Chapter 5

We begin this chapter by reviewing the elementary beam theory known as

the Euler-Bernoulli theory. In the case of beams lacking an axis of symmetry,
an applied bending moment about an axis would cause curvatures in both
the planes perpendicular to the beam axis. The beam axis, x, will be placed
along the neutral axis where the axial strain due to bending is zero. A cross
section of the beam will be in the y, z- plane. The deflections of the beam
in the y and z directions will be denoted by v and w, respectively. A sketch
of the undeformed and deformed beam with its projections on the xz-plane
and xy-plane are shown in Fig. 5.1.
y

v
x
w
z
Figure 5.1: Beam deflections in two directions.
The basic assumptions of the elementary beam theory are: (a) plane sections normal to the neutral axis remain plane and normal to the deformed
neutral axis and (b) the cross-sectional dimensions do not change. The original, undeformed straight beam undergoes deflections v and w. The curvatures of the projections of the deformed beam in the x, y plane and xz plane
1

CHAPTER 5. BENDING

are given by
y

v
1
=
,
Ry
[1 + (v )2 ]3 /2

1
w
=
,
Rz
[1 + (w )2 ]3 /2

(5.1)

where Ry and Rz are radii of curvatures and v = dv/dx, etc. We assume the
deflections are small compared to the cross-sectional dimensions of the beam
and slopes of the deflections are small compared to unity. These assumptions
allow us to drop the quadratic terms in the curvature expressions, to have
y = v ,

z = w .

(5.2)

First, let us consider the deflection in the y-direction.

Ry

Ry

Figure 5.2: Strain due to bending.

As shown in Fig. 5.2 an element of length x on the neutral axis deforms
into a curve of radius Ry without any change in length. Thus,
Ry = x.
An element AB at a height y from the neutral axis deforms into A B with
length (Ry y). Then the strain of this element is
=

(Ry y) x
(Ry y) Ry
=
.
x
Ry

(5.3)

3
This gives

y
= y y.
(5.4)
Ry
A similar contribution to the strain can be found from the curvature in
the xz-plane. The total strain is obtained as
=

= (y y + z z).

(5.5)

The axial stress in the beam is

= E = E(y y + z z).

(5.6)

As there is no axial force,

dA = 0,

(5.7)

independent of the curvatures. Here, A is the cross-sectional area of the

beam. This equation determines the properties of the neutral axis:
EydA = 0,
A

EzdA = 0.

(5.8)

Thus, y and z are measured from the modulus-weighted centroid of the crosssection. If E is a constant, the origin of y and z is the centroid. Keeping
composite beams in mind, we keep E as a variable. From Fig. 5.3, the
bending moments My and Mz are given by
z

dA
y
Mz

My

Figure 5.3: Moment components and normal stress.

CHAPTER 5. BENDING

My =

zdA,

Mz =

ydA.

(5.9)

Using Eq. (5.6), we obtain the moment-curvature relations

My = EI zy y EI yy z ,

Mz = EI zz y + EI yz z ,

(5.10)

where we have introduced the modulus-weighted second moment of areas,

EI yy =

Ez 2 dA,

EI zz =

Ey 2dA,
A

EI yz = EI zy =

EyzdA.
A

(5.11)
We can solve for y and z in terms of the bending moment components,
and substitute the values for y and z into Eq. (5.6), to find the stress as a
function of applied moment components.
If there is an axis of symmetry for the cross-section (including the modulus) EI y z = 0. Then an applied moment My produces a curvature in the
xz-plane, z , only. And, similarly, Mz produces y . This way, the momentcurvature relations will be uncoupled. For the sake of simplicity, let us assume the beam is homogeneous, that is E is a constant. Then, the neutral
axis passes through the centroid of the section and the moment-curvature
relations become
My = EIzy y EIyy z ,
Introducing the notation

Mz = EIzz y + EIyz z .

2
I2 = Iyy Izz Iyz
,

(5.12)
(5.13)

we may solve for the curvatures, to obtain

1
[Iyy Mz Iyz My ] ,
E I2
1
= 2 [Izz My + Iyz Mz ] .
EI

y =

(5.14)

(5.15)

In the case of symmetry Iyz = 0 and we obtain the uncoupled relations

Mz
,
EIzz
My
=
.
EIyy

y =

(5.16)

(5.17)

5
If we denote the deflections of the neutral axis by v(x) and w(x) as shown
in Fig. ??, the curvatures are given by
y =

v
,
[1 + (v )2 ]3/2

z =

w
.
[1 + (w )2 ]3/2

(5.18)

Under the assumption of small slopes, we may omit the quadratic terms in
the denominator, to obtain
y = v ,

z = w .

(5.19)

5.0.1 Principal axes of a section

The coordinate system can be rotated about the x-axis, to obtain a new
system: x = x,
y = y cos + z sin ,

z = y sin + z cos .

