Scribd
Upload
76 views7 pages

Lecture 37: Galerkin FEM Galerkin FEM

i) The domain is discretized into elements defined by nodes. ii) The global solution is approximated as a combination of local interpolating polynomials within each element. iii) The Galerkin method is used to derive the element equations by setting the weighted integral of the residual to zero and using the shape functions as the weighting functions. This results in a system of equations that can be solved for the nodal values.

Uploaded by

Tapas Kumar
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
76 views7 pages

Lecture 37: Galerkin FEM Galerkin FEM

i) The domain is discretized into elements defined by nodes. ii) The global solution is approximated as a combination of local interpolating polynomials within each element. iii) The Galerkin method is used to derive the element equations by setting the weighted integral of the residual to zero and using the shape functions as the weighting functions. This results in a system of equations that can be solved for the nodal values.

Uploaded by

Tapas Kumar
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/ 7

CE 601: Numerical Methods

Lecture 37

Galerkin FEM

Course Coordinator:
Course Coordinator:
Dr. Suresh A. Kartha,
Associate Professor,
Department of Civil Engineering,
IIT Guwahati. 
In the last class, we started the discussion on Galerkin FEM.
We were considering the boundary-value ODE
d2y
2
+ Qy = F ; with appropriate BCs
dx
I a nutshell,
In t h ll the
th procedure
d is
i
i) Discretise the global solution domain D( x) into elements. There
can be I − nodes and I − 1 elements.
ii) Approximate y ( x) ≈ y ( x) = combinations of interpolating functions.
d2y
e R( x) = 2 + Qy − F
iii) Substitute y ( x) in the expression for residual ii.e.
dx
iv) The weighted integral is to be evaluated I = ∫ W ( x) R ( x)dx
v) Determine element equations for each element.
vi) Assemble element equations to form system equations.
vii)
ii) Adjust
Adj t the
th system
t equations
ti b considering
by id i boundary
b d conditions.
diti
viii) Solve the adjusted system equation for the nodal values yi .
1) The Domain Discretisation
The domain D( x) discretised: I → nodes and I − 1 → elements.
d2y
The global solution of 2
+ Qy = F is y ( x).
dx
Let us approximate this global solution
y ( x) ≈ y ( x) = y (1) ( x) + y (2) ( x) + + y (i ) ( x) + + y ( I −1) ( x)
This means that y ( x) is sum of series of local interpolating polynomials
y (i ) ( x); i = 1, 2,3,… , I − 1. These local interpolating polynomials are valid
only within the element ' i ', elsewhere it is zero.

2) We write y ( i ) ( x) = yi N i( i ) ( x) + yi N ii+)1 ( x)
x − xi +1 ⎫
Recall N i(i ) ( x) = −
∆x (i ) ⎪⎪
⎬ Linear Shape functions
x − xi ⎪
N i +1 ( x) = − ( i )
(i )

∆x ⎪⎭
You can also use higher order shape functions as well.
You can see that the shape function N i( i ) ( x ) is valid only for element ' i ' and
elsewhere it is zero.

3) T
To use Galerkin
G l ki method
h d
b b
I ( y ( x)) = ∫ W j ( x) R( x)dx = ∫ W j ( x) [ y "+ Qy − F ] dx = 0
a a

Now we have seen in the last class that


b
I ( y ( x)) = ∫ ⎣⎡ − y 'W j '+ QyW
Q W j − FW j ⎦⎤ddx + yb 'W j (b) − ya 'W j (a ) = 0
a

i.e. I ( y ( x)) = I (1) ( y ( x)) + I (2) ( y ( x)) + + I ( I −1) ( y ( x))


+ yb 'WI (b) − ya 'WI (a ) = 0
xi +1

where I (i ) ( y ( x)) = ∫ ⎡⎣− y 'W


xi
j W j ⎤⎦dx
'+ QyW j − FW

y ( i ) ( x) is the interpolating polynomial in the element i.


