10.

Introduction to matrix methods in structural mechanics

Contents – 1st revision
10.1 Introduction and terminology ...................................................................................................................2
10.2 The direct stiffness method ........................................................................................................................3
10.3 The plane truss element...............................................................................................................................4
10.3.1 Governing differential equation and shape functions ...........................................................4
10.3.2 Stiffness matrix assembly ..................................................................................................................6
10.3.3 Calculated example: Compound cylindrical bar in tension .................................................7
10.3.4 Transformation matrix .......................................................................................................................9
10.4 The plane truss element............................................................................................................................ 14
10.4.1 Governing differential equation and shape functions ........................................................ 14
10.4.2 Transformation matrix and constrains ..................................................................................... 17
10.4.3 Line loads: Work equivalent nodal forces ............................................................................... 17
10.4.4 Modal analysis: buckling ................................................................................................................. 18
10.4.5 Modal analysis: undamped eigenfrequencies ........................................................................ 19
10.4.6 Calculated example: simple supported pipe shaped beam ............................................... 20
10.4.7 Calculated example: MrBeam........................................................................................................ 25
10.4.8 Calculated example: modal analyses.......................................................................................... 27
10.4.9 Frame structures ................................................................................................................................ 28
Problems ............................................................................................................................................................................. 29
References .......................................................................................................................................................................... 33

Lecture Notes
Introduction to Strength of Materials pp. 1
Prof. Dr. N.H. Østergaard
10.1 Introduction and terminology
The present chapter contains an introduction to application of matrix methods for structural
analysis of linear truss and beam structures. In classical mechanics of materials, the differential
equations governing deformations of such structures due to given loads have been derived and
considered on explicit form. This has provided a framework where stresses are calculated based
on equilibrium between internal and external forces, given constitutive equations (material
laws) and (for statically indeterminate structures) equations of compability ensuring that the
deformations are physically possible. This, however, leads to a framework where a solution to a
large number of differential- and algebraic equations on explicit form is required for larger
structures. The explicit solution form, where the differential questions are required fulfilled for
all points inside the considered structure (or domain) is often referred to as strong form and is
the conventional solution form applied for analytical solution of mechanical problems.
A method for structural analysis, which is based on polynomial interpolations of the solutions to
the governing differential equations, will now be introduced. Such a solution is often called a
weak form and has as major advantage, that the equations required to solve a mechanical
problem can be derived algebraically and therefore be solved applying linear algebra. While this
inevitably also often leads to a large number of equations, these can be shown to be formulated
in a modular fashion, which is very convenient for numerical analysis. The introduced methods
constitute the basis for linear finite element analysis (FEA) and are implemented in
commercially available software for structural analysis (like ANSYS and Abaqus). However, the
underlying theoretical framework is very useful in general, both for analytical and numerical
calculations.
Our scope will be limited to structures constituted by trusses/bars and beams. The considered
systems will be divided into a set of points, which are denoted nodes. Each node can move in a
number of pre-defined ways (translation/rotation). We refer to this number of ways as degrees
of freedom (DOF). The part of the structure between two nodes will be denoted an element.
External loads can only be applied in nodes, so each DOF has a corresponding load component
(force/moment). It is our objective to organize our equations in terms of a column vector
containing the nodal deformations ( ) and a column vector containing the corresponding
external forces ( ). These two vectors will be related by a number of stiffness terms organized
in a stiffness matrix , -
( ) , -( ) 1.
The deformations along the considered element due to the applied loads are expressed solely in
terms of the nodal deformations, which are interpolated by prescribed functions. The functions
applied for interpolation of a unit displacement of a DOF will be denoted shape functions. In
Figure 1, a two node beam element is shown. Each node can deform by translation in the
longitudinal direction, , by translation perpendicular to the longitudinal direction (deflection),
u and furthermore rotate, . One node therefore has tree DOF’s yielding a total of 6 DOF’s for the
entire beam element. These are organized on the following vectorial form for a single node

( ) ( + ( ) ( + ( ) ( + ( ) ( + 2.

For the entire element, we may now write

( ) ( ) 3.
( ) ( * ( ) ( *
( ) ( )

( ) ( ) ( )

Lecture Notes
Introduction to Strength of Materials pp. 2
Prof. Dr. N.H. Østergaard
Figure 1 DOF’s and nodal loads of a two-node 6 DOF beam element.

In the following, it will be demonstrated how to derive the stiffness matrix from chosen
interpolation functions and the governing differential equations derived in classical strength of
materials. Furthermore, it will be shown how multiple element-wise stiffness matrices can be
assembled to form equations for larger structures, and how supports are added as constrains to
the linear system of equations.
In the current chapter, numerical analysis will be performed using Matlab for the simple reason,
that Matlab is available and widely applied at HSRW. The learning objective is, however, that
students are able to formulate and solve structural mechanics problems both analytically and
numerically with the developed methods. Furthermore, two Matlab scripts are provided with
the lecture notes, to demonstrate how the developed formulations are implemented for
problems of larger scale:

 MyFirstFEACode.m : An ultra simple FEA-code for truss structures similar to the

computational engine running the epic computer game Bridgebuilder.
 MrBeam.m : A FEA-code for plane (2D) beam problems

These can be downloaded here [6].

A note of great importance: Students who hated doing Matlab scripting to solve numerical
problems, should stop hating it now. Programming is an
extremely useful engineering tool, and the reason we use
Matlab is that it has the least steep learning curve. There’s an
open source Matlab clone called Octave [7] , you can use for
most simple Matlab scripts, if you end up in a company
without Matlab. Another alternative is one of the Python
distributions (fx. Python x-y [7]), which are developed for
numerical analysis and datamining.

