Module 5

Lecture 32,33,34
19/04/24
Introduction to Matrix Method of Analysis
 Introduction
• Structural analysis, is an integral part of any structural engineering project.
• It is the process of predicting the performance of a given structure under a prescribed loading condition.
• The performance characteristics usually of interest in structural design are: (a) stresses or stress resultants
(i.e., axial forces, shears, and bending moments); (b) deflections; and (c) support reactions.
• Thus, the analysis of a structure typically involves the determination of these quantities as caused by the
given loads and/or other external effects (such as support displacements and temperature changes)
Classical, Matrix, and Finite-Element Methods of Structural Analysis
• Matrix and classical methods are based on the same fundamental principles—but that the fundamental
relationships of equilibrium, compatibility, and member stiffness are now expressed in the form of matrix
equations.
• Most classical methods were developed to analyze particular types of structures, and since they were
intended for hand calculations, they often involve certain assumptions (that are unnecessary in matrix
methods) to reduce the amount of computational effort required for analysis.
• Because of their specialized and intricate nature, classical methods are generally not considered suitable for
computer programming.
• In contrast to classical methods, matrix methods were specifically developed for computer implementation;
they are systematic (so that they can be conveniently programmed), and general (in the sense that the same
overall format of the analytical procedure can be applied to the various types of framed structures).
• As the analysis of large and highly redundant structures by classical methods can be quite time consuming,
matrix methods are commonly used.
• However, classical methods are still preferred by many engineers for analyzing smaller structures because
they provide a better insight into the behavior of structures.
• Classical methods may also be used for preliminary designs, for checking the results of computerized
analyses, and for deriving the member force-displacement relations needed in the matrix analysis.
• Furthermore, a study of classical methods is considered to be essential for developing an understanding of
structural behavior.
Matrix versus Finite-Element Methods
• Matrix methods can be used to analyze framed structures only.
• Finite-element analysis, which originated as an extension of matrix analysis to surface structures (e.g.,
plates and shells), has now developed to the extent that it can be applied to structures and solids of
practically any shape or form.
• From a theoretical viewpoint, the basic difference between the two is that, in matrix methods, the
member force-displacement relationships are based on the exact solutions of the underlying (Euler-
Bernoulli) differential equations, whereas in finite-element methods, such relations are generally derived
by work-energy principles from assumed displacement or stress functions.
• Because of the approximate nature of its force-displacement relations, finite-element analysis generally
yields approximate results. However, in the case of linear analysis of framed structures composed of
prismatic (uniform) members, both matrix and finite-element approaches yield identical results.
 Flexibility and stiffness methods
• Two methods used for the matrix analysis of structures: the flexibility method, and the stiffness method.
• The flexibility method, which is also referred to as the force or compatibility method, is essentially a
generalization in matrix form of the classical method of consistent deformations.
• In this approach, the primary unknowns are the redundant forces, which are calculated first by solving the
structure’s compatibility equations.
• Once the redundant forces are known, the displacements can be evaluated by applying the equations of
equilibrium and the appropriate member force-displacement relations.
• The stiffness method, which originated from the classical slope-deflection method, is also called the
displacement or equilibrium method.
• In this approach, the primary unknowns are the joint displacements, which are determined first by solving the
structure’s equations of equilibrium.
• With the joint displacements known, the unknown forces are obtained through compatibility considerations
and the member force-displacement relations.
• The flexibility method is generally convenient for analyzing small structures with a few redundants. It may also
be used to establish member force-displacement relations needed to develop the stiffness method.
• The stiffness method is more systematic and can be implemented more easily on computers; therefore, it is
preferred for the analysis of large and highly redundant structures.
• Most of the commercially available software for structural analysis is based on the stiffness method.
Truss Analysis using Stiffness Method
 Fundamentals of stiffness method
• Application of the stiffness method requires subdividing the structure into a series of discrete finite
elements, and then identifying their end points as nodes.
• For truss analysis, the finite elements are represented by each of the members of the truss, and the
nodes represent the joints.
• Member and Node Identification
• One of the first steps when applying the stiffness method is to identify the elements or members of
the truss and their nodes.
• We will specify each member by a number
enclosed within a square, and use a number
enclosed within a circle to identify the nodes.
• Also, the “near” and “far” ends of the member
must be identified. This will be done using an
arrow written along the member, with the head
of the arrow directed toward the far end.
• Global and Member Coordinates
• Since loads and displacements are vector quantities, it is necessary to establish a coordinate
system in order to specify their sense of direction.
• Here we will use two different types of coordinate systems.
• A single global or structure coordinate system, x, y, will be used to specify the sense of each of
the external force and displacement components at the nodes, Fig. a.
• A local or member coordinate system will be used for each member to specify the sense of
direction of its displacements and internal loadings. This system will be identified using x′, y′
axes with the origin at the “near” node and the x′ axis extending toward the “far” node. An
example for truss member 4 is shown in Fig. b.
• Kinematic Indeterminacy
• The unconstrained displacements for a truss joint represent the primary unknowns of any
displacement method, and therefore these must be identified.
• As a general rule, there are two degrees of freedom, or two displacements, for each joint (node).
• For application, each degree of freedom will be specified on the truss using a code number, shown
at the joint or node, and referenced to its positive global coordinate direction using an associated
arrow.
• For example, the truss in Fig. a has eight degrees of freedom, which have been identified by the
“code numbers” 1 through 8 as shown.
• The truss is kinematically indeterminate to the fifth degree because of these eight possible
displacements: 1 through 5 represent unknown or unconstrained degrees of freedom, and 6
through 8 represent constrained degrees of freedom. Due to the pin and roller, these displacements
are zero.
• For later application, the lowest code numbers will always be used to identify the unknown
displacements (unconstrained degrees of freedom) and the highest code numbers will be used to
identify the known displacements (constrained degrees of freedom).
• The reason for choosing this method of identification has to do with the convenience of later
partitioning the structure stiffness matrix so that the unknown displacements can be found in the
most direct manner.
 Member Stiffness Matrix
• We will now establish the stiffness matrix for a single truss member using
local x′, y′ coordinates, oriented as shown in Fig.
• The terms in this matrix represent the load-displacement relations for the
member.
• Since the loads on a truss member only act along the member, then the
displacements of the nodes are only along the x′ axis.
• To obtain the load-displacement relations we will apply two independent
displacements to the member. When a positive displacement dN occurs
on the near end of the member, while the far end is held pinned (fixed),
Fig. a, the forces developed at the ends of the member are

