder of this section.
9-9. What are the maximum stresses in Eq. (9-53), and where do they occur?
9-10. Solve Prob. 5-12 by the method of this section.
208 THERMAL STRESSES
fiy
way that every material completely surrounded
n)
is
1,
2,
(J
. . .
,
by Cy, the outside
Sj
by material So, and designate the boundary of
boundary being Co- In the region the physical constants are Ej, rrij,
Sj
and ay.
The materials are assumed to adhere on all junction surfaces so that the
displacements will necessarily be continuous on the curves Ci, . . . C„.
,
further condition along = obtained from the con
Cy
n)
A
is
1,
(J
. . .
,
tinuity of the normal stress across Cy, while on C0 the condition of Eq.
Cy
Vj
(9-27) must hold. Let be the biharmonic function in Sj, the heat
function in Sj, <fei(z) and &i(z) the analytic functions in
Sj
defined by
<t>
Eq. (9-28). Then, in terms of the functions and the above three
ip,
conditions can be expressed as follows,
+ z^i(z) = on Co
+ +
2,
j
1,
. . . (9-54)
,
(tfy l)*yi(z) + (Mj + 1)M«0 + «*'oi(2) + <Mz)
= RjZ + f0j(s) on Cj, = ,n
2,
j
1,
. .
.
where arc length on Cj, and
is
s
=
^ onC'
+
^(s)
laF
- -
^ _~ - _ -
4Gj(m0 ,
1)
1)
~~ 4gyfrny
1
^~bb>
m0(G„ Cy)
- my(Go G,)
^ _ - Gj)
GoGj(m0 my)
"
2'
* ' "
3
m„m,(G„
'
It
should be observed that the conditions (9-54) become very much
simpler when assumed that the elastic constants and m are the
is
it
•
•
S
*S„,
ficient of thermal expansion different in each region. Equation (9-54)
is
reduces to
+ foi(z) = /oo(s) on Co
+
<£oi(z) z4>'01(z)
-
=
-
Cj
^oi(z) Cy, . .
.
if
<fei(z) </>oi(z).
be determined from Eq. (9-35), then known that its first deriva
V
it
is
right-hand side of the third equation in Eq. (9-56) is zero and &i(z) is
the analytic extension of ^oi(z). Hence for this case the function U is
continuous throughout the region so that the problem is that of the
simply connected region of one material with the boundary condition
as the first equation in Eq. (9-56). If the region S be mapped con-
formally on the unit circle by a rational function and if foo satisfies con
ditions (9-33), then the functions <£oi(z) and ^oi(z) can be determined by
the method of Sec. 9-5.
Consider the case of a long circular cylinder of radius b with a con
centric core, of radius a, of a different material. Let B be the outside
circular boundary, and let A circular boundary between the two
be the
materials, while Si is the material of the core and $0 is the material of
the ring. Take Ei} rrii, ai, and ki} where = £fm,a,/(mi — 1), as the
constants of material S< (i = 0, 1). Assume the temperature T to be a
continuous function of the radius r, and designate by Ti(r) the temper
ature distribution in Si (i = 0, 1). As is customary, assume the external
surface forces to be zero. Use the mapping function z = s, and deter
mine the first derivatives of V from Eq. (9-38)
(9-57)
(9-58)
2a, + S - k j rTi{r) dr -
' (9-60)
/r„ =
r r jo § Ja/ rr0(r) dr T
rr,0 = 0
2a, - ^ + ^
§ Ja/ rT0(r) dr -
'
feo
=
/ rT,(r) dr + k0T0
r r jo ~
dr = 0
f'f.ir dr + j*ft(lr (9-61)
- k0T)
(9-62)
= A with A = £(4a,
feo
(l +
^
The more general problem of m + 1 concentric materials is considered
in Ref. 13.
