UNIT- 1

Matrix Method
INTRODUCTION TO MATRIX METHODS
MATRIX METHODS OF ANALYSIS :
Broadly the methods of analysis are categorised in two ways
1. Force Methods : Methods in which forces are made unknowns i.e Method
of consistent deformation and strain energy method. In both thesemethods
solution of number of simultaneous equations isinvolved.
2. Displacement Methods in which displacements are made unknowns i.e
slope deflection method, Moment distribution method and Kani’sMethod
(In disguise). In slope deflection method also, the solution of number of
simultaneous equations isinvolved.
In both of the above methods, for the solution of simultaneous equations
matrix approach can be employed & such Method is called Matrix method
of analysis.

FORCE METHOD :
Method of consistent deformation is the base and forces are made unknown

∆b = Upward Deflection of point B on primary structure due to all causes

∆bo = Upward Deflection of point B on primary structure due to applied
load(Redundant removed i.e condition Xb = 0)
∆bb = Upward Deflection of point B on primary structure due to Xb
( i.e Redundant )
δbb = Upward Deflection of point B on primary structure due to Xb = 1
∆bb = δbb . Xb
∴ ∆b = ∆bo +∆bb Substituting for ∆bb–
 ∆b = ∆bo + δbb.Xb Called Super positionequation
Using the compatibility condition that the net displacement at B = 0i.e

∆b = 0 we get Xb = – ∆bo / δbb

To Conclude wecansay, [ ∆ ] = [∆L ] +[ ∆R]

DISPLACEMENT METHOD:
This method is based on slope deflection method and displacements are made
unkowns
which are computed by matrix approch instead of solving simultaneous equations
and finaly unknown forces are calculated using slpoe deflection equations.
Mab = Mab + 2EI / L ( 2 θa + θb + 3δ / L)
Mab = Final Moment and may be considered as net force P at the joint
Mab = Fixed end moment i.e Force required for the condition of zero
displacements & is called locking force. ( i.e.P’)
The second term may be considred as the force required to produce the required
displacements at the joints. (i.e Pd )Therefore the above equation may bewritten
as [P] = [P’] +[Pd]
Thus, there are Two Methods in matrix methods

MATRIX METHODS
FORCEMETHOD DISPLACEMENTMETHOD

The force method is also called by the names 1) Flexibility Method 2)Static
Method 3)Compatibility.
Similarly the displacement method is also called by the names 1)Stiffness Method
2) Kinematic method 3) EquilibriumMethod.
In both force method & displacement method there are two different approaches 1)
System Approach 2) Element Approach.

To study matrix methods there are some pre-requisites :

i) Matrix Algebra - Addition, subtraction ,Multiplication & inversionof
matrices (Adjoint Method)
ii) Methods of finding out Displacements i.e. slope & deflection at anypoint
in a structure, such as a)Unit load method or Strain energy method b)
Moment area methodetc.
According to unit load method the displacement at any point ‘j’ is given by
∆j = ∫0 Mmjds/ EI
Where
M – B M due toapplied loads & mj – B M due to unit load atj
δij = ∫0 mi.mj.ds/EI
When unit load is applied at i andiscalled flexibility coefficient.
The values of ∆j and δij can be directly read from the table depending upon the
combinations of B M diagrams & these tables are called Diagram Multipliers.
iii) Study of Indeterminacies – Static indeterminacy & kinematicindeterminacy
 INDETERMINATE STRUCTURES
1. Statically IndeterminateStructure
2. Kinematically IndeterminateStructure
INDETERMINATESTRUCTURES
Statically Indeterminate Structure : Any structure whose reaction components
or internal stresses cannot be determined by using equations of staticequilibrium
alone, (i.e.ΣFx = 0, ΣFy = 0, ΣMz = 0) is a statically IndeterminateStructure.
The additional equations to solve statically indeterminate structurecome from
the conditions of compatibility or consistentdisplacements.

Roller Support : No.ofreactions,

r =1

Hinged Support : No.ofreactions, r =2

Fixed Support : No.ofreactions, r =3

1. Pin Jointed Structures i.e. Trusses
Internal static indeterminacy : (Dsi) No. of members required for stability is
given by – 3 joints – 3 members – every additional joint requires two additional
members.
∴ m’ = 2 (j– 3)+3 j = No. ofjoints
∴ m’ = 2j–3 Stable and statically determinate
Dsi = m – m’ Where m = No. of members in astructure
Dsi = m – (2j –3)
External static indeterminacy (Dse)
r = No. of reaction components
Equations of static equilibrium=3 (i.e.ΣFx = 0, ΣFy = 0, ΣMz =0)
∴ Dse = r –3
Total static indeterminacy Ds = Dsi+Dse ∴ Ds = m –(2j – 3)+(r –3)
∴ Ds = (m+r) – 2j
Rigid Jointed Structures : No. of reaction components over and above the no. of
equations of static equilibrium is called a degree of static indeterminacy.
Ds=r-3
Equations of static equilibrium=3
(i.e.ΣFx = 0, ΣFy = 0,ΣMz= 0)

Example 1

No. ofreactioncomponents r = 5 (asshown)

∴ Ds = r – 3 = 5 – 3=2 Ds =2

Introduce cut in the member BC as shown. At the cut the internal stresses are
introduced i.e. shear force and bending moment as shown.
Left part : No. of unknowns=5 Equations of equilibrium =3
∴ Ds = 5 – 3 = 2
Right Part : No. of unknowns = 4 Equations of equilibrium = 2
∴ Ds = 4 – 2 = 2
∴ Ds = Static Indeterminacy = 2
Example 2
Fig.(A) Fig.(B) Fig.(C)
Another Approach : For Every member in a rigid jointed structure there will be 3
unknowns i.e. shear force, bending moment, axial force.
Let r be the no. of reaction components and m be the no. of members
Total no. of unknowns = 3m + r
At every joint three equations of static equilibrium are available
∴ no. of static equations of equilibrium = 3j (where j is no. of joints)
∴ Ds = (3m + r) – 3j
∴ Intheexample r = 6, m = 6, j =6
∴Ds = (3 x 6 + 6) – (3 x 6 ) = 6
Kinematic Indeterminacy :
A structure is said to be kinematically indeterminate if the displacement
components of its joints cannot be determined by compatibility conditions alone.
In order to evaluate displacement components at the joints of these structures, it is
necessary to consider the equations of static equilibrium. i.e. no. of unknown joint
displacements over and above the compatibility conditions will give the degree of
kinematic indeterminacy.

Fixed beam : Kinematically determinate :

Simply supported beam Kinematically indeterminate

Any joint – Moves in three directions in a plane structure

Two displacements δx in x direction, δy in y direction,
θrotation about z axis as shown.

Roller Support :
r = 1, δy = 0, θ&δx exist – DOF=2 e=1

Hinged Support :
r = 2, δx = 0, δy = 0, θexists – DOF = 1 e = 2

Fixed Support :
r = 3, δx = 0, δy = 0, θ=0 DOF=0 e=3

i.e. reaction components prevent the displacements ∴ no. of restraints = no. of

reaction components.
Degree of kinematic indeterminacy :
Pin jointed structure :Every joint – two displacements components and no
rotation
 ∴ Dk = 2j–e
where,
e = no. of equations ofcompatibility
j = no. of reaction components

Rigid Jointed Structure : Every joint will have three displacement components,
two displacements and one rotation.
Since, axial force is neglected in case of rigid jointed structures, it is assumed
that the members are inextensible & the conditions due to inextensibility of
members will add to the numbers of restraints. i.e to the ‘e’ value.