• Here 𝑞𝐹′ is negative since for equilibrium it must act in the negative x′ direction.
• Likewise, a positive displacement dF at the far end, keeping the near end
pinned (fixed), Fig. b, results in member forces of
 =

By superposition, Fig.c, the resultant forces

caused by both displacements are The four elements
that comprise it are
called member
stiffness influence
′
where
coefficients, 𝑘𝑖𝑗 .

These load-displacement equations may This matrix, k′, is called the member stiffness matrix, and it
be written in matrix form as has the same form for each member of the truss.
 ′
• Physically, 𝑘𝑖𝑗 represents the force at joint i when a unit displacement is imposed at
joint j.
′
• For example, if i = j = 1, then 𝑘11 is the force at the near joint when the far joint is held
fixed, and the near joint undergoes a displacement of dN = 1, i.e.,

Likewise, the force at the far joint is determined from i = 2, j = 1 so that

• These two terms represent the first column of the member stiffness matrix.
• In the same manner, the second column of this matrix represents the forces in the
member when the far end of the member undergoes a unit displacement and the near
end is fixed
 Displacement and force transformation matrices
• Since a truss is composed of many members (elements), we must now
develop a method for transforming the member forces q and
displacements d defined in local coordinates to global coordinates.
• To do this we will define the direction of each member using the
smallest angles between the positive x, y global axes and the positive x′
local axis. These angles are 𝜃𝑥 and 𝜃𝑦 .
• The cosines of these angles will be used in the matrix analysis that
follows, where 𝜆𝑥 = cos 𝜃𝑥 and 𝜆𝑦 = cos 𝜃𝑦 .
• Numerical values for 𝜆𝑥 and 𝜆𝑦 can easily be generated by a computer
once the x, y coordinates of the near end N and far end F of the
member have been specified.
• For example, consider member NF of the truss shown.
• Here the coordinates of N and F are (xN, yN) and (xF, yF), respectively.
Therefore,
 Displacement transformation matrix
• In global coordinates each end of the member can have
two degrees of freedom or independent displacements;
namely, joint N has DNx and Dny and joint F has DFx and
DFy.
• When the far end is held pinned (fixed) and the near
end is given a global displacement DNx, Fig. a, the
corresponding displacement (deformation) along the
member is 𝐷𝑁𝑥 cos 𝜃𝑥
• Likewise, a displacement DNy will cause the member to
be displaced 𝐷𝑁𝑦 cos 𝜃𝑦 along the x′ axis, Fig. b.
• The effect of both global displacements causes the local
displacement