10.2 The direct stiffness method

Stiffness matrices are often in the literature derived on basis of energy- and variational methods,
which qualitatively speaking let work done by inner force (strain energy) balance work done by

Lecture Notes
Introduction to Strength of Materials pp. 3
Prof. Dr. N.H. Østergaard
outer forces (external loads). This approach to matrix methods is elegant, abstract and perfectly
suited for master students specializing in structural mechanics. In the present approach, the
stiffness matrices will be derived from the well-known governing equations for trusses and
beams (derived in chapter 2, 4 and 8). If you are curios and cannot wait to see how to do this in a
more general and abstract manner, a brief outline is presented in [1]. For a more detailed
description, FEM-guru and professor Carlos Felippa’s lectures notes [2] are best-in-class, when it
comes to online material.
For now, we will stick to the following framework, known as the direct stiffness method, in order
to derive our own stiffness matrices:
1. Identify the differential equation governing the deformation of the considered element
2. Guess a general solution, in the considered cases, a polynomial.
3. Apply a unit deformation (or rotation) one by one to each DOF with other DOF’s
deformations set to zero.
Note that for a unit deformation, the magnitude of the nodal forces will equal the
magnitude of the stiffness constants in linear mechanics
4. Use force- and moment equilibrium to calculate the forces acting on the node due to this
particular unit deformation
5. On basis of superposition, to write the total deformation as a sum (linear combination) of
the stiffness constants determined in the previous form.
Note that this when converted to matrix form will provide the stiffness matrix , -.
This method will in section 10.3.110.3 be applied to derive , - for a truss element and in section
10.4.1 to derive , - for a beam element.

10.3 The plane truss element

We recall that truss structures are constituted by pinned-pinned bars only loaded in the joints.
Each truss is subjected to either tension or compression with no bending stresses. Under such
boundary conditions, each truss act like a spring with stiffness k:
, 4.

The nodal forces are eventually easily obtained from the deformations and can be arranged on to
matrix form as follows
( ) 5.
} ( * 0 1( * , - 0 1
( )
The stiffness matrix for a truss element has just been derived. That was easy ! However, we
forgot entirely about the direct stiffness method introduced in section 10.2. In order to practice
this method, we will re-derive equation 5 in the following.

10.3.1 Governing differential equation and shape functions

Applyinig the direct stiffness method in section 10.2, we proceed step-wise
Step 1: The first thing we have to figure out, is which equation will govern the deformations.
Considering the bar in Figure 2, we recall that the strain is given by
6.

Recalling from chapter 6 how we used to develop a force quantity over an infinitesimal length
segment using a first order (linear) Taylor approximation, we write for a distributed
longitudinal load p(x)
( ) ( ) 7.

Lecture Notes
Introduction to Strength of Materials pp. 4
Prof. Dr. N.H. Østergaard
Rearranging this, dividing through with dx and substituting , the following differential
equation is obtained
( )
( ) ( ) 8.
( )
Now considering the special case where the distributed load is zero, since we limit ourselves to
considering the case where loads solely are applied in the nodes, we have as governing equation
9.
Step 2: This equation 10 has a solution on the form
( ) 10.
Step 3/4: We now in general decide, that denotes the stiffness of DOF i due to a unit
displacement of DOF j. For a unit longitudinal deformation of the first node (the first DOF) and
zero longitudinal deformation of the second node (the second DOF), the following hold:
( ) 11.
( )
This gives us the shape function .
We observe from Figure 2, that - Force equilibrium now gives
∑ 12.
For a unit longitudinal deformation of the second node, and zero deformation of the first node,
the following results are obtained:
( ) 13.
( )
This gives us the shape function .
Again, we observe from Figure 2, that - Force equilibrium now gives
∑ 14.
The deformation function throughout the element is now given by
( ) 15.
That pretty much gives us an idea of the shape functions as normalized interpolation
polynomials.
Step 5: The element stiffness matrix is obtained as
16.
, - [ ] 0 1
We have now obtained the element stiffness matrix for a single truss element oriented along the
global x-axis. It is noted that the stiffness matrix is symmetrical (that’s always the case). Each
node currently only has a single DOF and this holds as long as the truss is only loaded along it’s
longitudinal direction.

Lecture Notes
Introduction to Strength of Materials pp. 5
Prof. Dr. N.H. Østergaard
Figure 2, Left: Plane 2-node truss element with shape functions, Right: Equilibrium of a truss
element

10.3.2 Stiffness matrix assembly

When assembling the stiffness matrix for a system constituted by multiple truss-elements, the
unconstrained stiffness matrix , -( ) is square with dimension equal to the number of nodes
times the number of DOF. However, nodes which are included in more than one element, will
naturally have stiffness contributions from all of those elements.
There are two ways of implementing supports by constraining the stiffness matrix:
1. A quick and dirty one where a stiffness value, which is sufficiently large to prevent
deformations, but not so large that it will turn the stiffness matrix singular, is added to
corresponding entry in the stiffness matrix.
2. A theoretically proper one where the constraint DOF is removed from the stiffness
matrix and from the corresponding force vector and thereby reducing the dimension of
the system of equations when obtaining the constraint stiffness matrix , -( ) .
In general, the following holds:

This implies that a constraint is added, everytime a DOF in ( ) is set to zero. When programming
larger systems, it is crucial to remember, that everytime a row and a column are removed from
the stiffness matrix to add a constraint, the dimension of the stiffness matrix is altered making it
difficult to navigate by using node numbers. This is often overcome by removing the entries for
the constrains for nodes with highest node number first and proceeding backwards towards the
constrains for the lowest node number.
This might have been a bit abstract and hard to grasp. It’s often easier after having seen an
example.

Lecture Notes
Introduction to Strength of Materials pp. 6
Prof. Dr. N.H. Østergaard
10.3.3 Calculated example: Compound cylindrical bar in tension

Figure 3 Compound cylinder subjected to multiple longitudinal loads

Problem: The compound cylinder in Figure 3 has dimensions given by

The cylinder is made of aluminum with elastic modulus and is subjected to loads

Determine: a) the deformation of point D using a) classical strength of materials (see chapter
2), b) matrix methods
Solution: Initially, the cross-sectional areas will be calculated, since these will be needed in
both approaches:
( )
( )
Approach from classical mechanics: The reaction force at A is calculated based on the FBD in
Figure 3. This allows us to lay in sections AB and BD to calculate the internal forces:

The deformation is now calculated by converting everything to SI units and using the
expression
( )
∑
We could get the stresses as internal force by area. Instead, let’s try this out using matrices:
Matrix methods approach: As first step, the cylinder is divided into two elements defined by
three nodes. Note, that these always should be numbered before proceeding in order for us not
to loose overview. The system to solve is given by

( ) , -( ) ( ) ( ) ( + ( + ( ) ( + ( +

Lecture Notes
Introduction to Strength of Materials pp. 7
Prof. Dr. N.H. Østergaard
We just need to get , -( ) . Now the element-wise stiffness matrices are calculated by equation
16:
( ) ( )
, -( )
[ ( ) ( )
] 0 1 0 1
( ) ( )
, -( )
[ ( ) ( )
] 0 1 0 1

