Beruflich Dokumente
Kultur Dokumente
We consider a three dimensional elastic body in static equilibrium with external forces P 1 , P 2 , etc. Equilibrium implies that these forces and their
moments add up to zero.
P4
P3
A
P1
Tn
n
P2
P1
P2
T x
Ax
Tn
n
T z
Az
x
Ay
z
A
T
(1.2)
where f is the body force density (weight per unit volume) and V is the
volume of the tetrahedron. If we assume the maximum length of an edge
of the tetrahedron is L, the surface areas are of the order of L2 and the
volume, L3 .
As Ax , Ay , and Az are projections of A along the coordinate
3
y
yy
11
00
yz
11
00
11
00
1
zy 0
1
0
1
0
zz
1
0yx
1
0 xy
1
0
1
0
xx
zx
11 111
00 000
11
00
xz
z
111
000
111
000
11
00
11
00
L0
Ax Ay Az
,
,
A A A
= {nx , ny , nz },
lim
L0
V
= 0.
A
(1.3)
(1.4)
where xx is the normal stress and xy and xz are shear stresses. The rst
subscript indicates the plane and the second subscript the direction of the
force. Similarly, the other two traction vectors can be resolved as
T y = yx i + yy j + yz k
T z = zx i + zy j + zz k.
(1.6)
(1.7)
(xx i + xy j + xz k)nx
+(yx i + yy j + yz k)ny
+(zx i + zy j + zz k)nz .
(1.8)
Collectively, we need nine components of stress to create three traction vectors on coordinate planes; they in turn can give us the traction vector on any
inclined plane. We started with a cut and a point on the cut to introduce
the nine components of the stress. There are these nine components at every
point of our three dimensional body. These nine components form the stress
matrix or the stress tensor,
xx xy xz
[] = yx yy yz .
(1.9)
zx zy zz
The rows of the matrix indicate the planes and the columns the directions
of the forces. It is called a tensor based on how the components transform
under rotation of coordinates. Fig. 1.3 shows the nine components of stress.
If we write the unit normal to any inclined plane as a column vector
nx
ny
{n} =
,
(1.10)
nz
using matrix multiplication we get
n
xx yx zx nx
Tx
Tn
= xy yy zy ny
,
y
n
xz yz zz
nz
Tz
(1.11)
Figure 1.4: Components of the traction vector normal and parallel to the
plane.
where the column matrix on the left hand side represents the components of
T n.
With n being the unit normal to the inclined plane, please note that, in
general, the traction vector is not aligned with the normal. The component
of T n along n is the normal stress on this plane. This can be expressed in
multiple ways as
n
n
= n T n = nx Tx + ny Ty + nz Tzn = {n}T []T {n},
(1.12)
= xx n2 + yy n2 + zz n2
(1.13)
x
y
z
+(xy + yx )nx ny + (yz + zy )ny nz + (zx + xz )nz nx , (1.14)
where the superscript, T , indicates transpose of the matrix. The component
of T n parallel to the plane is the shear traction vector
= T n n.
1.0.2 Example: traction vector on a plane
Consider the stress matrix at a point (the units may be MPa)
3 3 0
[] = 3 4 2
0 2 5
(1.15)
Tx 1 3 3
Tn
= 3 4
y 3
n
0 2
Tz
0 2
1
2 2
=
3
5
1
12
16 .
1
T n = (12i + 16j + 9k).
3
The normal stress on this plane is the normal component of T n ,
= n Tn =
65
= 7.222.
9
1.1
Equilibrium Equation
(1.16)
T x + x T x
T x
y
x
f
z
T y
T z + z T z
Ay = zx,
Az = xy,
V = xyz.
(1.17)
As the areas of the positive and negative x-facets are the same, the net force
from these two facets is x T x Ax Including the pairs of traction vectors on
the y and the z -facets and the body force f V , we balance the forces to
get
x T x Ax + y T y Ay + z T z Az + f V = 0.
Dividing by V and simplifying, we have
x T x y T y z T z
+
+
+ f = 0.
x
y
z
Taking the limit, we get
T x T y T z
+
+
+ f = 0.
x
y
z
(1.18)
f = fx i + fy j + fz k,
(1.19)
Using
xx
xy
xx
y
x
xy
yx
yy
Figure 1.6: Moment balance in the z-plane
and the expressions for the traction vectors from 1.5-1.7, the above vector
equation can be written as
xx yx zx
+
+
+ fx = 0,
x
y
z
xy yy zy
+
+
+ fy = 0,
x
y
z
xz yz zz
+
+
+ fz = 0,
x
y
z
(1.20)
yz = zx ,
zx = xzx .
