Sei sulla pagina 1di 104

COURSE OUTLINE

FINITE ELEMENT METHODS


(16 ME 64)
Compiled by:
Dr P R Venkatesh
Associate Professor
ME Dept, RVCE, Bengaluru-59
Dr P R Venkatesh, ME Dept RVCE
PART A (Theory)
UNIT-1 (08 Hrs)
(One compulsory Question in SEE)
Introduction to Finite Element Methods:
• General description of FEM
• Steps involved in FEM
• Engineering applications of FEM
• Advantages & Disadvantages of FEM
• Basic Equations of Elasticity: Stress-strain relationship,
differential equations of equilibrium, plane stress and
strain conditions, Strain displacement relations
• Rayleigh Ritz Method (Variational method), Galerkin’s
Method,(WeightedDr Presidual method)
R Venkatesh, ME Dept RVCE
UNIT-2 (08 Hrs)
(Two Questions with internal choice in SEE)
One dimensional finite elements, Bar & Truss
elements;
• Linear elements, Principle of minimum potential
energy, admissible displacement function, stiffness
matrix, strain matrix, static analysis using elimination
method, penalty method, boundary conditions and
assembly of load vector,
• Convergence and Compatibility conditions, Shape
functions for 1D linear, quadratic and Truss elements

Dr P R Venkatesh, ME Dept RVCE


UNIT-3 (08 Hrs)
(Two Questions with internal choice in SEE)
Two Dimensional CST Elements :
• Iso, super and sub parametric representation, Shape functions.
• Element stiffness and load vectors
UNIT-4 (08 Hrs)
(One compulsory Question in SEE)
Analysis of Beam Elements:
• Hermitian shape functions, formulations of element stiffness
matrices, load vectors, Analysis bending moment and shear
force
Dynamic Analysis 1-D bar element:
• Equations of motion, mass and stiffness matrices, distributed
and consistent mass matrices, Eigen values and Eigen vectors.

Dr P R Venkatesh, ME Dept RVCE


UNIT-5 (08 Hrs)
(One compulsory Question in SEE)
Analysis of Heat Transfer 1-D element:
• Steady State Heat Transfer, Galerkin’s Formulation of
Element Equations for Heat Conduction, Heat flux
boundary condition , Analysis of composite slabs

• Analysis of thin Fin: Formulation of element equation


for Heat conduction, Heat flux boundary condition,
Analysis of heat transfer through fins

Dr P R Venkatesh, ME Dept RVCE


General description of FEM
• In the finite element method, the actual continuum or
body of matter like solid, liquid or gas is represented as
an assemblage of sub-divisions called finite elements.

• These elements are considered to be interconnected at


specified joints which are called ‘nodal points’ or nodes

• The nodes usually lie on the elements boundaries


where adjacent elements are considered to be
connected .

Dr P R Venkatesh, ME Dept RVCE


General description of FEM....
• Since the actual variation of the field variable
(displacements, stress, temperature, pressure, velocity)
inside the continuum is not known, we assume that, the
variation of the field variable inside a finite element can be
approximated by a simple function.
• These approximating functions (are called interpolation
models) are defined in terms of field variables at the nodes.
• When field equations for the whole continuum are written,
the new unknowns will be the nodal values of the field
variable.
Dr P R Venkatesh, ME Dept RVCE
General description of FEM....
• By solving the field equations, which are generally in the
form of matrix equation, the nodal values of the field
variables will be known.

• Once these are known, the approximating functions


define the field variables throughout the assemblage of
elements.

Dr P R Venkatesh, ME Dept RVCE


Steps involved in FEM
The solution of a general continuum problem by the FEM
always follows an orderly step by step process as follows:

STEP I: DISCRETIZATION OF THE STRUCTURE

• The continuum is separated by imaginary lines of


surfaces into a number of finite elements.

• The number, type, size and the arrangements of the


elements have to be decided based on the accuracy of
the solution required.
Dr P R Venkatesh, ME Dept RVCE
Steps involved in FEM...
STEP 2: SELECTION OF A PROPER INTERPOLATION OR
DISPLACEMENT MODEL:

• Since the displacement solution of a complex structure


under any specified load conditions cannot be predicted
exactly, we assume some suitable solution within
element to approximate the unknown solution.

• The assumed solution must be simple from


computational point of view.
Dr P R Venkatesh, ME Dept RVCE
Steps involved in FEM...
STEP 3: DERIVATION OF ELEMENT STIFFNESS MATRICES
AND LOAD VECTORS:

• From the assumed displacement model, the stiffness


matrix [ ke] and the load vector [ pe], of element ‘e’ are
to be derived by using either equilibrium conditions or a
suitable variational principle.

Dr P R Venkatesh, ME Dept RVCE


Steps involved in FEM...
STEP 4: ASSEMBLAGE OF ELEMENT EQUATIONS TO OBTAIN
THE OVERALL EQUILIBRIUM EQUATIONS:
Since the structure is composed of several finite elements , the
individual element stiffness matrices and load vectors are to be
assembled in suitable manner and overall equilibrium
equations have to be formulated as , [ K ]{q} = {p}
where ,
[k] = assembled stiffness matrix ,
{q} = Vector of nodal displacements ,
{p} = Vector of nodal forces for the complete structure.
Dr P R Venkatesh, ME Dept RVCE
Steps involved in FEM...
STEP 5: SOLUTION FOR THE UNKNOWN NODAL
DISPLACEMENTS

The overall equilibrium equations have to be modified to


account for the boundary conditions of the problem. After
the incorporation of the boundary condition, the
equilibrium equation can be expressed as :

[ K ] {Q} = {P}

Dr P R Venkatesh, ME Dept RVCE


Steps involved in FEM...
STEP 6: COMPUTATION OF ELEMENT STRAINS AND STRESSES
From the known nodal displacements {Q} , element strains and
stresses can be computed using equations of solid or structural
mechanics.
In an equilibrium problem, we need to find;
• The steady state displacement or stress distribution if it is a solid
mechanics problem
• Temperature or heat flux distribution if it is heat transfer
problem and,
• Pressure or velocity distribution if it is fluid mechanics problem.
Dr P R Venkatesh, ME Dept RVCE
Comparison between FEM & Continuum (classical) method
Finite Element Method Continuum or Classical method

Approximate but acceptable solutions are Exact solutions can be obtained for few
1
obtained to all kinds of problems standard problems only

FEM does not make assumptions but In case of complexities, classical methods
2 treats the problems as it is even in case of make drastic assumptions to get exact
complex problems solutions

FEM can handle problems with anisotropy Difficult to handle anisotropy, assumptions
3 (elastic properties vary in all directions) are made that the materials are
homogeneous & isotropic.

4 FEM can handle non linear problems easily Can not handle non linear problems

FEM can solve problems of composite Difficult to solve problems involving


5 structures (structures made of more than composite structures
one material)
Dr P R Venkatesh, ME Dept RVCE
ENGINEERING APPLICATIONS OF FEM
FEM has been successfully applied to solve several types of
problems in engineering, medical science & dental science.
Structural problems:
• Stress analysis of frames, trusses, stress concentration
problems, buckling, vibration analysis, etc.
Non structural problems: Heat transfer, fluid flow,
distribution of electrical & magnetic fields
Biomedical problems: Analysis of human skull, hip joints,
spinal cord, heart, teeth (dentures), etc.
Dr P R Venkatesh, ME Dept RVCE
ADVANTAGES OF FEM
• FEM can model irregularly shaped bodies.