In order to assemble the total stiffness matrix , -( ) , we add these two stiffness matrices while
accounting for which DOF’s the separate entries belong to. Each node only has a single DOF and
node 2 is the only node having stiffness contributions from both elements. We obtain
( ) ( )
( ) ( )
, -( ) * ( ) ( ) + * +
( ) ( )

[ ]
This is the unconstrained stiffness matrix. The only problem is that it’s singular. The reason is,
that it’s still unconstrained. The only support acting on the system is the wall support in A
yielding . This means that DOF1 is constrained. We implement this by removing the first
row and the first column from , -( ) and thereby also the first entry in ( ). The constrained
system is now given by

, -( )
* + ( )( ) ( *

The deformations are now obtained by

( ) ( * (, -( ) ) ( ( )
) . /
Which corresponds to the results we got using the classical approach. We could also have done
this using Matlab:

%Calculated example 2.4

clc; close all; clear all;
%Input %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
dAB=0.1; dBD=0.075; LAB=0.5; LBD=0.6;
E=70*10^9; FB=100*10^3; FD=200*10^3;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
A_AB=pi/4*dAB^2; A_BD=pi/4*dBD^2; %Calculate areas
I=[1,-1;-1,1];
kAB=E*A_AB/LAB*I; kBD=E*A_BD/LBD*I; %Element stiffness matrices
KU(1:3,1:3)=0; %Initialize global stiffness matrix
KU(1:2,1:2)=KU(1:2,1:2)+kAB; %Add elem1 to global stiffness matrix
KU(2:3,2:3)=KU(2:3,2:3)+kBD; %Add elem2 to global stiffness matrix
KC=KU(2:3,2:3); %Global constrained stiffnes matrix
dlt=inv(KC)*[FB;FD] %Solve for deformations

While this admittedly was not easier than the classical approach, actually to the contrary, it is
crucial to get how the stiffness matrix is constructed as sum of separate elementwise stiffness
matrices and constrained by removing DOF’s as entries and columns from the stiffness matrix.

Lecture Notes
Introduction to Strength of Materials pp. 8
Prof. Dr. N.H. Østergaard
10.3.4 Transformation matrix

Figure 4 Truss element with spatial orientation

This far, only bars and trusses with longitudinal axis corresponding to the global x-axis have
been considered. Furthermore, loads have solely been applied in the longitudinal direction. This
is far from always the case, and in order to account for a different spatial orientation of a truss
element, displacement, stiffness, and force terms must be transformed from the local element
coordinate system (denoted with primes in Figure 4) to the global Cartesian coordinate system.
From math, we know that the transformation matrix required to rotate a vector in an angle of
counter clockwise is given by
, - 0 1 17.

From the geometry in Figure 4, the directional cosine and sine terms are given by
√( ) ( ) 18.
It is now important to note, that when loads in any directions are applied to a truss, each node
can translate in both the x- and y- directions. Therefore, for spatially oriented trusses, each node
has two displacement DOF’s and for each element we have
19.
( ,
( )
In order to derive an expression for the stiffness matrix in the global coordinate system, we
utilize that forces and displacements can be transformed from a local to a global coordinate
system by
( ) , -( ) ( ) , -( ) 20.
However, since the transformation matrix in equation 17 is valid only for a single vector, and
equation 20. contains two vectors organized into a single column, the transformation matrix
must in this context be rearranged as

, - * + 21.

The following equation can now be written

( ) , -( )
, -( ) , - , -( ) 22.
( ) , - , - , -( ) , - , - for rotation matrices
( ) , -( ) with , - , - , - , -

Lecture Notes
Introduction to Strength of Materials pp. 9
Prof. Dr. N.H. Østergaard
We now have
, - , - , - , -

* + * +* +
23.

[ ]

This gives us the stiffness matrix for a truss element on component form. The internal forces can
be calculated by

( )( ) ( ) 24.
( )

10.3.5 MyFirstFEMCode.m
As example of a simple code for structural analysis of linear truss structures, a Matlab script was
written. However, it’s only actual functionality is that it assembles a stiffness matrix for a truss
structure on basis of a set of nodal coordinates on matrix form (nodes) and a set of elements
(elem) defined in terms of two node numbers referring back to two rows in the nodal matrix.
The moment of inertia (I) and the cross sectional area (Area) of the elements are constants for
all members and must be calculated prior to construction of the total stiffness matrix. The
constrains are implemented using the ‘quick and dirty’ method, where the k’th DOF is
constrained by replacing the k’th times k’th entry in the stiffness matrix with a large stiffness
term (BCStiffness) defined by the user. Furthermore, the external force array must be
initiated and defined manually (F). An example of the implemented visualization plots from the
code are shown in Figure 7.

Figure 5 Output from MyFirstFEACode.m for example model, bridge_geom2.m

Lecture Notes
Introduction to Strength of Materials pp. 10
Prof. Dr. N.H. Østergaard
10.3.6 Calculated example: Truss console

Figure 6 Truss console

Problem: The truss system shown in Figure 6 is constituted by members with rectangular
cross-sections with dimensions . The geometry is given by the following set of parameters:

The members are made of steel with elastic modulus . The structure is subjected to
a downwards load in point D of magnitude P=100 kN.
Determine: the normal stresses in the members using a) classical strength of materials (see
chapter 2), b) matrix methods

Solution: initially, the missing geometrical properties are calculated

( )( )
Classical mechanics approach (method of joints): Drawing a FBD of node D and applying
static equilibrium, we obtain:
∑ ∑
Solving these equations, the internal forces can be calculated
and .
The stresses in the members are

Lecture Notes
Introduction to Strength of Materials pp. 11
Prof. Dr. N.H. Østergaard
Matrix methods approach: The unconstrained system of equations to be solved is given by
( ) ( )

( ) ( )

( ) ( ) ( )
( ) , -( ) ( ) ( )
( ) ( ) ( )

( ) ( )
( )
( ) ( )
( )
( ) ( )
The element stiffness matrices are calculated using equation 23. The first element, AD, is
horizontal:

, -

[ ]
Member BD is obtained as rotated 60 deg. clockwise with respect to horizontal, i.e.