(5.20)

In this new system, the second moments of the area are given by

Iyy
=

Iyz

(z )2 dA =
A

[cs(z 2 y 2) + (c2 s2 )yz]dA = (Iyy Izz )cs + Iyz (c2 s2 ),

y z dA =

Izz
=

[s2 y 2 2csyz + c2 z 2 ]dA = Izz s2 2Iyz cs + Iyy c2 ,

(z )2 dA =
A

[c2 y 2 + 2csyz + z 2 s2 ]dA = Izz c2 + 2Iyz cs + Iyy s2 ,

A

where c = cos and s = sin . If we interchange y and z, these transformation

relations are similar to the stress transformation relations involving yy , zz

and yz . We choose to have Iyz

= 0, in order to find the principal axes of
a cross-section. This leads to the equation,
tan 2 =

2Iyz
,
Izz Iyy

for
In the principal coordinate system
Izz + Iyy Iyy Izz
+
cos 2 Iyz sin 2,
2
2
Izz + Iyy Iyy Izz

cos 2 + Iyz sin 2,

=
2
2

Iyy
=

Izz

(5.21)

CHAPTER 5. BENDING

In the new coordinate system the moment-curvature relations will be uncoupled. Of course, we also have to to find the components of the moment and
the curvature in the new system. Thus,

My = EIyy
z ,

Mz = EIzz
y .

(5.22)

5.0.2 Example: Angle section

For the angle section shown in Fig. 5.4 the Youngs modulus is homogeneous.
A moment component Mz = M0 is applied at this section. (a) Obtain the
location of the neutral axis, (b) find the second moments and the product
moment, (c) compute the stresses at the corner points A, B, and C. Also,
find the principal axes for this section and using them obtain the stresses at
the corner points. All dimensions are in centimeters.
z z
B

M0
y

10

Figure 5.4: Angle section beam.

When the Youngs modulus is constant, the neutral axis is through the
centroid. To find the centroid, we use a temporary coordinate system with
y, z as shown in Fig. 5.4. We divide the section into two area: A1 = 10 1
and the other A2 = 5 1. The total area is
A = A1 + A2 = 15cm2 .
Taking the moment of the areas about the z -axis
A
y = 10 0.5 + 5 3.5,

y = 1.5.

7
A
z = 10 5 + 5 9.5,

z = 6.5.

With these coordinates, we establish our coordinate system y, z at the centroid. The coordinates of the centroids of the two areas are obtained as
(1, 1.5) and (2, 3). The second moments are found using the shift theorem as
5 13
1 103
+ 10 1.52 +
+ 5 32 = 151.25,
12
12
1 53
10 13
+ 10 12 +
+ 5 22 = 41.25,
=
12
12
= 10 (1)(1.5) + 5 2 3 = 45.

Iyy =
Izz
Iyz

From the moment-curvature relations (5.10), we have

Iyz y + Iyy z = 0,

Izz y + Iyz z = M0 /E.

Using the numerical values for the second moments, these become
45y + 151.25z = 0,

41.25y + 45z = M0 /E.

Solving these equations, we get

y = 0.03589M0/E,

z = 0.01068M0 /E.

The bending stress distribution across this section is given by Eq. (5.6) as
= M0 (0.03589y 0.01068z).
To compute the stresses at the points A, B, and C, we note their coordinates
as
A : (1.5, 6.5), B : (1.5, 3.5), C : (4.5, 3.5),
to obtain
A = 0.0156M0 ,

B = 0.0912M0 ,

C = 0.1241M0 .

We may also solve this problem using the principal coordinates. The
needed rotation angle is obtained from
tan 2 =

90
2Iyz
=
= 0.8182,
Izz Iyy
41.25 151.25

CHAPTER 5. BENDING

which gives 2 = 0.6857 or = 0.3429 radians = 19.64 . There is a

second solution which is 90 away from this. The second solution just rotates
the coordinate orientations obtained using the first solution by 90 . Let y

and z represent the rotated coordinates. Of course, Iyz

= 0 and
151.25 + 41.25 151.25 41.25
+
cos 2 45 sin 2 = 167.31,
2
2
151.25 + 41.25 151.25 41.25

cos 2 + 45 sin 2 = 25.19,

=
2
2

Iyy
=

Izz

The components of the applied moment along the principal coordinates are
My = M0 sin = 0.3362M0 ,

Mz = M0 cos = 0.9418M0 .

curvatures are given by

Mz
= 0.9418M0/(25.19E) = 0.0374M0/E,

EIzz
My
= = 0.3369M0 /(167.31E) = 0.0020M0 /E.
EIyy

y =
z

The stress distribution is given by

= E(y y + z z ) = M0 (0.0374y + 0.0020z ).
The new coordinates of the corner points are
A : (1.5 cos 6.5 sin , 1.5 sin 6.5 cos ) = (0.7725, 6.6260),
B : (2.5894, 2.7920),

C : (3.0614, 4.8091).

With these coordinates, we obtain the same stress values at the corner points.
As can be seen, the method of principal axes involve considerably more calculations as the applied moment as well as the coordinates of the points have
to be resolved along the principal axes.

5.1

Equilibrium Equations

As shown in Fig. 5.5, the applied distributed loads, py along the y-direction
and pz along the z-direction, create shear forces Vy and Vz and bending
moments My and Mz .