∴ Dk = 3j–e
where,
e = no. of equations ofcompatibility
j= no. of reaction components + constraints due to in
extensibility
Example 1 : Find the static and kinematic indeterminacies

r = 4, m = 2, j = 3

Ds = (3m + r)–
3j
= (3 x 2 + 4) – 3 x 3 = 1
Dk = 3j –e
=3x3–6=3
i.e. rotations at A,B, & C i.e. θa, θb &θc are
the displacements.
(e = reaction components + inextensibility conditions = 4 + 2 = 6)
Example 2 :
D
s

(
3
m
+
r
)

–
3 =
j
6
m ,

= j

3 =
,
4
r ∴ Ds = (3 x 3 + 6) – 3 x 4 = 3

Dk = 3j– e e = no. ofreaction

components + conditions of
inextensibility
= 6+3 = 9
Dk = 3 x 4 – 9 = 3 i.e. rotation θb, θc & sway.
Example 3 :
D
s

(
3
m

r
)

3
j

6
,

1
0
,

9
 ∴Ds
= (3
x 10
+ 6)
–3x
9=9

Con
diti
ons
of
ine
xte
nsib
ility
:
Join
t:
B
C
EF
H
1 1 2 2 2 2 Total =10

Reactioncomponents r =6

∴ e = 10 + 6 = 16

∴Dk = 3j – e
= 3 x 9 – 16 = 11
 FORCE METHOD :
This method is also known as flexibility method
or compatibility method. In this method the degree of static indeterminacy of the
structure is determined and the redundants are identified. A coordinate isassigned
to each redundant.Thus,P1,P2 Pn are the redundants at thecoordinates
1,2, n.If all the redundants are removed , the resulting structure knownas
released structure, is statically determinate. This released structure is also known
as basic determinate structure. From the principle of super position the net
displacement at any point in statically indeterminate structure is some of the
displacements in the basic structure due to the applied loads and the redundants.
This is known as the compatibility condition and may be expressed by the
equation;

∆1 = ∆1L+∆1R Where ∆1 - - - - ∆n = Displ. At Co-ord.at 1,2 --n

∆2 = ∆2L+∆2R ∆1L ---- ∆nL = Displ.AtCo-ord.at 1,2 n
| | | Due to apliedloads
| | | ∆1R ----∆nR = Displ.AtCo-ord.at1,2 n
∆n = ∆nL+∆nR Due toRedudants

The above equations may be return as [∆] = [∆L]+ [∆R] (1)

∆1 = ∆1L + δ11 P1 + δ12P2+ δ1nPn
∆2 = ∆2L + δ21 P1 + δ22P2+ δ2nPn
| | | | |
| | | | | - - - - -(2)

∆n = ∆nL + δn1 P1 + δn2P2+ δnnPn

 ∴∆ = [∆ L] +[δ][P] (3)

∴[P]= [δ] {[∆] –[∆ L]}

-1
(4)

If the net displacements at the redundants are zero then

∆1,∆2 ∆n=0,
Then ∴ [P] = - [δ] -1[∆ L] (5)
TheredundantsP1,P2, Pn are Thusdetermined
DISPLACEMENT METHOD :
This method is also known as stiffness or equilibrium. In this
method the degree of kinematic indeterminacy (D.O.F) of the structure is
determined and the coordinate is assigned to each independent
displacement component.

In general, The displacement components at the supports and joints are treatedas
independent displacement components.Let1,2, n be thecoordinates
assigned to these independent displacement components∆1,∆2 ∆n.
In the first instance lock all the supports and the joints to obtain the
restrained structure in which no displacement is possible at the coordinates.Let
P’1, P’2 , - - - - P’n be the forces required at thecoordinates1,2, n inthe
restrained structure in which the displacements∆1,∆2 ∆n are zero. Next,Let
the supports and joints be unlocked permitting displacements∆1, ∆2 ∆nat
the coordinates. Let these displacements require forces inP1d,P2d, Pndat
coordinates1,2, nrespectively.
If P1, P2- - - - - - Pn are the external forces at thecoordinates1,2, n, thenthe
conditions of equilibrium of the structure may be expressed as:
P1 = P’1 +P1d
P2 =P’2+P2d (1)
| | |
| | |
Pn = P’n + Pnd
or [P ] = [P’] +[Pd] (2)

P1 = P’1 +K11 ∆1+ K12, ∆2+K13∆3+ K1n∆n

P2 = P’2 +K21 ∆1+ K22 ∆2+K23∆3+ K2n∆n
 | | | | | |
| | | | | ---------
(3)
Pn = P’n +Kn1 ∆1+ Kn2, ∆2+Kn3∆3+ - - - - - - - --Knn ∆n ∴
i.e [P ] = [P’ ] + [ K ] [∆] (4)
∴ [ ∆ ]= [ K ] –1 {[P]–[P’]} (5)
Where P = External forces
 P’ = Locking forces
Pd=Forces due to displacements
If the external forces act only at the coordinates the terms P’1,P’2, P’n
vanish. i. e the Locking forces arezero,then [
∆ ]= [ K ] –1 [P] (6)
On the other hand if there are no external forces at the coordinates then [P]=0then
[ ∆ ]= – [ K ] –1 [P’] (7)
Thus the displacements can be found out. Knowing the displacements the forces
are computed using slope deflection equations:
Mab= Mab+ 2EI / L (2θa+ θb+3δ /
L) Mba=Mba+ 2EI / L (θa+ 2θb+3δ /
L)
Where Mab& Mba are the fixed end moments for the member AB due to external
loading
FLEXIBILITY AND STIFFNESS MATRICES : .
SINGLECO-ORD.

D
=

D
=
P
L
3

/
3
E
I

P
∴
δ=
L3
/3
EI
δ=
Fl
exi
bil
ity
Co
eff
 P=K x D
P=K x PL3 / 3EI K=3EI / L3
K=Stiffness Coeff.

M=K x D =K x ML/EI
D= ML / EI
D= δ x M=ML/EI ∴ K=EI / L
 δ=L / EI K=Stiffness Coeff.
δ=Flexibility Coeff. δ X K= 1

TWO CO-ORDINATE SYSTEM

D1= δ11P1 + δ12 X P2 & D2= δ21P1+ δ22P2
∴D1 δ11δ12 P1
D2 δ21δ22 P2

L3/ 3EI L2/ 2EI L2/ 2EI

[δ ]=
L /EI

δ11=L3/3EI δ21=L2/2EI
Unit Force At Co-ord.(1)

Unit Force At Co-ord.(2) δ12=δ21 =L2/ 2EI δ22=L /EI

Develop the Flexibility and stiffness matrices for frame ABCD with reference to
Coordinates shown

The Flexibility matrix can be developed by

applying unit force successively at
coordinates (1),(2) &(3) and evaluating the
displacements at all the coordinates

δij = ∫ mi mj / EIxds δij =displacement at I due to unit load atj

INVERSING THE FLEXIBILITY MATRIX [ δ ]
THE STIFENESS MATRIX [ K ] CAN BE OBTAINED
UNIT-2
U
7777
 UNIT 3
Direct stiffness method and the global stiffness matrix
Although there are several finite element methods, we analyse the Direct Stiffness
Method here, since it is a good starting point for understanding the finite element
formulation. We consider first the simplest possible element – a 1-dimensional
elastic spring which can accommodate only tensile and compressive forces. For the
spring system shown, we accept the following conditions:

▪ Condition of Compatibility – connected ends (nodes) of adjacent

springs have the same displacements
▪ Condition of Static Equilibrium – the resultant force at each node is
zero
▪ Constitutive Relation – that describes how the material (spring)
responds to the applied loads

Model spring system

The constitutive relation can be obtained from the governing equation
for an elastic bar loaded axially along its length:

ddu(AEΔll0)+k=0(1)
Δll0=ε(2)
ddu(AEε)+k=0(3)
ddu(Aσ)+k=0(4)
dFdu+k=0(5)
dFdu=−k(6)
dF=−kdu(7)
The spring stiffness equation relates the nodal displacements to the
applied forces via the spring (element) stiffness. The minus sign
denotes that the force is a restoring one, but from here on in we use
the scalar version of Eqn.7.

Derivation of the Stiffness Matrix for a Single Spring Element

From inspection, we can see that there are two degrees of freedom in
this model, ui and uj. We can write the force equilibrium equations:
k(e)ui−k(e)uj=F(e)i(8)
−k(e)ui+k(e)uj=F(e)j(9)
In matrix form

[ke−ke−keke]{uiuj}={F(e)iF(e)j}(10)
The order of the matrix is [2×2] because there are 2 degrees of
freedom. Note also that the matrix is symmetrical. The ‘element’
stiffness relation is:
[K(e)]{u(e)}={F(e)}(11)
Where Κ(e) is the element stiffness matrix, u(e) the nodal
displacement vector and F(e) the nodal force vector. (The element
stiffness relation is important because it can be used as a building
block for more complex systems. An example of this is provided
later.)
Derivation of a Global Stiffness Matrix
For a more complex spring system, a ‘global’ stiffness matrix is
required – i.e. one that describes the behaviour of the complete
system, and not just the individual springs.