(1.21)
xz nx + yz ny + zz nz = Tzn ,
(1.22)
(1.23)
(1.24)
where the bar over T indicates that these are known (given) functions.
1.2
Plane Stress
If the elastic body is thin in one direction compared to the other two directions and if there is no force applied in that direction, we can orient our
coordinates to have the z-axis perpendicular to the thin plate and use the
approximation
zz = 0, xz = 0, yz = 0.
(1.25)
This is called the plane stress approximation. The remaining three stress
components satisfy the equations of equilibrium
xx,x + xy,y + fx = 0,
xy,x + yy,y + fy = 0,
(1.26)
(1.27)
(1.28)
(1.29)
As most of the aero structures are thin and the in-plane stresses are large
compared to the applied load perpendicular to these plates, we encounter
plane stress approximation frequently.
10
x
Figure 1.7: Curved panel
1.2.1 Curved Panels
If the curvature of a thin panel is small, we may, locally, use a Cartesian
system with coordinates x, s, and z, where the s coordinate changes in the
direction of curvature. This is shown in Fig. ??. We can adapt the plane
stress equations above with y replaced by s. Then
xx,x + xs,s + fx = 0,
xs,x + ss,s + fs = 0,
(1.30)
(1.31)
xsnx + ss ns = Tsn .
1.3
(1.32)
(1.33)
Fig. (??), let vx and vy be the components in the x, y-system and vx and vy
vx = vx cos + vy sin ,
vy = vx sin + vy cos .
(1.34)
11
vy
vy
vx
vx
(1.35)
(1.36)
We mentioned earlier that the stress matrix is also called the stress tensor.
In the plane stress case it is a 2 2-matrix,
[] =
xx xy
xy yy
(1.37)
In the rotated coordinate system, we expect the stress matrix to have the
form
xx xy
.
(1.38)
[ ] =
xy yy
We assume unit thickness for the wedge OAB and, rst, balance the forces
along the x and y -directions to get
(1.39)
12
y
B
xy
xy
xx
xx
O
yy
xy A
(1.40)
OA/AB = sin s,
we have
xx = xx c2 + yy s2 + 2xy cs,
The stress component yy is the normal stress on a plane 90 from the x plane. If we replace by + 90 , we get
cos( + 90 ) = s,
sin( + 90 ) = c,
(1.41)
and
yy = xx s2 + yy c2 2xy cs.
Thus,
xx = xx c2 + yy s2 + 2xy cs,
yy = xx s2 + yy c2 2xy cs.
(1.42)
13
vx
vy
vx
vy
v =
[Q] =
c
s
s c
(1.43)
[ ] = [Q][][Q]T .
(1.44)
1 + cos 2
,
2
s2 =
1 cos 2
,
2
sc =
sin 2
,
2
xx = m +
cos 2 + xy sin 2,
2
xx yy
cos 2 xy sin 2,
yy = m
2
yy xx
xy =
sin 2 + xy cos 2,
2
(1.45)
(1.46)
xx + yy
.
2
(1.47)
xx + yy
.
m =
2
(1.48)
Thus, the mean stress remains an invariant under the coordinate rotation.
1.3.1 Principal Stresses
Having a formula for the normal stress on an arbitrary plane, we are interested in nding the angle which makes the normal stress a maximum (or a
minimum). For this, we let
dxx
= 0.
d
14
This gives
xx yy
sin 2 xy cos 2 = 0.
2
Denoting the solution of this equation by p , we get
tan 2p =
xy
.
(xx yy )/2
(1.49)
This equation has two solutions for 2p , which are separated by 180 . Before
we discuss these solutions fully, we observe from Eq. (??),
xx yy
2
xy
xx yy
cos 2 + xy sin 2,
2
yy xx
=
sin 2 + xy cos 2.
2
xx yy
R =
2
2
[xy ]2
xx yy
=
2
+ [xy ]2 .
(1.50)
As shown in Fig.??, the angle in Eq. (??) can be expressed two ways:
xx yy
2
A
2p
R
xy
B
xx yy
,
2R
sin 2p =
xy
,
R
15
(1.51)
or
xx yy
xy
, sin 2p =
.
