Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.

ca

Euler-Bernoulli Beams
The Euler-Bernoulli beam theory was established around 1750 with contributions from
Leonard Euler and Daniel Bernoulli. Bernoulli provided an expression for the strain
energy in beam bending, from which Euler derived and solved the differential equation.
That work built on earlier developments by Jacob Bernoulli. However, the beam problem
had been addressed even earlier. Galileo attempted one formulation that aimed at
determining the capacity of beams in bending, but misplaced the neutral axis. Earlier,
Leonardo da Vinci also seems to have addressed the problem of beam bending. The two
key assumptions in the Euler-Bernoulli beam theory are:
• The material is linear elastic according to Hooke’s law
• Plane sections remain plane and perpendicular to the neutral axis
The latter is referred to as Navier’s hypothesis. In contrast, Timoshenko beam theory,
which is covered in another document, relaxes the assumption that the sections remain
perpendicular to the neutral axis, thus including shear deformation. In the following, the
governing equations are established, followed by the formulation and solution of the
differential equation. Thereafter, a section cross-section analysis describes the
computation of stresses and cross-section constants. The starting point is 2D beam
bending with the following sign conventions:
§ The z-axis is increases upwards
§ Displacement w is positive in the direction of the z-axis
§ Distributed load qz is positive in the direction of the z-axis
§ Bending moment that imposes tension at the bottom is positive
§ Clockwise shear force is positive
§ Counter-clockwise rotation q is positive, thus it can be interpreted as the slope of
the deformed beam element
§ Tensile stresses and strains are positive, compression is negative

Equilibrium
The equilibrium equations are obtained by considering equilibrium in the x-direction for
the infinitesimal beam element in Figure 1. Notice that the distributed load, q, acts in the
downward direction, while the z-axis is in the upward direction. The notation qz is
employed in other documents to identify the case where positive load acts in the positive
z-direction.
Vertical equilibrium yields:
dV
qz = (1)
dx
Moment equilibrium about the rightmost edge yields:
dM
V= (2)
dx

Euler-Bernoulli Beams Updated February 2, 2020 Page 1

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

In Eq. (2) the “second-order terms” that contain dx2 are neglected.

qz

M V V+dV M+dM

dx

Figure 1: Equilibrium for infinitesimally small beam element.

Section Integration
Integration of axial stresses over the cross-section:
M = ∫ −σ ⋅ z dA (3)
A

where the minus sign appears because it is compressive (negative) stresses in the positive
z-axis domain that gives a positive bending moment, i.e., bending moment with tension at
the bottom. Figure 2 is intended to explain this further.

•  Negative compression stress

z
•  Positive z-values

M M

x Minus sign in cross-section

integral is necessary to get
positive bending moment

•  Positive tension stress

•  Negative z-values

Figure 2: The reason for the minus sign in Eq. (3).

Material Law
The material law throughout linear elastic theory is Hooke’s law:

Euler-Bernoulli Beams Updated February 2, 2020 Page 2

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

σ = E ⋅ε (4)
In the context of two-dimensional theory of elasticity, the use of Eq. (4) implies a “plane
stress” material law. It implies that there is zero stress, i.e., air on the sides of the beam.
The alternative “plane strain” version of the two-dimensional Hooke’s law is more
appropriate in cases where the beam is only a strip of a long rectangular plate that is
supported along the two long edges. In that case the strain is restrained in the y-direction:
σ yy σ
ε yy = − ν ⋅ xx = 0 ⇒ σ yy = ν ⋅ σ xx (5)
E E
Which, substituted into the material law in the x-direction yields:

ε xx =
σ xx σ σ
− ν yy = xx − ν
(ν ⋅ σ xx ) = σ xx (1− ν 2 ) ⇒ σ xx =
E
⋅ ε xx (6)
E E E E E 1− ν 2
All the derivations and results in the following are based on the material law sxx=E.exx
from Eq. (4). However, the plain strain version is easily introduced by replacing the
Young’s modulus, E, in any equation by E/(1-n2).

Kinematics
The relationship between the axial strain and the transversal displacement of a beam
element is sought. It is first recognized that bending deformation essentially implies
shortening and lengthening of “fibres” in the cross-section. Fibres on the tension side
elongate, while fibres on the compression side shorten.
The starting point for the considerations is to link the axial strain to the change of length
of the imaginative fibres that the cross-section is made up of. The same consideration as
in kinematics of truss members, namely that strain is “elongation divided by original
length” yields:
du (7)
ε=
dx
z, w

d!

z
x, u

Figure 3: Navier’s hypothesis for beam bending.