• FEM Can handle several types of loading & bodies


composed of different materials.

• FEM can handle unlimited number & types of boundary


conditions.

• Altering the finite element model is easy & economical.

• FEM can handle non linear behavior resulting from large


deformations & material non lineraities.
Dr P R Venkatesh, ME Dept RVCE
DISADVANTAGES OF FEM
• FEM can give approximate solutions, and for exact
solutions, the cost & computational time is increased.
• FEM requires high end configuration of computers.
• FEM depends on the designer’s understanding and
hence skill & experience are required to obtain
acceptable results.
• Errors in formulating the mathematical model of the
physical system, selection of element type, etc. leads to
inaccurate results.
Dr P R Venkatesh, ME Dept RVCE
Types of finite elements
The selection of the element used in FEM depends upon
the problem type & analysis requirements.

Based on the shapes, the elements can be classified as;

• One dimensional elements

• Two dimensional elements

• Three dimensional elements

• Axisymmetric elements

Dr P R Venkatesh, ME Dept RVCE


One dimensional (1 D) elements
• When the configuration of elements can be expressed
in terms of one coordinate only, 1-D or line element can
be used. The coordinate is measured along the axis of
the element.
• It is represented as a straight line with two nodes, one
at each end. The field variable is chosen as the nodal
unknown (degrees of freedom)
• This element is used for structures that can be idealized
by line drawings such as bars, trusses, beams & frames.

Dr P R Venkatesh, ME Dept RVCE


Two dimensional (2 D) elements
• If the configuration of elements can be defined by two
coordinates the elements are called 2-D elements or
plane elements.
• It is represented as a triangles, rectangles or
quadrilaterals. These are used for plane stress, plane
strain & plate bending problems.
3 3
4 3 4

2
1 2 1 1
Rectangular element 2 Quadrilateral element
Triangular element

Note: The boundaries can have intermediate nodes & internal nodes also.
Dr P R Venkatesh, ME Dept RVCE
Three dimensional (3 D) elements
• If the configuration of elements can be defined by
three coordinates the elements are called 3-D elements
or solid elements.
• The commonly used 3 D elements are tetrahedron and
hexahedron (or brick element)

Tetrahedron Hexahedron or brick


Note: The boundaries can have intermediate nodes & internal nodes also.
Dr P R Venkatesh, ME Dept RVCE
Axisymmetric elements
• The axisymmetric elements are developed by rotation
of a planar surface (triangle or quadrilateral) about an
axis located in the plane of the element through 3600.
• These elements are used when the geometry & loading
of the problem are symmetric about the axis.
Ex: Shafts, Flywheels, Cylindrical storage tanks, rocket nozzles, etc.

Dr P R Venkatesh, ME Dept RVCE


Higher order elements

Higher order elements are those which have curved


boundaries & internal/intermediate nodes.

Dr P R Venkatesh, ME Dept RVCE


Simplex elements: In these elements, the approximating
polynomial consists of linear terms. They have only boundary
nodes. Ex: 3 noded (corners) Triangular element

Complex elements: In these elements, the approximating


polynomial consists of quadratic, cubic and higher order
terms in addition to constant & linear terms. They may have
additional boundary nodes & internal nodes also. Ex: 6 noded
CST element

Multiplex elements: These elements have boundaries


parallel to the coordinate axes to achieve inter element
continuity & approximating polynomials contain higher order
terms.
Ex: 9 noded quadrilateral element

Dr P R Venkatesh, ME Dept RVCE


Size of elements
• The size of the element plays an important role in the
accuracy of solution. If the element size is small, the solution
will be closer to the exact solution, but the computational
time will increase.
• Sometimes, different size of elements will be used in the
same body, if it has a hole, notch or some other
discontinuity, where the size of the element must be
smaller.
• Aspect ratio describes the size/shape of the element. It is
the ratio of largest dimension of the element to the smallest
dimension. Element with an aspect ratio of unity yields
best results.

Dr P R Venkatesh, ME Dept RVCE


Location of nodes
• If there is no abrupt change in geometry or material
property or boundary conditions, the body can be divided
into equal subdivisions and hence the spacing of the nodes
can be uniform.
• If there are changes in geometry, loading etc. the nodes
have to be introduced at the discontinuities.

Dr P R Venkatesh, ME Dept RVCE


Number of elements
• When the size of the elements is smaller, the number of
elements is greater. Accuracy of solution increases with
increase in number of elements upto a certain value.
• Beyond this, increase in number of elements will only
increase the computational time and strain on the computer
hardware, without any significant improvement of accuracy.

Dr P R Venkatesh, ME Dept RVCE


Nodes & Node numbering scheme
• Nodes are the points of connectivity between the elements & the
locations where the loads & boundary conditions are applied.

• Each node can have a maximum of six degrees of freedom (three


translational u, v, w & three rotational qx, qy, qz)

• The size of the global stiffness matrix resulting from the finite element
analysis depends upon the node numbering scheme & degrees of
freedom per node.

• To achieve minimum size of stiffness matrix & hence to minimize the


computer memory storage, a systematic way is adopted to number the
nodes & store the elements in a half band width instead of the storing
the whole matrix.

• The half band width ‘nbw’ or B=(D+1)f where D = Maximum difference


between the node numbers of an element & f is the number of
degrees of freedom at each node.

Dr P R Venkatesh, ME Dept RVCE


Ex: Three-bay frame with rigid joints. 20 storeys high

Let the dof per node is 3 which is fixed.


Case (a): Half band width B = (D + 1) f = ((6-2)+1). 3 = 15
Case (b): Half band width B = (D + 1) f = ((22-2)+1). 3 = 63
Thus, a shorter bandwidth can be obtained simply by continuous numbering
the nodes across the shortest dimension ofDept
Dr P R Venkatesh, ME the body.
RVCE
Banded matrix & Half band width: In a banded matrix, all the non-zero
elements are contained within a band; outside the band, all elements are
zero.
Bandwidth of  A is (2× nbw)+1, where‘nbw’ is called the half band width.
Here,bandwidth of [ A]  (2× 2)+1= 5

w
th nb nbw nbw
id
w
nd
Ba     
w     
nb
    
 
   

    
    
A
III diagonal     
    

II diagonal     
    

I diagonal
Dr P R Venkatesh, ME Dept RVCE III II I diagonal II III
Symmetric banded matrix:
• A symmetric matrix is one which remains the same even if the rows &
columns are interchanged, i.e [A]=[A]T
• If a matrix is symmetric, then its banded matrix is also symmetric about the
first diagonal. Such matrices require less storage space in computer memory.

nbw

       

   7    7

       
A
III diagonal
 7      

II diagonal
       

I diagonal

I diagonal II III
Dr P R Venkatesh, ME Dept RVCE
Skyline of matrix:
• If in a banded matrix, if there are zeroes at the top of the column,
only the elements starting from the first non-zero value need to be
stored.
• The line separating the top zeroes from the first non-zero elements is
called a skyline.