(1.52)
2R
R
Substituting these in the equation for xx in Eq. (??), we get the maximum
and minimum values,
cos 2p =
1 = m + R,
2 = m R.
(1.53)
These are the principal stresses; their planes, the principal planes, will be
referred as p-planes.
Substituting the angles in the equation for the shear stress xy shows
xy = 0.
(1.54)
We use the same procedure to maximize the shear stress xy in Eq. (??) to
get
xx yy
sin 2 = 0.
(1.55)
xy cos 2 +
2
The solution of this equation is denoted by s for the maximum shear planes
(s-planes); and we have
tan 2s =
xx yy
.
2xy
(1.56)
(1.57)
xx yy
,
2R
cos 2p =
xy
,
R
(1.58)
in xy and xx we get
xy = R,
xx = m .
(1.59)
16
(xx m )2 + (xy )2 = R2 ,
(1.60)
which shows that the normal stress = xx and the shear stress = xy are
coordinates of a point lying on a circle of radius R and center at (m , 0) in
a , -plane. The principal stresses 1 and 2 are the extreme values of
along a diameter where = 0. The maximum shear stresses correspond to
points (m , R).
We maintain the convention for measuring angles as counter clock-wise
being positive. Then, as shown in Fig ??, the state of stress on the x-plane
must be below the point corresponding to (1 , 0), subtending an angle 2p
at the center. The sign convention for shear stresses is as follows: If we view
the shear stress from below the plane to see it creating a clock-wise moment
then it is positive. On the x-plane xy creates a counter clock-wise moment
and hence it is negative. We have to keep in mind that in the Cartesian
system we see two planes x and y but on the Mohrs circle each point on the
circles is one plane. While the real planes dier by an angle , they dier by
2 on the Mohrs circle. As shown if Fig. ??, the arc BC subtends 2p at
s1
p2
F
2p
p1
C
B
xy
xx
s2
Figure 1.11: Mohrs circle showing principal stress planes, p1 , p2 , maximum
shear stress planes, s1 , s2 and the x-plane.
the center and p at F on the circumference. We can use the point F , called
the focus, to get real angles.
17
1.4. STRAIN
10
10
R = 5.
We draw the Mohrs circle as shown in Fig. ??. On the x-plane the normal
stress is = 10 MPa and the shear stress of xy = 4 MPa creates a clockwise moment and, so, it is positive. The x-plane is at (10, 4). We draw a line
parallel to the axis to locate the focus F . The angle 2 is found from
tan 2 =
1.4
xy
= 4/3;
(xx yy )/2
2 = 53.13 ,
= 26.57 .
Strain
(1.61)
18
s1
5
x
10
p2
p1
2
O
12
s2
4
2
4
(1.62)
(1.63)
In order to account for large displacement gradients and large rotation of elements we have to modify the denition of normal strain from the simple form
of change in length over original length.The normal strain of an element
r oriented in the direction of the unit vector n is dened by
(r )2 (r)2
,
r0
2r)2
nn = lim
(1.64)
where r is the magnitude of the vector r, etc. When the deformed length
is close to the original length and if the rotation of the element is small, this
denition reduces to the classical denition. In order to calculate the shear
strain as a change in angle between two elements which are initially at 90,
19
1.4. STRAIN
B
P
y
Q
r
Q
r
r
r
x
z
Figure 1.14: Displacement vector for deformation.
we choose two orthogonal elements r1 and r2 along the unit vectors n and
m and dene
r 1 r 2
.
(1.65)
nm = mn =
2r 1 r2
The dot product in this expression will be zero if the deformed elements are
still at 90 and we get zero shear strain.
These general denitions can be specialized by selecting r along the
coordinate directions. For example, if we choose r oriented along the xdirection
r
u
r
(r )2
(r )2 (r)2
xx
=
=
=
=
=
xi, r = x,
x u = [x ui + x vj + x wk],
[(x + x u)i + x vj + x wk],
(x + x u)2 + (x v)2 + (x w)2 ,
2xx u + (x u)2 + (x v)2 + (x w)2,
2xx u + (x u)2 + (x v)2 + (x w)2
= lim x 0
.
2(x)2
20
u 1
+
=
x 2
u
x
v
x
w
x
(1.66)
Similarly,
yy
v 1
=
+
y 2
u
y
zz
w 1
+
=
z
2
u
z
v
y
v
z
w
y
w
z
.