Next, the axial displacement u is related to the rotation of the cross-section. In particular,
consider the infinitesimal counter-clockwise rotation dq of the infinitesimally short beam

Euler-Bernoulli Beams Updated February 2, 2020 Page 3

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

element in Figure 3. In passing, it is noted that dq is equal to the curvature, k. Under the
assumption that plane sections remain plane and perpendicular to the neutral axis during
deformation, each fibre in the cross-section change length proportional to its distance
from the neutral axis.
The amount of shortening or elongation depends upon the rotation of the cross-section. A
geometrical consideration of to Figure 3 shows that the shortening and lengthening, i.e.,
axial displacement, of each infinitesimally short fibre is
du = −dθ ⋅ z (8)
Finally, the rotation q is related to the transversal displacement. For this purpose,
consider two points on a beam that is dx apart, as shown in Figure 4. The relative
displacement is dw, which is measure positive upwards. Consequently, a geometrical
consideration of Figure 4 shows that:
dw
tan(θ ) = ≈θ (9)
dx
where the equation is simplified by assuming that the deformations are sufficiently small
so that tan(q)≈q.
!


!
dw

dx
Figure 4: Rotation of the cross-section of a beam element.

Combination of the previous equations yields the following kinematic equation for beam
members:
du dθ d 2w
ε= =− ⋅z = − 2 ⋅z (10)
dx dx dx
This expression implies an approximation of the exact curvature of the beam.
Mathematically, curvature is defined as
1
κ≡ (11)
R
where R is the radius of curvature of the beam. In the Euler-Bernoulli beam theory that is
presented here, the curvature is approximated by
dθ d 2 w
κ≈ ≈ (12)
dx dx 2

Euler-Bernoulli Beams Updated February 2, 2020 Page 4

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

Notice that there are two approximation signs. The first alludes to the fact that
differentiation is carried out with respect to the x-axis. Unless the deformations are
negligible this is inaccurate; differentiation should be carried out with respect to the “s-
axis” that follows the curving beam axis. The second approximation is due to Eq. (9).
From that equation it is observed that the accurate expression for q is:
⎛ dw ⎞
θ = tan −1 ⎜ ⎟ (13)
⎝ dx ⎠
If this expression was utilized in the derivations above then the differentiation of the
inverse tan-function yields
⎛ d 2w ⎞
dθ ⎜⎝ dx 2 ⎟⎠
κ≈ = (14)
dx ⎛ ⎛ dw ⎞ 2 ⎞
⎜ 1 + ⎜⎝ dx ⎟⎠ ⎟
⎝ ⎠
which reduces to the expression in Eq. (12) when the slope dw/dx is small. However,
the curvature expression in Eq. (14) is still approximate because the differentiation is
carried out with respect to the x-axis and not the beam axis. From mathematics, the exact
curvature expression is:
⎛ d 2w ⎞
⎜⎝ dx 2 ⎟⎠
κ= 3
(15)
⎛ ⎛ dw ⎞ ⎞ 2 2

⎜ 1 + ⎜⎝ dx ⎟⎠ ⎟
⎝ ⎠

Differential Equation
The governing equations for beam bending, namely equilibrium, section integration,
material law, and kinematics are summarized in Figure 5. The differential equation is
obtained by combining them as follows:
dV d 2 M d2
qz = = = − 2 ∫ σ ⋅ z dA
dx dx 2 dx A
d2 d2 d 2w 2
dx 2 ∫A dx 2 ∫A
=− E ⋅ ε ⋅ z dA = E ⋅ ⋅ z dA (16)
dx 2
d 4w
= EI y
dx 4
where the modulus of elasticity is assumed constant over the cross-section and the
moment of inertia is defined:
I y = ∫ z 2 dA (17)
A

Euler-Bernoulli Beams Updated February 2, 2020 Page 5

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

In Eq. (16) it is assumed that the cross-section is homogeneous so that E is constant. For
composite cross-sections this assumption is invalid, and the revised version of Eq. (16) is
d 4w
dx 4 ∫A
qz = E ⋅ z 2 dA (18)

One approach to retaining the original definition of I is to first select a reference-value of

the Young’s modulus and assume that all parts of the cross-section has that E-value.
Next, the width of each part of the cross-section is modified if its Young’s modulus is
different from the reference value. The change in width is proportional to the difference
in E-value. E.g., if E is twice the reference value then the width should be doubled. If this
procedure is followed then Eq. (17) remains valid for the determination of I.
d 4w
qz qz = EI y w
dx 4
dV
qz =
dx
d 2w
dM M = EI y
V= dx 2
dx
d 2w
ε=− ⋅z
M dx 2

M = ∫ −σ ⋅ z dA
A

σ σ = E ⋅ε ε
Figure 5: Governing equations in Euler-Bernoulli beam theory.