, - [ ]

The total unconstrained stiffness matrix is obtained as

, -( )

[ ]
Since node 1 and node 3 are pinned to the world frame and both have two DOF’s, we eliminate
the 6th,5th,2nd, and 1st rows and columns, i.e. the constrained stiffness matrix is given by

, -( ) * +

Having obtained , -( ) , we may solve for deformations

( )
( ) ( ( )+ (, -( ) ) ( ( )
) (* +) . / . /

The internal forces are finally calculated by equation 24

( )( )
( ) ( )

( )
( )

( )( )
( ) ( )

( )
( )

Lecture Notes
Introduction to Strength of Materials pp. 12
Prof. Dr. N.H. Østergaard
Since the internal forces are equal to those obtained by analytical calculations, the stresses will
be too. We could also have done this using Matlab:

%Calculated example 2.5

%Input %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
h=0.1; b=0.05; LAD=2; theta=60;
E=210*10^9; P=100*10^3;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
A=h*b; LBD=LAD/cosd(theta);
I=[1,-1;-1,1];
c=1; s=0;
kAD=A*E/LAD*[c^2,c*s,-c^2,-c*s;c*s,s^2,-c*s,-s^2;-c^2,-c*s,c^2,c*s;-c*s,-s^2,c*s,s^2];
c=cosd(-60); s=sind(-60);
kBD=A*E/LBD*[c^2,c*s,-c^2,-c*s;c*s,s^2,-c*s,-s^2;-c^2,-c*s,c^2,c*s;-c*s,-s^2,c*s,s^2];
KU(1:6,1:6)=0; KU(1:4,1:4)=KU(1:4,1:4)+kAD; KU(3:6,3:6)=KU(3:6,3:6)+kBD;
KC=KU(3:4,3:4);
dlt=inv(KC)*[0;-P] %Calculate deformations
ForceAD=A*E/LAD*[-1,0,1,0]*[0;0;dlt]; %Internal force, AD
ForceBD=A*E/LBD*[-c,-s,c,s]*[0;

Or it could have been scripted as input for the truss code:

%Input for MyFirstFEACode

%Truss console geometry (input section)
%Input nodes
Lb=2.0; db=Lb*tan(60*pi/180); %Frame geometry
nodes(1,:)=[0.0,0.0];nodes(2,:)=[Lb,0];nodes(3,:)=[0.0,db];

%Input elements [first element node, second element node]

elem(1,:)=[1,2]; elem(2,:)=[2,3];

%Truss console Loads/BCDS (paste after setup of stiffness matrix)

%%%Define boundary conditions
K(1,1)=BCStiffness;
K(2,2)=BCStiffness;
K(5,5)=BCStiffness;
K(6,6)=BCStiffness;
%%%Define load vector
F(6)=0; %Initialize array
F(4)=-P; %Define load

Figure 7 Output from MyFirstFEACode.m, Left: Node numbers and scaled deformations,
Right: element stresses

Lecture Notes
Introduction to Strength of Materials pp. 13
Prof. Dr. N.H. Østergaard
10.4 The plane truss element
A plane two-node beam oriented along the global x-axis will now be considered, see Figure 8. As
long as nodal loads are only added in the transverse direction or as moments, the deformations
of the beam can be described in terms of a transverse displacement u and a rotation θ of each
node. Therefore, each node has two DOF’s yielding a total of four DOF’s for the entire element.

10.4.1 Governing differential equation and shape functions

From the Bernoulli-Euler beam theory for long and slender members, we have already obtained
a differential equation of fourth order governing the deformations of the beam. Considering the
case where no distributed load is applied, the right hand side of the equation is zero, and the
differential equations has a solution, which is a polynomial of third order
( ) 25.
Applying the direct stiffness method, our objective is for the beam element to derive a system of
equations on the form

26.
( , [ ]( )

Applying a unit displacement to the first DOF, while the three remaining DOF’s are set to
zero, the first shape function and the first line of the stiffness matrix is derived. The internal
bending moment in the beam is given by
∑ 27.
The following expression is now obtained by direct integration
∫ ∫
. / 28.
∫ ∫ . /
= . /
Applying the boundary conditions, the following four algebraic equations are now derived
( )
( )
( ) . / 29.

( ) . /
Solving the above equations, we obtain the values
30.
We can solve for the remaining stiffness terms using force equilibrium in the y-direction and
moment equilibrium
∑ 31.
∑ 32.

We have now obtained the first row of the stiffness matrix. The shape function for the current
case is given by
33.
( ( * ( * *

Lecture Notes
Introduction to Strength of Materials pp. 14
Prof. Dr. N.H. Østergaard
Figure 8 Two node beam element

Now applying a unit rotation to node A with all other displacement DOFs set to zero, the 2nd row
of the stiffness matrix can be derived. The internal bending moment is given by
34.
By direct integration, the following relations are obtained
∫ ∫
. / 35.
∫ ∫ . /
. /
Applying the boundary conditions, we obtain the following four equations
( )
( )
( ) . / 36.

( ) . /
Solving the above equations, we obtain the values
37.

Lecture Notes
Introduction to Strength of Materials pp. 15
Prof. Dr. N.H. Østergaard
Force equilibrium allows us to solve for the remaining two unknown stiffness terms
∑ 38.
∑ 39.

The corresponding shape function is

. . / . / / 40.
Solving the two remaining cases for a unit displacement of the right node and finally a unit
rotation of the right node, the two remaining rows (3rd and 4th) of the stiffness matrix can be
determined, and the two remaining shape functions are given by
41.

42.

You are to do those yourself in exercise 10.3.2. You might consider this an act of cruelty, but it is
usually necessary for students in order to understand the direct stiffness method.
The element stiffness matrix is now given by

, - * + 43.

As expected, the stiffness matrix can be observed to be symmetrical, and equation 26 can be
written

44.
( , * +( )

This system of equations is however obtained for a beam with no longitudinal DOF, i.e. it cannot
elongate. Adding the DOF’s for elongation of a truss from equation 16, the following expression
is obtained

45.

( ) ( )

[ ]
This system of equations is on the form required for transformation, when analyzing beams not
oriented along the global x-direction.

Lecture Notes
Introduction to Strength of Materials pp. 16
Prof. Dr. N.H. Østergaard
10.4.2 Transformation matrix and constrains
The transformation matrix derived in section 10.3.4 is still valid for transformation of the beam
element system in equation 45. However, it has to be accounted for that rotational DOFs (3rd and
6th) do not require transformation. For a 6DOF beam element, the transformation matrix will
therefore be arranged as follows

, - 46.

[ ]
The stiffness matrix in the global coordinate system can still be obtained using equation 23 and
the principle for assembly of the stiffness matrices known from bar element are generally valid.
Constrains for various common support types are shown in Figure 9.

