Professional Documents
Culture Documents
Nov. 8, 2011
Marks : 20
(a) Obtain the MoM solution for f(x) employing Galerkins approach and suitable basis
functions. Using this solution determine the expression for residual R g (x) and the
1
(b) Obtain the MoM solution for f(x) employing point matching and same basis functions
as in (a) above. Using this solution determine the expression for residual R p (x) and the
1
(c) Solve the differential equation by integrating it and determine the exact solution.
Compare with the solutions obtained using point matching and Galerkins methods.
(d) Comment on the errors g and p .
6,6,6,2
Solution
Since the operator is second-order differential type, the sub-domain basis should be
triangular or higher-order. We shall use triangular basis functions. Since these basis
occupy two consecutive sub-domains, only two basis functions (N = 2) can be
accomodated in 3 segments. This is shown in Fig. 1.
x1
x2
d2 f
+1+ x2
dx 2
1/5
f = n fn
(9.95)
n =1
where the constants n are unknown, and needs to be determined. The n are found
such that the residual R( x) is forced to zero in an average sense, that is, the weighted
integral of the residual is set to zero,
(9.96)
w( x) R( x)dx = 0
where w( x) is called the weight function. In order to determine the constants n
uniquely, we should have N simultaneous equations. For this, we choose N weight
functions wm ( x) , m = 1,2,3,N, and form the weighted integrals as
wm ( x) R( x)dx = 0
m = 1,2,3,N
(9.97)
wm ( x) is also called test function in MoM language. The solution of the resulting set of
equations determines n and therefore f ( x) . Using triangular basis we write
f n = T ( x xn )
n = 1,2
where xn is the mid-point of the nth basis function. The use of these basis functions in
(9.94) gives the following residual
N
d2
R ( x) = 1 + x 2 + n 2 T ( x xn )
n =1
dx
1 2
n [ ( x xn 1 ) 2 ( x xn ) + ( x xn +1 ) ]
x n =1
The weighted residual (9.97) now becomes,
1
1
1 2
n wm ( x) [ ( x xn 1 ) 2 ( x xn ) + ( x xn +1 ) ]dx = wm ( x)(1 + x 2 )dx
x n =1 0
0
= 1 + x2 +
(9.99)
m = 1,2
(9.100)
1 2
n T ( x x2 ) [ ( x xn1 ) 2 ( x xn ) + ( x xn+1 )]dx = T ( x x2 )(1 + x 2 )dx
x n =1 1/3
1/ 3
or
x2
2
1
1 2 = T ( x x1 )(1 + x 2 )dx
x
x
0
x
3
1
2
1 + 2 = T ( x x2 )(1 + x 2 )dx
x
x
x1
where x1 = 1/ 3, x2 = 2 / 3, x3 = 1, x = 1/ 3 .
2/5
x xm 1
x x
m m 1
x x
T ( x xm ) = m +1
xm +1 xm
0
for xm-1 x xm
for xm x xm +1
elsewhere
so that
for 0 x x1
x x
T ( x x1 ) = 2
for x1 x x2
x
elsewhere
0
x x1
for x1 x x2
1 x
for x2 x x3
T ( x x2 ) =
x
elsewhere
0
Integration gives
2/3
2
T ( x x2 )(1 + x )dx =
1/ 3
x23 0.1626
1 x22 x24 x12 x14 x32 x34
x
x
x
2
+
+
+
+
+
+
+
(
)
=
1
3 2
3
x 2 4 2 12 2 12
x
or
2
1
61 1
1 2 =
486 x
x
x
1
2
0.1626
1 + 2 =
x
x
x
or
1 = 0.1378, 2 = 0.1502
or
N
f g ( x) = n f n ( x) = 0.1378T ( x x1 ) + 0.1502T ( x x2 )
n =1
3/5
1 2
n [ ( x xn 1 ) 2 ( x xn ) + ( x xn +1 ) ]
x n =1
1
1 ( ( x x0 ) 2 ( x x1 ) + ( x x2 ) ) + 2 ( ( x x1 ) 2 ( x x2 ) + ( x x3 ) )
x
1
0.1378 ( ( x x0 ) 2 ( x x1 ) + ( x x2 ) ) + 0.1502 ( ( x x1 ) 2 ( x x2 ) + ( x x3 ) )
= 1 + x2 +
x
= 1 + x2 +
g = Rg ( x)dx
0
1
x +
2
2
1 2
lim
(
)
(
)
2
(
)
(
)
lim
x
+
x
dx
=
( x x2 )(1 + x 2 )dx
n
n 1
n
n +1 ]
2 [
0
0
x n =1
x2
x2
or
2
1
1 ( x x1 ) 2 ( x x1 ) = 1 + x12
x
x
1
2
1 ( x x2 ) + 2 ( x x2 ) = 1 + x22
x
x
where x1 = 1/ 3, x2 = 2 / 3, x3 = 1, x = 1/ 3
x2 x4
+ + c1 x + c2
2 12
x2 x4 7
+ x
2 12 12
4/5
0.18
0.16
0.14
0.12
f(x)
0.1
0.08
0.06
0.04
0.02
0
0.1
0.2
0.3
0.4
0.5
x
0.6
0.7
0.8
Comparing with
f g ( x) = 0.1378T ( x x1 ) + 0.1502T ( x x2 )
gives
x
fe(x)
fg(x)
0.1502
1
5/5
0.9