From inspection, we can see that there are two springs (elements) and
three degrees of freedom in this model, u1, u2 and u3. As with the
single spring model above, we can write the force equilibrium
equations:

k1u1−k1u2=F1(12)
−k1u1+(k1+k2)u2−k2u3=F2(13)
k2u3−k2u2=F3(14)
In matrix form
⎡⎣⎢k1−k10−k1k1+k2−k20−k2k2⎤⎦⎥⎧⎩⎨⎪⎪u1u2u3⎫⎭⎬⎪⎪=⎧⎩⎨
⎪⎪F1F2F3⎫⎭⎬⎪⎪(15)
The ‘global’ stiffness relation is written in Eqn.16, which we
distinguish from the ‘element’ stiffness relation in Eqn.11.
[K]{u}={F}(16)
Note the shared k1 and k2 at k22 because of the compatibility
condition at u2. We return to this important feature later on.
Assembling the Global Stiffness Matrix from the Element Stiffness
Matrices
Although it isn’t apparent for the simple two-spring model above,
generating the global stiffness matrix (directly) for a complex system
of springs is impractical. A more efficient method involves the
assembly of the individual element stiffness matrices. For instance, if
you take the 2-element spring system shown,

split it into its component parts in the following way

and derive the force equilibrium equations

k1u1−k1u2=F1(17)
k1u2−k1u1=k2u2−k2u3=F2(18)
k2u3−k2u2=F3(19)
then the individual element stiffness matrices are:
[k1−k1−k1k1]{u1u2}={F1F2}and[k2−k2−k2k2]{u2u3}={F2F3}(20)
such that the global stiffness matrix is the same as that derived
directly in Eqn.15:

(Note that, to create the global stiffness matrix by assembling the

element stiffness matrices, k22 is given by the sum of the direct
stiffnesses acting on node 2 – which is the compatibility criterion.
 Note also that the indirect cells kij are either zero (no load transfer
between nodes i and j), or negative to indicate a reaction force.)
For this simple case the benefits of assembling the element stiffness
matrices (as opposed to deriving the global stiffness matrix directly)
aren’t immediately obvious. We consider therefore the following
(more complex) system which contains 5 springs (elements) and 5
degrees of freedom (problems of practical interest can have tens or
hundreds of thousands of degrees of freedom (and more!)). Since
there are 5 degrees of freedom we know the matrix order is 5×5. We
also know that it’s symmetrical, so it takes the form shown below:

We want to populate the cells to generate the global stiffness matrix.

From our observation of simpler systems, e.g. the two spring system
above, the following rules emerge:

▪ The term in location ii consists of the sum of the direct stiffnesses of

all the elements meeting at node i
▪ The term in location ij consists of the sum of the indirect stiffnesses
relating to nodes i and j of all the elements joining node i to j
▪ Add a negative for reaction terms (–kij)
▪ Add a zero for node combinations that don’t interact
By following these rules, we can generate the global stiffness matrix:
This type of assembly process is handled automatically by
commercial FEM codes

Drag the springs into position and click 'Build matrix', then apply a
force to node 5. You will then see the force equilibrium equations, the
equivalent spring stiffness and the displacement at node 5.

Solving for (u)

The unknowns (degrees of freedom) in the spring systems presented
are the displacements uij. Our global system of equations takes the
following form:

To find {u} solve

u=F[K]−1(22)
Recall that [k][k]−1=I=IdentitiyMatrix=[1001].
Recall also that, in order for a matrix to have an inverse, its
determinant must be non-zero. If the determinant is zero, the matrix is
said to be singular and no unique solution for Eqn.22 exists. For
instance, consider once more the following spring system:
 We know that the global stiffness matrix takes the following form
⎡⎣⎢k1−k10−k1k1+k2−k20−k2k2⎤⎦⎥⎧⎩⎨⎪⎪u1u2u3⎫⎭⎬⎪⎪=⎧⎩⎨
⎪⎪F1F2F3⎫⎭⎬⎪⎪(23)
The determinant of [K] can be found from:
det⎡⎣⎢adgbehcfi⎤⎦⎥=(aei+bfg+cdh)−(ceg+bdi+afh)(24)
Such that:

(k1(k1+k2)k2+0+0)−(0+(−k1−k1k2)+(k1−k2−k2))(25)
det[K]=(k12k2+k1k22)−(k12k2+k1k22)=0(26)
Since the determinant of [K] is zero it is not invertible, but singular.
There are no unique solutions and {u} cannot be found. If this is the
case in your own model, then you are likely to receive an error
message!
1

Instructional Objectives
After reading this chapter the student will be able to
1. Derive member stiffness matrix of a beamelement.
2. Assemble member stiffness matrices to obtain the global stiffness matrix for a beam.
3. Write down global load vector for the beamproblem.
4. Write the global load-displacement relation for thebeam.

Introduction.
In chapter 23, a few problems were solved using stiffness method from
fundamentals. The procedure adopted therein is not suitable for computer
implementation. In fact the load displacement relation for the entire structure was
derived from fundamentals. This procedure runs into trouble when the structure is
large and complex. However this can be much simplified provided we follow the
procedure adopted for trusses. In the case of truss, the stiffness matrix of the entire
truss was obtained by assembling the member stiffness matrices of
individualmembers.
In a similar way, one could obtain the global stiffness matrix of a continuous beam
from assembling member stiffness matrix of individual beam elements. Towards
this end, we break the given beam into a number of beam elements.
Thestiffnessmatrixofeachindividualbeamelementcanbewrittenveryeasily.
For example, consider a ABCD The
continuous beam as shown in Fig. 27.1a.
given continuous beam is divided into three beam elements as shown in Fig.
27.1b. It is noticed that, in this case, nodes are located at the supports. Thus each
span is treated as an individual beam. However sometimes it is required to
consider a node between support points. This is done whenever the cross sectional
area changes suddenly or if it is required to calculate vertical or rotational
displacements at an intermediate point. Such a division is shown in Fig. 27.1c. If
the axial deformations are neglected then each node of the beam will have two
degrees of freedom: a vertical displacement (corresponding to shear) and a
rotation (corresponding to bending moment). In Fig. 27.1b, numbers
enclosedinacirclerepresentsbeamnumbers.ThebeamABCDisdividedinto
three beam members. Hence, there are four nodes and eight degrees of freedom.
The possible displacement degrees of freedom of the beam are also shown in the
figure. Let us use lower numbers to denote unknown degrees of freedom
(unconstrained degrees of freedom) and higher numbers to denote known
(constrained) degrees of freedom. Such a method of identification is adopted in
this course for the ease of imposing boundary conditions directly on the structure
stiffness matrix. However, one could number sequentially as shown in Fig. 27.1d.
This is preferred while solving the problem on acomputer.
In the above figures, single headed arrows are used to indicate translational and
double headed arrows are used to indicate rotational degrees of freedom.

Beam StiffnessMatrix.
Fig. 27.2 shows a prismatic beam of a constant cross section that is fully restrained
at ends in local orthogonal co-ordinate system x' y' z' . The beam ends
are denoted bynodes j and k .The x' axis coincides with the
centroidal axis of
the member with the positive sense being defined from j to k . Let L be the length
of the member, A area of cross section of I zz is the moment of
the member and
inertia about z'axis.

Two degrees of freedom (one translation and one rotation) are considered at each
end of the member. Hence, there are four possible degrees of freedom for this
member and hence the resulting stiffness matrix is of the order 4 × 4 . In this
method counterclockwise moments and counterclockwise rotations are taken as
positive. The positive sense of the translation and rotation are also shown in the
figure. Displacements are considered as positive in the direction of the co- ordinate
axis. The elements of the stiffness matrix indicate the forces exerted on
the member by the restraints at the ends of the member when unit displacements
are imposed at each end of the member. Let us calculate the forces developed in
the above beam member when unit displacement is imposed along each degree of
freedom holding all other displacements to zero. Now impose a unit
displacement along y' axisat holding all other
j end of the member while
displacements to zero as shown in Fig. 27.3a. This displacement causes both shear
and moment in the beam. The restraint actions are also shown in the figure. By
definition they are elements of the member stiffness matrix. In particular they form
the first column of element stiffnessmatrix.
In Fig. 27.3b, the unit rotation in the positive sense is imposed at j end of the beam
while holding all other displacements to zero. The restraint actions are
showninthefigure.Therestraintactionsatendsarecalculatedreferringto
tables given in lesson …
In Fig. 27.3c, unit y' axis at end k is imposedand
displacement along
corresponding restraint actions are calculated. Similarly in Fig. 27.3d, unit
rotation about z' axis at end k corresponding stiffness
is imposed and
coefficients are calculated. Hence the member stiffness matrix for the beam
member is

1 2 3 4
⎡12 EIz 6 EIz 12 EIz 6 EIz ⎤1
⎢ L2 L2⎥
⎢ L3 L3
⎥2
⎢ z6EI 4EIz z 6EI 2EI
z⎥

[k ]=⎢ L2 L L2 L ⎥ (27.1)
⎢  12EIz 6 EIz 12 EI z 6 EI z ⎥3
⎢ L3 ⎥
L3 L2 L2
⎢
⎢ 6EI Lz 2 2EI z  6 EI z 4EI z ⎥⎥4
⎣ L L2 L ⎦