Support type Constraint equation(s) Nodal DOF-numbers constrained

1,2

1,2,3

Figure 9 Nodal constraint equations corresponding common support types

10.4.3 Line loads: Work equivalent nodal forces

This far, only external loads applied in the nodes have been considered. However, structural
analysis is often performed of structures with line loads. A constant line load would increase the
order of the required polynomial in equation 25 by one. However, using energy methods (which
we won’t do in the present course), we may replace the distributed load with nodal loads that
exerts the same work as the distributed load would. If doing so, loads as shown in Figure 10 can
be derived.
You are to try this out on your own when solving problem 10.3.5.

Figure 10 Work equivalent nodal loads replacing a distributed load

Lecture Notes
Introduction to Strength of Materials pp. 17
Prof. Dr. N.H. Østergaard
10.4.4 Modal analysis: buckling
In chapter 9, Euler buckling was considered by formulating the moment equilibrium of a beam in
the deformed shape (assumed to be harmonic), see Figure 11-Left. The following differential
equation was derived and we would for given boundary conditions solve this for P yielding the
critical load .
47.

This type of buckling problem is usually referred as a bifurcation problem, since any
compressive load larger than will cause the beam to snap from the initial perfectly straight
state to a deformed state. For various boundary conditions, the critical load was obtained in
terms of the Euler length , see Figure 11-Right.
48.
( )

A.

B.

C.

D.

Figure 11, Left: Equilibrium of a long and slender beam in the deformed state, Right: Euler
lengths for simple cases of supports
If we now apply the direct stiffness method with polynomial shape functions of 3rd order to
equation 47, the following equation can be derived
( ) , -( ) , -( ) 49.
in which , - is the geometrical (or stress) stiffness matrix, which can be derived to be on the
form

, - * + 50.

It is noticed, that , - depends solely on geometrical parameters and not on materials. Setting
and considering the case where external transverse forces and moments are zero, we
obtain the following equation
(, - , -)( ) 51.

Lecture Notes
Introduction to Strength of Materials pp. 18
Prof. Dr. N.H. Østergaard
This is recognized as an eigenvalue problem, which can be solved setting the determinant of the
total stiffness matrix equal to zero
|(, - , -)| 52.
The corresponding values for ( ) are the mode shapes. These represent the shape, but not
magnitude, of the buckled beam geometry.
This technique for modal analysis is generally valid for buckling analysis of linear elastic
structures, and geometrical stiffness matrices may be derived for other structural elements (fx.
plates and shells). This turns out to be immensely useful for numerical analysis, since analytical
stability analysis often gets quite complicated (the best books on the topic is probably still
Timoshenkos epic classic textbook on elastic stability containing a large number of analytical
solutions [3]).

10.4.5 Modal analysis: undamped eigenfrequencies

The technique applied for modal analysis to obtain the buckling loads and mode shapes in the
previous chapter, can also be applied to determine the eigenfrequencies and mode shapes of
beams vibrating without damping. The reader should notice, that this is the first time vibrations
of continuous systems have been considered. The differential equation governing the undamped
vibrations of a beam can be derived to be on the form
53.
( )
We note, that y is a function of both position x and time t, which is why partial derives are now
required. The differential equation has a harmonic solution with eigenfrequency given by
54.
√
in which S denotes mass per unit length (and can be calculated as density times cross-sectional
area). The constant C is dependent on the boundary conditions and is given in Figure 12 for
various cases of supports.

Boundary conditions and mode shape Constant C

A.

B.
15.3

C.
22.4

D.
3.51

Figure 12 Mode shapes and coefficient for eigenfrequency calculation for simple beam cases

Lecture Notes
Introduction to Strength of Materials pp. 19
Prof. Dr. N.H. Østergaard
Again, applying the direct stiffness method to equation 53. with third order polynomials as
shape functions, the following linear system is derived
, -* ̈ + , -* + 55.
In which , - is the mass matrix of the system. The mass matrix usually comes on two forms: an
inaccurate, but simple form called the lumped mass matrix, in which the mass is assumed
concentrated in the nodes, and a more accurate, but complex form called the consistent mass
matrix, derived on basis of a distributed mass

, - * + , - * + 56.

Assuming the solution harmonic for all DOF’s, we obtain

* + * + ( ) * ̈+ ( ) * + ( ) 57.
Substituting this into the dynamics equilibrium, the following equation is derived
* + * + ( ) * ̈+ ( ) * + ( )
, -( ( ) * + ( )) , -(* + ( )) 58.
(, - ( ) , -)(* + ( ))
The derived equation will have a non-trivial solution when
(, - ( ) , -)* + 59.
Which is recognized as an eigenvalue problem, that can be solved using determinants.
|(, - ( ) , -)| 60.
Again, this technique is generally valid for other linear structures. The best book on structural
dynamics is probably Humar’s epic classic [4]– that’s the best place to look if you need more
information about this topic.

10.4.6 Calculated example: simple supported pipe shaped beam

(With Matlab – the longest example ever !!!)

Figure 13 Simply supported pipe with central load

Lecture Notes
Introduction to Strength of Materials pp. 20
Prof. Dr. N.H. Østergaard
Problem: The system shown in Figure 13 has geometry defined by the following set of
parameters

The pipe is made of steel with material parameters

A load of is applied at mid-span.

We will calculate:
a) the unconstrained global stiffness matrix
b) the constrained stiffness matrix
c) the equation of the elastic curve
d) the internal moment curve in the beam
e) the maximum deformation of the setup if geometry, loads and BCDs are rotated 90 deg
Solution: First of all, we will define all input parameters
%Input section %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%Material parameters
E=2.1*10^11; %Module of elasticity [N/m^2]
rho=7800; %Density in [kg/m^3]
%Geometry
d=0.1; %Radius
t=0.01; %Wall Thickness
L=2; %Beam length
%Loads
P=-10*10^3; %End load [N]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

Then we calculate the cross-sectional constants and set the element length
%Geometry postprocessing;
A=pi/4*(d^2-(d-2*t)^2); %cross-sectional area, rectangular cross-section
I=pi/64*(d^4-(d-2*t)^4); %second order area moment of inertia, rectangular
cross-section
L_elem=L/2; %set element length
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

In Figure 13, nodes and elements are defined. Is noticed that the displacement vector is given by
( ) , -
a) Now wanting to determine the global stiffness matrix, the element-wise stiffness matrices are
calculated from 43. Since both elements are horizontal, the local and global x-axis coincide
and there is no need to transform those to change the spatial orientation