PSfrag
9

5.1. EQUILIBRIUM EQUATIONS

py
y

00
11
00
11
00
11
00
11
11
00
00
11
00
11
00
11
0
1
0
1
0
1
0
1

py x
My
Vy + Vy
Vy

pz

Vz
Mz

Mz
x

Mz + Mz
Vy x

z
Figure 5.5: Force distribution and the equilibrium of an element.
Force and moment balances of an element of the beam of length x give
the equilibrium equations
dVy
+ py = 0,
dx
dMz
+ Vy = 0,
dx

dVz
+ pz = 0
dx
dMy
Vz = 0.
dx

(5.23)
(5.24)

The negative sign in the last equation can be understood if we rotate the
figure to have pz pointing upward. This rotation results in My pointing away
from the page, unlike Mz .
In integrating these equations to find the bending moment, the boundary
conditions on the beam have to be considered. A given end of the beam
can have (a) prescribed shear force and prescribed moment, (b) prescribed
vertical displacement and prescribed slope, (c) prescribed shear force and
prescribed slope, and (d) prescribed bending moment and prescribed displacement.
If the shear force and bending moment distributions can be found using
only the equilibrium equations, we have a statically determinate problem. If
the moment curvature relations
EIyy w = My ,

EIzz v = Mz ,

(5.25)

have to be integrated to find the shear force and bending moment, we have
a statically indeterminate problem.

10

CHAPTER 5. BENDING

Let us examine some examples of shear force and bending moment calculations for statically determinate problems first. For simplicity we assume
symmetry about the z-axis, and use the notations: I = Izz , V = Vy , and
M = Mz .
5.1.1 Example: Elliptic lift distribution
We assume a uniform wing is loaded by a lift distribution which has the
shape of an ellipse,
p(x) = p0 1 (x/L)2 ,
where p0 is the maximum load density at x = 0. This is shown in Fig. 5.6.

p0

L
Figure 5.6: Elliptic loading on a wing.
The shear force distribution is found from
dV
= p0 [1 (x/L)2 ]1/2 .
dx
A change of variable
x
= sin , dx = L cos d,
L
facilitates the integration of the load distribution.

1 + cos 2
d,
2
/2
/2
/2 sin 2
p0 L
= p0 L
=
+
[2 + sin 2].
2
4
4
= p0 L

cos2 d = p0 L

11

5.1. EQUILIBRIUM EQUATIONS

The differential equation for the bending moment,
dM
= V,
dx
gives, after integration by parts,
M =
=

p0 L2
4

[2 + sin 2] cos d,
/2

p0 L2
2
(2 ) sin + 2 cos cos3 .
4
3

The maximum values of the shear force and the bending moment are at
x = 0, given by
p0 L
p0 L2
Vmax =
, Mmax =
.
4
3
5.1.2 Concentrated forces and moments
In many beam problems the applied distributed force density p(x) consists of
functions with multiple discontinuities in the form of steps or it is idealized
as a concentrated force or concentrated moment. The Macaulay bracket
notation is a simple way to represent these situations.

a
Figure 5.7: Step function
The step function shown in Fig. 5.7 is represented as
xa

1, x > a,
0 x < a.

12

CHAPTER 5. BENDING

The integral of this function is, of course, zero up to any x < a and it is
linearly increasing with slope of unit for x > a. We write this as

xa

x a 0 dx =

x a, x > a,
0
x < a.

(5.26)

Continuing this process, we have

1
xa
n+1

x
n+1

x a dx =

(xa)n+1
,
n+1

x > a,
,
x < a.

n 0.

(5.27)

We know concentrated forces cause a jump in the shear force diagram. To

represent them as a distributed force we have to use a generalized function
which is the derivative of the step function. Using the Macaulay bracket
notation we write this as

xa

d
x a 0,
dx

(5.28)

which is also called a singularity function or Dirac delta function as it goes

to infinity when x = a and is zero every where else. We may go one step
further and also include concentrated moments using the derivative of the
Dirac delta function,

xa

d
xa
dx

(5.29)

To emphasize the fact that this is just a notation, we do not use a minus sign
in front of the derivative.

13

5.1. EQUILIBRIUM EQUATIONS

P0
M0
a

RB

RA
M1

V1

M2

M3

V2

V3

Figure 5.8: Simply-supported beam loaded with a concentrated force and

a moment with free-body diagrams to calculate shear forces and bending
moments.
First using the second diagram in Fig. 5.8, let us obtain the shear force
and the bending moment distributions using free-body diagrams. Let RA
and RB be the reactions from the supports. Setting the moment about A
and about B, respectively, to zero,
RB =

M0 P0
,
3a
3

RA =

M0 2P0

.
3a
3

The three cuts shown in the third diagram in Fig. 5.8 have to be considered
sequentially. From the first cut,
V1 =

M0 2P0
+
,
3a
3

M1 =

M0 2P0
x.
+
3a
3

The other two cuts give

M0 P0
,
3a
3
M0 P0
=
,
3a
3

V2 =
V3

M0 2P0
x + P0 (x a),
+
3a
3
M0 P0
(3a x).
M3 =

3a
3
M2 =

14

CHAPTER 5. BENDING

We note that
V2 V 1 = P0 ,
and at x = 2a
M3 M2 = M0 .
The shear force and the bending moment diagrams are shown in Fig. 5.9.
We have assumed M0 > P0 a for drawing the diagrams.

P0

M0

V1

V2

Figure 5.9: Shear force and bending moment diagrams.

can be represented as
p(x) = P0 x a

+ M0 x 2a

. 0 < x < 3a.

