Scribd
Upload
28 views12 pages

6th Lecture FEM 2022

The document discusses requirements for parameter functions used in finite element analysis. It describes two primary requirements: compatibility and completeness. Compatibility requires parameter functions and their derivatives to be continuous across element boundaries for C0 and C1-continuous problems. Completeness requires parameter functions to be able to represent constant field variables and derivatives as the element size decreases. The document provides examples of parameter functions that meet these requirements for one-dimensional problems.

Uploaded by

hudasaad
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
28 views12 pages

6th Lecture FEM 2022

The document discusses requirements for parameter functions used in finite element analysis. It describes two primary requirements: compatibility and completeness. Compatibility requires parameter functions and their derivatives to be continuous across element boundaries for C0 and C1-continuous problems. Completeness requires parameter functions to be able to represent constant field variables and derivatives as the element size decreases. The document provides examples of parameter functions that meet these requirements for one-dimensional problems.

Uploaded by

hudasaad
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 12

Restrictions on the Parameter Functions:

The assumed parameter functions must meet two primary requirements:


compatibility and completeness.
Compatibility:
The compatibility requirement may be stated as follows. For CO-continuous
problems, the parameter function itself (not its derivatives) must be continuous along
the boundaries of the element. For Cl-continuous problems, the parameter function
and its first derivative must be continuous, and not necessarily zero, along the
boundaries of the element. Note that for the CO-continuous problem, such as the
uniaxial stress member or one-dimensional heat transfer in a fin, the field variable
itself is continuous at the interface between any two elements (in this case, at each
node). On the other hand, in C1-continuous problems, the parameter function itself
and its first derivative must be continuous on the interface between any two
elements.
Elements that obey the compatibility requirement are said to be conforming, and
those that do not obey this requirement are said to be nonconforming.

One-dimensional clement with (a) C0-continuous parameter functions and (b)


C1-continuous parameter functions. In (a) only the parameter function itself is
continuous at each element interface; in (b) the parameter function and its slope are
continuous at each element interface.
Completeness:
For C0-continuous problems, the completeness requirement may be stated as
follows. The parameter function must be capable of representing both a constant
value of the field variable and constant first partial derivatives as the element size
decreases to a point. For Cl-continuous problems, the parameter function must be
capable of representing both a constant value of the field variable and constant first
and second partial derivatives as the element size decreases to a point. This may be
generalized as follows. For Cn-continuous problems, the parameter function must
be capable of yielding a constant value of the field variable as well as constant
partial derivatives of up to order n + 1 as the element size decreases to a point.
Let us examine what this requirement implies for CO-continuous problems in
stress analysis. For the two-node lineal element, we have been assuming a function
of the form

Note that if C2 is zero, we have u = C1 which is a constant as required. This is


referred to as the rigid body mode since the body should be able to undergo a rigid
body displacement without straining (when C2 takes on a value of zero). In ddition,
du/dx = C2' which is also a constant as required (regardless of the value of (C1).
Since the derivative of the displacement is the strain, this second condition requires
the displacement function to allow a constant strain in the element. This notion is
readily extended to two- and three-dimensional analyses.
Order of Numerical Integration:
Evaluate the integral
by using Gauss-Legendre quadrature.
Solution: since the integrand is a polynomial, we can determine the order of the
quadrature that will result in an exact evaluation from (2n - 1 = 4), where the order
of the polynomial is 4, then n=2.5 which must be rounded to give n = 3

It can readily be shown that the exact result is 2/5


In two dimentional case we have
In three dimensions
Evaluate the integral
the integral is of order 2 in r and of order 5 in s. Therefore, in the r direction we have 2n - 1 = 2
or n = Jh and in the s direction 2m - 1= 5 or m = 3. Hence we take n = 2 (3/2 rounds to 2) and m
= 3, or two Gauss points in the r direction, and three points in the s direction. It is instructive to
show the six sampling points (i.e., 2 x 3) on a typical element as shown in Figure.

Let us denote the integrand by f(r,s), where

This compares favorably with the exact evaluation of I = 4/15


The element nodal force vector from surface tractions,
the elemental area dS around the boundary may be expressed as: dS = t dC
where t is the element thickness and dC is the infinitesimal arc length, dC is given by:

since we have x = x(r,s) and y = y(r,s), we may write the differentials dx and dy as:
and
where rand s are the serendipity coordinates. The boundary integrals need to be evaluated around
the element boundary.
Let us derive the appropriate form of the expression
for dC by restricting the development to the legs of
the element over which the serendipity coordinate s is
constant; i.e., s = ± 1 on these faces. Using the following Equations:
and

on faces where s = ± 1
The elemental surface area dS or t dC on sides of constant s becomes:

In a similar fashion, it can be shown that on faces of constant r (i.e., faces 1-2 and 3-4), the
elemental surface area dS is given by:

The degree of distortion that is possible before the mapping breaks down (and is no longer one-
to-one) is now given. For a linear-order quadrilateral element, a one-to-one mapping is assured if
the maximum angle formed by any two sides of the quadrilateral is less than 1800 as shown in
Figure. For a quadratic-order quadrilateral element, not only must the same angle condition be
satisfied but also the midside nodes must be in the middle one-third of the distorted sides.

Serendipity family

Four elements of the serendipity family: (a) linear, (b) quadratic, (c) cubic, and (d) quartic.
To achieve the shape function for the first element it is obvious that a product of
linear Lagrangian polynomials may again be used. Indeed this element is identical to
the Lagrangian one with n = 1 and again all the shape functions may be written as
one expression:

“Quadratic” element

The above functions were originally derived by inspection, and progression to yet higher
members is difficult and requires some ingenuity

Systematic generation of 8 nodes “serendipity” shape functions.


Rigid translation l=0.0 Rigid rotation l=0.0
Rigid body mods

displacement mode.
𝟒

𝒖 = ∑ 𝑵𝒊 𝒖𝒊 + 𝑷𝟏 𝜶𝟏 + 𝑷𝟐 𝜶𝟐
𝒊=𝟏

𝒗 = ∑ 𝑵𝒊 𝒗𝒊 + 𝑷𝟏 𝜶𝟑 + 𝑷𝟐 𝜶𝟒 (8.50)
𝒊=𝟏

𝑷𝟏 = (𝟏 − 𝒓𝟐 ).

𝑷𝟐 = (𝟏 − 𝒔𝟐 )
The element ’ s stiffness matrix can be partitioned into parts corresponding to {U(e)} and parts
corresponding to the additional parameters {a(e)}:

We can solve for {a(e)} to obtain:

You might also like