[ ] [ ] , - , -
, - [ ]
, - , -

[[ ] [ ] ]

Lecture Notes
Introduction to Strength of Materials pp. 21
Prof. Dr. N.H. Østergaard
 [ ] [ ] , - , -
, - [ ]
, - , -

[[ ] [ ] ]
The unconstrained global stiffness matrix is now obtained as
, - , -
( )
, - *, - , - , - , - +
, - , -

%a) Setup unconstrained element stiffness matrix (local)

k_elem(1:6,1:6)=0; %Initialize array
k_elem(1,1)=E*A/L_elem; k_elem(1,4)=-E*A/L_elem;
k_elem(2,2)=12*E*I/L_elem^3; k_elem(2,3)=6*E*I/L_elem^2;
k_elem(2,5)=-12*E*I/L_elem^3;k_elem(2,6)=6*E*I/L_elem^2;
k_elem(3,2)=6*E*I/L_elem^2; k_elem(3,3)=4*E*I/L_elem;
k_elem(3,5)=-6*E*I/L_elem^2; k_elem(3,6)=2*E*I/L_elem;
k_elem(4,1)=-E*A/L_elem; k_elem(4,4)=E*A/L_elem;
k_elem(5,2)=-12*E*I/L_elem^3;k_elem(5,3)=-6*E*I/L_elem^2;
k_elem(5,5)=12*E*I/L_elem^3; k_elem(5,6)=-6*E*I/L_elem^2;
k_elem(6,2)=6*E*I/L_elem^2; k_elem(6,3)=2*E*I/L_elem;
k_elem(6,5)=-6*E*I/L_elem^2; k_elem(6,6)=4*E*I/L_elem;

disp('The element stiffness matrix is'); k_elem

%Setup the global global unconstrained stiffness matrix

K(1:9,1:9)=0; %Initialize array
K(1:6,1:6)=K(1:6,1:6)+k_elem; %Add 1st element
K(4:9,4:9)=K(4:9,4:9)+k_elem; %Add 2nd element

b) The constrained stiffness matrix is obtained by removing the rows and columns
corresponding to the constrained DOF’s in accordance with the constraint equations given in
Figure 9. The fixed support to the left will fix both displacement coordinates, so and
. The moving support to the right will fix the vertical displacements only, so .
The DOF’s to be constraint have numbers 1,2 and 8. We therefore eliminate the
corresponding rows and columns in the stiffness matrix (starting with the largest number) to
obtain , -( )

%Determine the constrained stiffness matrix by removing DOFs (eliminate row and
columns in k_elem starting with the largest)
K(8,:)=[]; K(:,8)=[]; %Right moving support (fix y)
K(2,:)=[]; K(:,2)=[]; %Left fixed support (fix y)
K(1,:)=[]; K(:,1)=[]; %Left fixed support (fix x)
%Note: nDOF = NumCor - nConstrains = 3*3 - 3 = 6

Lecture Notes
Introduction to Strength of Materials pp. 22
Prof. Dr. N.H. Østergaard
c) Having obtained the constrained stiffness matrix, we may solve for deformation using the
nodal force vector
( ) , -
, -
The deformations are now obtained by
( ) , - (, -( ) ) ( ( ) )
, -
The analytical value for the max. deflection at the midpoint is
( )

which is equal to the value calculated by numerical analysis.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%c) Determine the maximum deflection
F=[0;0;P;0;0;0]; %Define load vector,
X=K\F; %Solve for deformations
yMax=X(3); %Maximum numerical deformation in y-direction (for x=L)
yMaxA=P*L^3/(48*E*I); %Calculate maximum analytical deformation in y-direction
% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

In order to calculate the equation of the elastic curve, we will utilize that the deflections over
each element are a polynomial of third order. We have

Now considering the first element and inserting ( ) ( ), ( ) . / . /

( ) and . / . /, the following linear system of equations is obtained and solved for
the polynomial coefficients a,b,c and d

( , ( )

[ ]
These are in the Matlab implementation simply contained in a vector , -
The determined polynomial can be compared with the analytic expression
( ) ( )

% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Calculate the equation of the elastic curve by cubic interpolation from x=0 to
x=L/2
x1=0; y1=0; tht1=X(1); %BCDS in left end (wall-section)
x2=L/2; y2=X(3); tht2=X(4); %Extract values from solution (left)

CoefMat(1:4,1:4)=0; %Initialize array

%Setup coefficient matrix and right-hand-side vector
CoefMat(1,:)=[x1^3,x1^2,x1,1];
CoefMat(2,:)=[x2^3,x2^2,x2,1];
CoefMat(3,:)=[3*x1^2,2*x1,1,0];
CoefMat(4,:)=[3*x2^2,2*x2,1,0];
B=[y1;y2;tht1;tht2];
Z=CoefMat\B;

Lecture Notes
Introduction to Strength of Materials pp. 23
Prof. Dr. N.H. Østergaard
 %Evaluate and plot the obtained solution over the length of the beam on a grid
of n+1 points
n=10;
for i=1:n+1
x(i)=(L/2)/n*(i-1); %X-along the beam
y(i)=Z(1)*x(i)^3+Z(2)*x(i)^2+Z(3)*x(i)+Z(4); %interpolated deflection
yA(i)=-P/(48*E*I)*(4*x(i)^3-3*L^2*x(i));
end

figure; plot(x,y,'b.',x,yA,'b-'
xlabel('Length position along beam element [m]')
ylabel('Deflection [m]')
legend('Interpolation from FEM solution','Analytical solution')
grid on

d) The internal bending moment in the beam is obtained by differentiation of the rotation
response
( ) ( )

%d) Calculate the internal bending moment curve

for i=1:n+1
M(i)=E*I*(6*Z(1)*x(i)+2*Z(2)); %interpolated deflection
MAna(i)=-P/2*x(i);
end

figure; plot(x,M,'b.',x,MAna,'b-');
xlabel('Length position along beam [m]')
ylabel('Internal bending moment [Nm]')
legend('Interpolation from FEM solution','Analytical solution')
grid on

The obtained results are shown in Figure 14 and Figure 15.