In a similar manner, positive displacements DFx and

DFy successively applied at the far end F, while the
near end is held pinned (fixed), Figs. c and d, will
cause the local displacement
 which can be written in matrix form as

where

From the above derivation, T transforms the four global x, y displacements D into the two local x′
displacements d. Hence, T is referred to as the displacement transformation matrix.
 Force transformation matrix
• Consider application of the force qN to the near end of the member, with
the far end held pinned (fixed), Fig. a.
• Here the global force components of qN at N are

• Likewise, if qF is applied to the member, Fig. b, the global force

components at F are

In matrix form

In this case this force transformation matrix TT transforms the

two local (x′) forces q acting at the ends of the member into the
four global (x, y) force components Q.
TT is the transpose of the displacement transformation matrix.
 Member global stiffness matrix
We will now determine the stiffness matrix for a member in terms of its global force components Q and its global
displacements D
member’s forces q are expressed in terms of
Substituting into the global displacements D

Substituting in

or
The matrix k is the member
stiffness matrix in global
where coordinates.
Since TT, T, and k′ are known, we have

Member
stiffness
matrix in
global
coordinates
• The location of each element in this 4 × 4 symmetric matrix is referenced with each global degree
of freedom associated with the near end N, followed by the far end F.
• This is indicated by the code number notation along the rows and columns, that is, Nx, Ny, Fx, Fy.
• Here k represents the force-displacement relations for the member when the components of force
and displacement at the ends of the member are in the global or x, y directions.
• As a result, each column of the matrix represents the four force components developed at the ends
of the member when one degree of freedom at an end undergoes a unit displacement while the
other degrees of freedom are restrained.
• For example, a unit displacement DNx = 1 will create the four force components on the member
shown in the first column of the matrix.
 Truss stiffness matrix
• Once all the member stiffness matrices are formed in global coordinates, it then becomes necessary to
assemble them in the proper order so that the structure stiffness matrix K for the truss can be found.
• This process of combining the member stiffness matrices depends on careful identification of the row and
column of each element in the member stiffness matrices as noted by the four code numbers Nx, Ny, Fx,
Fy.
• For all the truss members the structure stiffness matrix will then have an order that will be equal to the
highest code number assigned to the truss, since this represents the total number of degrees of freedom
for the truss.
• To form the structure stiffness matrix, K, it is therefore necessary to take each member’s elements in k
and place them in the same row and column designation in K.
• If some elements are assigned to the same location, then they must be added together algebraically.
• This is because each element of the k matrix represents the resistance of the member to a unit
displacement at its end, and so, adding the resistances of two or more connected members in the x or y
direction determines the total resistance of each joint to a unit displacement in that direction.
• This method of assembling the member stiffness matrices to form the structure stiffness matrix will now
be demonstrated by two numerical examples. Although this process is somewhat tedious when done by
hand, it is rather easy to program on a computer.
Example
Determine the structure stiffness matrix for the two-member truss
shown in Fig. AE is constant.
The calculations can be checked in
part by noting that k1 is symmetric.
The rows and columns in k1 are
identified by the x, y code numbers at
the near end, followed by those at
the far end, that is, 1, 2, 3, 4, Fig. b.
This is done in order to identify the
elements for later assembly into the
K matrix.
 Here the rows and columns are
identified as 1, 2, 5, 6, since
these numbers represent,
respectively, the x, y code
numbers at the near and far
ends of the member.

Structure Stiffness Matrix.

