0% found this document useful (0 votes)
366 views160 pages

2 D Elements

2D finite elements can be triangular or rectangular in shape. Triangular elements are commonly used with 3 or 6 nodes, while rectangular elements are used with 4, 8, or 9 nodes. The document discusses shape functions and interpolation functions for various 2D elements. It also covers natural coordinate transformations, which allow representing element geometry in a dimensionless ξ-η coordinate system to simplify higher order elements. Stress analysis considerations for plane stress and plane strain conditions using triangular elements are also presented.

Uploaded by

eafz111
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
366 views160 pages

2 D Elements

2D finite elements can be triangular or rectangular in shape. Triangular elements are commonly used with 3 or 6 nodes, while rectangular elements are used with 4, 8, or 9 nodes. The document discusses shape functions and interpolation functions for various 2D elements. It also covers natural coordinate transformations, which allow representing element geometry in a dimensionless ξ-η coordinate system to simplify higher order elements. Stress analysis considerations for plane stress and plane strain conditions using triangular elements are also presented.

Uploaded by

eafz111
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 160

2D Elements

2-D elements
•Triangular elements
•Rectangular or quadrilateral
elements.

A triangular element can be taken with three nodes at the three


vertices.
Since the problem is a 2-D problem, we have two degrees of
freedom at each nodes.
Just as a 1-D element can be considered with more than two
nodes with a single degree of freedom (D.O.F) like for example, a
quadratic element, we can also take a triangular element with
four or six nodes.

Similarly, a rectangular element can be taken with four or five


nodes
Triangular Element with Three Nodes at Vertices
Consider a triangular element
ABC as shown in fig

Let

As the problem is 2-D, we can


express general polynomial to
represent the solution of the
problem as
If U1, U2 and U3 are the values of the primary variable U at
nodes 1, 2 and 3, we have

Expressing in matrix form


Let

where [N] is the matrix formed by the cofactors of elements of


matrix [M].
Where
Now, |M| = 2 A where A is the area of the triangle ABC.

Substituting in equation
Substituting in equation
Shape functions for a triangular element with three nodes as
vertices are given by

Since first derivatives of viz. are constants, all strains

computed for a linear triangular element are constant and hence,


when used for plane elasticity problem, such an element is called
as constant strain triangular (CST) element.
Calculate linear interpolation
functions for linear triangular
element shown in figure .
Now
Interpolation functions are given by
The triangular element used for ground
water flow simulation is shown in fig.
The nodal coordinates are (x1 = 1, y1 = 2),
(x2 = 4, y2 = 0.5), (x3 = 3, y3 = 4). The
nodal values of hydraulic heads {Φ} at
these nodes are [3.5, 2.2, 4.4] respectively.
Find the value of the hydraulic head Φ at
point (2.5, 2.5).

Let
Solving the equations we get
Rectangular Element with Four nodes at Vertices

Consider a rectangular element


ABCD with four nodes at vertices
as shown in fig.

For convenience, consider local


coordinate system x-y as shown in fig.

Since ABCD is a rectangle, we can


write
On solving
Substituting in equation

The shape functions may be expressed as


Alternate Method : Serendipity Method
Shape functions by this method are found using the fact that
Φi = 1 for ith node and zero for any other node.

We can have pilot function as


Coordinates of nodes of a
quadrilateral element are as shown
in the figure. Temperature
distribution at each node is
computed as T1 = 100 °C, T2 = 60 °C,
T3 = 50 °C and T4 = 90 °C. Compute
temperature at point P(2.5, 2.5)
convert the coordinate of the elements into local coordinates as
shown

From the figure, we find that l = 2 and h = 1

P(2.5, 2.5) can be represented as P(0.5, 0.5) in local coordinates


x y
and .
A typical four-noded rectangular element is
shown in figure. Derive the shape
function for the temperature field.
If the nodes 1, 2, 3 and 4 are located at the
coordinate points (3, 3), (6, 3), (6, 5) and
(3, 5) respectively and the temperature
distribution has been computed at each
node as T1= 120 °C, T2 = 65 °C, T3 = 40 °C
and T4 = 80 °C. Derive the shape function
and compute temperature at x=4 and y=4.
Convert x,y coordinates into local
coordinates taking point 1 as origin.

Point (4, 4) is given by (1, 1)

Shape functions at nodes are given as


From the figure, we find that l = 3 and h = 2.
Stress Analysis Considering Triangular Element
From structural mechanics, we know that strain displacement
relations are given by

In matrix form

We know
Substituting these values
Note that we have considered two degree of freedoms ui
and vi at each node.