The stiffness matrix is symmetrical. The stiffness matrix is partitioned to separate

the actions associated with two ends of the member. For continuous beam
problem, if the supports are unyielding, then only rotational degree of freedom
shown in Fig. 27.4 is possible. In such a case the first and the third rows and
columns will be deleted. The reduced stiffness matrix will be,

⎡4EIz 2EIz⎤
⎢ ⎥
[k ]=⎢2EIL L
4EI ⎥ (27.2)
⎢z z⎥
⎣L L⎦
Instead of imposing unit y' at j end of the member in
displacement along Fig.
a, apply ent u'1 y' at member as
adisplacem shown in
along j end of the
Fig. 27.5a, holding all other displacements to zero. Let the restraining forces
developed be denoted by q11 , q21 , q31 and q41 .
 The forces are equal to,
q11 =k11u'1 ; q21 =k21u'1 ; q31 =k31u'1 ; q41 =k41u'1
(27.3)

Now, give u'1 , u'2 ,u'3 and u'4 simultaneously

displacements along displacement
degrees of freedom 1,2,3 and 4 respectively. Let the restraining forces developed
at member ends be q1,q2,q3and q4 respectively as shown in
Fig. 27.5b along
respective degrees of freedom. Then by the principle of superposition, the force
displacement relationship can be written as,

⎡q ⎤
⎡12EIz 6EIz − 12EIz 6EIz⎤ ⎡u' ⎤
1 2 ⎥
⎢ ⎥ ⎢⎢ L3 L2 L ⎥ ⎢ 1
L3
⎢ ⎥ ⎥
⎢ ⎥ ⎢ 6EI 4EI 6EI 2EI
⎥

⎢2q⎥ ⎢ z z − z z ⎥⎢u'⎥ 2
⎢ ⎥ ⎢ L
2
L L 2
L ⎥⎢ ⎥
⎢ ⎥=⎢ ⎥⎢ ⎥ (27.4)
12EIz
⎢q⎥ ⎢−12EIz −6EIz ⎥⎢u' ⎥
⎥
−6EIz
⎢3⎥ 2
⎢ L3
LL L3
2
⎢ ⎥ ⎢
⎢ ⎥ ⎢
⎢6EI 2EI 6EI
⎢
 ⎢ ⎥
4EI ⎥
⎢ q⎥ z ⎥ ⎢ u' ⎥
⎣ 4 ⎦ ⎣ L 2 z − z
z
⎣ 4 ⎦
L ⎦ L2 L

This may also be written

in compact form as,
(27.5)
{
q
}
=
[
k
]
{
u
'
}
 Beam (global) StiffnessMatrix.
The formation of structure (beam) stiffness matrix from its member stiffness
matrices is explained with help of two span continuous beam shown in Fig. 27.6a.
Note that no loading is shown on the beam. The orthogonal co-ordinate system xyz
denotes the global co-ordinatesystem.

Forthecaseofcontinuousbeam,the x-and x'-axesarecollinearandother

axes(yandy', zandz')areparalleltoeachother.Henceitisnotrequiredto
transformmemberstiffnessmatrixfromlocalco-ordinatesystemtoglobalco
ordinate system as done in the case of trusses. For obtaining the global stiffness
matrix, first assume that all joints are restrained. The node and member numbering
for the possible degrees of freedom are shown in Fig 27.6b. The continuous beam
is divided into two beam members. For this member there are six possible degrees
of freedom. Also in the figure, each beam member with its displacement degrees
of freedom (in local co ordinate system) is also shown. Since the continuous beam
has the same moment of inertia and span, the member stiffness matrix of element 1
and 2 are the same. They are,
G f1 2 3
l 1 2 3
o ⎡ k '11 k '12 k '13
b
a
l

.
o
.

L
o
c
a
l

.
o
.
 4 4 k '14 ⎤ 1 1
⎢ ⎥ k '24 ⎥ 2 2 7.
⎢k '  ⎢21 k '22 k k '23
6
(2 a)
⎢k '31 k '32 k '33 k '34 ⎥ 3 3
⎢ ⎥
k '42 ⎣ k k '43 k '44 ⎦ 4 4
41

Global d .o. f 3 4 5 6
L 1 2 3 4
oc
al ⎡⎢kk221
211 k 212 k 213 k 214 ⎤ 1 3
k 224 ⎥ 2 4
k 222 k 223
d.
o.
f
⎢k 2  ⎢ ⎥ 7
k 232 k 233 k 234 ⎥ 3 5 .
k 31 ⎢
( 6
2 b
)
⎢2 ⎥
41 k 242 ⎣k k 243 k 244 ⎥⎦ 4 6

The local and the global degrees of freedom are also indicated on the top and side
of the element stiffness matrix. This will help us to place the elements of the
element stiffness matrix at the appropriate locations of the global stiffness matrix.
The continuous beam has six degrees of freedom and hence the stiffnessmatrix
isoftheorder6×6.Let [K]denotesthecontinuousbeamstiffnessmatrixof
order 6 × 6 . From Fig. 27.6b, [K ]may be written as,
 Member AB (1)
⎡ ⎤
k1 ⎢ ⎥
11
1 ⎢ ⎥
⎢ k ⎥ (27.7)
[]= 12

K ⎢ k1 ⎥
⎢ 13 ⎥
⎢ k1 ⎥
⎢ ⎥
⎣⎢ ⎥⎦
Member BC (2)
The receives (AB) and
4 × 4 upper left contribution from
hand section member 1
4 × 4 lower right hand section of global stiffness matrix receives contribution
from member 2. The element of the global stiffness matrix corresponding to global
degrees offreedom3and4[overlappingportionofequation (27.7)] receives
element from both members 1 and 2.

Formation of loadvector.
Consider a continuous beam ABC as shown in Fig. 27.7.

We have two types of load: member loads and joint loads. Joint loads could be
handled very easily as done in case of trusses. Note that stiffness matrix of each
member was developed for end loading only. Thus it is required to replace the
member loads by equivalent joint loads. The equivalent joint loads must be
evaluated such that the displacements produced by them in the beam should be the
same as the displacements produced by the actual loading on the beam. This is
evaluated by invoking the method of superposition.
The loading on the beam shown in Fig. 27.8(a), is equal to the sum of Fig. 27.8(b)
and Fig. 27.8(c). In Fig. 27.8(c), the joints are restrained against displacements and
fixed end forces are calculated. In Fig. 27.8(c) these fixed end actions are shown in
reverse direction on the actual beam without any load. Since the beam in Fig.
27.8(b) is restrained (fixed) against any displacement, the displacements produced
by the joint loads in Fig. 27.8(c) must be equal to the displacement produced by
the actual beam in Fig. 27.8(a). Thus the loadsshown
in Fig. 27.8(c) are the p1 , p2 , p3 , p4 , p5 and p6 be the
equivalent joint loads .Let,
equivalent joint loads acting on the continuous beam along displacement degrees
of freedom 1,2,3,4,5 and 6 respectively as shown in Fig. 27.8(b). Thus the global
load vectoris,

⎧ Pb ⎫
 ⎪ ⎪
L⎪ ⎪
⎧p1⎫ ⎪
− Pab
2
⎪
⎪ ⎪ ⎪ L2 ⎪
⎪p2⎪ ⎪ ⎪
⎪ ⎪
⎪ ⎪ −⎛Pa +wL ⎞
⎪p⎪ ⎪ ⎜ ⎟⎪
⎪ ⎪=⎪
3
⎝L 2 ⎠⎪
(27.8)
⎨ ⎬ ⎨ ⎛wL2 ⎬
p Pba 2⎞
⎪ ⎝ ⎪ ⎪ ⎪
⎪4⎪ L ⎠−⎜ − ⎟
⎪ ⎪
⎪p5⎪ ⎜12 2 ⎟
⎪ ⎪ ⎪ ⎪
⎛wL ⎞
⎪ ⎪
⎪
−⎜ +2P⎟
⎪⎩p 6⎪⎭ ⎪ ⎠⎪
⎝2
⎪
⎪ wL2 ⎪

⎩⎪ 12 ⎪⎭

Solution of equilibriumequations
After establishing the global stiffness matrix and load vector of the beam, the load
displacement relationship for the beam can be writtenas,

{P}=[K ]{u}
(27.9)

where global load is is the

vector,
{u} displacement
[ K]
{P}is the vector and
global stiffness matrix. This equation is solved exactly in the similar manner as
discussed in the lesson 24. In the above equation some joint displacements are
known from support conditions. The above equation may be written as