     

   8 
Skyline

   
A
  
Symmetric
 

Dr P R Venkatesh, ME Dept RVCE


Square matrix: A matrix whose number of columns & rows are equal.
Diagonal matrix: A diagonal matrix is a square matrix with non-zero
elements only along the principal diagonal.
Identity matrix: It is a diagonal matrix whose diagonal elements are
unity.
Symmetric matrix: It is a square matrix in which the elements on either
side of the principal diagonal are identical. It can also be defined as a
matrix which remains the same even after interchanging its rows &
columns. i.e [A]=[A]T
Upper & Lower triangular matrix:
• Upper triangular matrix is a square matrix whose elements below the
principal diagonal are zero.
• Lower triangular matrix is a square matrix whose elements above the
principal diagonal are zero.
Dr P R Venkatesh, ME Dept RVCE
Transpose of a matrix: It is the matrix obtained by interchanging the rows &
columns of a matrix.
Co factor matrix: It is the matrix obtained by the minors of a square matrix.
(Minor is the determinant of the matrix obtained by eliminating ith row & jth
column of a square matrix)
The co factor matrix cij  (1)i  j M ij
Adjoint of a matrix: It is the transpose of the cofactor matrix
Inverse of a matrix: It is obtained by dividing the adjoint of a matrix by the
determinant of the matrix.

Dr P R Venkatesh, ME Dept RVCE


Techniques to improve accuracy of FEM solution
(i) h-refinement: (or mesh refinement)
In machine elements which involve abrupt change in cross section
such as notches, keyways, FEA solution does not converge to exact
solution.
Hence convergence is achieved by judiciously (so as not to increase
computational time) selecting more elements in the FE model &
making a fine mesh at the critical sections.
(ii) p-refinement:
It is also known as polynomial refinement in which more number of
nodes are chosen at critical sections to achieve realistic solutions.
As the number of nodes is increased, the order of the polynomial
increases, better convergence is possible, but care must be taken to
balance the cost & computational time of analysis.

Dr P R Venkatesh, ME Dept RVCE


Single point & Multi point boundary conditions:
• To solve the general equation of equilibrium of a finite element  K Q  F  ,some
constraints are to be specified, say Q=a. It means that Q is considered as an
identified degree of freedom at a particular node. Hence the equilibrium equation
is known as SPC (single point constraint)
• In case of inclined supports, shrink fit assemblies, etc. multipoint constraints
(MPC‟s) are specified at nodes.
Homogenous & heterogeneous boundary conditions:
• Homogeneous boundary conditions occur at locations that are completely
prevented from movement.
• Heterogeneous bc‟s occur where finite non zero values are specified.
Essential & Natural boundary conditions:
• Essential or geometric bc‟s pertain to the geometrical values of displacement,
slope, etc.at the nodes.
• Natural or Force bc‟s pertain to the load or forces acting at the nodes.
Dr P R Venkatesh, ME Dept RVCE
Gauss Elimination Method : Gaussian elimination is a method of solving
a linear system of equations Ax  b by forward elimination method.
Solve the linear system by Gauss elimination method :
3x  y  4 z  1
 x  4 y  2 z  10
 2 x  2 y  2 z  2
In the matrix form, the equations can be written as;
 3 1 4   x   1   a11 a12 a13   x   b1 
 1 4 2   y   10    a    
a23   y   b2 
       21 a22
 2 2 2   z  2   a31 a32 a33   z  b3 
(1) (1)
nd rd
Eliminating the x term from 2 & 3 rows, i.e. a21 & a31 will be zero.
(1)
 The new values of 2 nd
row elements are; a21  0
(1) a   1 
a22  a22   21   a12  4     1  4.333
 a11   3 
(1) a   1 
a23  a23   21   a13  2     4  3.333
 a11   3 
(1) a   1 
b2  b2   21   b1  10     1  10.333
 a11   3  Dr P R Venkatesh, ME Dept RVCE
(1)
The new values of 3 row elements are; a31  0
rd

(1)  a31   2 
a32  a32     a12  2     1  2.67
 a11   3 
(1)  a31   2 
a33  a33     a13  2     4  0.67
 a11   3 
(1)  a31   2 
b3  b3     b1  2     1  1.33
 a11   3 
In the matrix form, the equations can be written as;
 a11 a12 a13   b1 
3 1 4  x  1     x  
0 4.33 3.33   y   10.33    0 a(1) (1)
    (1)
a23  y   b2 

    

22

  
0 2.67 0.67   z   1.33   (1)    (1) 
a33    b3 
(1)
z
 RVCE
Dr P R Venkatesh, ME Dept
0 a32
(2)
rd
Eliminating the 'y' term from 3 row, i.e. a32 will be zero.
(2)
The new values of 3 row elements are; a32  0
rd

 (1)
 (1)
(2) (1)
a  2.67 
a33  a33   (1)32   a23  0.67     3.33  1.383
   4.33 
 22 
a
 (1)
 (1)
(2) (1)
a  2.67 
b3  b3   (1)32   b2  1.33     10.33  7.7
   4.33 
 22 
a
In the matrix form, the equations can be written as;
3 1 4  x  1 
0 4.33 3.33   y   10.33  1.383 z  7.7
    
0 0 1.383  z   7.7 
 z = 5.57 Substituting in 4.33 y  3.33 z  10.33, we get y = -1.898
Substituting for y & z in 3DrxP RVenkatesh,
y  4ME z 1,RVCE
Dept we get x = -6.46
FE Analysis stages
Preprocessor : Within the preprocessor the model is set up. It includes a
number of steps and usually in the following order:
• Build geometry
• Define materials
• Generate element mesh
Processor : Here all specified information about the problem is gathered:
• Apply loads & boundary conditions
• Obtain solution
Postprocessor: In this part of the analysis, one can;
• Visualize the results: For example, plot the deformed shape of the
geometry or stresses.
• List the results: tabular listings or file printouts may be obtained.
Commercial FE Analysis Packages
• ANSYS
• MSC NASTRAN
• NISA
• COSMOS
• LS DYNA
Dr P R Venkatesh, ME Dept RVCE
BASIC EQUATIONS OF ELASTICITY
There are three basic sets of equation included in theory of
elasticity. These must be satisfied if an exact solution to a
structural mechanics problem is to be obtained. They are us
under;
(a)The differential equations of equilibrium in terms of stresses
acting on a body.
(b)The strain/displacement and compatibility equations.
(c)The stress/strain or material constitutive law (Generalized
Hooke‟s law) {s}=[D]{e, where [D] is the material constant
matrix
Dr P R Venkatesh, ME Dept RVCE
Equilibrium equation in elasticity subjected to body force,
surface traction and point load:
Fig shows a three dimensional
body having volume V, surface S
T on Cartesian coordinates x, y

Fy
and z. The boundary is
dv
v constrained at some region
y
Fx
where displacement is
Fz u Pi
specified.
w
The following forces are acting
V
on the body;
S
(i) Body force (F)
x
(ii) Surface traction force (T)
(iii) Point load (P)
z
Dr P R Venkatesh, ME Dept RVCE
(i) Body force (F): It is defined as the distributed force per unit
volume. (N/m3). (ex: weight/unit volume) Consider a small
element of volume dv having the body force along x. y & z
directions as Fx, Fy & Fz respectively.
 Fx 
 
