CEC-529

FEM IN
GEOTECHNICAL
ENGINEERING
Instructor- Dr. AKANKSHA TYAGI
[email protected]
Fundamentals of Discretization
Local and Global Numbering
• Once the domain of the problem is
discretized by elements, a unique element
number identifies each element and a
unique node number identifies each node
in the domain.
• Specific sequence (usually
counterclockwise) is based on the node
numbering at the element level
• Nodes are numbered within each element,
and are called local node numbers.
• The unique node numbering within the
entire domain is called global node
numbering.

Field variables are calculated at every node from governing equation. Values
between the nodes and within the elements are calculated using interpolation
functions, which are sometimes called shape or base functions.

2
Fundamentals of Discretization

Summary of the common

element types utilized in a
finite element analysis (FEA)

3
Element Approximation Functions
Variation of the field variable, φ(e),
over an element is approximated by
an appropriate choice of function
• Selection of these functions is the core of
the finite element method.
• Approximation functions should be reliable
that as the mesh becomes more refined,
the approximate solution should converge
to the exact solution monotonically.
• Oscillatory convergence is unreliable
because it is possible to observe an increase
in error with the refined mesh.
• Common approximation functions are
usually polynomials as their differentiation
and integration are rather straightforward
compared to other functions.
• To achieve a monotonically convergent
solution, the polynomials chosen as
approximation functions must satisfy four
requirements.
4
Approximate Functions- Compatibility
& Completeness
Requirement 1 Continuous behavior of the approximation function within
the element—no kinks or jumps
Graph for a function that’s smooth without any holes, jumps, or
asymptotes is called continuous.

https://www.dummies.com/education/math/calculus/how-to-determine-whether-a-function-is-continuous/
 Requirement 2 Compatibility along the common nodes, boundaries or
surfaces between adjacent elements—no gaps between elements
Along element boundaries, the field variable and its partial derivatives
up to one order less than the highest-order derivative appearing in the
integral formulation of the element equations must be continuous.
• The element is referred to as C0 continuous when only the field variable
(none of its derivatives) maintains continuity along its boundary.
• If the field variable and its rth derivative maintain continuity, the element is
Cr continuous

For a truss/rod element, the displacement must be continuous across

element boundaries, but none of the displacement derivatives is required to
be continuous across such boundaries.
The truss element is a constant strain element, so the first derivative is, in
general, discontinuous at the boundaries

The beam element formulation, includes the second derivative of

displacement, and compatibility requires continuity of both the displacement
and the slope (first derivative) at the element boundaries

6
Requirement 3 Completeness, permitting rigid body motion of the
element and ensuring (constant) variation of φ and its derivatives within
the element.
The completeness requirement ensures that a displacement field within a
structural element can take on a constant value, representing rigid body motion,

• Consider a cantilever beam under a

concentrated load in the middle
• Deformation occurs only to the left of
the load due to loading
• Section of the beam to the right of
the load experiences only rigid-body
translations and rotations (constant
displacements and zero strain), i.e.,
no stresses and strains occur.
• Element approximation functions
must permit such behavior.
• Complete polynomials satisfy these
requirements

7
Complete polynomials

8
 The order of the polynomial as an approximation function is
dictated by the total number of nodes in an element, i.e.,

Number of coefficients, αi, in approximation function = Number of nodes in the element

9
Requirement 4 Geometric isotropy for the same behavior in each direction.
Mathematical function satisfies geometric isotropy if the functional form
does not change under a translation or rotation of coordinates. – property
exhibited by complete polynomials
If the required degree of completeness does not provide a number of terms equal
to the number of nodes, then this requirement can be satisfied by disregarding the
non-symmetrical terms.

Thus, Incomplete polynomials also exhibit geometric isotropy if the incomplete

polynomial is symmetric.

Pascal triangle for polynomials in two dimensions

10
Coordinate Systems

Generalized Coordinates

• The coefficients of the approximation functions, αi, are referred to as

the generalized coordinates.