⎧⎪{pk}⎫⎪ ⎡[k11] [k12]⎤⎧⎪{uu}⎫⎪

⎨ ⎬=⎢ ⎥⎨ ⎬ (27.10)
⎪⎩{pu}⎪⎭ ⎢⎣[k21][k22]⎥⎦⎪⎩{uk}⎪⎭

where{pk }and{uk}denoterespectivelyvector of known forces andknown

displacements.And{pu }, {uu }denoterespectivelyvector of unknownforces and

unknown displacements respectively. Now expanding equation (27.10),
 {pk}=[k11]{uu}+ [k22]{uk}
[k12]{uk} (27.11a)
(27.11b)
{pu}=[k21]{uu}+

Since is known, from equation

{uk} 27.11(a), the unknown
joint displacements
can be evaluated. And support reactions are evaluated from equation (27.11b),
after evaluating unknown displacement vector.
Let R1 ,R3 and the constrained degrees of
freedom as
R5 be the reactions along
shown in Fig. 27.9a. Since equivalent joint loads are directly applied at the
supports, they also need to be considered while calculating the actual reactions.
Thus,

⎧R1⎫ ⎧p1⎫
⎪ ]{u }
(27.12)
⎪

⎪
R
=
−
p
+
[K
⎨3⎬ ⎨3⎬ 21 u
⎪ ⎪ ⎪ ⎪
⎪⎩R5⎪⎭ ⎪⎩p5⎪⎭

The reactions may be calculated as follows. The reactions of the beam shown in
Fig. 27.9a are equal to the sum of reactions shown in Fig. 27.9b, Fig. 27.9c and
Fig. 27.9d.
From the method of superposition,
Pb
R= +K u +K16 u6 (27.13a)
1 L
144

R
u +K u (27.1
Pa 34 4 36 6
= 3b)
+K
3
L
 R
u +K56 u6 (27.13c)
wL 54 4
=
+2P
+K
5
2
or
 ⎧R ⎫
⎧ ⎫
1 Pb ⎡K K⎤
⎪
⎪ ⎪ L⎪ ⎢14 ⎪
16
⎧u ⎫
⎪R Pa +⎢K K ⎥⎪4 ⎪
⎪=⎪ (27.1
⎥ 4a)
⎨3⎬ ⎨ L⎬ ⎢34 36 ⎥⎨ ⎬
⎪ ⎪ ⎪wl ⎪ ⎢ ⎪ u⎪
⎥ ⎩ 6 ⎭
⎪ +2P⎪ K K
⎪⎩R5⎪⎭ ⎣ 54 56 ⎦
⎩2 ⎭

Equation (27.14a) may be written as,

⎧ ⎫ K
⎧R1⎫ ⎪Pb L⎪ K16 ⎤
⎡ 14 ⎧u⎫
⎪ Pa L ⎪ ⎢ )
+ K K ⎥⎨4 ⎬
R (2
⎪
7
= .
1
− 4
⎪
b
⎨3⎬ ⎨
⎪ ⎬ ⎢34 36 ⎥u
⎪R⎪ ⎪wl ⎢K K ⎥⎩6 ⎭
⎩5 ⎭ ⎪ +2P⎪⎭ ⎣54 56 ⎦
⎩2

Member end actions the first element 1. q1 , q2 , q3 , q4 are calculated as follows.

For example consider
⎧ Pb ⎫
⎧q1⎫ ⎪ ⎧0 ⎫
L ⎪
⎪ ⎪
⎪ ⎪ Pab 2
⎪ ⎪
⎪q2⎪ ⎪ ⎪u2 ⎪
⎪ +[ K ]
⎪
⎬ (27.1
6)
⎪
⎪
2

= L
⎨ ⎬ ⎨ ⎬ element1 ⎨
⎪q
⎪ 3⎪ Pa ⎪0 ⎪
⎪ ⎪ ⎪ L ⎪ ⎪ ⎪
⎪ ⎪ Pa b⎪ 2 ⎪
⎪⎩q4 ⎭  ⎪ ⎪ ⎪⎩u4 ⎭
⎩

L2⎭

In the next
lesson few
problems are
solved to
illustrate the
method so far
discussed.
 UNIT -4

Introduction

Inengineeringproblemstherearesomebasicunknowns.Iftheyarefound,thebehaviouro
ftheentirestructure can be predicted. The basic unknowns or the Field variables
which are encountered in the engineering
problemsaredisplacementsinsolidmechanics,velocitiesinfluidmechanics,electricand
magneticpotentials in electrical engineering and temperatures in heat
flowproblems.
In a continuum, these unknowns are infinite. The finite element procedure
reduces such unknowns to a
finitenumberbydividingthesolutionregionintosmallpartscalledelementsandbyexpre
ssingtheunknown
fieldvariablesintermsofassumedapproximatingfunctions(Interpolatingfunctions/
Shapefunctions)within each element. The approximating functions are defined in
terms of field variables of specified points called nodes or nodal points. Thus in
the finite element analysis the unknowns are the field variables of the nodal points.
Once these are found the field variables at any point can be found by using
interpolationfunctions.
Afterselectingelementsandnodalunknownsnextstepinfiniteelementanalysisistoa
ssembleelement properties for each element. For example, in solid mechanics, we
have to find the force-displacement i.e. stiffness characteristics of each individual
element. Mathematically this relationship is of the form
 [k]e {δ}e = {F}e

where [k] is element stiffness matrix, {δ}e is nodal displacement vector of the
e

element and {F} is nodal force vector. The element of stiffness matrix k represent
e ij

the force in coordinate direction ‘i’ due to a unit

displacementincoordinatedirection‘j’.Fourmethodsareavailableforformulatingthese
elementproperties
viz. direct approach, variational approach, weighted residual approach and energy
balance approach. Any
oneofthesemethodscanbeusedforassemblingelementproperties.Insolidmechanicsvar
iationalapproach is commonly employed to assemble stiffness matrix and nodal
force vector (consistantloads).
Element properties are used to assemble global properties/structure properties to get system
equations
[k] { }  {F}. Then the boundary conditions are imposed. The solution of
these simultaneous equations give the nodal unknowns. Using these nodal
values additional calculations are made to get the required values e.g. stresses,
strains, moments, etc. in solid mechanics problems.
Thus the various steps involved in the finite element analysis are:
(i) Select suitable field variables and the elements.
(ii) Discritise the continua.
(iii) Select interpolation functions.
(iv) Find the element properties.
(v) Assemble element properties to get global properties.
(vi) Impose the boundary conditions.
(vii) Solve the system equations to get the nodal unknowns.
(viii) Make the additional calculations to get the required values.
(ix) A BRIEF EVPLANATION OF FEA FOR A STRESS ANALYSIS PROBLEM

Thestepsinvolvedinfiniteelementanalysisareclarifiedbytakingthestressanalysisofate
nsionstripwith
fillets(referFig.1.1).Inthisproblemstressconcentrationistobestudiesinthefilletzone.Si
ncetheproblem
ishavingsymmetryaboutbothxandyaxes,onlyonequarterofthetensionstripmaybecons
ideredasshown
inFig.1.2.Aboutthesymmetricaxes,transversedisplacementsofallnodesaretobemade
zero.Thevarious steps involved in the finite element analysis of this problem are
discussed below:
Step 1: Four noded isoparametric element (refer Fig 1.3) is selected for the
analysis (However note that 8 noded isoparametric element is ideal for this
analysis). The four noded isoparametric element can take
quadrilateralshapealsoasrequiredforelements12,15,18,etc.Asthereisnobendingofstri
p,onlydisplacement
continuityistobeensuredbutnottheslopecontinuity.Hencedisplacementsofnodesinxa
ndydirectionsare taken as basic unknowns in theproblem.
Fillet t
P
b2 C
b1
A

B D

Fig. 1.1 †ypical tension flat

 Introduction 3

A1 5 9 13 17 21 242933 37 41 45
C
2 1 4 7 1 1 1 1 2 25 28 3
2 5 8 11 14 4 3 6 9 12 15 1821
172023
3 32 P
26 29
B 12 16 24 27 30
8 33 D
20 24
32
2836 40 44 48

Fig. 1.2 Discretisation of quarter of tension flat

6 10 4 3

xP
5

7
(a) El
em 11 1 2
ent (b) Typicalelement
no.
5

Fig. 1.3

Step2:Theportiontobeanalysedistobediscretised.Fig.1.2showsdiscretisedportion.Fo
rthis33elements
havebeenused.Thereare48nodes.Ateachnodeunknownsarexandycomponentsofdispl
acements.Hence in this problem total unknowns (displacements) to be determined
are 48 × 2 =96.
Step 3: The displacement of any point inside the element is approximated by
suitable functions in terms of the nodal displacements of the element. For the
typical element (Fig. 1.3 b), displacements at P are

u= ∑N u =N1u1 +N2u2 +N3u3 +N4u4

i i

and v =∑N v =N v +N v +N v +N v
i i 1 1 2 2 3 3 4 4 …(1.2)
The approximating functions N are called shape functions or interpolation
i

functions. Usually they are

derivedusingpolynomials.Themethodsofderivingthesefunctionsforvariouselements
arediscussedinthis text in latterchapters.
Step 4: Now the stiffness characters and consistant loads are to be found for each
element. There are four nodes and at each node degree of freedom is 2. Hence
degree of freedom in each element is 4 × 2 = 8. The
relationshipbetweenthenodaldisplacementsandnodalforcesiscalledelementstiffness
characteristics.Itis of theform
[k]e {δ}e = {F}e , as explained earlier.