Body force vector F    Fy  or F   Fx , Fy , Fz 
T

F 
 z
(ii) Surface traction force (T): It is the distributed surface force per
unit area. (N/m2) (ex: contact force, action of pressure, etc.)
Tx 
 
T   Ty  or T   Tx , Ty , Tz 
T

T 
Dr zP RVenkatesh, ME Dept RVCE
(iii) Point load (P): The load acting at a point ‘i’ is represented by
three components as;  Px 
 
P   Py  or P  Px , Py , Pz 
T

P 
 z

Stresses: As a result of application of point loads & surface traction,


stresses are induced in the elemental volume dv. There are six
components of stresses ; three normal stresses & three shear
stresses, denoted by;
s x  xy  yz 
s    yx s y  zx  or s   s x , s y , s z , xy , yz , zx 
  T

 zy  xz s z 
(as  xy   yx ,  yz   zy &  zx   xz , complimentary shear)
Dr P R Venkatesh, ME Dept RVCE
where;
s x
ds x  . dx
x
s y
ds y  . dy
y
s z
ds z  . dz
z
 xy
d xy  . dx
x

d yz  yz . dy
y

d zx  zx . dz
z

d yx  yx . dx
x
 zy
d zy  . dy
y

d xz  xz . dz
Dr P R Venkatesh, ME Dept RVCE z
Displacements: The displacement of a point due to the action of the
forces is given by its components in x, y & z directions as;
 u
 
  v    
  
T
U or U u , v , w

 w

Differential equations of equilibrium in three dimensions:
Fig shows the stresses acting on an elemental volume. The
equilibrium equations can be obtained by summing up the forces in
x, y & z directions respectively.
Summing up all the forces in x - direction, we get;
 s x    yx 
s x  dx  dydz  s x dydz   yx  dy  dxdz   xy dxdz 
 x   y 
  zx 
 zx  dz  dxdy   xz dxdy  Fx (dxdydz )  0
 z  Dr P R Venkatesh, ME Dept RVCE
As dx  dy  dz = dv, the elemental volume &  xy   yx ,  yz   zy ,  zx   xz
cancelling out the terms positive & negative terms, the equation becomes;
s x  xy  xz
   Fx  0
x y z
Similarly, the force balance in y & Z directions gives
s y  yz  xy
   Fy  0
y z x
s z  xz  yz
   Fz  0
z x y
The above equations are known as equilibrium equations in 3D stress system .

Dr P R Venkatesh, ME Dept RVCE


Strain-Displacement relations
The strains at a point may be represented in the vector form as;
e  e x , e y , e z ,  xy ,  yz ,  zx  where e x , e y & e z are normal strains &
T

 xy ,  yz &  zx are the shear strains. Fig shows the deformation of the
y
dx - dy face for small deformations.
u
Considering other faces, the strain
dy
y relations are;
u v w
ex  , ey  , ez 
v
v
y
dy u x y z
y

 u v
 xy  
xy
2

u v
v
dx y x
dy x
x
w v
v
 yz  
dx y z
u
u  dx
x w u
x  zx  
Deformed elemental surface Dr P R Venkatesh, ME Dept RVCE x z
Knowing the three components of displacements u, v & w,
the six components of strain can be determined by
differentiating the displacements.

If the six strain compnents are known, it is required to


determine the three displacements u,v & w which can be
obtained by integrating the strains.

As there are six equations and three unknowns, the solution


is not unique. For the values f u, v & w to be unique, certain
additional equations known as 'compatibilty conditions'
are to be satisfied. Dr P R Venkatesh, ME Dept RVCE
Strain - Stress - relations
If six components of stresses namely s x , s y , s z , xy , yz &  zx are
known, the six components of strain namely e x , e y , e z ,  xy ,  yz &  xz
may be determined.
sx 
Strain in x - direction : e x   (s y  s z )
E E
where E = Elastic modulus, = Poisson's ratio
sy 
Strain in y - direction : e y   (s x  s z )
E E
sz 
Strain in z - direction : , e z   (s x  s y )
E E
 xy  yz  zx
Shear strains are :  xy  ,  yz  ,  zx 
G G G
E
Shear modulus ,G =
2(1  )
2(1  ) xy 2(1  ) yz 2(1  ) xz
  xy  ,  yz  ,  zx 
E E ME Dept RVCE
Dr P R Venkatesh, E
Strain - Stress relations......
Representing the above in a matrix form , e  =  B s 
where B is the symmetric (6  6) material matrix.

ex   1   0 0 0  s x 
e    1  0 0 0  s y 
e  1  
y
 1 0 0 0  s z 
 z    0 0 0 2(1  ) 0 0   xy 
 xy  E  0 2(1  ) 0   yz 
 yz 
0 0 0
 0 0 0 0 0 2(1  )   
 zx     zx 

Dr P R Venkatesh, ME Dept RVCE


Stress - Strain relations
If six components of strains namely namely e x , e y , e z ,  xy ,  yz &  xz
are known, the six components of stress s x , s y , s z , xy , yz &  zx
may be determined.
sx 
Strain in x - direction : e x   (s y  s z )
E E
sy 
Strain in y - direction : e y   (s x  s z )
E E
sz 
Strain in z - direction : , e z   (s x  s y )
E E
(s x  s y  s z )
Adding;  e x  e y  e z   (1  2 )
E
Putting (s x  s y  s z )  I1 (the first stress invariant)
&  e x  e y  e z   J1 (the first strain invariant)
I
we can write J1  1 (1  2 )
E Dr P R Venkatesh, ME Dept RVCE
 Invariant is a quantity which remains constant irrespective of the 
axis considered 
 
I1 J1
Re arranging the terms,  (i )
E (1  2 )
s x s y s z s x
Strain in x - direction e x    Adding & subtracting ,
E E E E
s x s y s z s x s x
ex     
E E E E E
sx  sx  I1
ex  (1  )  (s x  s y  s z )  (1   ) 
E E E E
I1 sx  J1
Substituting for from (i), e x  (1   ) 
E E (1  2 )
sx  J1
 (1  )  e x 
E (1  2 )
Dr P R Venkatesh, ME Dept RVCE
exE  J1 E
s x   Substituting J1  (e x  e y  e z ),
(1  ) (1  2 )(1  )
exE e x E e y E e z E
sx    
(1  ) (1  2 )(1  ) (1  2 )(1  ) (1  2 )(1  )
exE    e y E e z E
sx  1   
(1  )  (1  2 )  (1  2 )(1  ) (1  2 )(1  )
ε x E(1 - ν) νε y E νε z E
σx = + + Similarly,
(1 + ν)(1 - 2ν) (1 - 2ν)(1 + ν) (1 - 2ν)(1 + ν)
ε y E(1 - ν) νε z E νε x E
σy = + +
(1 + ν)(1 - 2ν) (1 - 2ν)(1 + ν) (1 - 2ν)(1 + ν)
ε z E(1 - ν) νε x E νε y E
σz = + +
(1 + ν)(1 - 2ν) (1 - 2ν)(1 + ν) (1 - 2ν)(1 + ν)
Dr P R Venkatesh, ME Dept RVCE
Shear stress
Rigidity modulus G   Shear stress  G  Shear strain
Shear strain
E Shear strain  E
Also, E  2G (1   ) or G  ,Shear stress 
2(1  ) 2(1  )
γ xy E γ yz E γ E
 τ xy = , τ yz = , τ zx = zx
2(1 + ν) 2(1 + ν) 2(1 + ν)
The six components of stress can be represented in the matrix form as;
 1    ex 
   0 0 0 
s x   
  
 
s   1    0 0 0  ey
 
 y   
s 
 z E    1   0 0 0  e z 
    1  2   

 xy  (1   )(1  2 )  0 0 0   0 0   xy 
    2   1  2   
  yz
 0 0 0 0   0   yz 
 zx    2  1  2   
 