Substituting the above values


In short this can be denoted as
…………………Eq (1)
From structural mechanics, we know that for plane stress
conditions

In matrix form, these equations can be written as


Plane stress condition can be
visualized as thin plate with
stresses acting along its plane.
There is no stress acting
perpendicular to the plane

Plain strain can be


idealized as long wire with
stresses acting
perpendicular to its length.
Therefore the strain or
displacement along the
length is zero.
Plane strain condition
For plane strain conditions, we have

..……………..Eq(2)
Considering equations (1) and (2), we can write stiffness matrix [K]e
as

where t = Uniform thickness of the plate


A = Area of the element.

The element matrix equation for a 2-D element can now be


written as

For 2-D problem, thickness t is considered to be constant.

Plane stress conditions are considered when thickness is very


small compared to the size of domain, whereas plane strain
conditions are considered when thickness is very large
compared to the size of domain.
For the triangular element shown in
figure the nodal values of
displacement are :
u1 = 2.0, u2 = 3.0, u3 = 5.0
v1 = 1.0, v2 = 2.0, v3= 3.0
Obtain the displacement (i.e. u,v) of
point P(2, 2) within the element.

Number the nodes as shown

Find u and v for point (2, 2)

we have
Interpolation functions are given by

Now
Evaluate the stiffness matrix for the
CST element shown below.
Coordinates are in mm.
Assume plain stress conditions.

Nodal displacements are given as


u1 = 1 mm ; u2 = 0.5 mm ; u3 = 2 mm
v1= 1 mm ; v2 = 0; v3 = 1 mm
Also find the stresses in the element.

Number the nodes as shown


stiffness matrix K is given by
Six noded triangular element
•Six noded triangular element is taken
when we want to choose a polynomial
function of higher order viz. order 2
for higher accuracy.
•In such an element, three
nodes are located at the three vertices
and three nodes on the edges, usually
at mid points as shown in fig.
• The numbering of the nodes is done
as per the convention followed as
shown in the figure.
Since we have six nodes we must
have six parameters in
approximate function.
Thus, the function will be of the
form
Natural coordinates and coordinates transformation

So far we have used cartesian coordinates x and y to describe


the position of nodes of an element, i.e. we have described
the vertices of a triangular element as (x1,y1),(x2, y2 ) and
(x3,y3).

It becomes tedious when a quadrilateral element is considered


or a polynomial of higher order or a triangular element with 6
nodes is considered for an approximate solution.

In order to simplify this procedure, we convert these


coordinates into natural coordinates.
In one dimensional problems, the natural coordinate is
designated as ξ and its limits in the natural coordinate system is
-1 to 1.

The two coordinate systems are shown in the fig.


Coordinate Transformation

It can be noted that natural coordinate ξ is


dimensionless.
Alternative methods of deriving Shape Functions

(1) Linear Element


At node 2, term (ξ- 1) vanishes

At node 2,

We can write shape functions for a linear clement in natural


coordinate systems as
(2) Quadratic Element
At node 3,
Coordinates of the nodes of a finite element are given by A(4, 0) and B(8, 0).
Find expression of x in terms of ξ when
(i) Third node C is taken at (6, 0)
(ii) Third node C is taken at (5, 0)
Comment on your answer.

Take a quadratic element with 3 nodes.


Comment : When C is taken at midpoint of the element the
transformation between x and ξ is linear, but when C is taken away
from the midpoint the transformation becomes nonlinear.
•Such a transformation is useful to formulate finite elements having
curved edges so that curved structural geometry can be modelled.
Natural Coordinates-Quadrilateral elements

Consider a quadrilateral element ABCD as shown in fig (a).


Let coordinates of vertices A, B, C and D b e (x1,y1), (x2,y2),(x3,
y3) and (x4, y4).

This quadrilateral can be transformed into a rectangle in natural


coordinates i.e. in ξ –η coordinates as shown
Since the element is a quadrilateral element with 4 nodes each
having 2 DOFs, we can write the equations,

In order to determine the coordinates of any point P(x,y)


within the element.
Method of serendipity is used to determine Φj.

We shall adopt a similar procedure and use basic properties of


shape functions viz.
(1) Value of Φi at the node i is 1.
(2) Value of Φi at any node j is 0 where j  i.
i.e. The shape function Φi vanishes at such nodes j where j  i.
Similarly we can find
Compute coordinates of point P in natural
coordinates ξ -η . Hence find values of
shape functions at P and temperature at P.
T1 = 100 °C, T2 = 60 °C, T3 = 50 °C and T4 =
90 °C.