(5.30)

Note that we have excluded the reactions at the supports. Integrating p(x),
we get
V (x) = P0 x a 0 + M0 x 2a 1 + C1 ,
(5.31)

15

5.2. SHEAR STRESSES IN BEAMS

where C1 is a constant of integration. Integrating again,
M = P0 x a

+ M0 x 2a

+ C1 x + C2 ,

(5.32)

where C2 is a second constant. For the simply supported beam, the boundary
conditions are: M(0) = 0 and M(3a) = 0. The first of these conditions give
C2 = 0.

(5.33)

or C1 = (2aP0 + M0 )/(3a).

(5.34)

The second condition gives

P0 (2a) M0 + 3aC1 = 0,
Then
2P0 M0

,
3
3a
2P0 M0
x,
+
M = P0 x a 1 + M0 x 2a 0
3
3a

= P0 x a

+ M0 x 2a

A comparison of these expressions with the ones obtained using the free-body
diagrams shows they are identical. Thus, the Macaulay bracket notation
allows us to skip the free-body diagrams and to write the load p(x) in terms
of singularity functions, which can be integrated to get the shear force and
the bending moment .

5.2

Shear Stresses in Beams

The calculation of shear stresses due to bending in an arbitrary cross-section

beam is a complicated problem in elasticity and there are no general formulas.
In special cases such as a rectangular cross-section we have the parabolic
distribution of shear stresses. Here we consider thin-walled open- and closedcell tubes. Consider a section of the beam of length x as shown in Fig. 5.10.
Under the thin-wall assumption we approximate the shear stress distribution
across the thickness t is uniform.

16

CHAPTER 5. BENDING

5.2.1 Open-cell tubes

Q+

F
F+
F

q
x
x
Figure 5.10: Shear flow and bending stress balance in a beam element.
Then,
q = t.

(5.35)

An element taken at s, with width s has normal force F on one side and
F + F on the other. The shear flow creates a force Q at the lower edge and
Q + Q at the upper edge. We have
F = ts,

Q = qx.

(5.36)

The force balance relation

F + Q = 0,

(5.37)

can be written as
ts + qx = 0 or

q
t+
= 0.
x
s

(5.38)

17

5.2. SHEAR STRESSES IN BEAMS

At the two edges of the open cell there are no shear stresses and q = 0 at
these edges. Using this, we can integrate this equation along s to find q.
For example, for an open section which is symmetric about the z-axis, using
M = Mzz , I = Izz , V = Vy , we have
=

M
y,
I

M y
Vy
=
=
.
x
x I
I

and
q=

(5.39)

V
I

ytds.

(5.40)

s=0

Obviously, at the edge, s = 0, the shear flow is zero. At the other edge,
s = smax , shear flow will be zero by virtue of the fact that y is measured
from the centroid.
5.2.2 Channel section
Consider the channel section shown in Fig. 5.11. We assume uniform thickness t.
qB

qB

A
s

2a

qmax

qB

qB
Figure 5.11: Shear flow in a channel section.

18

CHAPTER 5. BENDING

Starting from the top right edge as the origin for s, the shear flow in the
top segment is given by
Vt
q = as,
I
where I is given by
I = 2a3 t3 + 2a2 bt = 2a3 t 1 +

3b
/3.
a

Here, the thin wall assumption allows us to neglect the t3 -terms. At the top
corner B where s = b,
Vt
qB = ab.
I
Along the vertical segment, it is convenient to use a new coordinate y with
B being at y = a. Then in BC,
q = qB

Vt
I

y(dy) =
a

y 2 a2
Vt
ab
.
I
2

The magnitude of the shear flow, which varies as a parabola in BC, is a

maximum at y = 0.
Vt
qmax = [ab + a2 /2].
I
When y = a, we have
qC = qB .
Further, along the lower segment from C to D it decreases to be zero at
the free edge, D. A plot of the variation of the shear flow is superposed
in Fig. 5.11. Along the mid-surface the shear flow direction is indicated
using arrows. As s is taken counter clock-wise the negative values of q show
that the shear flow is clock-wise. The net force from the horizontal segments
cancel each other resulting zero horizontal force. The vertical force is given
by
y=a

(q)dy =
y=a

Vt
y 2 a2
2V t 2
a3
ab
dy =
a b
a3 = V.
I
2
I
3

This verifies the fact that the applied shear force is the resultant of the
distributed shear flow.

19

5.2. SHEAR STRESSES IN BEAMS

5.2.3 Center of shear

In the example of a channel section, we saw the shear flow distribution creates
a vertical force equal to the applied shear force. A question remains: Where
is the location along the z-axis where the shear force V has to applied to
create the same moment as that is due to the shear flow distribution? This
location is called the shear center. Remember the normal stress and the
shear flow distribution are derived on the basis that the beam bends in the
y-direction without any rotation. We may compute the moment produced
by the shear flow distribution about the point O. The force due to the shear
flow in the upper segment is given by the area of the triangle in Fig. 5.11,
F =

Vt 2
ab .
2I

There is an equal and opposite force due to the shear flow in the lower
segment. Then the torque is given by T = 2aF , or
T =

Vt 2 2
ab.
I

To create the same torque the shear force has to be applied at a distance e
to the left of O, such that
V e = T.
From this the distance to the shear center S is given by
e=

a2 b2 t
3b2
b
=
=
.
I
2a [1 + 3b/a]
2[1 + a/(3b)]

This expression shows that e = 0 when b = 0 and e = b/2 when b is very

large compared to a.
In the general case of a cross-section with symmetry about the z-axis,
we compute the torque produced by the shear flow distribution about a
convenient point O along the z-axis and find the distance to the shear center
e from
Ve =

qpds,

(5.41)

where p is the perpendicular distance to the tangent at a point s, and e is

measured along the positive z-direction.