2
r 2 = yj.
(1.67)
r 2 = yj + y u.
(1.68)
These give
r 1 = xi + x u,
1 u v u u v v w w
+
+
+
+
2 y x x y x y
x y
(1.69)
Similar calculations using element vectors along the y and z directions and
z and x directions result in
1
2
1
=
2
yz =
zx
v w u u v v w w
,
+
+
+
+
z
y
y z y z
y z
w u u u v v w w
.
+
+
+
+
x
z
y z y z
y z
(1.70)
(1.71)
xx xy xz
[] = xy yy yz .
(1.72)
xz yz zz
21
1.4. STRAIN
When the magnitude of the displacement derivatives and the rotation of the
elements during deformation are small, we neglect the nonlinear terms and
use the linear small strain expressions
xx =
u
x
v
y
yy =
zz =
w
,
z
(1.73)
1 u v
1 v w
1 w u
xy = [
+
], yz = [ +
] zx = [
+
].
(1.74)
2 y x
2 z
y
2 x
z
If we superimpose the origins of the undeformed elements xi and yj and
y
D
B
2
1
A
x
OA OC = x(1+
xx =
u
),
x
OC OA
u
=
OA
x
1 =
v
CA
=
OC
x
OB = y,
yy =
2 =
OB OD = y(1+
OD OB
v
=
.
OB
y
DB
u
=
.
OD
y
v
),
y
22
1 + 2
1 u v
= [
+
].
2
2 y x
The shear strains we have dened are known as mathematical shear strains
which are calculated using the average of 1 and 2 . Engineering shear strains
are dened as the total change is angle, 1 + 2 . These are denoted by xy ,
etc. We have
xy = 2xy , yz = 2yz , zx = 2zx .
(1.75)
1.4.1 Plane Strain
If
zz = 0,
zx = 0,
zy = 0,
the state of strain is referred to as plane strain. We may visualize a body kept
at constant length in the z-direction using rigid blocks and with a uniform
cross section. The loading on the surface should not vary along z to achieve
plane strain. conditions. In this case, elasticity problems involve three nonzero strains: xx , yy and xy , given by
xx =
u
,
x
yy =
v
,
y
1 u v
xy = (
+
).
2 y x
(1.76)
.
1.4.2 Strain Transformation
Under a coordinate rotation by an angle from the x, y-system to x , y system, in the context of stress transformation, we have seen the transformation law for vector components:
x = cx sy ,
y = sx + cy ,
u = cu + sv,
v = su + cv,
where c and s are cos and sin , respectively. The normal strain in the
x -direction is dened as
u
x
y
= [cu + sv] = [
+ ][cu + sv],
x
x
x x x y
= [c
+ s ][cu + sv],
x
y
2
2
= c xx + s yy + 2csxy ,
xx =
(1.77)
(1.78)
(1.79)
23
1.4. STRAIN
which, interestingly, is the same law for the stress components. This derivation is based on geometry while the stress transformation is based on statics.
We may complete the list as
yy = s2 xx + c2 yy 2csxy ,
xy = cs(yy xx ) + (c2 s2 )xy .
(1.80)
(1.81)
Because of this transformation law, we refer to the collection of strain components as the strain tensor.
Now, we can apply the same approach we used for stresses to nd the
direction and magnitude of the maximum normal strain, those of the maximum shear strain and to draw the Mohrs circle. The sign convention for
shear strains in the Mohrs circle representation would be clear if we associate it with the shear stress. In the Cartesian form, we associate positive
shear strain with positive shear stress, which for the x-plane is negative and
for the y-plane is positive.
From the expression for xx , cos and sin are expressed as nx and ny ,
the components of a unit vector, the strain in the direction n can be written
as
nn = nT []n.
(1.82)
This is analogous to the normal stress expression obtained using the stress
matrix and the unit vector.
1.4.3 Displacement Boundary conditions
In elasticity problems, certain surfaces of the body may have prescribed displacements. For example, for a cantilever beam one end has zero displacements. In general
u = u, v = v ,
(1.83)
where, as before, u and v are given functions.
24
(1.85)
Thus, we cannot choose three strains arbitrarily; they have to satisfy the
compatibility equation. In the three dimensional case there are six such
equations.