x − xi +1 x − xi
y ( x) = yi N ( x) + yi +1 N ( x) = −
(i ) (i )
yi + − (i ) yi +1
(i )
i +1
∆x ∆x
i (i )

As suggested earlier Galerkin recommended use of shape functions


as weighing functions.
A N i( i ) ( x) = 00.0
As 0 ffor x > xi +1 & x < xi
N i(+i1) ( x) = 0.0 for x > xi +1 & x < xi
W can say that
We th t the
th total
t t l integral
i t i W j = N i(i )
l using
xi +1
⎡ (i ) ⎤
b
d
I ( y ( x)) = ∫ ( )dx = ∫ ⎢ − y ' ( N i
(i )
( x )) + QyN i
(i )
− FN i ⎥dx = 0
a xi ⎣ dx ⎦
∵ N i( i ) ( x) = 0 everywhere else other than xi < x < xi +1
Al N i( i ) (a ) = 0 and
Also d N i( i ) (b) = 0
Similarly for W j = N i(+i1) ( x)
xi +1
⎡ d (i ) ⎤
I ( y ( x)) = ∫ − + − i +1 ⎥dx = 0
(i ) (i )
⎢ y ' ( N i +1 ( x )) QyN i +1 FN
xi ⎣ dx ⎦
We now get two element equations for the element ' i '.
x − x dN (i )
1
Now since, N i ( x) = −
(i ) i +1
∴ i
= − (i )
∆x (i )
dx ∆x
dN i(+i1) 1
Smilarly, = (i ) .
dx ∆x
∴ In the two element equations:
xi +1
⎡ d (i ) ⎤
I ( y ( x)) = ∫ − + − i ⎥dx = 0
(i ) (i )
⎢⎣ y ' ( N i ( x )) QyN i FN
xi
dx ⎦
⎡ ⎛ x − xi +1 ⎞ ⎛ x − xi +1 ⎞ ⎤
xi +1
⎛ 1 ⎞
= ∫
xi
⎢ − y '× ⎜ − ∆x ( i ) ⎟ + Qy × ⎜ − ∆x ( i ) ⎟ − F × ⎜ − ∆x ( i ) ⎟ ⎥dx = 0
⎣ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠⎦

1 ⎡ xi+1 xi +1 xi +1
⎤
i.e. ⎢ ∫ y 'dx − ∫ Qy ( x − xi +1 )dx + ∫ F ( x − xi +1 )dx ⎥ = 0
∆x (i ) ⎢⎣ xi xi xi ⎥⎦
( x − xi ) ( x − xi ) ⎤
xi +1
⎡ 1
Similarly, I ( y ( x)) = ∫
xi
⎢ − y '× ∆x ( i ) + Qy × ∆x ( i ) − F × ∆x ( i ) ⎥dx = 0
⎣ ⎦

1 ⎡ xi+1 xi +1 xi +1
⎤
or, ⎢ ∫ − y 'dx + ∫ Qy ( x − xi )dx − ∫ F ( x − xi )dx ⎥ = 0
∆x ( i )
⎢⎣ xi xi xi ⎥⎦
Using Q( x) and F ( x) as average values for each element
i Q (i ) ( x) andd F ( i ) ( x),
i.e. ) we gett
⎡ 1 ∆x (i )
⎤ ⎡ 1 ∆ x (i )
⎤ ∆x (i )
− yi ⎢ (i ) − Q (i ) +
⎥ i +1 ⎢ ( i )
y + Q (i )
⎥ − F (i )
=0
⎣ ∆x 3 ⎦ ⎣ ∆x 6 ⎦ 2
⎡ 1 ( i ) ∆x
(i )
⎤ ⎡ 1 ( i ) ∆x
(i )
⎤ ( i ) ∆x
(i )
yi ⎢ (i ) + Q ⎥ − yi +1 ⎢ ( i ) − Q ⎥−F =0
⎣ ∆x 6 ⎦ ⎣ ∆x 3 ⎦ 2
These are the two element equations in algebraic form.
Assemble these element equations
eq ations for all the elements starting from
element 1.

You might also like