• Not identified with particular nodes.

• Independent parameters that specify the magnitude of the prescribed

distribution of the field variable.

• No direct physical interpretation, but rather are linear combinations of

the physical nodal degrees of freedom.

11
Coordinate Systems

Global Coordinates

• Convenient for specifying the location of each node, the orientation of

each element, and the boundary conditions and loads for the entire
domain.

• Solution to the field variable is generally represented with respect to the

global coordinates.

• Approximation functions described in terms of the global coordinates

are not convenient to use in the evaluation of integrals necessary for the
construction of the element matrix
12
Coordinate Systems

Local Coordinates
• A local coordinate system whose origin is located within the element is
introduced in order to simplify the algebraic manipulations in the derivation of
the element matrix.
• Use of natural coordinates in expressing the approximation functions is
particularly advantageous because special integration formulas can often be
employed to evaluate the integrals in the element matrix.

13
 Shape Functions
The field variable φe (x), can be expressed within the element through the use of
its nodal values φi (i=1,m) in the form,

(𝑒)
 ( 𝑥 )= 𝑁 1  1+ 𝑁 2  2 +…+ 𝑁 𝑚 − 1 𝑚 − 1+ 𝑁 𝑚  𝑚

( 𝑒)
( 𝑥 )= 𝑵 𝑻 
where = { 𝑁 1 𝑁3 … 𝑁𝑚 }
𝑇
𝑁 𝑁2

= { 1 }
𝑇
 2 3 … 𝑚

in which Ni (i=1,m) are referred to as shape functions. Also known as interpolation

functions
• These functions are associated with node i and must have a unit value at node i
and a zero value at all other nodes.
• They must have the same degree of polynomial variation as in the element
14
approximation function.
The explicit form of the shape functions can be determined by solving for
the generalized coordinates, αi, in terms of the nodal coordinates, xi, and
nodal values, φ (i=1,2,….m) and rearranging the resulting expressions in
the form
At each node, the field variable φ(e)(x) is evaluated as

In matrix form

15
 Hence,

Properties of Shape Functions

1. Ni = 1 at node i and Ni=0 at all other nodes

16
Interpolation functions- One dimensional Element

Polynomial forms

The displacement field is expressed via the first-degree polynomial

17
If u1 = u2, the element displacement field corresponds to rigid body motion and
no straining of the element occurs.

First derivative of Equation with respect to x, i.e. axial strain yields a constant value.
Hence, the truss element satisfies the completeness requirement, since both
displacement and strain can take on constant values regardless of element size

Element satisfies the compatibility requirement automatically, since only displacement

is involved, and displacement compatibility is enforced at the nodal connections via
the system assembly procedure.

Inclusion of the constant term a0 ensures the possibility of rigid body motion, while
the first-order term provides for a constant first derivative.

Further, only two terms can be included in the representation, as only two boundary
conditions have to be satisfied, corresponding to the two element degrees of freedom.

Conversely, if the linear term were to be replaced by a quadratic term α2x2, the
coefficients could still be obtained to mathematically satisfy the nodal displacement
conditions, but constant first derivative (other than a value of zero) could not be 18
obtained under any circumstances
 Interpolation functions- Quadratic 1-D
element
Equation 1

For convenience, x1=0 and x2=L

Substituting values of a0, a1 and a2 in Equation 1

19
Interpolation functions- Quadratic 1-D
element

20
 Interpolation functions- Triangular
Aelement
general 3-node triangular
element referred to global
coordinates.

21
A= area of triangle element
22
 Constant strain triangle

3 3
Graphical displays of shape
functions for (a) u1 and (b) u2

2 2
1 1 L2=1
L1=1

• It is possible for the field variable to take on a constant value, as per the
completeness requirement, and that the first partial derivatives with respect
to the independent variables x and y are constants.
• The latter shows that the gradients of the field variable are constant in both
coordinate directions.
• For a planar structural element, this results in constant strain components.
• In fact, in structural applications, the three-node triangular element is
commonly known as a constant strain triangle (CST)

23
• When expressed in Cartesian coordinates, the interpolation functions for the
triangular element are algebraically complex !
• Further, the integrations required to obtain element characteristic matrices are
cumbersome.
• Considerable simplification of the interpolation functions as well as the
subsequently required integration is obtained via the use of area coordinates.