PROBLEMS
9-11. Show that the constant A in Eq. (9-62) has the value
B = —
E0H)mi(ma + 1) 4- E0(m,i — ro0)]
HEmnnJUEi
-
[HEam0(mi + 1) -
£,fn,(m0 + l)][H£om0(m, 2) 2£lm,(m0 - - - 1)] (9-63)
'-'-5
9-12. If the inside material is steel with m1 = 3, = 3(107), a, = 6(10-«) and the
outside material is aluminum alloy with too = 3, 2?o = 10', <*o = 12(10-«), show that
for T constant
fn = fei = AH /ro = AH on r = a
= A(2 — H) on r =
- 2H)T
feo a
, _ „, 540H(3
SOLVING THE THERMAL-STRESS EQUATIONS IN TWO DIMENSIONS 211
9-13. Solve Prob. 9-12 for b2/a2 = 4, and determine the maximum stresses.
9-14. Compare the axial stresses in Prob. 9-13 with those given by the formulas of
Sec. 1-2. What causes the difference in the results?
9-15. Evaluate the stresses in Eq. (9-60) for the case of both cylinders of the same
material but with 2\ and T0 different constant temperatures.
9-16. Show that the radial and axial strains for the case of constant temperature
[Eq. (9-62) and Prob. 9-11] are
mi
= AH(mi - 2) e, mi + 1
<*\i
rnjan
7, 7, 1 (9-00)
r 2m16i mi mi
»0
r 2Go L \ f I ™oJ m0 m0
9-17. Find the displacements in Prob. 9-16 for inside-material steel, outside-
material aluminum alloy, both at 500°F above datum. Take o = 0.26.
9-18. Suppose there is a screw-thread joint between the cylinders. What happens
to the stresses in this case, particularly if the joint is loose? If there is a radial gap of
0.005 in. and an axial gap of 0.010 in. in the joint, what temperature change is per
missible in Prob. 9-17 without producing stress? Use o = 0.25 in. and the length of
the thread as 2 in.
z = =
(9-66)
maps the region on two circular regions concentric to each other, the
circle C0 going into a circle B of radius b and the circle Ci into a circle A
of radius a concentric to B. Designate by Si the region in the s plane
which Si is the transform of and by s0 the transform of the region S0.
It is easily checked that the point 1/c is outside both circles A 'and B
so that the mapping function [Eq. (9-66)] has no pole in the region. Let
the elastic constants of the two materials be the same, but let the coef
ficients of thermal Assume the temperature T
expansion be different.
to be constant. Then a particular integral of V2F = kT is V = \kTr2,
whence in the s plane
<7oo
= ykoTw(t) for t on B
0oi = \kaTw(t) for t on A
ffii =
%kiTw{t) for t on A
/<3
where
Eq. (9-55). To obtain the biharmonic functions Uo and Ui in regions
So and Si, respectively, use Eq. (9-56), which determines the functions
4>i(s)
= 4>n(z), <tt>o(s) = tf>oi(z), ^i(s) = ^n(z), iAo(s) = iAoi(z). The func
tions <t>i(s) and ^i(s) will have Laurent-series developments in s0. From
Eq.(9-47) write <£0(s) = + iKi(s). and \p0(s) = iAop(s)
+
<t>op(s) <*xw(s)
Now integrate the equations in Eq. (9-56) and their conjugates by the
method of integration discussed in previous sections to obtain the follow
ing results:
= = in Si
0
= - - -
(p
+
k0
fci) WW
s)|
<t>oP(s) (fc0 2(fco
^'(cfr2)]
^
|
. a2
,
n
,
lAod(s)
= (fc0
- fci)Tw ^i(s) = ^03,(s) in Si
By using Eqs. (9-40) and (9-42) the stresses can be calculated in terms
of the coordinates and of the plane,
r
a2c2p
_ -
- a2)cr cos +
+
2b4[b2 (fr2
0
a2c2r2]2|
- cWr2)(s -
2cra262 cos
+
(6* a4cV2)2
0
,
- Tn
-2^-
,,a2L - 62\
- - cr2)2 __.