For the element under consideration, k is 8 × 8 matrix and δe and F are vectors
e e

of 8 values. In solid
mechanicselementstiffnessmatrixisassembledusingvariationalapproachi.e.byminim
izingpotentialenergy.
Iftheloadisactinginthebodyofelementoronthesurfaceofelement,itsequivalentatnodal
pointsaretobe found using variational approach, so that right hand side of the
above expression is assembled. This process is called finding consistantloads.
4 Finite ElementAnalysis

Step 5: The structure is having 48 × 2 = 96 displacement and load vector

components to be determined. Hence global stiffness equation is of the form
[k] {δ} = {F}
96 × 96 96 × 1 96 × 1
Eachelementstiffnessmatrixistobeplacedintheglobalstiffnessmatrixappropriatel
y.Thisprocessis called assembling global stiffness matrix. In this problem force
vector F is zero at all nodes except at nodes
45,46,47and48inxdirection.Forthegivenloadingnodalequivalentforcesarefoundandt
heforcevector F isassembled.
Step 6: In this problem, due to symmetry transverse displacements along AB and
BC are zero. The system equation [k] {δ} = {F} is modified to see that the solution
for {δ} comes out with the above values. This modification of system equation is
called imposing the boundary conditions.
Step7:Theabove96simultaneousequationsaresolvedusingthestandardnumericalproc
edureslikeGauss- elimination or Choleski’s decomposition techniques to get the 96
nodaldisplacements.
Step 8: Now the interest of the analyst is to study the stresses at various points. In
solid mechanics the relationship between the displacements and stresses are well
established. The stresses at various points of
interestmaybefoundbyusingshapefunctionsandthenodaldisplacementsandthenstress
escalculated.The
stressconcentrationsmaybestudiesbycomparingthevaluesobtainedatvariouspointsint
hefilletzonewith the values at uniform zone, far away from the fillet (which is
equal toP/b t).
2

FINITE ELEMENT METHOD VS CLASSICAL METHODS

1. In classical methods exact equations are formed and exact solutions are
obtained where as in finite element analysis exact equations are formed
but approximate solutions are obtained.
2. Solutions have been obtained for few standard cases by classical methods,
where as solutions can be obtained for all problems by finite element
analysis.
3. Whenever the following complexities are faced, classical method makes
the drastic assumptions’ and looks for the solutions:
(a) Shape
(b) Boundary conditions
 (c) Loading
Fig. 1.4 shows such cases in the analysis of slabs (plates).
To get the solution in the above cases, rectangular shapes, same boundary
condition along a side
andregularequivalentloadsaretobeassumed.InFEMnosuchassumptionsarem
ade.Theproblem is treated as itis.
4. When material property is not isotropic, solutions for the problems
become very difficult in classical method. Only few simple cases have
been tried successfully by researchers. FEM can handle structures with
anisotropic properties also without any difficulty.
5. If structure consists of more than one material, it is difficult to use
classical method, but finite element can be used without any difficulty.
6. Problems with material and geometric non-linearities can not be handled
by classical methods. There is no difficulty in FEM.
 HenceFEMissuperiortotheclassicalmethodsonlyfortheproblemsinvolvinganumb
erofcomplexities
whichcannotbehandledbyclassicalmethodswithoutmakingdrasticassumptions.Foral
lregularproblems, the solutions by classical methods are the best solutions. Infact,
to check the validity of the FEM programs developed, the FEM solutions are
compared with the solutions by classical methods for standard problems.

(a) Irregularshaper (b) Irregular boundarycondition

(c) Irregular loading

Fig. 1.4

FEM VS FINITE DIFFERENCE METHOD (FDM)

1. FDM makes pointwise approximation to the governing equations i.e. it
ensures continuity only at the node points. Continuity along the sides of
grid lines are not ensured.
FEMmakepiecewiseapproximationi.e.itensuresthecontinuityatnodepoint
saswellasalong the sides of theelement.
2. FDM do not give the values at any point except at node points. It do not
give any approximating function to evaluate the basic values (deflections,
in case of solid mechanics) using the nodal values.
FEM can give the values at any point. However the values obtained at
points other than nodes are by using suitable interpolation formulae.
3. FDM makes stair type approximation to sloping and curved boundaries as shown in Fig.
1.5.
FEMcanconsidertheslopingboundariesexactly.Ifcurvedelementsareused,
eventhecurved boundaries can be handledexactly.
4. FDM needs larger number of nodes to get good results while FEM needs fewer nodes.
5. With FDM fairly complicated problems can be handled where as FEM can
handle all complicated problems.
6 Finite ElementAnalysis

Fig. 1.5 VDW approximation of shape

A BRIEF HISTORY OFFEM

Engineers, physicists and mathematicians have developed finite element method independently. In
1943 Courant
[1] made an effort to use piecewise continuous functions defined over triangular domain.
Afterthatittooknearlyadecadetousethisdistributionidea.Infiftiesrenewedinteresti
nthisfieldwas
shownbyPolya[2],Hersh[3]andWeinberger[4].ArgyrisandKelsey[5]introducedthec
onceptofapplying energy principles to the formation of structural analysis
problems in 1960. In the same year Clough [6] introduced the word ‘Finite
ElementMethod’.
Insixtiesconvergenceaspectofthefiniteelementmethodwaspursuedmorerigorousl
y.Onesuchstudy
byMelesh[7]ledtotheformulationofthefiniteelementmethodbasedontheprinciplesof
minimumpotential
energy.SoonafterthatdeVeubeke[8]introducedequilibriumelementsbasedontheprinc
iplesofminimum potential energy, Pion [9] introduced the concept of hybrid
element using the duel principle of minimum potential energy and minimum
complementaryenergy.
In Late 1960’s and 1970’s, considerable progress was made in the field of
finite element analysis. The
improvementsinthespeedandmemorycapacityofcomputerslargelycontributedtothep
rogressandsuccess
ofthismethod.Inthefieldofsolidmechanicsfromtheinitialattentionfocusedontheelasti
canalysisofplane stress and plane strain problems, the method has been
successfully extended to the cases of the analysis of three dimensional problems,
stability and vibration problems, non-linear analysis. A number of books [10 –
20] have appeared and made this fieldinteresting.

NEED FOR STUDYINGFEM

Now,anumberofusersfriendlypackagesareavailableinthemarket.Henceonemayaskth
equestion‘What is the need to study FEA?’.
Theaboveargumentisnotsound.Thefiniteelementknowledgemakesagoodenginee
rbetterwhilejust
userwithouttheknowledgeofFEAmayproducemoredangerousresults.TousetheFEAp
ackagesproperly, the user must know the following points clearly:
1. Which elements are to be used for solving the problem in hand.
2. How to discritise to get good results.
3. How to introduce boundary conditions properly.
 4. How the element properties are developed and what are their limitations.
5. How the displays are developed in pre and post processor to understand their limitations.
6. To understand the difficulties involved in the development of FEA
programs and hence the need for checking the commercially available
packages with the results of standard cases.
Unless user has the background of FEA, he may produce worst results and may go with overconfidence.
Hence it is necessary that the users of FEA package should have sound knowledge of FEA.

WARNING TO FEA PACKAGEUSERS

Whenhandcalculationsaremade,thedesigneralwaysgetsthefeelofthestructureandgetr
oughideaabout
theexpectedresults.Thisaspectcannotbeignoredbyanydesigner,whateverbethereliabi
lityoftheprogram, a complex problem may be simplified with drastic assumptions
and FEA results obtained. Check whether
expectedtrendoftheresultisobtained.Thenavoiddrasticassumptionsandgetmorerefine
dresultswithFEA
package.Usermustrememberthatstructuralbehaviourisnotdictatedbythecomputerpro
grams.Hencethe
designershoulddevelopfeelofthestructureandmakeuseoftheprogramstogetnumerical
resultswhichare close to structural behaviour.