The area coordinates of P are defined as

P is not a node

We will learn the formulation using area coordinates in upcoming slides

24
Coordinate Systems

Natural Coordinates
• A local coordinate system that permits the specification of a point within
the element by a dimensionless parameter whose absolute magnitude
never exceeds unity

• Natural coordinates are dimensionless

• Defined with respect to the element rather than with reference to the
global coordinates.

• These are functions of the global coordinates in which the element is

defined

25
Natural Coordinates &
Isoparametric
• We adopt natural (normalized)Elements
coordinates in FE formulation:
- to provide general forms of interpolation functions
- to facilitate numerical integration
- to standardize computer implementation of FEM
• The idea is to standardize the formulation of finite element properties
via mapping the actual element coordinates to the natural coordinates.
• This makes the FE procedure “problem and coordinate independent”,
facilitating FE operations to be standardized and hence suitable for
programming and implementation in computers that are, nowadays,
available to any practicing engineers.
• The advantages of applying natural coordinates in formulating 1-D
elements are not obvious but the approach is beneficial and essential
for a systematic formulation of 2-D and 3-D elements.

26
Natural Coordinates &
Isoparametric Elements

• Iso-parametric concept - when element geometry and displacement

components are described by the same set of interpolation functions.

• The shape (or geometry) and the field variable of these elements are
described by the same interpolation functions of the same order.

• The modeling of domains involving curved boundaries by using straight-

sided elements may not provide satisfactory results.

• However, the family of elements known as “isoparametric elements” is

suitable for such boundaries

27
Finding Natural Coordinates
Coordinate Mapping
Linear Isoparametric Rod element

1
𝑥= ( 1+η ) ( 𝑥2 − 𝑥 ¿ ¿ 1)+ 𝑥 1 ¿
2

By rearranging terms 𝟏 𝟏
𝒙= ( 𝟏 −𝜼 ) 𝒙𝟏 + ( 𝟏+ 𝜼) 𝒙 𝟐
𝟐 𝟐
N1(η) N2(η)

28
Finding Natural Coordinates
Coordinate Mapping x=0 x=L/2 x=L
Quadratic Isoparametric Rod element
η=-1 η=0 η=1

Assuming both sets of geometric coordinates follow a linear mapping, then

the relationship between η and x is derived earlier as follows:
𝐿
𝑥= (1+ η )
2

Interpolation functions

29
Finding Natural Coordinates

30
31
2. Natural Coordinates in Two
Dimensions
Area Coordinates Location of a point P, denoted by (x,y) (global
coordinates), on the element can be expressed as linear
combinations of the global nodal coordinates, (x1,y1),
(x2,y2) and (x3,y3) and the area coordinates, ξ1, ξ2 and ξ3, as

ξ1=A1/A, ξ2 = A2/A and ξ3=A3/A

A is the area of triangle

ξ1 + ξ2 + ξ3 =1 & ξ1, ξ2 and ξ3 are not independent of each other, thus,

In matrix form,

32
ξi(xj) = δij
where δij =1 for i =j and δij =0 for i≠j

33
References
• Erdogan Madenci and Ibrahim Guven, The finite element method and
Applications in Engineering Using ANSYS, Second Edition, Springer-
2015.
• David V. Hutton, Fundamentals of Finite Element Analysis, The McGraw-
Hill Companies, 2004.
• Linear Finite Element Analysis-Lecture notes by Professor Somsak
Swaddiwudhipong, National University of Singapore.

34