 0 0 0 0 0     zx 
  2 

Dr P R Venkatesh, ME Dept RVCE


Problems in solid mechanics can be modeled as 3-dimensional
& analysis can be carried out by using FEM. In such cases
computational time increases due to more number of equations.
hence analysis becomes costlier.
Accurate results can be obtained in most cases by treating the
problem as 2- dimensional.
In 2-D case, problem is modeled as 'Plane stress' & 'Plane strain'
conditions.
(i) Plane stress : A thin planar body subjected to in-plane loading on
its edge surface is said to be in 'plane stress' condition.
Ex : A ring press fitted on a shaft, thin plates,
Thin walled cylinder filled with a fluid under pressure

Dr P R Venkatesh, ME Dept RVCE


PLANE STRESS Examples:
1. Thin plate with a hole  xy sy
 xy
sx

2. Thin cantilever plate

3. Ring press fitted onto a shaft

Dr P R Venkatesh, ME Dept RVCE


(i) Plane stress : Here, stresses s z , yz &  zx are set as zero.

e x  s x s y  , e y  s y s x  , e z   (s x  s y )
1 1
E E E
 xy 2(1  )
 xy  , But E  2G (1   )   xy   xy
G E
 1  
Adding , e x  e y  s x s y   s y s x     s x  s y 
1 1
E E  E 
 E 
sx 
E
1 
 e x  e y  s y , Similarly, s y    e x  e y   s x
 1  
Substituting for s y (in terms of s x ) in equation of e x ,
s x   E  
e x  s x s y    e x  e y   s x 
1
 
E E E  1   
sx       
ex   ex   e y  s x
E  1    1 DrPR Venkatesh,
 E
ME Dept RVCE
   sx   
ex   e
 x  1       e y Taking LCM & simplifying ,
 1   E  1  
sx
 ex  1   e
2

2  x
ε + νε y 
E
y E
   σx =
 1   1   1 - ν 
2  y
E E
Similarly, s y  e  e Also we have    xy
1   x xy
2(1  )
Representing s x , s y &  xy in the matrix form,
 1 
s x    0  ex 
s  E    e y 
 y  1 0

 xy  (1   2
) 1    xy 
   0 0  
 2 
Note : Though for plane stress case, the stress in the z - direction is zero,

( s z = 0) strain e z is non zero and is equal to 
E
 s x  s y .

(From 3 - D strain - stress relation matrix)


Dr P R Venkatesh, ME Dept RVCE
(ii) Plane strain : If a long body of uniform cross section is subjected to
transverse loading along its length, a small thickness in the loaded area
can be treated as subjected to 'plane strain'
Ex : A long cylinder filled with a fluid under pressure
Retaining wall of a dam
Here, strain e z ,  yz &  zx are set as zero.
s x   1    0 0 0  ex 
s    1   0 0 0  e y 
s y
E    1   0 0 0  e 
 z    0    
z

 xy  (1  )(1  2 )  0
0 0 0.5 0 0
0 0.5    xy 

 yz   yz 
0 0 0
 0 0 0 0 0 0.5    
 zx     zx 
Eliminating the third, fifth & sixth rows & columns corresponding to e z ,  yz
& zx from the general 3D matrix, we obtain the stress - strain relations for
plane strain case.
Dr P R Venkatesh, ME Dept RVCE
PLANE STRAIN Examples:
1. Dam Slice of unit
thickness
1

 xy sy
y  xy
sz sx

x
z
2. Long cylindrical pressure vessel subjected to internal/external
pressure and constrained at the ends

Dr P R Venkatesh, ME Dept RVCE


Eliminating the third, fifth & sixth rows & columns corresponding to
e z ,  yz &  zx from the general 3D matrix,
we obtain the stress - strain relations for plane strain case as ;
s x   1    0  ex 
s  E    e y 
  y    
1   0   
 xy  (1  )(1  2 )  0 0 0.5    xy 
    
The above matrix give the stress - strain relations for the plane strain case.
Note : Though for plane strain case, the strain in the z - direction is zero,
( e z = 0) stress s z need not be zero.
E
** From the general 3D stress - strain matrix, s z 
(1  )(1  2 )
 ex ey 

Dr P R Venkatesh, ME Dept RVCE


Problem 1
In a plane strain problem, sx=20 MPa, sy=-10 MPa, E=200
GPa and Poisson’s ratio =0.3. Determine the value of sz.

According to generalized Hooke's law, strains in x & y directions are


sx sy 20 (10)
ex     0.3  1.15  10 4
&
E E 200 10 3
200 10 3

sy sx (10) 20
ey     0.3   0.8  10 4

E E 200 103 200 103


Stress in the z-direction s z under plane strain condition is
E 0.3  200 103 104
sz 
(1  )(1  2 )
 ex ey  
(1  0.3)(1  2  0.3)
1.15  0.8  4.04 MPa

Dr P R Venkatesh, ME Dept RVCE


Exact vs. approximate solution methods
The equilibrium equations in
T
3D stress system are;
Fy

s x  yx  zx
dv
v

   Fx  0
y
Fx
Fz u Pi x y z
w

s y  zy  xy
V
   Fy  0
S y z x
s z  xz  yz
x

   Fz  0
z z x y
• In mechanics of solids, the problem posed is to obtain the displacements strains &
stresses in the body shown in fig, using the equations of equilibrium.
• This requires the solution of partial differential equations which is available only for
simple geometries & loading conditions. (Exact solutions)
• For problems involving complex geometries & general boundary & loading
conditions, exact solution becomes impossible.
• Hence approximate methods such as potential energy or variational methods
Dr P R Venkatesh, ME Dept RVCE
are used which are simple to solve & obtain solutions.
Potential Energy & Equilibrium
The total potential energy of an elastic body  is the sum of strain energy (U)
& the work potential (WP).
  Strain energy (U) +Work Potential (WP). For linear elastic materials,
1
the strain energy (resilience) per unit volume in the body is  s T e dV
2V
 Resilience per unit volume = area under linear portion of stress strain curve 
The work potential is
WP =   u T f dV   u T T dS   uiT Pi
V S i

 Work done by body force = body force / unit volume  volume  displacement 
 
 Work done by surface traction = traction force / unit area  surface area  displacement 
 Work done by point load = point load  displacement 
 
The total potential energy for the general elastic body is;
1 T

2 V
s e dV  V u T
f dV  S u T
T dS  i i Pi
u T

Dr P R Venkatesh, ME Dept RVCE


In FEM, we consider conservative systems. A conservative system is a
system in which work done by a force is independent of path taken &
completely reversible. (i.e. there is no loss or gain of energy)

Principle of minimum potential energy: For conservative systems, of all


the kinematically admissible displacement fields, those corresponding to
equilibrium extremize the total potential energy. If the extremum
condition is a minimum, the equilibrium is stable.