Convert x,y coordinate to local x


and y coordinates.
Take point (2, 2) as (0, 0). Then
the element in local coordinates
will be as shown in the fig
Temperature at P is given by
Calculate the linear interpolation
function for the linear rectangular
elements shown in fig.
On solving
QUADRILATERAL ELEMENT WITH EIGHT NODES

Serendipity elements are the rectangular elements which have


no interior nodes i.e. all nodes lie on the boundary of the
element only.

Consider quadrilateral element with eight nodes as shown in fig.


To determine Φ1 we note that Φ1 vanishes
along lines
Similarly, we can show that
Quadrilateral Elements with Nine Nodes

Find the shape function for two dimensional nine noded


rectangular elements mapped into natural coordinates.
A nine node quadrilateral element is similar to eight node
quadrilateral element but with an additional node at the centre
i.e. at ξ = η = 0 as shown in the figure.
Using serendipity concept, find
the shape functions for the
element shown in the fig.
For Φ1, note that Φ1 vanishes along 2-3, 3-4 at point 5 and 6.

At node 1
For Φ6
Natural coordinates for triangular elements
(1) Each of the natural coordinates varies from 0 to 1.
(2) The three natural coordinates are not independent of each
other but bound by the relation

(3) Value of L1 is 1 at node 1 and 0 at nodes 2 and 3, similarly


value of L2 is 1 at node 2 and 0 at nodes 1 and 3 and value of
L3 is 1 at node 3 and 0 at nodes 1 and 2.
We can, therefore, define shape functions for a triangular
elements with 3 nodes as
Alternatively we can determine shape functions as per our
conventional method.
Consider triangular element with six nodes.
Note the numbering of nodes, three nodes at vertices are
denoted by 1, 2, 3 as in three noded triangular element.
Other three nodes 4, 5, 6 are located at mid points of the
three sides as shown in figure.
Consider a triangular element ABC as shown in fig.
Let coordinates of vertices A, B and C be (x1,y1 ), (x2, y2) and
(x3,y3) respectively. This triangle can be transformed into a
triangular element in the form of right angled isosceles
triangle A'B'C'as shown in fig.
Now since the element is a triangular element with 3 nodes
each having 2 DoF, write the equations
Substituting values of Φ1,Φ2 and Φ3
Expressing ξ and η in terms of α,β and γ
Jacobians for Triangular Element
Jacobian matrix [J] gives a relation between the derivatives in
two different coordinate systems viz. cartesian i.e. x - y and
natural i.e. ξ and η.

Using the relations

We get
Putting i = 1, 2, 3 and using relations
In general when quadratic or higher order triangular elements
are used, the derivatives are obtained by using

natural coordinates Li.

Where
STRESS SNALYSIS USING FOUR NODE QUADRILATERAL ELEMENT
We need to use natural coordinate and coordinate
transformation to simplify the analysis considering
quadrilateral element.
• The nodal displacements are in x-y coordinates only and not in
ξ and η coordinates.
•Also ξ and η axes may not be orthogonal when transformed into
x-y axes.
Hence, we write the strain-displacement relations in x-y
coordinates as given in equation.
Since u and v are functions of x and y, where x and y are functions of
ξ and η after transformation, we can use chain rule for partial
differentiation as

In general, we can write


known as the Jacobian Matrix which gives
a relation between the derivatives in two
different coordinate systems.

we know that
We have the equation

Now, if we write
Substituting the value of J-1 in the equation
We have the strain equation

Which can be written as


We have the equation

Which can be written as


Now, we know that

For a quadrilateral element with four nodes we can write

In matrix form
Differentiating w.r.t. ξ and η , we get
Combining the three
equations, we get
Where,

and
Stress-strain relation may be written as

The element stiffness matrix is given as


Using Gauss Legendre Quadrature for double integration, we
get

where i and j are the number of Gauss points along ξ and η coordinates and
wi and wj are respective weight coefficients.
Gauss Quadrature Method
In Gauss Quadrature method, we convert the given function
in terms of another variable ξ such that the interval a to b is
converted into (-1) to ( + 1).
The new coordinates thus formed are called as natural
coordinates and the conversion of the function is called as
normalization of the function.
Step 1: Convert the variable x to another variable £ such that
the interval a to b is changed to -1 to +1. The relation between x
and ξ can be deduced as shown in fig.

where h is the length of the interval.

Substituting in the equation

Thus we can convert f(x) into F(ξ).


Step 2 : Divide the interval (-1 to +1) into a number of sub
The sub intervals are not of equal length.
The coordinates of the nodes, forming the sub intervals, depend on
the number of nodes.
Number of nodes

where p is the degree of polynomial.