QUESTIONS

1. Explain the concept of FEM briefly and outline the procedure.

2. Discuss the advantages and disadvantages of FEM over
(i) Classical method
(ii) Finite difference method.
3. Clearly point out the situations in which FEM is preferred over other methods.
4. When there are several FEM packages are available is there need to study this method?
Discuss.

References
1. R.Courant,“VariationalMethodsfortheSolutionsofProblemsofEquilibriumandVibrations”,
Bulletin of American Mathematical Society, Vol. 49, 1943.
2. G. Polya, Estimates for Eigen Values, Studies presented to Richard Von Mises, Academic Press,
New York,1954.
3. J.Hersch,“EquationsDifferentiellesetFunctionsdecellules”,C.R.Acad.Science,Vol.240,1955.
4. H.F.Weinberger,“UpperandLowerBoundsforEigenValuesbyFiniteDifferenceMethod”,Pure
Applied Mathematics, Vol. 9, 1956.
5. J.H. Argyris and S. Kelsey, “Energy Theorems and Structural Analysis”, Aircraft Engineering,
 Vol. 27,1955.
6. R.W. Clough, “The Finite Element Method in Plane Stress Analysis”, Proceeding of 2nd ASCE
Conference on Electronic Computation, Pittsburg, PA, September, 1960.
7. R.J. Melosh, “Basis for the Derivation for the Direct Stiffness Method”, AIAA Journal, Vol. 1,
1963.
8. B.FraeijsdeVeubeke,“UpperandLowerBoundsinMatrixStructuralAnalysis”,AGARDograph
72, B.F. de Veubeke (ed). Pergaman Press, New York, 1964.
9. T.H.H.Pian,“DerivationofElementStiffnessMatrices”,AIAAJournal,Vol.2,1964.pp.556–57.
10. O.C.Zienkiewicz,TheFiniteElementMethodinEngineeringScience,McGraw-Hill,London1971.
11. K.H. Huebner, The Finite Element Methods for Engineers, John Wiley and Sons, 1971.
12. DesaiandAbel,IntroductiontotheFiniteElementMethod,CBSPublishers&Distributors,1972.
13. H.C. Martin and G. F. Carey, Introduction to Finite Element Analysis- Theory and Applications,
Tata McGraw-Hill Publishing Company Ltd., New Delhi, 1975.
14. K.L. Bathe and E.L. Wilson, Finite Element Methods, Prentice Hall,1976.
15. Y.K.CheunyandM.F.Yeo,APracticalIntroductiontoFiniteElementAnalysis,PitmanPublishers,
1979.
16. R.D. Cook, D.S. Makus and M.F. Plesha, Concept and Applications of Finite Element Analysis,
John Wiley and Sons,1981.
17. J.N Reddy, An introduction to the Finite Element Method, McGraw-Hill International Edition,
1984.
18. C.S. Krishnamoorthy, Finite Element Analysis, Theory and Programming, Tata McGraw-Hill
Publishing Company Ltd., New Delhi,1987.
19. T.R. Chandrapatla and A.D. Belegundu, Introduction to Finite Elements in Engineering, Prentice
Hall,1991.
20. S. Rajasekharan, Finite Element Analysis in Engineering Design, Wheeler Publisher,1993.
 INTRODUCTION

This chapter summarizes the results from theory of elasticity which are useful in
solving the problems in structural and continuum mechanics by the finite element
method.

STRESSES IN A TYPICALELEMENT

In theory of elasticity, usually right hand rule is used for selecting the coordinate
system. Fig. 2.1 shows
variousorientationsofrighthandruleofthecoordinatesystems.Equationsderivedforany
onesuchorientation hold good for all other orientationsof
z

z y

y x x
(b)
y
x
z
(a) (
c
)

Fig. 2.1

coordinate system with right hand rule. In this Chapter orientation shown in Fig.
2.1(a) is used for the
explanation.Fig.2.2showsatypicalthreedimensionalelementofsizedx×dy×dz.Faceab
cdmaybecalled asnegativefaceofxandthefaceefghasthepositivefaceof
xsincethexvalueforfaceabcdislessthanthat for the face efgh. Similarly the face aehd
is negative face of y and bfgc is positive face of y. Negative and positive faces of z
are dhgc andaefb.
The direct stresses σand shearing stresses τacting on the negative faces are
shown in the Fig. 2.3with
suitablesubscript.Itmaybenotedthatthefirstsubscriptofshearingstressistheplaneandth
esecondsubscript is the direction. Thus the τxy means shearing stress on the plane
where x value is constant and y is the direction.
 z z
dy
a e
x
dx

dz xy
yx

b f

y xz
d h y y
yz zx

zy

z
c
g
x x

Fig.2.2 Fig.2.3

In a stressed body, the values of stresses change from face to face of an

element. Hence on positive face the various stresses acting are shown in Fig. 2.4
with superscript ‘+’.
All these forces are listed in table 2.1
Note the sign convention: A stress is positive when it is on positive face in positive
direction or on negative face in negative direction. In other words the stress is + ve
when it is as shown in Figs 2.3 and 2.4.
z

zy

zx yz

y
xz y

zx

xy

x
Fig. 2.4
Table 2.1 Stresses on a typical element

Fac Stress on –ve Face Stresses on +ve Face

e
∂σx
x σx σ+=σ+ dx
x x ∂x
τxy ∂τxy
τ+=τ + dx
xy xy ∂x
τxz
∂τxz
τ+=τ+ dx
xz xz ∂x
+ ∂σy
σy σ = σ+ dy
y
y y ∂y
τ ∂τyx
yx τ+=τ + dy
yx yx ∂y
τyz ∂τyz
τ+=τ+ dy
yz yz ∂y
∂σz
σz σ+=σ+ dz
z
z z ∂z

τzx ∂τzx
τ+=τ+ dz
zx zx ∂z
τzy
∂τzy
τ+=τ+ dz
zy zy ∂z

Y
x y

dz Fig. 2.5

X
 Let the intensity of body forces acting on the element in x, y, z directions be X,
Y and Z respectively as
showninFig2.5.Theintensityofbodyforcesareuniformoverentirebody.Hencethetotal
bodyforceinx, y, z direction on the element shown are givenby
(i) X dx dy dz in x – direction
(ii) Y dx dy dz in y – direction and
(iii) Z dx dy dz in z – direction

Matrix Displacement Formulation

INTRODUCTION

Though mathematicians, physicists and stress analysts worked independently in

the field of FEM, it is the
matrixdisplacementformulationofthestressanalystswhichleadtofastdevelopmentofF
EM.Infacttillthe
wordFEMbecamepopular,stressanalystworkedinthisfieldinthenameofmatrixdisplac
ementmethod.In matrix displacement method stiffness matrix of an element is
assembled by direct approach while in FEM though direct stiffness matrix may be
treated as an approach for assembling element properties (stiffness matrix as far as
stress analysis is concerned), it is the energy approached which has revolutionized
entire FEM.
Hence in this chapter, a brief explanation of matrix displacement method is
presented and solution techniques for simultaneous equations are discussed briefly.

MATRIV DISPLACEMENT EQUATIONS

The standard form of matrix displacement equation is,
[k] {}  {F}
where [k] is stiffnessmatrix
{} is displacement vector and
{F} is force vector in the coordinate directions
The element kij of stiffness matrix maybe defined as the force at coordinate i
due to unit displacement in coordinate direction j.
The direct method of assembling stiffness matrix for few standard cases is briefly given in this
article.