Although solving the differential equation for beam bending is rarely done in everyday
engineering practice, it is instructive to study its solution for simple reference cases. In
particular, the solution of the differential equation is the starting point for the selection of
shape functions in the finite element method. Those shape functions are often
approximate, while the solution of the differential equation reveals the exact shape when
the member deforms. The general solution of the differential equation reveals whether the
finite element shape functions are exact or not. For beam members, the general solution
of the differential equation when the load qz is uniformly distributed along the beam is
obtained by integrating four times:
1 qz 4
w(x) = ⋅ ⋅ x + C1 ⋅ x 3 + C2 ⋅ x 2 + C3 ⋅ x + C4 (19)
24 EI y
It is observed that under uniform load the displaced shape of a beam is a fourth-order
polynomial. Without distributed load the displaced shape is a third-order polynomial. To

Euler-Bernoulli Beams Updated February 2, 2020 Page 6

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

obtain the solution for a specific beam problem it is necessary to specify boundary
conditions. To prescribe a rotation, shear force, or bending moment, the following
equations are useful, obtained by combining the governing equations established earlier:
dw
θ= (20)
dx
d 2w
M = EI y (21)
dx 2
d 3w
V = EI y (22)
dx 3

Cross-section Constants
Centroid
The centroid, whose coordinates are denoted yo and zo, is the location about which the
first moments of area must be zero. Throughout this document the y- and z-axes are
assumed to originate from the centroid. For the sake of these derivations, suppose another
coordinate system ( y! , z! ) has its origin elsewhere. That means the condition of zero first
moments of area read

∫ y dA = ∫ ( y! − y )dA = ∫ y! dA − ∫ y
A A
o
A A
o dA = ∫ y! dA − yo ⋅ A = 0
A
(23)
∫ z dA = ∫ (z! − zo )dA = ∫ z! dA − ∫ zo dA = ∫ z! dA − zo ⋅ A = 0
A A A A A

Solving for the centroid coordinates yo and zo yields:

∫ y! dA
yo = A
A
(24)
∫ z! dA
zo = A
A
In words, the location of the neutral axis is obtained by summing “area multiplied by
distance” for all parts of the cross-section, and then dividing the sum by the total area.
For homogeneous cross-sections the neutral axis along the beam coincide with the
geometrical centroid of the cross-section. For composite cross-sections, i.e., cross-
sections composed of different materials, it is possible to scale the area of each cross-
section part about its local centroid. The scaling is proportional to the E-value relative to
a reference value. Subsequently, the centroid of the scaled cross-section is determined,
which is the sought neutral axis.
Moment of Inertia
The cross-section of any beam member has two moments of inertia, Iy and Iz. If y and z
are the principal axes of the cross-section, then these two moments of inertia are the

Euler-Bernoulli Beams Updated February 2, 2020 Page 7

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

largest and smallest among all possible orientations of the orthogonal y and z axes. The
subscripts on Iy and Iz indicate the axis about which the cross-section rotates under
bending. As a result, their definitions are:
I y = ∫ z 2 dA
A
(25)
I z = ∫ y 2 dA
A

For general cross-sections it is often convenient to divide the cross-section into parts,
each with local moment of inertia denoted by Ii about its local centroid. The contributions
to the global moment of inertia from each part, about the global centroid of the cross-
section, are summed in accordance with the parallel axis theorem, sometimes referred to
as “Steiner’s sats:”

(
I y = ∑ I y,i + zi2 ·Ai )
(26)
Iz = ∑( I z,i + y ·A )
2
i i

where yi and zi are distances from the centroid of the entire cross-section, i.e., the neutral
axis along the beam, to the centroid of the part. In cases where Iy and Iz about the
principal axes are sought, but y and z are not the principal axes, Iy,principal and Iz,principal can
either be calculated from scratch once the orientation, qprincipal, of the principal axes are
determined, or they can be determined by transformation of the original moments of
inertia (Megson 2020):

I y, principal =
Iy + Iz 1
2
+ ⋅
2 (( I − I ) + 4 ⋅ I )
z y
2 2
yz
(27)
I z, principal
I +I 1
= y z− ⋅
2 2 (( I − I ) + 4 ⋅ I )
z y
2 2
yz

Product of Inertia
The product of inertia, Iyz=Izy, is a quantity that is non-zero only for cross-sections
without any axis of symmetry. If a symmetry axis exists then it is a principal axis, and
thus both principal axes are immediately known. The product of inertia should not be
confused with the polar moment of inertia, which is related to torsion. The product of
inertia appears in the derivation of the governing equations for bi-axial bending, where it
is defined as:
I yz = ∫ y ⋅ z dA (28)
A

The coordinates y and z in Eq. (28) represent distances from the neutral axis, which is a
point in the cross-section, to the area elements. For cross-sections with a symmetry axis,
Iyz=0. The parallel axis theorem, i.e., Steiner’s formula, for products of inertia is:

(
I yz = ∑ I yz,i + yi ·zi ·Ai ) (29)

Euler-Bernoulli Beams Updated February 2, 2020 Page 8

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

where yi and zi are distances from the global centroid to the local centroid. For example,
the value of Iyz for a triangular cross-section with dimensions b and h is
b2h2
I yz = (30)
72
Principal Axes
The identification of the principal axes of a cross-section has several benefits; it
simplifies the equations for stress and strain and it implies that Iy and Iz are the greatest
and smallest moments of inertia of the cross-section. The product of inertia is employed
to determine the principal axes; once Iyz is determined the following conclusions can be
made in regards to the principal axes:
• If Iyz=0 then y and z are the principal axes
• If Iyz≠0 then y and z are NOT the principal axes
The orientation of the principal axes relative to the original axes is
1 ⎛ I ⎞
θ principal = ⋅ arctan ⎜ 2 ⋅ yz ⎟ (31)
2 ⎝ Iy − Iz ⎠
If Iy=Iz then that formula cannot be evaluated, but then only two options exist: The first is
that Iyz=0, in which case qaxes=0; the only other possibility is that qaxes=±45°.

