CH 5 Tutorial 2014

CE4257 Chapter 5 - 2014 (C G Koh)
p. 1
NATIONAL UNIVERSITY OF SINGAPORE

DEPARTMENT OF CIVIL ENGINEERING
CE4257 Chapter 5 INTEGRAL METHODS

Questions 2 and 5 will be graded (25 + 25 marks)
Submit to CEE Dept Office (E1A-07-03) by 5:30 pm Oct 21st (Tues).
1. A bar of uniform axial rigidity EA and length L is subjected to a distributed
axial load which varies linearly, i.e. T(x)= Cx (force per unit length). It is
fixed at the left end and free at the right end.
T(x) = Cx
The governing equation is given by

EA
d 2u
dx 2
T 0
with the boundary conditions

u0
at x 0
and
EA
du
0
dx
at x L
(a) Derive the above equation and boundary conditions based on structural
mechanics.
(b) Derive also the exact solution by integrating the governing equation.
(c) The following quadratic polynomial is used as the trial solution:
a0 a1 x a2 x 2 . Make use of the boundary conditions to reduce the
number of unknown coefficients and simply the trial function.
(d) Use the following methods to obtain an approximate solution for the axial
displacement and compare with the exact solution graphically.
(i) Collocation method (use midpoint as the collocation point)
(ii) Least square method
(iii) Galerkin method (use the trial function in part c as the weighting
function)
p. 2
2. An one-dimensional engineering problem is governed by the following

differential equation with boundary conditions u ( 1) u (1) 0 .
x 2 u" ( x ) 12 u( x ) 12 where ()' d() / dx and 1 x 1 .

Let the trial solution be u( x ) A B x C x 2 . Determine the three
coefficients A, B and C, by using the appropriate boundary condition(s) and
the following numerical methods.
(i)
(ii)
(iii)
(iv)
Subdomain method based on x=0 to x=1

Collocation method based on x=0
Least square method
Garlerkins method
Plot the above solutions to compare with the exact solution u = x4-1.
3. Consider a tapered bar subjected to an end load F as follows. The right

end is fixed.
E = constant; A is linear function of x
x
F
x= 5
x= 15
Figure 1
The governing equation is written as
EA( x )
d 2u
dx 2
(1)
(a) Adopt a two-node rod element with a linear displacement function, i.e.
u = N1u1 + N2u2
Use Galerkin method to formulate the finite element matrices (or
equations) for this problem. In other words, for a typical finite element
of length L, derive the element matrices such that
p. 3
k e ue f e
(b)
Divide the tapered bar into three finite elements of equal length. Use
the element matrices that you have derived to obtain the axial
displacement for the following case.
F = 100, A(x) = 2.5+0.5x, E = 200 x 103
Note that the origin of x is offset from the bar. The properties are
assumed to be constant in a finite element. Use the average value of
EA (across the element length) in each finite element, but the EA value
may vary from element to element.
4.
The one-dimensional equation for a convection-diffusion problem can be
written in terms of unknown C, which stands for concentration of a substance (e.g.
salt), as follows.
dC( x ) d
V (x)
dx
dx
dC ( x )
D( x ) dx KC( x ) m
x [ L, L]
where V is velocity, D diffusivity, K reaction rate, and m an external source

of mass. The boundary conditions are
dC
b (a non - zero constant) at x = L and C=0 at x = L .
dx
(a)
(b)
(c)
Construct a weak formulation based on Garlerkin method. State any

requirement(s) for the weighting function.
Applying Galerkin method, establish a system of equations K a = P and
derive the coefficients in terms of shape functions Ni and Nj and system
parameters.
Is matrix K symmetric?
5.
The following differential equation is to be solved by Galerkin method:
3 x u( x) u "( x) 2 3x3 0
(a)
p. 4
for
0 x 1
Divide the domain into several elements. For a typical element, let
n
u ( x ) Nj ( x ) a j where n is the number of trial functions to be used in

j1
each element. Derive the symmetric system of n equations k u f at

the element level (integration by part is needed where appropriate to
achieve symmetry). In this part, do not impose any essential boundary
condition and do not assume any specific function for trial function.
(b)
Now two equal-length elements are used. Adopt two linear trial functions
N1(x) and N2(x) for each element and let a represent the solutions for u at
the nodes (as in typical finite element analysis). For each typical element,
N1(x) =1-x/L and N2(x)=x/L, where x is the local coordinate defined from
the left node of the element and L is the element length. Assemble the
element solutions to form the global system of equations: K u F . Use
exact integration where possible.
(c)
Impose the following boundary conditions and solve the equations to

obtain the solution for the whole domain.
(A)
u 0 at x 0
and
(B)
u 1 at x 1 .

CH 5 Tutorial 2014

Uploaded by

Copyright:

Available Formats

CH 5 Tutorial 2014

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

CH 5 Tutorial 2014

Uploaded by

Copyright:

Available Formats

CE4257 Chapter 5 - 2014 (C G Koh)

NATIONAL UNIVERSITY OF SINGAPORE

CE4257 Chapter 5 INTEGRAL METHODS

The governing equation is given by

with the boundary conditions

CE4257 Chapter 5 - 2014 (C G Koh)

2. An one-dimensional engineering problem is governed by the following

x 2 u" ( x ) 12 u( x ) 12 where ()' d() / dx and 1 x 1 .

Subdomain method based on x=0 to x=1

3. Consider a tapered bar subjected to an end load F as follows. The right

CE4257 Chapter 5 - 2014 (C G Koh)

where V is velocity, D diffusivity, K reaction rate, and m an external source

Construct a weak formulation based on Garlerkin method. State any

CE4257 Chapter 5 - 2014 (C G Koh)

The following differential equation is to be solved by Galerkin method:

u ( x ) Nj ( x ) a j where n is the number of trial functions to be used in

each element. Derive the symmetric system of n equations k u f at

Impose the following boundary conditions and solve the equations to

You might also like