1.5
The stress and strain descriptions we have seen so far are independent of
the material. The relations between stress and strain are called constitutive
relations. An elastic material is called linear elastic if the stresses are linearly related to the strains. Generalizing the simple one-dimensional case
where stress is proportional to strain, we have, for a linear, isotropic, elastic
material,
xx =
yy =
zz =
xy =
yz =
xz =
1
[xx (yy + zz )],
E
1
[yy (zz + xx )],
E
1
[zz (xx + yy )],
E
xy
,
2xy =
G
yz
2yz =
,
G
xz
2xz =
,
G
(1.86)
(1.87)
(1.88)
(1.89)
(1.90)
(1.91)
where E is the Youngs modulus, is the Poissons ratio, and G is the shear
modulus which can be written as
G=
E
.
2(1 + )
(1.92)
25
This set of relations with just two independent material constants is dictated
by the requirement of isotropy. If we rotate the coordinate system stresses
and strains would change, but the Hookes law should be independent of the
coordinates used. We will discuss this further when we present anisotropic
relations.
1.5.1 Hookes law for plane stress
Under the plane stress assumption zz = xz = yz = 0, and the Hookes law
reduces to
1
[xx yy ],
(1.93)
xx =
E
1
[yy xx )],
(1.94)
yy =
E
1
xy =
xy ,
(1.95)
G
The strain in the z-direction can be obtained as
(1.96)
zz = [xx + yy ].
E
1.5.2 Hookes law for plane strain
From zz = 0, we nd
zz = [xx + yy ].
(1.97)
[xx
yy ],
(1.98)
E
1
1 2
[yy
xx )],
(1.99)
yy =
E
1
1
xy =
xy .
(1.100)
G
The plane stress and plane strain constitutive relations can be cast in a
similar form if we introduce the eective Youngs modulus E and Poissons
ration ,
xx =
(1.101)
26
P
x
x=
2c
b
My
,
I
M = P ( x),
2
I = bc3 .
3
(1.102)
(1.103)
xy
P
= y,
y
I
(1.104)
gives
P 2
y + K,
2I
(1.105)
where K is a constant. At the top and bottom surfaces of the beam there are
no shear stresses. That is xy = 0 when when y = c. Then K = P c2 /(2I)
and
P
xy = (c2 y 2 ).
(1.106)
2I
27
(c2 y 2 )bdy = I.
(1.107)
v
xx
=
,
y
E
v
P
=
( x)y.
y
EI
(1.108)
P
y2
v(x, y) =
[(x) +g(x)], (1.109)
EI
2
where f (y) and g(x) are the arbitrary functions. Next, we substitute u and
v in the shear strain constitutive relation,
xy =
xy
,
G
x2
dg
y2
E c2 y 2
df
+ x
+ =
.
dy
2
dx
2
G 2
Separating the variables,
df
y 2 E c2 y 2
=
+ C,
dy
2
G 2
dg
x2
= x
+ C,
dx
2
where C is a new constant. Integrating these two equations
y 3 E c2 y y 3
+ Cy + D1 ,
6
G 2
6
x2 x3
+ Cx + D2 .
g =
2
6
f =
(1.110)
(1.111)
(1.112)
28
EI
2
6
G
2
6
y 2 x2 x3
P
[( x) +
+ Cx + D2 ].
(1.114)
v =
EI
2
2
6
u =
Although, we would like to satisfy the boundary conditions at the xed end,
namely, u = 0, v = 0, at x = 0 for all y between c and c, these solutions
do not allow it for all values of y. The next step is to try to meet these
requirements at x = 0 and y = 0. This gives D1 = 0 and D2 =0. That leaves
C. The contributions due to C can be viewed as those due to a small counter
clock-wise rigid body rotation of the beam about the point (0, 0). We may
set
u
= 0, at x = 0, y = 0.
(1.115)
y
This leads to
C=
Ec2
,
2G
+
x].
EI
2
2
6
2G
u =
(1.116)
(1.117)
29
where, our present approximation gives, k = 1.5. This is attributed to Timoshenko. The value of k = 6/5 was obtained by Reissner for static cases and
k = 12/ 2 by Mindlin for dynamic cases.
1.6
Anisotropic Materials
1.7
Non-homogenous structures
30
(a)
Ez
(b)
Ex
Ey
where
1.8
xy
yx
=
,
Ex
Ey
xz
zx
=
,
Ex
Ez
yz
zy
=
.
Ey
Ez
(1.119)
(1.120)
31
Exercises