Axial Stress
For uniaxial bending, a convenient approach for obtaining the axial stress in terms of the
bending moment is to combine material law and kinematics equations, which yields:
d 2w
σ = −E ⋅ 2 ⋅ z (32)
dx
Then substitute the differential equation without equilibrium equations to obtain:
M
σ =− ⋅z (33)
I
The minus sign means that a positive bending moment, i.e., tension at the bottom,
correctly yields negative stresses at the top, i.e., compression, where z is positive. This is
the same reason that was given for the minus sign in Eq. (33), which also correctly gives
positive tension stresses at the bottom when a positive moment acts on the cross-section.
For biaxial bending, Eq. (33) is valid for bending about both the y and z axes, as long as
they are the principal axes. If they are not the principal axes but their orientation, f,
relative to the original axes is known, then it is possible to develop formulas for stress in
terms of f and the original axes. These formulas will be more complicated than the
fundamental ones because they include the product of inertia, Iyz. The starting point is the
expression for strain at an arbitrary location in the cross-section:

Euler-Bernoulli Beams Updated February 2, 2020 Page 9

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

d 2v d 2w
ε = εo − 2 ⋅ y − 2 ⋅ z (34)
dx dx
The material law yields the preliminary expression for stress:
d 2v d 2w
σ = E ⋅εo − E ⋅ ⋅ y − E ⋅ ⋅z (35)
dx 2 dx 2
Integration of axial stress yields the bending moments My and Mz about the y and z axes,
respectively. First consider the bending moment about the y-axis:
d 2v d 2w
M y = − ∫ σ ⋅ z dA = −E ⋅ ε o ⋅ ∫ z dA + E ⋅ 2 ∫
⋅ y ⋅ z dA + E ⋅ 2 ∫
⋅ z 2 dA (36)
A A
dx A dx A
The first integral vanishes because z originates at the neutral axis, while the last term is
the ordinary bending moment. As a result, Eq. (36) and its counterpart for Mz can be
written as:
d 2v d 2w
M y = −EI yz ⋅ 2 − EI y ⋅ 2
dx dx
2
(37)
d w d 2v
M z = EI yz ⋅ 2 + EI z ⋅ 2
dx dx
where the product of inertia, Iyz, has been defined as:
I yz = ∫ y ⋅ z dA (38)
A

Similarly, Iyz appears in the expression for Mz. Eq. (37) represents kinematics, material
law, and section integration. In contrast, Eq. (35) represents only kinematics and material
law. Thus, combination of Eq. (35) and Eq. (37) facilitates the isolation of the section
integration relationship, namely the relationship between axial stress and bending
moment. This is done by solving for the curvatures in Eq. (37) and substituting them into
Eq. (35). Because the axial strain is N/EA the result is

)( ) ( )( )
N My Mz
σ= + ⋅ I z ⋅ z − I yz ⋅ y − ⋅ I y ⋅ y − I yz ⋅ z (39)
A (
I y ⋅ I z − I yz
2
I y ⋅ I z − I yz2

Neutral Axis
The line in the cross-section along which the axial stress is zero is here sought, assuming
zero axial force, N. In uni-axial bending this task is trivial; the answer is the axis around
which the bending occurs. Thus, this section addresses the problem of bi-axial bending.
The location of the sought axis depends on the loading, and consider first the case of a
moment, M, applied at a clockwise positive angle, q, relative to the horizontal y-axis. The
decomposition of M yields the following moments about each principal axis:
M y = M ⋅ cos (θ )
(40)
M z = M ⋅sin (θ )

Euler-Bernoulli Beams Updated February 2, 2020 Page 10

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

Because y and z are the principal axes, the axial stress in the cross-section is found by the
following formula from fundamental beam theory:
My M
σ= ⋅z + z ⋅y (41)
Iy Iz

In passing, it is noted that the value of q that would yield the largest axial stress at a
particular location (y, z) can here be determined. The sought neutral axis is characterized
by s=0, which according to Eqs. (40) and (41) yields the following condition:
M ⋅ cos (θ ) M ⋅sin (θ )
⋅z + ⋅y = 0 (42)
Iy Iz
Solved for z, Eq. (42) yields:
Iy
z=− ⋅ tan (θ ) ⋅ y (43)
Iz
which is the sought result. From Eq. (43), the angle, y, between the y-axis and the neutral
axis is solved from the equation
z I
tan (ψ ) = = − y ⋅ tan (θ ) (44)
y Iz