1. Bar Element
Common problems in this category are the bars and columns with varying cross
section subjected to axial forces as shown in Fig. 3.1.
 For such bar with cross section A, Young’s Modulus E and length L (Fig. 3.2 (a))
extension/shortening 
is given by
  PL
EA
 P1

L1

P2

A3
P3 A2 P2 A1 x
L2
P 1

L3 L2 L1
P3

Fig. 3.2

 P  EA 
L
 If   1, P  EA
L
Bygivingunitdisplacementincoordinatedirection1,theforcesdevelopmentintheco
ordinatedirection 1 and 2 can be found (Fig. 3.2 (b)). Hence from the definition of
stiffnessmatrix,
 EA L   EA
k11 and k21 L
Similarly giving unit displacement in coordinate direction 2 (refer Fig. 3.2 (c)), we get
 k12   EA 
L and k22 E
A
L
EA µ 1 1y
Thus, L ¡µ1 1 ¡j …
k (3.5
)

2. Truss Element
Membersofthetrussesaresubjectedtoaxialforcesonly,buttheirorientationintheplanem
aybeatanyangle to the coordinate directions selected. Figure 3.3 shows a typical
case in a plane truss. Figure 3.4 (a) shows a typical member of the truss with
Young’s Modulus E, cross sectional area A, length L and at angle  to x-axis

k21  P sin  EA cos sin

k L
  P cos   EA cos2 
31
L
  P sin   EA cos sin
k41 L
 4
P P

3
2
1

1 (a) P P
P 1 1
P
(b) (
c
)

  P sin   EA sin2 
L

k42
(iii) Unit displacement in coordinate direction 3,
Extension along the axis is 1  sin and hence the forces developed
are as shown in the Fig. 3.4(d)
 P  EA cos
L
k13
  P cos   EA cos2 
L
 k23   P sin   EA cos sin
k L
 P cos  EA cos2 
33
L
 P sin  EA cos sin
k43 L
(vi) Due to unit displacement in coordinate direction 4,
Extension of the bar is equal to 1 Fig. 3.4 (e).
 sin, and hence the forces developed are as shown in
 P  EA sin 
L
  P cos   EA sin cos
L
  P sin   EA sin2 
L
 P cos  EA sin cos
L
 P sin  EA sin2 
L
¡µ cos2  cos sin y¡
cos sin cos2 
EA ¡ cos sin sin2  cos sin –sin2  ¡
k
L ¡ –cos2  cos sin cos2  cos sin ¡
¡µcos sin sin2 ¡j
–sin2  cos sin

µ¡ l2
lm l2 lm y¡

 EA ¡ lmm2 lm m2 ¡

L ¡l2 lm l2 lm ¡ …(3.6)
¡µlm m2 lm m2 ¡j

Where l and m are the direction cosines of the member i.e. l = cos and m = cos (90 –  ) = sin  .
(v) Beam Element
In the analysis of continuous beams normally axial deformation is
negligible (small deflection theory) and hence only two unknowns may be
taken at each end of a element (Fig. 3.5). Typical element and the
coordinates of displacements selected are shown in Fig. 3.5 (b). The end
forces
 developedduetounitdisplacementinallthefourcoordinatedirectionsareshown
inFig.3.6(a,b, c,d).
y 1 3 5 7 9
E1, I 1 E2, I 2 E3, I 3 E4, I 4

x
2 L1 4 L2 6L 8 3 L
4 10

(a)
z

1 3
E,I, L

From the definition of stiffness matrix and looking at positive senses indicated, we can write
(a) Due to unit displacement in coordinate direction 1,

 12EI  6EI   12EI  6EI

k11 L3 k21 L2 k31 L3 k41 L2
(b) Due to unit displacement in coordinate direction 2,
 6EI  4EI   6EI  2EI
k12 L2 k22 L k32 L2 k42 L
(c) Due to unit displacement in coordinate direction 3,
  12EI   6EI  12EI   6EI
k13 L3 k23 L2 k33 L3 k43 L2
 (d) Due to unit displacement in coordinate direction 4,
 6EI  2EI   6EI  4EI
k14 L2 k24 L k34 L2 k44 L
µ¡ 12 6L 12 6L y¡
 k  FI ¡ 6L 6L 2L2 ¡
4L2
µ 6L 6 4L2 j …
6L L3
12
2L2 6L
If axial deformations in the beam elements are to be considered as in
case of columns of frames, etc. (Fig. 3.7), it may be observed that axial
force do not affect values of bending moment and shear force and vice
versa is also true. Hence stiffness matrix for the element shown in Fig.
3.8 is obtained by combining the stiffness matrices of bar element and
beam element and arranging in proper locations. For this case
µ¡ EA y
0 0  EA
0 0 ¡
¡ L L
0 12EI L3 6EI L2 0  12EI L3 6
6EI 4EI 0  6EI L2 E
¡ 0 L2 L I
k¡
¡
L
2
¡ 2EI
L¡
¡ EA 0 0 ¡
E¡
A
0
0
L L 6EI ¡ …
06EI 0 12EI L3
L2
12EI ¡  L2 ¡ (
L
 ¡ 3
¡ .
8
)
¡ 6EI 2EI 0  6EI 4EI ¡
µ L2 ¡ 0
L
L2
L j
(a) (b)

Fig. 3.7

2 5
1 4
3 6

Fig. 3.8
 The following special features of matrix displacement equations are worth noting:
(i) The matrix is having diagonal dominance and is positive definite. Hence
in the solution process there is no need to rearrange the equations to get
diagonal dominance.
(ii) The matrix is symmetric. It is obvious from Maxwell’s reciprocal theorem.
Hence only upper or lower triangular elements may be formed and others
obtained using symmetry.
(iii) The matrix is having banded nature i.e. the nonzero elements of stiffness
matrix are concentrated near the diagonal of the matrix. The elements
away from the diagonal are zero. Considerable saving is effected in
storage requirement of stiffness matrix in the memory of computers by
avoiding storage of zero values of stiffness matrices. The banded nature of
matrix is shown in Fig. 3.9.

I
I
I

Fig. 3.9 I
I I
In this case instead of storing N ×N size matrix only N × B size I V matrix can be stored.

V
I

V
no force in the z-direction and no variation of any forces in z-direction. I V Hence
 z   xz   yz  0 I II
I
The conditions  xz   yz  0 give  xz   yz  0 and the condition  z  0
gives,
IX
 z   x   y  1    z  0
 Basic Equations in Elasticity 17
i.e.  
 
z 1x y

If this is substituted in equation 2.13 the constitutive law reduces to

j¡ x y¡ µ¡1  0 y¡ j¡ x y¡

E
¡y¡ 2 ¡ 1 0¡¡y¡ .1
…
( 4
2 )
¡ ¡ 1   ¡0 0 1   ¡ ¡ ¡
t xy 1 µ¡ 2 ¡j t xy 1

(c) (d)

no force in the z-direction and no variation of any forces in z-direction. Hence

 z   xz   yz  0
The conditions  xz   yz  0 give  xz   yz  0 and the condition  z  0 gives,
 z   x   y  1    z  0
i.e.  
 
z 1x y

If this is substituted in equation 2.13 the constitutive law reduces to

j¡ x y¡ µ¡1  0 y¡ j¡ x y¡

E
¡y¡ 2 ¡ 1 0¡¡y¡ .1
…
( 4
2 )
¡ ¡ 1   ¡0 0 1   ¡ ¡ ¡
t xy 1 µ¡ 2 ¡j t xy 1

x
x

z z
(c) (d)
Fig. 2.7
z
(e)

Axi-
symmetricstructuresarethosewhichcanbegeneratedbyrotatingalineorcurveaboutana
xis.Cylinders (refer Fig. 2.8) are the common examples of axisymmetric
structures. If such structures are subjected to axisymmetric loadings like uniform
internal or external pressures, uniform self weight or live load uniform over
thesurface,
thereexistsymmetryaboutanyaxis.Theadvantageofsymmetrymaybemadeusetosimpl
ifytheanalysis. In these problems cylindrical coordinates canbe used
advantageously. Because of symmetry, the stress components are independent of
the angular (  ) coordinate. Hence all derivatives with respect to  vanish i.e. in
these cases.
v   r    z   r    z  0

Axi-Symmetric Problems
 z, w

z, w

In these cases j r y
stress-strain 0 y j r y
relation is
j1    
¡ z E 0 ¡ ¡ z ¡
¡ ¡
1 
 2) 0 ¡  1
7
(1  
¡ )
…
(
)(1 1 2
.

¡ ¡ ¡ 1 2 ¡ ¡  ¡
t rz 1 y 2 j t rz 1

Fig. 2.7
 z
(e)

Axi-
symmetricstructuresarethosewhichcanbegeneratedbyrotatingalineorcurveaboutana
xis.Cylinders (refer Fig. 2.8) are the common examples of axisymmetric
structures. If such structures are subjected to axisymmetric loadings like uniform
internal or external pressures, uniform self weight or live load uniform over
thesurface,
thereexistsymmetryaboutanyaxis.Theadvantageofsymmetrymaybemadeusetosimpl
ifytheanalysis. In these problems cylindrical coordinates canbe used
advantageously. Because of symmetry, the stress components are independent of
the angular (  ) coordinate. Hence all derivatives with respect to  vanish i.e. in
these cases.
v   r    z   r    z  0

Axi-Symmetric Problems
 z, w

z, w

In these cases
J1−μ μ μ
stress-strain
relation is y
0
Jε y
r

Jσ y
r

¡Eσ ¡
z ¡ 0
¡ ¡ε ¡
z
1−μ μ
2 1
σ 7

0 ¡ε )
1
¡
+μ
…
(1 (
2
1 −μ .
¡ ¡ ¡
¡¡¡
1−2μ θ
tτ 1
rz y 2 Jtγrz1