Note: Kinematically admissible displacements are those which satisfy the


compatibility conditions (unique value of displacements) & boundary
conditions.

Dr P R Venkatesh, ME Dept RVCE


Problem 2

Determine the displacements of nodes 1 & 2 in the spring


system shown in fig using minimum potential energy
principle.
k1 = 60 N/m

k3 = 100N/m
1 2 80 N
100 N

k2 = 75 N/m

Dr P R Venkatesh, ME Dept RVCE


k1 = 60 N/m
Note : The potential energy of a spring
1
k3 = 100N/m
2 80 N is = Average force  extension of spring
100 N
 0+k  1 2
=     k
k2 = 75 N/m
 2  2

The total potential energy of the system is;


1 1 1
  k11  k2 2  k3 32  F1u1  F2u2
2 2

2 2 2
Let u1 & u2 be the displacements of nodes 1 & 2.
The extensions of the springs are;
1  u1 ,  2  u1 &  3  u2  u1
1 2 1 1
  k1u1  k2u1  k3 (u2  u1 ) 2  F1u1  F2u2
2

2 2 2
Dr P R Venkatesh, ME Dept RVCE
For stable equilibrium, potential energy must be a minimum.

i.e.  0  (k1  k2  k3 )u1  k3u2  F1 (i )
u1

Also  0  k3 (u2  u1 )  80 or -k3u1  k3u2  F2 (ii )
u2
 k1  k2  k3 k3   u1   F1 
In the matrix form,     
  k3 k3  u2   F2 
 235 100   u1  100 
Substituting for k1 , k2 , k3 , F1 & F2     
 100 100  u2   80 
235 u1  100 u2  100 & 
  solving, u1  1.333, u2  2.333
 100 u1  100 u2  80 
 u1  1.333 
In the matrix form,     
u2 Dr P RVenkatesh,
2.333MEDept RVCE
Problem 3
Obtain the equations of equilibrium & the nodal
displacements for the spring system shown in fig using
minimum potential energy principle. Take k1=40 N/mm,
k2=40 N/mm, k3=80 N/mm, k4=90 N/mm, F1=60 N, F2=0,
F3=50 N.

Note: Nodal displacements may


Dr P R Venkatesh, also
ME Dept be denoted by q
RVCE
Problem 3...

The total potential energy of the system is;


1 1 1 1
  k11  k2 2  k3 3  k4 4 2  F1q1  F3q3
2 2 2

2 2 2 2
Let q1 , q2 & q3 are nodal displacements of nodes 1,2 & 3.
From the arrangement, the extensions of the springs are;
1  (q1  q2 ),  2  q2 ,  3  (q3  q2 ) and  4 =  q3
1 1 1 1
  k1 (q1  q2 )  k2 q2  k3 (q3  q2 )  k4 (q3 ) 2  F1q1  F3q3
2 2 2

2 2 2 2
Dr P R Venkatesh, ME Dept RVCE
For stable equilibrium, PE must be a minimum.

i.e.  0, i = 1,2,3
qi

 0  k1 (q1  q2 )  F1  0 (i )
q1

 0   k1 (q1  q2 )  k2 q2  k3 (q3  q2 )  0 (ii )
q2

 0  k3 (q3  q2 )  k4 q3  F3  0 (iii )
q3
 k1 k1 0   q1   F1 
In the matrix form,  k1 k1  k2  k3  k3  q    0 
 2  
 0  k3 k3  k 4   q3   F3 

Dr P R Venkatesh, ME Dept RVCE


 k1 k1 0   q1   F1 
 k1 k1  k2  k3  k3  q    0 
  2  
 0  k3 k3  k 4   q3   F3 
Substituting for k1 , k2 , k3 & F1 , F2 , F3 ;
 40 40 0   q1  60 
 40 160 80  q    0 
  2  
 0 80 170   q3  50 
 40q1  40q2  60
 40q1  160q2  80q3  0
 80q2  170q3  50
Solving , q1 = 2.514 mm, q2 = 1.014 mm, q3 = 0.771 mm
Dr P R Venkatesh, ME Dept RVCE
Rayleigh-Ritz Method
• For complex physical problems, it is difficult to obtain exact analytical
solution. For such cases, a numerical or approximate analytical method
can be used to obtain an approximate solution.
• Most of the continuum problems can be solved by any one of the two
approximate methods.
(i) Variational method (Rayleigh Ritz method)
(ii) Weighted Residual method (Galerkin’s method)

Steps in Rayleigh-Ritz Method:


1. Formulate the potential energy functional.
2. Assume a displacement function satisfying the boundary conditions.
The displacement function can be a polynomial or trigonometric
function.
3. Substitute the displacement function into the Potential energy
functional.
4. Minimize the potential energy functional to get equilibrium equations.
5. Solve these equations to obtain unknown displacements, stresses &
strains.
Dr P R Venkatesh, ME Dept RVCE
Problem 3
Find the displacement of the midpoint & stress in the rod
loaded as shown in fig using Rayleigh Ritz method.

1
Note : Strain energy (SE)for one dimensional axial load   se dV
2V
2
 du   du   du 
Also e    & s  Ee  E    se  E   and dV  Adx
 dx   dx   dx 
2
EA  du 
 SE     dx
2 L  dx  Dr P R Venkatesh, ME Dept RVCE
Rayleigh Ritz method:

1 T
Step 1 : Potential energy functional is   
2V
s e dV   u T
f dV   u T
T dS  
i
u T
i Pi
V S

Here, no body force & traction is present, only strain energy & point load terms
1
are considered. Also as its is one dimensional case    se dV  (u  P)
2V
2 2
 du   du   du   du 
Also e    & s  Ee  E    se  E   , =   ( E  1)
 dx   dx   dx   dx 
and dV  Adx  dx ( A  1), P  2 units
Dr P R Venkatesh, ME Dept RVCE
2
1  du 
2
     dx  2u.
2 0  dx 
Step 2 : Assuming a displacement function & apply bc's :
u  a0  a1 x  a2 x 2 , The boundary conditions are;
@ x  0, u  0,  0  a0  a1 (0)  a2 (0),  a0  0
Also @ x  2, u  0,  0  0  2a1  4a2 ,  a1  2a2
Displacement function u = 0  (2a2 ) x  a2 x 2  a2 ( x 2  2 x)
At midpoint, x = 1, u1 = a2
2
du  du 
Also  a2 (2 x  2)  2a2 ( x  1)     4a2 2 ( x 2  2 x  1)
dx  dx 
Dr P R Venkatesh, ME Dept RVCE
Step 3 : Substitute the displacement function in PE functional :
2
2 x 
2 3
1 4 2
         a2  2 a2
2 2 2
4 a2 ( x 2 x 1) dx 2 a 2 = 2a 2  x x  4 a 2
20  3 0 3
4 2
Step 4 : Minimize the PE functional :   a2  2a2
3
 8 6
 0  a2  2  0  a2    0.75
a2 3 8
Step 5 : Determine the displacements, strains & stresses :
Displacement u  a2 ( x 2  2 x)  0.75( x 2  2 x)  0.75(2x - x 2 )
At x  1, u  0.75(2  1)  0.75, At x  0, u  0, At x  2, u  0
du
Strain e =  0.75(2  2 x)  1.5(1 - x)
dx
du du du
At x  1,  1.5(1  1)  0, At x  0,  1.5, At x  2,  1.5
dx dx dx
Stress s = Ee = 1×1.5(1- x)  1.5(1 - x)
At x  1, s  1.5(1  1)  0, AtDr xP RVenkatesh,
0, s ME1.5, RVCEx  2, s  1.5
DeptAt
Dr P R Venkatesh, ME Dept RVCE
Problem 4