If degree of polynomial p is even, correct r to the next higher
integer.
Values of ξ for different values of r and corresponding weight
coefficients are given in the table.
Step 3 : Carry out the numerical integration and compute the
value of the integral using formula.

The exact value of the integral is obtained when


Solve

Solution

And
Thus the integral reduces to

by Gauss Quadrature formula

As the maximum power of the polynomial is 3, we have, for


exact solution
Gauss Quadrature Method in Two-Dimensions
Gauss Quadrature can also be used to evaluate an integral over a
rectangular or aquadrilateral element.
However, as in one-dimension problems we shall have to convert
the integration function into natural coordinates ξ and η.
Question
Given a quadrilateral element as shown in fig, evaluate the
Jacobian matrix and also find

Number the nodes as shown in fig


Now Jacobian matrix is given by
Evaluate stresses σx , σy and τxy for a rectangular element shown in
fig. at ξ = 0 and η = 0. Assume plain stress conditions. Take E = 2 x
105 N/mm2 and ν = 0.3.
Displacements at vertices are given as
u1 = 0 ; u2 = 0.05 mm ; u3 = 0.15 mm ; u4 = 0
v1 = 0; v2 = 0.075 mm ; v3 = 0.08 mm ; v4 = 0
Compatibility

The first step in Finite Element Analysis is discretization which


means dividing a domain into a number of elements.
In 1- D problems the two adjacent elements have one node
common for both elements as shown in fig.

Deformation or displacement at node 2 considering it on


element number 1 is same as displacement at same node
considering it on element 2.

In other woids, we can say that displacements in 1-D


elements are compatible i.e. there is no separation.
However, in 2-D problems, elements havecommon edges and
hence to provide compatibility, we have to ensure that
displacement at each point on the edge should remain the same
while we consider the edge on element 1 or element 2.
Consider two bilinear rectangular
elements sharing a common edge
2-3 and the nodes 2 and 3.
Now we can show that
compatibility of displacements is
ensured along edge 2-3 as shown
in fig.

Nodes 2 and 3 are common


Hence displacements at nodes 2
and 3 being compatible,
displacement at any intermediate
point p on edge 2-3 is given by

For a rectangular element, for edge 2-3


i.e. up and vp are functions of nodal displacements of the edge 2-3.

Since nodal deflections remain same for the two elements we can
conclude that up and vp will remain the same whether we consider
the edge on element 1 or 2.
Thus, we can say that compatibility of displacements can be
ensured for bilinear rectangular elements.
Convergence Criteria
The solution obtained from Finite Element Analysis is always an
approximate one and not exact.
However, it is possible to reduce error to a minimum acceptable
level.
When this happens, we say that solution converges.

To ensure convergence following criteria have to be fulfilled.


These criteria relate to the polynomial which we assume to be
the approximate solution to the problem in hand

1. Polynomial should be a complete polynomial i.e. it must


contain all the degrees from 0 to the highest order of the
derivative used in the weak form.
This requirement is necessary to capture all possible states
of the actual solution.
2. The polynomial should be continuous over the element and
also differentiable upto the order of the derivatives in the
weak form.
This requirement ensures non-zero coefficients in the
solution.
3. The approximate solution should be interpolation function
of the primary variables at the nodes of the finite element.
This requirement is essential to ensure compatibility of the

solution.
Besides these criteria, convergence can be improved by reducing
the size of the elements i.e. increasing number of elements in
which case we call h-convergence or by increasing the degree of
polynomial in which case we call p-convergence.
Iso-parametric element 
Generally it is very difficult to represent the curved boundaries
by straight edge elements.
A large number of elements may be used to obtain reasonable
resemblance between original body and the assemblage.
In order to overcome this drawback, iso-parametric elements
are used.
If the number of nodes used for defining the geometry is same
as number of nodes used defining the displacements, then it is
known as iso-parametric element.
Subparametric element 
If the number of nodes used for defining the geometry is less than
number of nodes used for defining the displacements, then it is
known as subparametric element.
Super parametric element 
If the number of nodes used for defining the geometry is more
than number of nodes used for defining the displacements, then
it is known as super parametric element.
Patch T est
Patch test is a simple test to ensure that the element which we
have formulated satisfies all the convergence criteria.
Following procedure is followed to carry out the patch test.
1. Assemble a small number of elements to form a patch.
2. See that at least one node, shared by two or more elements, lies

within the patch.


3. Constrain the patch at some nodes so as to prevent rigid
motion.
4. Apply loads to other boundary nodes and compute stresses.
5. If computed stresses are equal to exact stresses for the physical
problem modeled, we can say that the element passes the patch
If accurate results are not obtained, then the accuracy can be increased by refining the
mesh test.
further.

You might also like