20

CHAPTER 5. BENDING

For sections without any symmetry axis, we can apply arbitrary shear
forces Vy and Vz to compute the shear flow q. Then the torque due to q is
found about any point O to have
Vy ez Vz ey =

qpds,

(5.42)

where ey and ez are the coordinates of the shear center. Fig. 5.12 shows a
schematic of this case.
Vy

ez
Vz

ey
O

ds
q
Figure 5.12: Shear flow and the shear center location.
5.2.4 Center of shear and the center of twist
In Chapter 4 we encountered the center of twist as the point about which a
thin-walled cross-section rotates under a torque. The center of shear is the
point through which shear force has to be applied so that the beam bends
without any rotation. We can show that these two points coincide, by using
the energy principles.
We begin by assuming the shear center S and the center of twist C are
two distinct points, say point 1 and point 2, as shown in Fig. 5.13. Next we
apply a vertical load P at 1 and a torque T at 2.

21

5.2. SHEAR STRESSES IN BEAMS

P
T

Figure 5.13: Center of shear 1 and center of twist 2.

The vertical deflection v under the load P and the rotation can be
written as

v = c11 P + c12 T,
= c21 P + c22 T,

(5.43)
(5.44)

where From our reciprocal theorem, c12 = c21 . If we had P alone, it will
produce a rotation = c21 P . As 1 is the shear center, the beam should
deflect without rotation. Then = 0. This implies c12 = 0.
Now, let us try applying T alone. As the vertical deflection, v, due to T
is given by v = c12 T , we will have v = 0. But the beam is rotating about the
point 2. In order to have no deflection at 1, the points 1 and 2 must be the
same.
Considering the aircraft wing as a thin-walled beam, we may analyze two
scenarios: (a) The lift load passes through a point in front of the shear center
and (b) it acts through a point behind the shear center. In the first case if
the lift increases by a small amount it produces a torque and the angle of
attack would also increase. This results in more lift. This is a destabilizing
process straining the wing to the point of failure. In the second case, any
increase in lift would reduce the angle of attack and decrease the lift, which
is a stabilizing process. This example shows the importance of determining
the shear center accurately during the design process so that the lift acts
behind the shear center.

22

CHAPTER 5. BENDING

5.3

Multi-cell Tubes

Unlike an open-cell tube, closed-cell tubes have no stress-free edges, s = 0,

to begin the shear flow evaluation using the formula
V
q=
I

ytds.

(5.45)

s=0

In such cases, we proceed in two steps: (a) we make sufficient number cuts at
convenient points (the cut cells must have certain symmetry if possible) to
transform all cells to open cells and compute the varying shear flow distribution, q v , using Eq. (5.45) with q v = 0 at the newly created stress-free edges,
and (b) apply unknown constant shear flows, qi0 , around the ith cell. In fact,
these are the shear flows at the cuts. Finally, we solve for the unknown shear
flows, qi0 , using the requirement that there is no rotation during bending with
the shear force acting through the shear center. Recall that the rate of twist
of the ith cell is given by
1
qds
=
.
(5.46)
2Ai G i t
Fig. 5.14 shows a single-cell tube before and after making a cut.
s
y
s=0

Figure 5.14: A single-cell tube with and without a cut.

With one cut we have one unknown q 0 and the rate of twist relation
becomes
q v ds
ds
0
=
,
(5.47)
q
t
t
where q v is given by Eq. (5.45).

23

5.3. MULTI-CELL TUBES

5.3.1 Shear flow transfer at wall intersections

When three or more walls meet at a node, as shown in Fig. 5.15, a force
balance of the node in the axial direction establishes a relation among the
shear flows at the node. For example, at node 2.
v
v
v
v
q23
= q12
+ q82
+ q92
.

(5.48)

The constant shear flows qi0 are handled in the same way as we did in torsion.
2

3
7

10

11

12

q10

1
6

q20

q30

q40

Figure 5.15: A four-cell tube with and without cuts.

We use the electrical current analogy and obtain a shear flow of q10 q20 in
the wall between cell 1 and cell 2. This is also shown in Fig. 5.15.
5.3.2 Example: Two-cell tube
Obtain the shear flows and the shear center for the two-cell tube shown in Fig.
5.16. Assume a thickness of t for the middle web and 2t for the remaining
walls. The smaller cell is 2a 2a and the larger cell is 4a 2a in dimension.
The value of I under the thin-wall assumption is obtained as
I=2

2a3 2t 2a3 t
46
+
+ 2 2ta a2 + 2 4ta a2 = a3 t.
3
3
3

24

CHAPTER 5. BENDING

q10

q20

2
s
10

1
9

11

Figure 5.16: A two-cell tube without and with cuts.