•
}62(62s2
(U
T2,)
V1
- - - caV2)2(S ca2)2
\
_ _
L(l["(1
s2
- c2b2r2
- fr2c2)2
- b4(l cs)2
- a2)(l - -
(s
_ 2cfr4s(b2
- - cs)(s cr2))
(s
c62)(62 ca2s)3
(
It clear that =
then the mapping function [Eq. (9-66)] w(s) =
is
if
is
0
s,
c
and the stresses reduce to those for the concentric cylinders of Sec. 9-7.
The stresses in the region Si can be calculated in way similar to those
a
for the region sa. The stresses in the coordinates of the plane can be
z
SOLVING THE THERMAL-STRESS EQUATIONS IN TWO DIMENSIONS 213
• 4
calculated by transforming the functions of Eq. (9-67) to the z plane by
s = z/(l + cz) and using Eq. (9-40).
It should be remarked that the most important restriction on the
complex-variable method in Secs. 9-5 to 9-8 is that the mapping function
be rational. The primary purpose of this restriction has been to provide
one with a system of equations that could be solved explicitly for the
desired functions, from which the stresses in the body could be calculated
in explicit form. On the other hand, if the given contour C, enclosing a
simply connected region
»S,
sufficiently regular, then can be mapped
it
is
on unit circle to any degree of approximation by function of the form
a
a
n
= for finite
S
a
0
In this event the solution of the thermoelastic problem can be obtained,
then for some region S' which differs from
S,
S
little as desired. Hence, from the practical point of view, the thermo
elastic problem can always be solved by means of such integrals of Cauchy
as are discussed in Sec. 9-5, the evaluation of which requires only ele
mentary algebraic operations. The situation, naturally, vastly more
is
involved the region multiply connected.
is
if
ing on the external surface forces can be carried along without introducing
any complications.
Finally knowledge of the temperature function has been assumed
T
a
throughout the above discussion in Secs. 9-5 to 9-8. But the tem
if
perature follows the law of heat conduction [Eq. (2-3)], then in many
instances may be possible to solve the stress problem by the foregoing
it
solution for the heat-conduction equation for the region under consider
ation in this section.
REFERENCES
29. Sharma, B.: Thermal Stresses in Infinite Elastic Disks, ASME Paper 56-APM-19,
June, 1956.
30. Weiner, J.: An Elastoplastic Thermal-stress Analysis of a Free Plate, ASME
Paper 56-APM-6, June, 1956.
31. Hoff, N. J., and others: Theory and Experiment in the Solution of Structural
Problems of Supersonic Aircraft, WADC TR 55-291, Wright-Patterson Air Force
Base, Ohio, 1956.
32. Zizicas, G. A.: Transient Thermal Stresses in Thin Isotropic Elastic Plates, Univ.
Calif., Los Angeles, Rept. 52-7, 1952.
APPENDIX A
is,
and derivatives of functions in the differential calculus. That
if
= F(x,y,z,t) (A-l)
F
then „
dF =
dF
— ,
dx
. dF ,
dy
. dF
+ — dz
,
+
. dF
— ,t
dt
+
dx dy dz dt
(A-2)
d^dy
fzdz
y
as
a
the coordinate axes, then all the components in Eq. (A-3) are directed
along that axis, the subscripts merely indicating which variable pro
is
^_
■F+[dF)x
4^
F+{dF)y F
dF dF , dF
— , dF , dF
—
-57
=
5— Vx + vy + -r- v, + (A-4)
dt ox by bz bt
derived from Fig. A-2. The forces acting on the tetrahedron can be
determined by multiplying the stress components by the areas of the
faces. If A is the area of the face ABC, then the areas of the three
other faces are
k = (B-l)
3'6°p°fC'
219
220 THERMAL STRESSES
h T a V p m(IO')
16,000
- 1 98 1,054 1,147 0 0 001448 3.390
17,000
- 1,050 1,101 1 3.369
1 59 0 001401
3.347
-
18,000 5 15 1,046 1,056 9 0 001355
19,000 8 72 1,041 1,014 0 0 001311 3.326
20,000 -12 28 1,037 972 6 0 001267 3.305
h T a V p m(107)
-
240,000 25 3 1,080 0 1699 0 00000020 3 533
250,000 8 2 1,042 0 1139 0 00000015 3 333
255,905 -28 0 1,019 0 0742 0 00000012 3 212
260,000 -28 0 1,019 0 0742 0 00000010 3 212
t Ames Laboratory Staff, Notes and Tables for Use in the Analysis of Supersonic
Flow, NACA TN 1428, December, 1947.