Shear Stress
When approaching shear stresses in bending, an anomaly in Euler-Bernoulli beam theory
is observed. The theory assumes that plane sections remain plane and perpendicular to the
neutral axis. In other words, the only strain that takes place is the axial shortening or
elongation of the fibres in the cross-section. Effectively, this prevents shear strain. With
no shear strain there is no shear stress, which adds up to zero shear force. In other words,
shear strain and shear force are not part of the Euler-Bernoulli beam theory. This is an
anomaly, because shear force will develop even in simple beams that are subjected to
transversal load. The anomaly is addressed by recovering the shear force by equilibrium,
once the bending moment is computed. This is seen in the simple beam theory, where the
shear force is equal to the derivative of the bending moment; this is the equilibrium
equation that recovers the shear force. Equilibrium considerations will also be employed
in this document, to determine the shear stress and “shear flow” at any point in the cross-
section. Another document on Timoshenko beam theory describes an approach to extend
Euler-Bernoulli beam theory to include shear deformation in the beam deflection. To
obtain the most popular expression for the shear stress, t, in terms of the shear force, V,
consider the infinitesimally short beam element in Figure 6. Furthermore, consider a
“cut” in the cross-section and let qs denote the “shear flow” at that location.

Euler-Bernoulli Beams Updated February 2, 2020 Page 11

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

σ σ + dσ

qs
M V+dV
V

M+dM

dx

Figure 6: Shear flow by equilibrium of infinitesimal beam element.

The shear flow, qs, is the force per unit length of the beam that ensures equilibrium with
the axial stresses, which are greater on one side than the other due to dM:
dM
qs ⋅ dx = ∫ dσ dA = ∫
As As
I
⋅ z dA (45)

where As is the cross-sectional area outside the cut. Given V=dM/dx this yields
V
qs = ⋅Q (46)
I
where the first moment of area, Q, has been defined as

Q = ∫ z dA (47)
As

To ease the evaluation of Q in practice, the cross-section is often discretized into several
parts with area Ai and distance zi from the neutral axis to the centroid of the part. Then
N
Q = ∑ zi Ai (48)
i=1

where N is the number of cross-section parts. Once Q is determined at a particular

location, the shear stress is calculated by distributing the shear flow over the thickness, t,
of the cross-section at the particular location:
V ⋅Q
τ= (49)
I ⋅t
It is noted that Eqs. (46) and (49) are not necessarily exact. Both are approximations
unless the beam is prismatic, i.e., straight along the x-axis with a uniform cross-section
along the x-axis. Even for prismatic beams Eq. (49) may not be exact because the shear
stresses may not be uniformly distributed over the thickness, t. Consider a shear force
applied in the z-direction. In that case, Eq. (49) is an approximation unless the edges of
the cross-section are parallel with the z-axis, so that the shear stresses align with the z-
axis. In other words, the cross-section for which Eq. (49) is most accurate is rectangular,
high, and thin. In contrast, it is more difficult to determine the shear stress distribution

Euler-Bernoulli Beams Updated February 2, 2020 Page 12

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

within a triangular cross-section. That being said, it is often possible to apply Eq. (49) as
an approximation, and it is pretty good even for circular and triangular cross-sections
(Timoshenko and Goodier 1969). For example, for a solid circular cross-section the edges
are locally parallel to the z-axis near the centroid. Eq. (49) can therefore be used as an
approximation to determine the shear stress at that location in the cross-section. Another
type of cross-sections where Eq. (49) is utilized is “thin-walled” cross-sections. For such
problems, a new s-axis is defined, which follows the contour of the cross-section. It is
assumed that the shear flow and shear stress follow that direction, thus satisfying the
aforementioned requirement. With that assumption, Eq. (49) can be applied without
modifications. For thin-walled cross-sections that are “closed,” i.e., consisting of one or
more “cells,” the shear stress cannot be solved by equilibrium alone, i.e., they are
statically indeterminate, as discussed next.
Closed Thin-walled Cross-sections
The determination of shear stress and shear centre for closed cross-sections, i.e., cross-
sections with cells, can be carried out in two ways. In the approach addressed first, the
shear flow is determined before the shear centre. This approach explicitly recognizes that
the calculation of shear flow in a closed cross-section is a “statically indeterminate”
problem. In other words, equilibrium equations alone are insufficient to determine the
sought forces. Each cell is associated with one redundant. Similar to the flexibility
method in fundamental structural analysis, the solution approach involves removing the
capacity of the structure to carry the sought forces, i.e., to introduce “cuts,” and then to
enforce compatibility equations that are solved for the unknown forces. In the following,
a thin-walled cross-section with one cell is considered, and one cut is introduced to make
it an open cross-section. At the cut there will develop a discrepancy in u-displacement at
each side of the cut. The compatibility equation requires this displacement to be zero:

∫ du = 0
u= (50)

where du are the infinitesimal contributions that are integrated along the s-axis, which
runs through the middle thickness along the cross-section. Figure 7 shows an
infinitesimal part of the cross-section of a beam element, seen from the side of the beam.
The figure illustrates the kinematic relationship between the shear strain, g, and the
sought quantity du:
du
γ = (51)
ds

Euler-Bernoulli Beams Updated February 2, 2020 Page 13

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

s
x, u

dx
qs

ds γ

qs
du
Figure 7: Kinematic relationship between du and shear strain g.