Figure 14 Elastic curve, 1st element Figure 15 Bending moment, 1st element

Lecture Notes
Introduction to Strength of Materials pp. 24
Prof. Dr. N.H. Østergaard
e) For a 90 deg. rotation, we assemble a global rotation matrix and apply equation 22 to obtain
the global stiffness matrix in the rotated frame

alpha=90*pi/180; %Define angle of rotation

%Setup the transformation matrix for a beam element
T(1:3,1:3)=0; %Initialize array
T(1,1)=cos(alpha); T(1,2)=sin(alpha);
T(2,1)=-sin(alpha); T(2,2)=cos(alpha);
T(3,3)=1;
TG=[T,zeros(3,6);zeros(3,3),T,zeros(3,3);zeros(3,6),T];
TG(8,:)=[]; TG(:,8)=[]; %Right moving support (fix y)
TG(2,:)=[]; TG(:,2)=[]; %Left fixed support (fix y)
TG(1,:)=[]; TG(:,1)=[]; %Left fixed support (fix x)
K_ROT=TG'*K*TG;
F(1:6)=0; F(2)=P;
X=K_ROT\F; %Solve for deformations
xMax=X(2);

The obtained deflection is as expected equal to the value obtained in b).

10.4.7 Calculated example: MrBeam

The example in the previous section can also be solved using the plane beam FEA code MrBeam1
on basis of the following input commands ran after the input block from a) in the previous
example. The deformed shape in Figure 18 and the moment curves in Figure 16 and Figure 17
look as expected. The model is contained in the example input file
ExModel_SimplySuppertedPipe.m

%Horisontal simply supperted pipe

Nodes=[0,0;L/2,0;L,0];
BeamElem=[1,2,A,I,E;2,3,A,I,E];
PointLoads=[2,2,P];
%Constraint simply support
Constrains=[1,2;3,2];

Figure 16 Moment curve, 1st element Figure 17 Moment curve, 2nd element

1 Name invented by former mech. Eng. Student Daniel Neubauer. This is highly appreciated by the author,
since the code inevitably would have been after a weird fantasy character with it’s own web-comic, should
I have made the decision

Lecture Notes
Introduction to Strength of Materials pp. 25
Prof. Dr. N.H. Østergaard
Figure 18 Beam model and deflected shape

Figure 19 The MrBeam logo

Lecture Notes
Introduction to Strength of Materials pp. 26
Prof. Dr. N.H. Østergaard
10.4.8 Calculated example: modal analyses

Figure 20 Critical load of simply supported pipe shaped beam

Problem: for geometry and material properties like those given in section 10.4.6, determine the
critical load of the beam.
Solution: Since the beam is oriented horizontally, the local and global x-axes coincide and we do
not need the DOF-corresponding to extension. We may therefore divide the structure into a
single element, and obtain the stiffness matrix along with the geometric stiffness matrix using
equations 43 and 50 respectively. After constraining both matrices, the standard eigenvalue
solver in Matlab, eigs, will be applied to determine the eigenvalue, which can be compared to
the analytically calculated critical load.
clear k_elem; L_elem=L;
k_elem(1:4,1:4)=0; %Initialize array
k_elem(1,1)=12*E*I/L_elem^3; k_elem(1,2)=6*E*I/L_elem^2;
k_elem(1,3)=-12*E*I/L_elem^3;k_elem(1,4)=6*E*I/L_elem^2;
k_elem(2,1)=6*E*I/L_elem^2; k_elem(2,2)=4*E*I/L_elem;
k_elem(2,3)=-6*E*I/L_elem^2; k_elem(2,4)=2*E*I/L_elem;
k_elem(3,1)=-12*E*I/L_elem^3;k_elem(3,2)=-6*E*I/L_elem^2;
k_elem(3,3)=12*E*I/L_elem^3; k_elem(3,4)=-6*E*I/L_elem^2;
k_elem(4,1)=6*E*I/L_elem^2; k_elem(4,2)=2*E*I/L_elem;
k_elem(4,3)=-6*E*I/L_elem^2; k_elem(4,4)=4*E*I/L_elem;

kG_elem(1:4,1:4)=0; %Initialize array

kG_elem(1,1)=6/(5*L_elem); kG_elem(1,2)=1/10;
kG_elem(1,3)=-kG_elem(1,1); kG_elem(1,4)=kG_elem(1,2);
kG_elem(2,1)=kG_elem(1,2); kG_elem(2,2)=2*L_elem/15;
kG_elem(2,3)=-kG_elem(1,4); kG_elem(2,4)=-L_elem/30;
kG_elem(3,1)=-kG_elem(1,1); kG_elem(3,2)=-kG_elem(1,2);
kG_elem(3,3)=kG_elem(1,1); kG_elem(3,4)=-kG_elem(1,2);
kG_elem(4,1)=kG_elem(1,4); kG_elem(4,2)=kG_elem(2,4);
kG_elem(4,3)=kG_elem(3,4); kG_elem(4,4)=kG_elem(2,2);

K=k_elem; KG=kG_elem;
K(3,:)=[]; K(:,3)=[]; KG(3,:)=[]; KG(:,3)=[];
K(1,:)=[]; K(:,1)=[]; KG(1,:)=[]; KG(:,1)=[];

[Modes,EigVals]=eigs(K,KG) %Solve eigenvalue problem

Pcr=EigVals(1,1) %Extract critical load
PcrA=pi^2*E*I/L^2 %Compare to analytical value

Lecture Notes
Introduction to Strength of Materials pp. 27
Prof. Dr. N.H. Østergaard
10.4.9 Frame structures
One of the major advantages of the developed methods is, that these enable analysis of frame
structures by solving systems of equations using linear algebra. The techniques demonstrated in
section 10.4.6 can easily be applied to develop larger structures. This section should contain a
calculated example, but the author decided to turn this into exercise problems (as an act of
cruelty). However, an example of a frame structure is shown below in Figure 21.