We begin by cutting the right web and the middle web to create an open
cell. Next, we number the nodes as shown in the figure. In computing
qijv =

V
I

ytds,

as the integral is cumulative, we assume a new s coordinate starting from

the node i. Thus,
v
q12
=

V
I

y2tds =
0

Vt
I

2sds =
0

When s = a, this shear flow has a value

v
q12
|2 =

V ta2
,
I

Vt 2
s.
I

25

5.3. MULTI-CELL TUBES

which is to be added to the shear flow calculation from 2 to 3.
v
q23
=

Vt 2
a +
I

2ads =
0

v
q103
=
v
q34
=

Vt 2
a + 2as ,
I

V ts2
,
2I

V t a2
+ 5a2 +
I
2

v
q23
|3 =

5V ta2
.
I

V ta2
.
2I
V t a2
a2ds =
+ 5a2 + 2as .
I 2
v
q103
|3 =

v
q34
|4 = f rac15V ta2 2I.

Along 45, the value of y is (a s) and

v
q45

V t 15a2
=
+
I
2

s
0

V t 15a2
+ a2 (a s)2 ,
(a s)2ds =
I
2

17V ta2
.
2I
For the lower half of the beam the value of y is negative and the integrals
contribute negative values to q v . Due to this, we get the same values for q v as
in on the upper half, except the direction of the shear flow is opposite of what
is in the upper half, that is, from left to right or from top to bottom. Fig
5.17 shows the magnitudes (not to scale) and directions of the shear flows,
q v , after removing a factor V ta2 /I.
The shear forces produced by these shear flows are obtained as
v
q45
|5 =

q v ds,

Fij =
i

which are the areas under the magnitude curves shown in Fig 5.17. All
our computed shear flows have a minus sign in front. We include this in
computing the shear forces:
F21 =

1 V ta3
,
3 I

F32 = 6

V ta3
,
I

F43 =

13 V ta3
,
2 I

1 V ta3
49 V ta3
, F310 =
.
6 I
6 I
As a check, we may compute the net vertical force due to the shear flows.
F54 =

1 49 1 1 V ta3
)
= V.
2(F12 + F54 + F103 ) = 2( +
3
6
3 6 I

26

CHAPTER 5. BENDING
7.5

5.5
5
1

7.5

0.5
8.5

7.5

5.5
7.5
Figure 5.17: The shear flow distribution after making the cuts (the magnitudes are not to scale).
Again, the contributions from the lower half have the same magnitude
but the opposite sign.
The condition that the rate of twist of each of the two cells must be zero,
that is,
qds/t = 0,
is used to obtain the shear flows at the cuts.
q10

2a
2F54 2F43 F310
4a 2a
q20
=
+
+
+
,
2t
t
t
2t
2t
t

which can be simplified to get

4q10 2q20 = 15

V ta2
.
I

From the second cell,

q10

2F310 2F32 2F21

,
+
+
t
2t
2t

27

5.4. SHEET-STRINGER CONSTRUCTION

or
2q10 + 5q20 = 6

V ta2
.
I

Solving the two simultaneous equations,

q10 =

87 V ta2
,
16 I

q20 =

27 V ta2
.
8 I

The net shear flow in any wall is the sum of q 0 and q v . Thus
q12 =
q23 =
q103 =
q34 =
q45 =

27
8
19
8
87
16
25
16

s 2 V ta2
19 V ta2
, q12 |2 =
,
a
I
8 I
s V ta2
13 V ta2
2
, q23 |3 =
,
a
I
8 I
27 1 s 2 V ta2
25 V ta2

, q103 |3 =
,
8
2 a
I
16 I
13
s V ta2
33 V ta2

2
, q34 |4 =
,
8
a
I
16 I

33
+1
16

sa
a

V ta2
,
I

q45 |5 =

17 V ta2
.
16 I

Having obtained all the shear flows, we are in a position to compute the
shear center. For this we choose a convenient point to compute the torque
produced by the shear flows. The point 5 is appropriate for this as the shear
force F45 passes through it. The net torque is
T = 2A1 q10 + 2A2 q20 2
=

13
2 1 V ta4
263
,
2
3 3
I

257
257 V ta4
=
V a.
12 I
184

The applied vertical force V must applied through a point, 257a/184 to the
right of the node 5.

5.4

Sheet-Stringer Construction

In aircraft structures, except for the spars, solid section beams are seldom
used. The bending stresses are carried by stringers (long metal bars with

28

CHAPTER 5. BENDING

cross-sections in the shapes of L, Z, or T.). Sometimes these are also

called longerons. These are kept in position by outer metal sheets made of
aluminum. In a preliminary design the exact shape of the stringer crosssection is not important but their areas and the location of centroid are.
We use circular cross-sections, which make them look like the booms used
to support sails in boats. The stress analysis of this simplified structure is
called boom analysis. Our basic assumption is that the sheets carry shear
stresses and the booms carry normal stresses.
As a simple example, let us consider two booms of area B connected by
a sheet of thickness t and height 2c. Under the elementary beam theory
assumptions the axial strain is distributed as
= y,

(5.49)

which gives the stress at the centroid of the boom 1,

= c,

(5.50)

= c.

(5.51)

and at the centroid of boom 2,

The axial force in the top boom is

F = B.

(5.52)

The applied moment M is given by

M = 2cF = 2cB = 2c2 B.