It is noted that generally there would be another contribution to the shear strain, namely
(df/dx)(h), where f is the rotation of the cross-section and h is the distance from the
neutral axis to the tangent of the cross-section part. However, it is understood that the
rotation of the cross-section must be zero if the shear force acts through the shear centre.
Thus df/dx=0. In short, the change in axial displacement at two points located a distance
ds apart is, from kinematics:
du = γ ⋅ ds (52)
Material law provides the following expression for the shear strain in terms of the shear
stress and thus the shear flow:
τ q
γ = = s (53)
G G ⋅t
where G=E/(2(1+n)) is the shear modulus and t is the thickness of the cross-section-wall,
which may vary around the circumference. Substitution of Eq. (53) into Eq. (52) yields
qs
du = ⋅ ds (54)
G ⋅t
Integration around the cell yields the total gap opening at the cut, which is enforced to
zero by compatibility:
qs
u = ∫ ds = 0 (55)
G ⋅t

Euler-Bernoulli Beams Updated February 2, 2020 Page 14

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

The unknown shear flow, i.e., the redundant, at the cut is denoted qo. Because the cut
cross-section is open, the shear flow at other locations is:
V
qs (s) = qo + ⋅Qdet (s) (56)
I
where Qdet is the first moment of area with zero value at the cut, evaluated according to
s s

Q = ∫ z ⋅ dA = ∫ z ⋅t ⋅ ds (57)
0 0

Substitution of Eq. (56) into Eq. (55) yields

qo V ⋅Qdet
∫
u=!
G ⋅t ∫
ds + !
I ⋅G ⋅t
ds = 0 (58)

Solving for qo yields:

Q
V ! ∫ Gdet⋅t ds
qo = − ⋅ (59)
I 1
!∫ G ⋅t ds
By comparing Eq. (59) and Eq. (46) it becomes clear that the integral fraction in Eq. (59)
represents the value of the first moment of area at the cut:
Qdet Qdet
!∫ G ⋅t ds t !∫
ds
Qo = − =− (60)
1 1
!∫ G ⋅t ds !∫ t ds
where constant shear modulus G is assumed in the last equality. According to Eq. (60),
the final Q-diagram for the cross-section can be obtained as follows:
1. Draw the Qdet-diagram for the statically determinate (cut) cross-section by evaluating
Eq. (57)
2. Obtain the numerator in Eq. (60) by integration of the statically determinate Q-
diagram divided by respective thicknesses; this integration can sometimes be
cumbersome but is essentially the integral of Eq. (57)
3. Obtain the denominator in in Eq. (60), which is straightforward because, for example
for a rectangular closed cross-section with width b, height h, and thickness t, it is
simply b/t+b/t+h/t+h/t
4. Obtain the final Q-diagram by adding the constant Qo to the statically determinate one
from the first step, rememberingt the minus-sign in Eq. (60)
Another way of evaluating Eq. (60) is to first apply integration by parts to the numerator,
which essentially moves the integration from Q to 1/Gt:

Euler-Bernoulli Beams Updated February 2, 2020 Page 15

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

⎛ 1⎞ ⎛s ⎞ ⎛ 1⎞
!∫ (Q ) ⋅ ⎜⎝ Gt
⎟⎠ ds = ∫
! ⎜
⎝0
∫ z ⋅t ⋅ ds" ⎟ ⋅ ⎜ ⎟ ds
⎠ ⎝ Gt ⎠
⎡⎛ s ⎞ ⎛s 1 ⎞⎤ ⎛s 1 ⎞
= ⎢⎜ ∫ z ⋅t ⋅ ds ⎟ ⋅ ⎜ ∫ ⋅ ds ⎟ ⎥ − !
" " ∫ ( z ⋅t ) ⋅ ⎜⎝ ∫ Gt ⋅ d "
s ⎟⎠ ds (61)
⎢⎣⎝ 0 ⎠ ⎝ 0 Gt ⎠ ⎥⎦ o 0