This matrix has an order of 6 × 6 since there are six designated degrees of freedom for the truss, Fig. b.
Corresponding elements of the member stiffness matrices are now added algebraically to form the
structure stiffness matrix.
The assembly process is easier to see if the missing numerical columns and rows in k1 and k2 are filled
with zeros to form two 6 × 6 matrices.
Structural stiffness matrix

If a computer is used for this operation,

generally one starts with K having all zero
elements; then as the member global
stiffness matrices are generated, they are
placed directly into their respective
element positions in the K matrix, rather
than developing the member stiffness
matrices, storing them, then assembling
them.
Example
• Determine the structure stiffness matrix for the truss shown in Fig. AE
is constant

Although the truss is statically indeterminate to the first

degree, this will present no difficulty for obtaining the structure
stiffness matrix.
Each joint and member are arbitrarily identified numerically, and the near and far
ends are indicated by the arrows along the members. The unconstrained
displacements are code numbered first. There are eight code numbers for the
truss, and so K will be an 8 × 8 matrix. In order to keep all the joint coordinates
positive, the origin of the global coordinates is chosen at joint 1 .
Application of stiffness method to truss analysis
 we can obtain a direct solution for all the unknown joint displacements

Then from the the support

second equation reactions are
with
Member forces can be determined using
Procedure for analysis
• Notation
• Establish the x, y global coordinate system. The origin is usually located at the joint for which the
coordinates for all the other joints are positive.
• Identify each joint and member numerically, and arbitrarily specify the near and far ends of each member
symbolically by directing an arrow along the member with the head directed towards the far end.
• Specify the two code numbers at each joint, using the lowest numbers to identify unconstrained degrees of
freedom, followed by the highest numbers to identify the constrained degrees of freedom.
• From the problem, establish Dk and Qk.
Structure Stiffness Matrix
• For each member determine 𝜆𝑥 and 𝜆𝑦 and the member stiffness matrix.
• Assemble these matrices to form the structure stiffness matrix.
• As a partial check of the calculations, the member and structure stiffness matrices should be symmetric.
Displacements and Loads
• Partition the structure stiffness matrix
• Determine the unknown joint displacements Du, the support reactions Qu and each member force qF
 Example
• Determine the support reactions and the force in each member of the
two-member truss. AE is constant.
• The origin of the global coordinates x, y and the numbering of the joints and members are shown.
• Also, the near and far ends of each member are identified by arrows, and code numbers are used
at each joint.
• By inspection the known external displacements are D3 = D4 = D5 = D6 = 0.
• Also, the known external loads are Q1 = 0, Q2 = -2 kN

Structure Stiffness Matrix: done in previous example

Displacements and Loads.
Unknown displacements Member forces

Support reactions
 Example
• Determine the support reactions and the force in member 2 of the
truss shown. AE is constant.
Structure Stiffness Matrix : Done in previous example
 Displacements and Loads.
For this problem Q = KD is

Unknown displacements
 Support reactions

The force in member 2

Example
• Determine the force in member 2 of the assembly in Fig. if the
support at joint 1 settles downward 25 mm. Take AE = 8(103 ) kN.
Structure Stiffness Matrix

Assembling
Force in member 2
 Lecture 35
20/04/24
Beam Analysis using Stiffness Method
 Member and Node Identification
• In order to apply the stiffness method to beams, we must first determine how to subdivide the
beam into its component finite elements.
• In general, each element must be free from load and have a prismatic cross section, and so the
nodes of each element are located at a support or at points where members are connected
together, where an external force is applied, where the cross-sectional area suddenly changes,
or where the vertical or rotational displacement is to be determined.
• Four nodes are specified numerically within a circle, and the three elements are identified
numerically within a square. Finally, the “near” and “far” ends of each element are identified
by the arrows written alongside each element.
 Global and Member Coordinates
• The global coordinate system identified using x, y, z axes generally have their origin at a node and
are positioned so that the nodes at other points on the beam all have positive coordinates.
• The local or member x′, y′, z′ coordinates have their origin at the “near” end of each element,
and the positive x′ axis is directed towards the “far” end. (Fig b)
• In both cases we have used a right-handed coordinate system, so that when the fingers of the
right hand are curled from the x (x′) axis towards the y (y′) axis, the thumb points in the positive
direction of the z (z′) axis, which is directed out of the page. Notice that for each beam element
the x and x′ axes will be collinear and the global and member coordinates will all be parallel.
Therefore, unlike the case for trusses, here we will not need to develop transformation matrices
between these coordinate systems.
 Code Numbers
• Once the elements and nodes have been identified, and the global coordinate system has been
established, the degrees of freedom for the beam and its kinematic determinacy can be
determined.
• If we consider the effects of both bending and shear and neglect axial deformation, then each
node on a beam can have two degrees of freedom, namely, a vertical displacement and a rotation.
• As in the case of trusses, these linear and rotational displacements will be identified by code
numbers. The lowest code numbers will be used to identify the unknown displacements
(unconstrained degrees of freedom), and the highest numbers to identify the known
displacements (constrained degrees of freedom).