A cantilever beam of span „L‟ is subjected to a point load „P‟ at free


end. Determine the maximum deflection by using Rayleigh – Ritz
Method assuming the displacement function as a Trigonometric
function. P

y
L max

s2
Note : Strain energy (SE)for beam under bending    dV
V
2E

where s 
Mc
I
& dV  dA  dx, SE   2
L
I
M 2c 2
 2 E
 dA  dx  
L
2
M2
EI 2  c 
  dA  dx
2

But  c 2  dA  I , moment of inertia of beam sction about centroidal axis.


2
M 2 2
d y EI  d 2 y 
 SE of a beam =  dx, Putting M = EI 2 , SE =   2  dx
L
2 EI dx
Dr P R Venkatesh, ME Dept RVCE
2 L  dx 
P

y
L max

x=0 x=L
y=0 y=ymax
dy/dx=0

EI  d 2 y 
L
Potential energy functional fo a beam  =SE+WP=   2  dx  Pymax
2 0  dx 
x dy x    
Let the displacement polynomial y  a sin   a cos   
2L dx  2L   2L 
d 2x x    
2
and  a sin   2 
 2L   4L 
2
dy
dy
Using boundary conditions at x  0, y  0,  0 & at x  L, y  ymax  a
dx
2
EI    x     EI a 2 4 2 x 
L 2 L
= 
2 0
  a sin  
  2 
 2L   4L  
dx  Pa 
32 L 04 
sin   dx  Pa
 2L 
Dr P R Venkatesh, ME Dept RVCE
P

y
L max

x=0 x=L
y=0 y=ymax
dy/dx=0

x 
2 4 L 1  cos 2  
EI a   2L   1  cos 2q 
4 
= dx  Pa  sin q 
2

32 L 0 2  2 
EI a 
2 4 L
x 
=
64 L4 0 1  cos   dx  Pa
 L 
L
EI a  
2 4
  x  EI a 2 4
= 4 
x  sin     Pa   L  Pa
 L 0
4
64 L  64 L
Dr P R Venkatesh, ME Dept RVCE
P

y
L max

x=0 x=L
y=0 y=ymax
dy/dx=0


Minimising the PE functional, 0
a
  EI a 2 4 
   Pa   0
a  64 L 3

EI  4 PL3
 3
 2a  P  0  a  ymax 
64 L 3.044 EI
Dr P R Venkatesh, ME Dept RVCE
Interpolation models
• Interpolation models are defined as the appropriate
mathematical model or trial function which
represents the displacement variation within the
element.

• The following types of interpolation models are used


in Variational methods/FEM.

1. Trigonometric functions
2. Polynomial function

Among the above, polynomial models are most widely


used due to ease of formulating, calculating
(differentiating & integrating) & better level of
accuracy (with higher order polynomials)
Dr P R Venkatesh, ME Dept RVCE
Polynomial form of interpolation model
A polynomial type of interpolation model assumed to represent the
displacement variation within an element, then the dispalcement can
for be expressed as ;
u(x)= a0  a1 x  a2 x 2  a3 x 3 .......... (for 1-D element)
u(x, y)= a0  a1 x  a2 y  a3 x 2  a4 xy  a5 y 2  a6 x 2 y  a7 xy 2  .........
(for 2-D element)
If in the above polynomials, terms upto x1 & y1 are considered, it is said to
be a linear model.
If terms upto x 2 & y 2 are considered, it is said to be a quadratic model &
if terms upto x 3 & y3 are considered, it is said to be a cubic model & so on.

Dr P R Venkatesh, ME Dept RVCE


Convergence Criteria
• Convergence implies results obtained by FEA solution reaches the
exact solution. It depends on the proper selection of displacement field
variable & order of the interpolation polynomials.
• The convergence of the finite element solution can be achieved if the
following three conditions are fulfilled by the assumed displacement
function.
1. The displacement function must be continuous within the elements.
This can be ensured by choosing a suitable polynomial. For example, for
an n degrees of polynomial, displacement function in 1-D problem can be
chosen as:

u(x)= a0  a1 x  a2 x  a3 x .......... an x
2 3 n

Dr P R Venkatesh, ME Dept RVCE


Convergence Criteria....
2. The displacement function must be capable of rigid body
displacements of the element. The constant term used in the
polynomial (a0) ensures this condition. (Even for x=0, the
displacement will be equal to ao)
3. The displacement function must include the constant strains
states of the element. As element becomes infinitely small, strain
should be constant in the element. Hence, the displacement
function should include terms for representing constant strain
states. The second term used in the polynomial (a1) ensures this
condition. (As differentiation of a1x will be a1 , a constant)

Dr P R Venkatesh, ME Dept RVCE


Compatibility
• Displacement should be compatible between adjacent
elements. There should not be any discontinuity or
overlapping when deformed.
• The adjacent elements must deform without causing
openings, overlaps or discontinuities between the
elements.
Elements which satisfy all the three convergence
requirements and compatibility condition are called
Compatible or Conforming elements.

Dr P R Venkatesh, ME Dept RVCE


Criteria for selection of order of interpolation polynomial
• The number of generalized coordinates should be equal to the
number of degrees of freedom of the elements.

• The pattern of variation of the polynomial should be independent of


the local coordinate system. (Geometric or spatial isotropy or
Geometric invariance).

• The interpolation polynomial should satisfy the convergence


requirements.

• Displacement shape should not change with a change in local


coordinate system. This can be achieved if polynomial is balanced in
case all terms cannot be completed.