(5.53)

I = 2Bc2 ,

(5.54)

We may set
to have

M
Mc
MBc
, =
, F =
.
(5.55)
I
I
I
As shown in Fig. 5.18, for an infinitesimal length dx, the change in force
=

dF = dB = dMBc/I = qdx,

(5.56)

and

V Bc
dM Bc
=
.
dx I
I
The shear flow is a constant between the two booms.
q=

(5.57)

29

5.4. SHEET-STRINGER CONSTRUCTION

dx

B
t

dF

2c

B
Figure 5.18: Two booms connected by a web with the shear flow produced
by a varying axial stress.
5.4.1 Example: Open cell
Consider the open cell tube shown in Fig. 5.19. The horizontal walls have a
thickness t and the vertical wall has a thickness of 2t. The boom areas are
shown in the figure. We want to compute the shear flows in all the walls and
the location of the shear center. For this cell,
a

2a

2B

Figure 5.19: An open cell beam under the boom approximation.

I = 10Ba2 .
Using the node numbering shown in Fig. 5.19,
q12 =

V
V Ba
=
,
I
10a

q23 =

2V
,
10a

q43 =

V
,
10a

30

CHAPTER 5. BENDING

5V
2V
=
.
10a
10a
Fig. 5.20 shows the shear flows as multiples of V /(10a).
q36 = q23 + q43

5
O
1

Figure 5.20: Shear flow distribution as multiples of V /(10a).

To find the shear center for this beam, let us take the moment due to the
shear flows about the point O on the neutral axis. This gives a clock-wise
moment of,
4V a
V ta4
=
.
(5.58)
M = 2q23 a2 = 4
I
10
The shear force has to be acting through a point 2a/5 to the left of the
vertical web to create this moment.
5.4.2 Example: A two-cell beam under boom approximation
Consider the two-cell beam with a uniform wall thickness t and a constant
boom areas B shown in the left sketch in Fig. 5.21. The cells are numbered
from left to right. Our objectives are to find the shear flows in the walls
and the location of the shear center. The middle sketch shows the node
numbering after applying two cuts. For this geometry we have
I = 20Ba2 .
Starting from the node 1,
v
q12
= 0,

v
q23
=

V Ba
,
I

v
q v 123 = 0, q34
=

2V Ba
,
I

31

5.4. SHEET-STRINGER CONSTRUCTION

B
2a

a 6

12
13

9
B

2a B

2a

2a B

2
1
11
10

44
56

48

11

44

34
11

Figure 5.21: Two-cell beam: geometry, node numbers, and shear flows in
multiples of V Ba/(23I).
6V Ba
4V Ba
v
, q56
=
.
I
I
We can extend these to the lower half as we have done before. The areas of
the cells are: A1 = 8a2 and A2 = 4a2 . The constant shear flows at the cuts
are found from
qds/t = 0,
v
q45
=

for each cell. Thus,

V Ba2
44V Ba2
[6 4a + 4 4a + 2 2a] =
,
I
I
4V Ba2
V Ba2
[1

4a]
=
.
=
I
I

12aq10 2aq20 =
2aq10 + 8aq20

Solving these equations give

q10 =

90V Ba
,
23I

q20 =

34V Ba
.
23I

Adding the constant shear flows q 0 to the varying shear flows q v , we get
34V Ba
,
23I
44V Ba
,
=
23I

11V Ba
56V Ba
, q123 =
,
23I
23I
2V Ba
48V Ba
=
, q56 =
.
23I
23I

q12 =

q23 =

q34

q45

Here, if qij is positive the shear flow is directed from node i to node j. The
right sketch in Fig. 5.21 shows the shear flows in direction and magnitudes
in multiples of V Ba/(23I).

32

CHAPTER 5. BENDING

In computing the location of the shear center, it is more convenient to

keep q 0 and q v separate. If we take an anti-clockwise moment about the node
6, we find its total value is
M = 2A1 q10 + 2A2 q20

35V a
V Ba3
[32 + 8 + 4] =
.
I
23

This shows the shear force V has to be applied through a point which is
located inside cell 1, at a distance 35a/23 from the node 6.

5.5

Combined Bending and Torsion

When a shear force V does not pass through the shear center there will be
bending and twisting of the beam. To analyze this case, we may consider
two approaches: (a) Find the shear center and separate the problem into a
bending problem with the shear force through the shear center and into a
torsion problem with a torque equal to the shear force times the distance
between the shear center and the actual point of application of the load or
(b) Find the shear flow distribution which introduces a rate of twist and zero
moment about the point of application of the shear force. In the example
below, we consider the case (b).
5.5.1 Example: Single-cell Beam
The single-cell beam shown in Fig. 5.22 has boom areas B and web widths 2a.
The wall thickness of the left wall is 3t and all other walls have thicknesses
of t. Assume a uniform shear modulus of G. A shear force V is applied
vertically through the left wall. The second moment of the area for the
booms is
I = 4Ba2 .
After introducing a cut in the right wall,
v
q12
= 0,

v
q23
=

V Ba
,
I

v
q34
=

2V Ba
.
I

We extend these anti-symmetrically to the lower half of the beam. A shear

flow of q 0 is added to close the cut. The moment about the point 4 is found
as
V Ba
.
M = 8a2 q 0 2 2a2
I