h = height above sea level, ft
T = temperature, °P
a = speed of sound, ft /sec
p = pressure, lbf/ft2
p = mass density, slug/ft3
H
= coefficient of viscosity, slug/ft-sec
222 THERMAL STRESSES
T cp c„ 7 a *(10') Pr
f Reprinted with permission from J. H. Keenan and J. Kaye, "Gas Tables," 1948,
John Wiley & Sons, Inc.
T = absolute temperature, °F
cp = specific heat at constant pressure, Btu/(lb)(°F)
c„ = specific heat at constant volume, Btu/(lb)(°F)
7 = cp/c„
a = speed of sound, ft/sec
n = coefficient of viscosity, lb-m/sec-ft
Pr = Prandtl number
PROPERTIES OF AIR, STANDARD ATMOSPHERE, AND MATERIALS 223
0.10 10 13 75 0.22
0.065 6.5 15 45 0.25
Steel 0.28 29 7 22 0.16
0.16 16 5.5 10 0.15
0.30 31 8 15 0.12
p = density, lb /in. 3
E = modulus of elasticity, psi
a = coefficient of thermal expansion, in./(in.)(°F)
k = coefficient of thermal conductivity, Btu/(ft)(hr)(°F)
c„
= specific heat, Btu/(lb)(°F)
REFERENCES
1. Ames Laboratory Staff: Notes and Tables for Use in the Analysis of Supersonic
Flow, NACA TN 1428, December, 1947.
2. Schapiro, Ascher H.: "The Dynamics and Thermodynamics of Compressible Fluid
Flow," 2 vols., The Ronald Press Company, New York, 1954.
3. Langley Laboratory Staff: Manual of the ICAO Standard Atmosphere, NACA TN
3182, May, 1954.
4. Keenan, J. H., and Joseph Kaye: "Gas Tables," John Wiley & Sons, Inc., New
York, 1948.
5. Eckert, E. R. G.: Survey of Heat Transfer at High Speeds, W ADC TB 54-70,
Wright-Patterson Air Force Base, Ohio, 1954.
6. Hoyt, S. L.: "Metal Data," Reinhold Publishing Corporation, New York, 1952.
APPENDIX G
The literature cited at the ends of the chapters in this book does not
constitute a complete bibliography on the thermal-stress problem. How
ever, several literature lists have been published which can be consulted
for more information.
For a listing of literature on all phases of the thermal-stress problem
see Brahtz and Dean, An Account of Research Information Pertaining
to Aerodynamic Heating of Airframe, WADC TR 55-99, 1955, which
includes a five-volume Part II on Bibliography (Ref. 19 in Chap. 9).
Volume 1 is on External Environment, Vol. 2 on Materials, Vol. 3 on
Configurations, Vol. 4 on Methods of Analysis (much of the thermal-
stress literature in English is listed in this volume), and Vol. 5 on Methods
of Test. Abstracts are given for many of the papers.
For the literature on the two-dimensional thermal-stress problem see
literature lists in Refs. 1, 11, and 12 of Chap. 9. In the author's thesis
(Ref. 11 in Chap. 9) comments upon the literature up to 1939 are given.
Melan and Parkus (Ref. 1 in Chap. 9) list the literature up to 1953,
including German and Russian papers. Chapter 9 of the present text
lists some of the literature since 1953.
225
INDEX