⎛s 1 ⎞
∫ ( z ⋅t ) ⋅ ⎜⎝ ∫ Gt ⋅ ds" ⎟⎠ ds
= −!
0

where the boundary term vanishes because the first moment of area for the entire cross-
section is zero when z originates at the neutral axis. Substitution into Eq. (60) yields:
⎛s 1 ⎞
"∫ ⎜⎝ ∫ Gt ⎟⎠ ⋅ z ⋅t ⋅ ds
⋅ d !
s
Qo = 0
(62)
1
"∫ G ⋅t ds
If the cross-section has some open parts, i.e., flanges that stick out from the part that
encloses the cell, then the following must be noted: The parts of Eq. (60) that relates to
the closed integral around the cell do not pick up contributions from the protruding
flanges, but Q does. In other words, the integration of z.t along s in the numerator in Eq.
(62) picks up contributions from the protruding flanges. This fact is reiterated shortly. For
now, notice that when the material is homogeneous, so that G is constant throughout the
cross-section, the expression in Eq. (62) simplifies to:
⎛s1 ⎞
"∫ ⎜⎝ ∫ t ⋅ ds! ⎟⎠ ⋅ z ⋅t ⋅ ds
Qo = 0
(63)
1
"∫ t ds
To reorganize this expression for practical computations it is useful to define the function
s
1
∫ t ⋅ ds
g(s) = 0
(64)
1
∫ t ds
where the denominator is a constant and the numerator varies with s. Substitution of g(s)
into Eq. (63) yields the following final expression for the first moment of area at the cut:

∫ g(s)⋅ z ⋅t ⋅ ds
Qo = ! (65)

It is reiterated that g(s) is unaffected by protruding open parts of the cross-section, while
the integration of z.t.ds must include contributions from those parts:

(
∫ g(s)⋅ z ⋅t ⋅ ds + ∑ Q flange # i ⋅ gi
Qo = ! ) (66)

Euler-Bernoulli Beams Updated February 2, 2020 Page 16

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

where Qflange is the first moment of area of the flange and gi is the value of the g-function
where the flange attaches to the cell. Once Qo is computed, the shear flow at other
locations is determined as for open cross-section, relative to the location of the cut:
V
qs (s) = ⋅ (Qo + Q(s)) (67)
I
For multi-cell cross-sections, multiple cuts are introduced to make the cross-section
statically determinate. Specifically, one cut and one compatibility equation is introduced
for each cell. At each cut, the value of the first moment of area, Qo, is determined either
by Eq. (60), i.e., direct integration of the statically determinate Q-diagram, or by Eq. (65),
e.g., integration of the auxiliary g-function. Both approaches are viable for multi-cell
cross-sections, although the use of Eq. (65) can now be somewhat more error prone. The
reason is the additional complication arising from the shear flow in the separation-walls
that have cells on both sides.
The use of Eq. (60) is first addressed. The step-wise procedure that was described above
for single-cell cross-sections is adopted here, with an important modification; the integral
of Q must include both the earlier statically determinate diagram and now also the
statically indeterminate shear flow in the separation walls:
Q Qdet (s) ⎛ Q ⎞
!∫ t ds = !∫ ds + ∑ ⎜ hwall # i ⋅ o,cell # j ⎟ (68)
t ⎝ t wall # i ⎠
where Qdet is the first moment of area for the statically determinate (cut) cross-section,
hwall and twall are the length and thickness of a wall that separates two cells, respectively,
and Qo is the redundant in the cell on the other side of the wall compared to the cell
around which the integral is conducted. In other words, the total Q-value that is integrated
consists of the statically determinate values, Qdet, plus constants Qo around each cell.
Because the shared walls couple the redundants in neighbouring cells, a system of
equations is formed, which is solved for the unknown Qo values at the cuts.
The main challenge in the use of Eq. (60) is to integrate a Q-diagram. For single-cell
cross-sections, that challenge was addressed by the introduction of the g-function and
evaluation of Eq. (65). This approach is also possible for multi-cell cross-sections, but it
is now easier to commit sign errors. In the application of Eq. (65) it is once again noted
that g(s) is unaffected by protruding flanges, while z.t.ds must include such parts. In fact,
this is how the neighbouring cells enter the continuity integral around a cell. As before,
the g-function is established around each cell, varying from zero to unity according to Eq.
(64). Eq. (65) is then evaluated, wall-by-wall around the cell, adding contribution from
the “protruding flanges” from neighbouring cells:

((
∫ g(s)⋅ z ⋅t ⋅ ds + ∑ Q flange# j ± Qo, j ⋅ g j
Qo = ! ) ) (69)

where Qflange is the statically determinate first moment of area, Qo is the contribution
from the redundant in that cell, and g is the value of the g-function where the “flange”
from the neighbouring cell attaches to the cell around with the compatibility integration is
taking place. In short, the integration around each cell includes the parts around the other
cell(s) as if they were protruding flanges.