Figure 21 Example of frame structure modeled using MrBeam

Lecture Notes
Introduction to Strength of Materials pp. 28
Prof. Dr. N.H. Østergaard
Problems
Bars and trusses
Problem 10.2.1
The compound cylinder ABD is fixed against
deformations in section A and subjected to axial
loads and as shown in
the figure. The geometry is given by the
following set of parameters:

The cylinder is made of structural steel with

elastic modulus of 210 GPa. Determine the
deformation of the points B and D using a)
classical mechanics, b) matrix methods
Ans:
Problem 10.2.2
For the compound cylinder in the previous
problem, the load in D is replaced with a wall, so
both ends A and D are fixed against
deformations. Determine the deformation of
point B and the reactions in A and D using a)
classical mechanics, b) matrix methods
Ans:

Problem 10.2.3
The frame ADB is constituted by truss elements
with cross-sectional areas and
length pinned in A and B at a distance
. Point D is subjected to a vertical load
. The trusses are made of structural
steel with elastic modulus of 210 GPa.
Determine the internal forces in both members
using a) classical mechanics, b) matrix methods,
c) draft the input file for the Matlab truss script
MyFirstFEACode.m
Ans:
Problem 10.2.4
The truss console structure ACB is made up of
elements with cross-sectional areas
and is subjected to a vertical
load in point C. The dimensions of
the trusses are given by the parameters
and . Determine the internal forces in
both members using a) classical mechanics, b)
matrix methods, c) draft the input file for the
Matlab truss script MyFirstFEACode.m
Ans:

Lecture Notes
Introduction to Strength of Materials pp. 29
Prof. Dr. N.H. Østergaard
 Problem 10.2.5
For the truss structures to the left determine the
number of
a) Nodes
b) Elements
c) DOF’s for the unconstrained system
d) DOF’s for the constrained system
Furthermore, determine the size of the
constrained stiffness matrix

Problem 10.2.6
The frame shown is loaded by a vertical force
P=245 kN in point B. The cross sectional areas
are 2500 mm2 and 1800 mm2 for trusses AB and
AD. All trusses are made of structural steel with
elastic modulus E=210 GPa. The frame is of
height h=3 m and length l=10 m.
For all members in the frame, determine a) the
internal forces using a) classical mechanics, b)
matrix methods, c) draft the input file for the
Matlab truss script MyFirstFEACode.m
HINT: you’ll need Matlab from question b)
a) FAB=-238.1 kN, FAD=204.2 kN

Beams
Problem 10.3.1
A simply supported beam of length L=2m is
subjected to some loads. These are
unfortunately unknown. However, we have
measured the rotations in each end to be of
magnitude . If the beam is modeled
using a single element a) Determine the
equation of the elastic curve, b) Estimate the
average curvature, c) how would you estimate
the maximum curvature of the beam?
Ans.:

Problem 10.3.2
Carefully review the derivation of the two first
rows of the stiffness matrix in section 10.4.1,
before you (entirely on your own) derive the
two last rows (3 and 4).

Ans: see equation 44

Lecture Notes
Introduction to Strength of Materials pp. 30
Prof. Dr. N.H. Østergaard
 Problem 10.3.3
A cantilevered rectangular hollow beam with
height , width , wall
thickness , and length is
subjected to a concentrated end load of
magnitude P=10kN. The beam is made of
structural steel with elastic modulus of 210 GPa.
Determine the maximum deflection, the
equation of the elastic curve and the equation of
the internal bending moment curve using a)
classical mechanics, b) matrix methods
Ans:

Problem 10.3.4
A beam defined by the same set of parameters
as in the previous problem (double length) is
subjected to a central load P as double
cantilever. Determine the support moments and
reactions using a) classical mechanics, b) matrix
methods
Ans: M=-5000Nm, V=-5000N
Problem 10.3.5
A beam with geometry like in problem 10.3.3 is
subjected to a distributed load .
Determine the end deflection and rotation using
a) classical mechanics, b) matrix methods
Ans:

Problem 10.3.6
For the plane beam structures to the left
determine the number of
a) Nodes
b) Elements
c) DOF’s for the unconstrained system
d) DOF’s for the constrained system
Furthermore, determine the size of the
constrained stiffness matrix

Lecture Notes
Introduction to Strength of Materials pp. 31
Prof. Dr. N.H. Østergaard
Modal analysis
Problem 10.3.7
Determine the critical compressive load of a
cantilever beam using a single beam element in
terms of E,I, and L. Calculate the error with respect
to the analytical critical Euler load
Hint: yes, that’s right – you have to solve an
eigenvalue problem
Ans: 0.8%
Problem 10.3.8
Determine the critical compressive load of an end
supported cantilever beam using a single beam
element in terms of E,I, and L. Calculate the error
with respect to the analytical critical Euler load
Ans: 49%
Problem 10.3.9
Determine the lowest eigenfrequency of a
cantilever beam using a single beam element in
terms of E,I,L, and S. Calculalte the error with
respect to the analytical eigenfrequency
Ans: 0.6%
Problem 10.3.10
You’ll need Matlab for this problem !
A compound cylinder is defined by the following
geometrical set of parameters:
, ,
The beam is made of structural steel with elastic
modulus of 210 GPa.
Determine the critical compressive load of the
cylinder. Check your results against the solution
tables in Roark, pp. 718 [5]

Hint for Problem 10.3.10: define a function of E,I and L which returns the element wise
stiffness matrix. In a similar fashion, define a function of L which returns the geometrical
stiffness matrix. This makes life so much easier
Problem 10.3.11 – solve with Matlab
How would you, for all modal problems in the current section, check the validity of the obtained
eigenvalues

Lecture Notes
Introduction to Strength of Materials pp. 32
Prof. Dr. N.H. Østergaard
Frames
In the following problems, the cross-sectional dimensions are the same as in problem 10.3.3.
The load (if any) is downwards and the moment (if present) is .
The length dimension is L=2 m and the height dimension is h=3 m.
Determine the nodal deflections and the internal bending moments along the lengths of all
members using:
a) Your own Matlab scripts (Hint: remember that functions are handy)
b) The MrBeam.m Matlab script

Problem 10.3.12 Problem 10.3.13

Problem 10.3.14 Problem 10.3.15

References
[1] Lecture notes on finite element analysis (Introduction to FEA theory for dummies), NH
Østergaard, HSRW, 2015
[2] Introduction to finite element methods (download here), Online lecture notes by Carlos
Felippa, University of Colorado at Boulder, 2005
[3] Theory of elastic stability, SP Timoshenko & JM Gere, Courier Corporation, 2012
[4] JL Humar, Dynamics of structures, CRC-Press, 2012
[5] Roark’s formulaes for stress and strain, WC Young & RG Budynas, McGraw-Hill, 2002
[6] Course FEA scripts for download:
https://drive.google.com/drive/folders/1fpAf_pt-i1-WJtKz4mJIpMAJtwddw-VG
[7] GNU Octave project page
[8] Python x-y project page

Lecture Notes
Introduction to Strength of Materials pp. 33
Prof. Dr. N.H. Østergaard