• This „balanced‟ representation can be achieved with the help of


Pascal triangle in case of a 2 D polynomial. The geometric
invariance can be ensured by the selection of the
corresponding order of terms on either side of the axis of
symmetry. Dr P R Venkatesh, ME Dept RVCE
Geometric invariance (or Spatial isotropy); Pascal’s
triangle

𝛼6

Pascal’s triangle
Dr P R Venkatesh, ME Dept RVCE
Geometric invariance (or isotropy); Pascal’s triangle
Ex : If a cubic model is assumed, displacement polynomial
using Pascal’s triangle is ;
U ( x, y )  a0  a1 x  a2 y  a3 x 2  a4 xy  a5 y 2  a6 x 3  a7 y 3
or
U ( x, y )  a0  a1 x  a2 y  a3 x 2  a4 xy  a5 y 2  a6 x 2 y  a7 xy 2

• In the above polynomials, if we interchange x & y terms,


the pattern does not change.
• In both the equations, the same variable occur even after
interchanging. These polynomials are known as
“Balanced Polynomials”
Dr P R Venkatesh, ME Dept RVCE
Coordinate systems

• Co ordinate system is a space where configuration of a


body is represented.
Ex: Cartesian Coordinate system, Polar Coordinate system

• In FEM, these general coordinate systems are further


classified as;
1. Global Coordinate system
2. Local Coordinate system
3. Natural coordinate system

Dr P R Venkatesh, ME Dept RVCE


Global Coordinate system

• The global coordinate system corresponds to the entire


body and used to define the points on the entire body.
• Fig shows method of representation in global coordinate
system.
y(v)
x ( x4, y 4 ) ( x 3, y 3 )
x(u) 4 3

1 2
x1 1 2
( x2 - x1 ) ( x 1, y 1 ) ( x 2, y 2 )
x2
x(u)

1-D Global coordinate system 2-D Global coordinate system


Dr P R Venkatesh, ME Dept RVCE
Local Coordinate system
• A local coordinates system whose origin is located within
the element in order to simplify the algebraic manipulations
in the derivation of the element matrix.
• Local coordinate system corresponds to a particular
element in the body , and the numbering is done to that
particular element neglecting the entire body .

s s
r y
r
4 3
x1 x2 x(u)
L/2 r
1 L 2 2
1
r

x
x

1-D Local coordinate system 2-D Local coordinate system


Dr P R Venkatesh, ME Dept RVCE
Natural Coordinate system

• Natural coordinate system - Similar to local coordinate


system but a node is expressed by a dimensionless set of
numbers whose magnitude never exceeds unity.

y(v) 

 ( ,  ) ( ,)
4 3

  
1 2 1 2

(  , ) (  , )
x x(u)

1-D Natural coordinate system 2-D Natural coordinate system


Dr P R Venkatesh, ME Dept RVCE
Relation between global & Consider an one dimensional bar
Natural Coordinate system element represented in natural
coordinates as shown in fig.

Also the variation of natural

 coordinate is as shown in fig.
1 2 From similar triangles
AB BE
 D ABE & ACD, 
E  AC CD
x  x1   1
  
x2  x1 2
A B
2  x  x1 
C
x1
2 i.e.   1 
 x2  x1 
1
x
x2 2  x - x1 
ξ = -1
Dr P R Venkatesh, ME Dept RVCE
 x 2 - x1 
Shape Functions
• Shape functions are defined as the interpolation functions
used to interpolate the value of the field variable (ex:
displacement) at any point within the element in terms of
nodal values.
Mathematically, displacement at any point within the element
n
is given by u ( x)   N i ui ; where ' n ' is the number of nodes
i 1

N i are the shape functions & ui are the nodal displacement


components in x-direction. Thus, For a two noded element,
u ( x)  N1u1  N 2u2 where N1 & N 2 are the shape functions &
u1 & u2 are the displacements at node 1 & 2 respectively.
Dr P R Venkatesh, ME Dept RVCE
For a two dimensional model, displacement at any point is;
 n
 
u   N i ui
u ( x, y )      n
i 1
For a three noded triangular element,
 v    N i vi
 i 1

u ( x)  N1u1  N 2u2  N 3u3
v( x)  N1v1  N 2v2  N 3v3
where N1 , N 2 & N 3 are the shape functions, u1 , u2 & u3 & v1 , v2 & v3
are the nodal displacements in x and y directions.

Dr P R Venkatesh, ME Dept RVCE


Shape Functions for Consider a 1-D bar element of
1 D bar element In
terms of Cartesian length le with a node at each end,
coordinates & each node has one DOF.
x
The variation of displacement inside
u1 u2
N1 N2
the element is given by u  ao  a1 x
O
x1 1
2 where ao & a1 are the generalized
x
x2
coordinates to be found from BC ' s
l e = ( x2 - x1 )
At x  x1 , u  u1 & At x  x2 , u  u2
 u1  ao  a1 x1 & u2  ao  a1 x2
Thus,  u2  u1   a1 ( x2  x1 )

or a1 
 u2  u1 
( x2  x1 )
Dr P R Venkatesh, ME Dept RVCE
Substituting the value of a1 into equation of u1 ; u1  ao 
 u2  u1 
x
( x2  x1 )
1

 ao  u1 
 u2  u1  x  u1 x2  u2 x1 
1  Substituting the values of a0 & a1
( x2  x1 ) ( x2  x1 )

into equation of u, we get u 


 u1 x2  u2 x1   u2  u1 
 x
( x2  x1 ) ( x2  x1 )

u
 u1 x2  u2 x1   u2  u1 
 x where l  ( x2  x1 ) is the length of the 1 D
e
le le

bar element. Re-arranging the terms, u 


 u1 x2  u2 x1   u2 x  u1 x
le

u
 x2  x 
u 
 x  x1 
u Also u  N1u1  N 2u2 Comparing the two equations;
1 2
le le

N1 
 x2  x 
,N 
 x  x1 
Thus, values of shape functions at nodes 1 & 2 are
2
le le
  x 2 - x   x - x1  
 N  =  N1 N2  =  , 
 l e l e  ME Dept RVCE
Dr P R Venkatesh,
x -x1
x 2 -x N2 =
N1 =
le
le
N1 =1
N2 =1

1 2 1 2

N2 u
N1

u = N1 u1 + N2 u2 u2
u1
1 1 x
2 2
Variation of shape function for 1 D bar element

Dr P R Venkatesh, ME Dept RVCE


Shape Functions for Consider a 1-D bar element of
1 D bar element In
terms of Natural length le with a node at each end,
coordinates & each node has one DOF.
x
The variation of displacement inside
u1  u2

the element is given by u  ao  a1 x
O  
x1 1
2 where ao & a1 are the generalized
x
x2
coordinates to be found from BC ' s
l e = ( x2 - x1 )
At node 1;   1, u  u1
At node 2,,   1, u  u2
 u1  ao  a1 & u2  ao  a1

Thus, ao 
 u1  u2 
& a1 
 u2  u1 
2 2
Dr P R Venkatesh, ME Dept RVCE
Substituting the values of ao & a1 into equation of u;

u
 u1  u2   u2  u1 
  Re-arranging the terms, u 
1  
u 
1  
u 1 2
2 2 2 2

Also u  N1u1  N 2u2 , Comparing the two equations; N1 


1  
, N2 
1  
2 2
 1    1    
Values of shape functions at nodes 1 & 2 are  N    , 
 2 2 
1-  N2 =
1 
N1 = 2
2
N1 =1
N2 =1

1 2 1 2
N1 N2 u

N1 =1 N2 =1 u1 u2
u= N1 u1 + N2 u 2
1 2 1 2 
         

Variation of shape function forDr1PDR bar


Venkatesh,
elementME Dept RVCE

Potrebbero piacerti anche