Notice that the internal

hinge at node 3 deflects the There are five
same for both elements 2 unknown
and 3; however, the rotation deflection
at the end of each element is components
different. For this reason labeled with
three code numbers are used the lowest code
to show these deflections. numbers.
 Beam –member stiffness matrix

• We will now develop the stiffness matrix for a beam element or member having a constant cross-
sectional area and referenced from the local x′, y′, z′ coordinate system.
• The origin of the coordinates is placed at the “near” end N, and the positive x′ axis extends toward
the “far” end F.
• There are two reactions at each end of the element, consisting of shear forces qNy′ and qFy′ and
bending moments qNz′ and qFz′.
• Notice that these loadings all act in the positive coordinate directions, where the moments qNz′ and
qFz′ are positive counterclockwise, since by the right hand rule the moment vectors are then
directed along the positive z′ axis, which is out of the page.
• Linear and angular displacements associated with these loadings also follow this same positive sign
convention.
• We will now impose each of these displacements separately and then determine the loadings
acting on the member caused by each displacement.
y’ Displacements

• When a positive displacement dNy′ is imposed on the element, while other possible
displacements are prevented, the required shear forces and bending moments that have to
occur are shown in Fig. a.
• Likewise, when dFy′ is imposed, the necessary shear forces and bending moments are shown in
Fig. b.
 z' Rotations
• In order to impose a positive rotation dNz′ while all other displacements are prevented, the required
shear forces and bending moments are shown in Fig. a.
• Likewise, when dFz′ is imposed, the resultant loadings are shown in Fig. b
 Member stiffness matrix - local and global coordinates
If the above results in (y’ displacements and z’ rotations) are added, the resulting four load-
displacement relations for the member can then be expressed in matrix form as

member stiffness matrix

(symmetric)

• The 16 influence coefficients kij that comprise it account for the load on the member when the
member undergoes a specified unit displacement.
• For example, if dNy′ = 1, while all other displacements are zero, the member will be subjected only
to the four loadings indicated in the first column of the k matrix. In a similar manner, the other
columns are the member loadings for unit displacements identified by code numbers listed above
the columns.
 Beam structure stiffness matrix
• Once all the member stiffness matrices have been found, we can then assemble them into the
structure stiffness matrix K.
• Since the rows and columns of each k matrix are identified by the two code numbers at the
near end of the member (Ny′, Nz′) followed by those at the far end (Fy′, Fz′), then, like a truss,
when assembling the matrices, each matrix element must be placed in the same location of the
K matrix.
• In this way, K will have an order that will be equal to the highest code number assigned to the
beam, since this represents the total number of degrees of freedom.
 Application of stiffness method for beam analysis
• After the structure stiffness matrix is determined, the loads Q at the nodes of the beam can
then be related to the displacements D using the structure stiffness equation

• Partitioning the matrices into the known and unknown elements of load and displacement,
we have
Expansion yields