33

5.5. COMBINED BENDING AND TORSION

B

2a

2a

2a

Figure 5.22: Single-cell beam: geometry, node numbers, and shear flows in
multiples of V Ba/(23I).
This moment is zero as the shear force passes through the point 4. Then
V Ba
.
2I

q0 =
The rate of twist is obtained from
=

1
2AG

qds
,
t

as

V
.
16Ga2 t
As we have q positive in the counter clock-wise direction, the negative sign
in front of implies a clock-wise rate of twist.
If the shear force were to pass through the shear center, there will not be
any rate of twist. In this case we compute q 0 using
=

qds
= 0,
t

q0

6a 2a
V Ba 4a 4a
=
.
+
+
t
3t
I
t
3t

This gives
q0 =

4 V Ba
.
5 I

34

CHAPTER 5. BENDING

The moment about the node 4 is now obtained as

M = 2Aq 0

V Ba 2
4a ,
I

M=

3V a
.
5

Thus, the shear center for this beam is 3a/5 to the right of node 4. In original
problem, the shear force passes through the node 4. This is equivalent to a
force of V through the shear center and a counter clock-wise torque
T = 3V a/5.
Using the formula
=

T
4A2 G

ds
,
t

we get value of
V
,
16Ga2 t
which is the same as we obtained earlier.
=

5.6

Analysis of Tapered Beams

Most tubular beams in aircraft made of the sheet-stringer construction are

tapered to accommodate variable bending moments, shear forces and aerodynamic shapes. We present an approximate analysis of the effect of taper
below. To simplify the analysis, we assume the beam is of uniform width in
the z-direction, symmetric about the z-axis and it tapers in the y-direction.
We further assume the tapered booms will converge to a single point if extended.
5.6.1 Quadrilateral Shear panels
In the case of tapered sheet-stringer construction, we encounter quadrilateral
sheets (shear panels) pictured in Fig. 5.23. We make the assumption that
the tapered stringers meet at a common point O. With the node numbers
shown in the figure, we may denote the side lengths by hij where i and j are
the node numbers. Taking the moment about O,
q12 h12 p1 = q34 h34 p3 ,

q34 = q12

h12 p1
p2
= q12 12 ,
h34 p3
p3

(5.59)

PSfrag
35

5.6. ANALYSIS OF TAPERED BEAMS

q23

q12

q34
2
O

q41

p3

p1

Figure 5.23: Quadrilateral shear panel with assigned nodes.

where we have used a property of similar triangles. Given q12 , this formula
can be used to find q34 along any vertical cut and it is inversely proportional
to the distance to O.
By balancing the horizontal forces,
q23 h23 cos 2 = q41 h41 cos 1 ,

q23 = q41 .

(5.60)

As we will see q23 and q41 are average shear flowstheir local values vary from
node-to-node. Balancing the vertical forces,
q41 h41 sin 1 + q34 h34 = q12 h12 + q23 h23 sin 2 ,

(5.61)

where the lengths hij are related in the form

h34 h12 = h41 sin 1 h23 sin 2 .

(5.62)

Using this, the vertical force balance gives,

q41 = q12

p1
.
p3

(5.63)

This expression shows that the average value of the shear flow depends on
the distances of the end nodes from O.
5.6.2 Shear flow due to bending in tapered beams
Let us consider a beam consisting of N booms, all converging to a line through
the point point O in the x, y-plane. The strain formula
= y,

(5.64)

36

CHAPTER 5. BENDING

gives the change of length in the x-direction for a boom segment of length
unity. If the boom segment is oriented along a unit vector
n = nx i + ny j,

(5.65)

then the boom elongation for the ith boom is yi /nx , where yi is the coordinate of the boom. The x-axis forms the neutral axis. The stress in the
boom is Eyi /nx . This creates a force
Pi = EBi yi /nx ,

(5.66)

where Bi is the boom area. Its x-component will be

Pix = EBi yi .

(5.67)

M =

(5.68)

Using
we find
=
and
Pix =

M
,
EI

Pix yi ,

I=

MBi yi
,
I

Pi =

Bi yi2 ,

(5.69)

MBi yi
.
Inx

(5.70)

From Fig. 5.24, the equilibrium equation relating bending moment and shear
force is obtained as
N
dM
= V
Pjy ,
(5.71)
dx
j=1
where Pjy = Pix tan i is the y-component of the force in the j t h boom. As
Pi + Pi

Pi y i
O

i
M

V + V M + M

x
Figure 5.24: Element of a tapered beam.

37

5.6. ANALYSIS OF TAPERED BEAMS

Pi

q3

q2
q1

Figure 5.25: Element of a boom at the intersection of three webs.

shown in Fig. 5.25, a force balance on the boom segment of length x/nx ,
shows
dPix
.
(5.72)
Pi = (q1 + q2 + q3 )x/nx , q1 + q2 + q3 =
dx
Using Eq. (5.70) and assuming Bi are constants, we find
dPix
dM Bi yi MBi dyi MBi yi dI
=
+

.
dx
dx I
I dx
I 2 dx
Our assumption that all the booms converge to a point O, implies

yi = x tan i .

(5.73)

(5.74)

Then,

dPix
dM Bi yi MBi
MBi yi 2Bj yj tan j
=
+
tan i
dx
dx I
I
I2
,
Bi yi MBi yi
MBi yi
= V
Pjx tan j
+
2
,
I
Ix
Ix
MBi yi
V Bi yi MBi yi
Bj yj tan j
+
,
=
2
I
I
Ix
V Bi yi MBi yi I
MBi yi
=
+

,
I
I2 x
Ix
V Bi yi
.
(5.75)
=
I

38

CHAPTER 5. BENDING

Finally
q1 + q2 + q3 =

V Bi yi
.
I

(5.76)