Euler-Bernoulli Beams Updated February 2, 2020 Page 17

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

If the shear centre is known a priori then another approach is possible. Assuming the
cross-section has one cell, the shear flow is determined as follows. Also in this approach,
a “cut” is introduced, which yields an open cross-section. The coordinate s originates at
the cut and traces the cross-section around the cell. The unknown shear flow at the cut is
denoted qo, and the shear flow at all other locations is determined relative to qo in
accordance with Eq. (46), so that
V
qs (s) = qo + ⋅Q(s) (70)
I
Once qs is determined at all locations of the opened cross-section, the moment of the
shear flow about the known shear centre is computed as

T = ∫ qs ⋅ h ds
⎛ V ⎞
= ∫ ⎜ qs + ⋅Q ⎟ ⋅ h ds (71)
⎝ I ⎠
V
= ∫ qo ⋅ h ds + ∫ ⋅Q ⋅ h ds
I
where the integrals are made around the cell, starting at s=0, and h(s)=distance from the
shear centre to the tangent line of the contour of the cross-section at s. By definition the
moment, T, about the shear centre must be zero, and solving for qo yields

V ∫ Q ⋅ h ds = − V ⋅ Q ⋅ h ds
qo = − ⋅ ∫
2 ⋅ Am ⋅ I 
(72)
I  ∫ h ds
where the last equality is obtained by recognizing that the integral of h around the cross-
section is twice the cell area, Am. Having the value of qo, the shear flow is determined at
other locations with Eq. (70).
Finally, a note on shear stresses in bi-axial bending. The formulas above are derived for
uni-axial bending, i.e., bending about one of the principal axes of the cross-section. The
problem of bi-axial bending can be decomposed into two cases of uni-axial bending by
determining the principal axes and consider bending about each axis separately.
However, an alternative is to leave the principal axes unknown and rather develop stress
formulas that include the product of inertia, Iyz, which is non-zero unless y and z are the
principal axes. The expression for axial stress in Eq. (39) yields the following expression
for shear stress, again considering equilibrium, as done above for open cross-sections:
Vy Qz ⋅ I y − Qy ⋅ I yz Vz Qy ⋅ I z − Qz ⋅ I yz
τ =− ⋅ − ⋅ (73)
t I y ⋅ I z − I yz2 t I y ⋅ I z − I yz2
For reference, this expression reverts to the summation of the ordinary shear stress
expressions when the y and z axes are the principal axes:
Vy ⋅Qz Vz ⋅Qy
τ =− − (74)
I z ⋅t I y ⋅t

Euler-Bernoulli Beams Updated February 2, 2020 Page 18

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

Shear Centre
To understand the shear centre concept, think of a cantilevered beam completely fixed at
one end, with a point load applied at the free end. Examples of the cross-section that may
be seen at the free end is shown in Figure 8. The shear centre of a cross-section,
sometimes called the centre of twist, is the point in the cross-section through which the
point load, P, must act to avoid rotation of the cross-section. In other words, if the point
load does not act through the shear centre then the cantilevered beam will experience
torsion. The coordinates of the shear centre are denoted ysc and zsc, and there are several
techniques to determine them. The simplest case is double-symmetric cross-sections; for
these cross-sections the shear centre coincides with the centroid. Conversely, if the cross-
section has one axis of symmetry then the shear centre is located somewhere on that
symmetry axis.
z

(ysc, zsc) y

Shear centre P Centroid

Figure 8: Shear centre.

The document on warping torsion describes one approach to determine ysc and zsc for
thin-walled cross-sections, utilizing “omega diagrams” and the document on St. Venant
torsion presents an approach for generic solid cross-section. A simpler approach is
available for thin-walled cross-sections: The “shear flow” must be in equilibrium about
the shear centre. The determination of shear stress and shear flow is addressed shortly in
this document. Using the shear flow, the following procedure to determine the
coordinates of the shear centre is suggested, provided y and z are the principal axes
through the centroid of the cross-section:
1. Select an arbitrary point as trial shear centre, and let ysc and zsc denote the
coordinates of the shear centre relative to the centroid; in other words, let ysc and
zsc denote the distances from the centroid to the trial shear centre
2. Determine the shear flow in the cross-section due to a shear force in the z-
direction, using the formulas presented above
3. Write the equation that expresses the moment of the shear flow about the trial
shear centre; in general, both ysc and zsc will appear in this expression
4. Determine the shear flow in the cross-section due to a shear force in the y-
direction, again using the formulas presented above
5. Write the equation that expresses the moment of the shear flow determined in the
previous item about the trial shear centre; in general, both ysc and zsc will appear in
this expression
6. Set the equations from Items 3 and 5 equal to zero and solve them for the two
unknowns ysc and zsc

Euler-Bernoulli Beams Updated February 2, 2020 Page 19

Professor Terje Haukaas The University of British Columbia, Vancouver terje.civil.ubc.ca

Only one moment equation is needed for single-symmetric cross-sections, in which case
the procedure simplifies to:
1. Select an arbitrary point along the symmetry axis as trial shear centre, and let e
denote the distance from the centroid to that point
2. Determine the shear flow in the cross-section due to a shear force in the direction
perpendicular to the axis of symmetry
3. Write the equation that expresses the moment of the shear flow in Item 2 about
the trial shear centre; e will appear in this expression
4. Set the equation from Items 3 equal to zero and solve for e

References
Megson, T. H. G. (2020). Structural and Stress Analysis. Elsevier / Butterworth-
Heinemann.
Timoshenko, S. P., and Goodier, J. N. (1969). Theory of Elasticity. McGraw-Hill.

Euler-Bernoulli Beams Updated February 2, 2020 Page 20