• The unknown displacements Du are determined from the first of these equations. Then
using these values, the support reactions Qu are calculated from the second equation.
 Intermediate loadings
• For application, it is important that all the finite elements of the beam be
free of loading along their length. This is necessary since the stiffness
matrix for each element was developed for loadings applied only at its
ends. Oftentimes, however, beams will support a distributed loading,
and this condition will require modification in order to perform the
matrix analysis.
• First we will apply its fixed-end moments and reactions to the nodes
which will be used as external loadings in the stiffness method, Fig.b.
• These loadings will give the correct displacements and internal
reactions (shear and moment) at the ends of members 1 and 3.
However, to obtain the correct internal reactions for member 2 we
must add the reverse fixed-end loadings back on this member.
• For example, if the matrix analysis
produces shear forces qNy′, qFy′ and
moments qNz′, qFz′ for member 2, Fig. c,
then the loadings in Fig. d must be added
to these loadings to determine the final
results, Fig. e.
• superposition can be expressed as
where q0 represents the reversed fixed-end loadings
Procedure
for analysis
Example
• Determine the reactions at the supports of the beam shown in Fig.
EI is constant
Member Stiffness Matrices
• Displacements and Loads
Ex: K11 = 0 + 2 = 2, K55 = 1.5 + 1.5 = 3
Example
• Determine the reaction at the roller of the compound beam shown in Fig. EI
is constant.
Member Stiffness Matrices
Displacements and Loads
Example
Beam is shown in Fig below is subjected to two couple moments. If the
Member Stiffness Matrices
Displacements and Loads
 Example
• Determine the moment developed at support A of the beam shown in Fig..
Assume the roller supports can pull down or push up on the beam. Take E
= 200 GPa, I = 216(106) mm4
 The matrix analysis requires that the external loading be
applied at the nodes, and therefore the distributed and
concentrated loads are replaced by their equivalent fixed-end
moments.

No external loads are placed at node ① and no external vertical

forces are placed at nodes ② and ③,since the reactions at code
numbers 3 through 6 are to be unknowns in the load matrix. Using
super position, the results of the matrix analysis for the loading in
Fig.b will later be modified by the fixed-end loads shown in Fig. c.

Member Stiffness Matrices

Member 1.
• Member 2

The actual moment at A must include the fixed-supported

reaction of +96 kN m shown in Fig. c, along with the calculated
result for Q3. Thus,
MAB = 12 kN.m + 96 kN. m = 108 kN .m
• We can determine the internal moment and shear at B by considering, for example,
member 1, node 2
Example
• Determine the deflection at ① and the reactions on the beam shown in Fig. EI is
constant.
 A rotational displacement D4 Unknown displacements
does not occur because of the
roller constraint.

Member Stiffness Matrices

EI is constant and the members
are of equal length, the member
stiffness matrices are identical.
Plane Frame Analysis Using Stiffness Method
 Frame member stiffness matrix
• We will develop the stiffness matrix for a prismatic frame member referenced from local x′, y′, z′ coord system.
• Here the member is subjected to axial loads qNx′, qFx′, shear loads qNy′, qFy′, and bending moments qNz′, qFz′ at its
near and far ends, respectively.
• These loadings all act in the positive coordinate directions along with their associated displacements.
• As in the case of beams, the moments qNz′ and qFz′ are positive counterclockwise, since by the right-hand rule
the moment vectors are then directed along the positive z′ axis, which is out of the page.

Member stiffness matrix

–local coordinate system
Each coefficient- Load on the
member when the member
undergoes a specified unit
displacement.
We have established each of the load-displacement
relationships caused by these loadings in the previously (for
trusses and beams).
By superposition, if these results are added together, then
the resulting six load-displacement relations for the member
can be expressed in matrix form as
 Displacement and force transformation matrices
• As in the case for trusses, we must be able to transform the internal member loads q and
deformations d from local x′, y′, z′ coordinates to global x, y, z coordinates.
• For this reason transformation matrices are needed.
Displacement Transformation Matrix
A global coordinate displacement DNx creates local coordinate
displacements

Likewise, a global coordinate displacement DNy, creates local

coordinate displacements of

Finally, since the z′ and z axes are coincident, that is, directed out of
the page, a rotation DNz about z causes a corresponding rotation dNz′
about z′. Thus,
• In a similar manner, if global displacements DFx in the x direction, DFy in the y direction,
and a rotation DFz are imposed on the far end of the member, the resulting
transformation equations become

Letting represent the direction cosines of the member, we

can write the superposition of displacements in matrix form as

T transforms the six global x, y, z displacements D into the six local x′, y′, z′ displacements d.
Hence T is referred to as the displacement transformation matrix.