Rayleigh-Ritz Method Overview
Rayleigh-Ritz Method Overview
• Implementation:
n
ux, y, z qi fi x, y, z
i1
n
Let vx, y, z ri gi x, y, z
i 1
n
w x, y, z s i hi x, y, z
i1
i, ri and si are called the generalized coordinates and are the unknown coeff
i(x,y,z), gi(x,y,z) and hi(x,y,z) are user-chosen arbitrary admissible basis
t is an admissible function?
- must have a sufficient number of derivatives
- must satisfy the essential boundary conditions
2. 1
MEEM 5150
Rayleigh-Ritz Method
Parenthesis : what is an essential boundary condition?
Examples
f(x) P
1) axially loaded bar x
u0 0 ?
BC :
EAu L P ?
2. 2
MEEM 5150
Rayleigh-Ritz Method
p(x)
2) beam bending
V
M x
2. 3
MEEM 5150
Rayleigh-Ritz Method
Substitute the approximate displacement field into the total
potential energy to get
q1 ,..., qn , r1 ,..., rn ,s 1 ,..., sn
0 E, A L
2)
z
3L/4 P
x
0 L
E, I
2. 5
MEEM 5150
Application 1 - Exact Solution
0.5
0.45
= kL/EA = 0
0.4
0.35
= 0.5
E Au x 0.3
2
po L 0.25 = 1
0.2
0.15
= 10
0.1
= 10,000
0.05
0
0 0.2 0.4 0.6 0.8 1
x/L
2. 6
MEEM 5150
Application 1 - First attempt
0.5
0.45
Exact
0.4 = 0
0.35 RRM
0.3
E Au x
2 0.25
po L = 1
0.2
0.15
0.1
0.05
0
0 0.2 0.4 0.6 0.8 1
x/L
2. 7
MEEM 5150
Application 1 - Second attempt
0.5
0.45
Exact
0.4
= 0
0.35
= 1
E Au x
0.3
2 0.25
po L
0.2
0.15 RRM
0.1
0.05
0
0 0.2 0.4 0.6 0.8 1
x/L
2. 8
MEEM 5150
Application 2 - First attempt
EIw(x)/(P L^3)
0.016
0.014
wexact(x)
w1(x)
0.012
0.01
0.008
0.006
0.004
0.002
0
0 0.2 0.4 0.6 0.8 1
x/L
2. 9
MEEM 5150
Application 2 - Second attempt
EIw(x)/(P L^3)
0.016
w2(x)
0.014
w1(x) wexact(x)
0.012
0.01
0.008
0.006
0.004
0.002
0
0 0.2 0.4 0.6 0.8 1
x/L
2.10
MEEM 5150
RRM: final notes
Conclusions
• Advantages
– Simplicity
– “keep old terms” when adding new ones
• Disadvantages
– Basis functions are hard to find for complicated
geometries (especially in 2-D and 3-D cases)
– qi have no physical significance
– Convergence is hard to quantify
2.11
MEEM 5150
3. Basic concepts of FEM: solution of 1-D
bar problem
Table of contents
2.12
MEEM 5150
3.1 Basic concepts
• The FEM is also based on the RRM, but
– the basis functions are easy to find : interpolation
– the qi have a physical significance : nodal displacements
• Basic idea
– Discretize the domain with a finite element mesh composed
of nodes and elements
NODE
ELEMENT
po (N/m)
x
0Stiffness L
E
Cross-section A(x)=Ao(1-x/aL) with a > 1
(w
Exact solution d x du
GDE: EAo 1 po for 0 x L
dx aL dx
u0 0
BC : du
EA
L 0
dx L
x x
o ex o
EA u x p L2
a
L
a 2
a ln1
aL
x
Solution : 1
x E duex po L L
ex
dx Ao 1 x
aL
2.14
MEEM 5150
FEA of axially loaded bar
Finite element solution
We will use the PMPE. The total potential energy for this problem is
L
1 du 2 L
U V
2
EA dx
dx po u x dx
0 0
Step 1 : Discretization
local coordinate s
•a E, A, l •b
local numbering
u˜ s
c idea of the FEM is to write the approximate displacement in an ele
ation between the (so far unknown) nodal values Ua and Ub
•a
s
E, A, l •b
2.16
MEEM 5150
FEA of axially loaded bar
Na(s) and Nb(s) play an important role in FEA and are called
interpolation or shape functions
Na(s)=1-s/l Nb(s)=s/l
1
• •1
•a l •b
ave the approximate expression of u(s) u˜ (s) on the generic element
e (approximate) contribution of the element to the total potential energy
e 1 l
e d˜u2 l
˜ E A ds
po u˜ s ds o
20 ds 0
but u˜ s Ua Na s U b Nb s
du˜ s dNa s dN b s
thus Ua Ub
ds ds ds
1 l dNa dN 2 l
po Ua N a s Ub Nb s ds
e
˜ EAU
Substitute in o Ub b
ds
2 0 a ds ds 0
2.17
MEEM 5150
FEA of axially loaded bar
Expand the square term :
U a
˜e 1 U
Ub k Ua Ub r
2 a
U b
2.18
MEEM 5150
Since we know that
s s
N a s 1 and N b s
l l
Find
2.19
MEEM 5150
FEA of axially loaded bar
p 3 : Assemble the elements (i.e., put the pieces together)
D U1 U 2
he objective is to find the 4 nodal displacements U 3 U4
ese are determined, we can use the shape functions to interpolate the displa
nside each element
~
u(s)
U1 U2 U3 U4
•1 •2 •3 •4
2.20
MEEM 5150
FEA of axially loaded bar
u˜ s solution
Note : the approximate displacement is
continuous within and between elements
– Within an element :why?
– Between elements : why?
These equations will be obtained through the PMPE for the whole
structure.
The first thing we need to do is to write the (approximate)
expression of the total potential
energy of the structure 1 by addingU the
a
contribution of all three
˜ e
result
elements (using the U a of k 2).
Ub Step Ua Ub r
2
U b
with
Recall that, for a generic
EA 1 2-node
1 element of1/length 2 l :
k r po l
l 1 1 1/ 2
ctive : write the (approximate) total potential energy as
˜ e 1 D K D D R
˜ TOT
e 2
2.21
MEEM 5150
FEA of axially loaded bar
Let’s add the contribution of all three elements to the total potential energy
Its contribution to the total
potential energy is :
59EAo 59EAo po L
20L 0 0U1 6
20L
1 59EA 59EAo
U 2
po L
˜
TOT
U1 U 2 U 3 U 4 o
0 0 U1 U 2 U 3 U 4
2 20L 20L U 3 6
0 0 0 0
U 4 0
0 0 0 0
0
2.22
MEEM 5150
FEA of axially loaded bar
ement #2 : l=L/3 a=2 b=3 A=57Ao/60
57EAo 57EAo po L
˜ (2) 1 U2
U3 20L 20L U2 U U 6
2 57EAo 57EAo U3 2 3
po L
20L 20L 6
2.23
MEEM 5150
FEA of axially loaded bar
ment #3 : l=L/3 a=3 b=4 A=55 Ao/60
55EAo 55EAo po L
U3
˜ (3) 1 U3
U4 20L 20L U U 6
2 55EAo 55EAo U4 3 4
po L
20L 20L 6
6 6
po L
6
2.24
MEEM 5150
FEA of axially loaded bar
We have thus achieved our objective :
˜ TOT 1
D K D D R
2
where [K] is the GLOBAL STIFFNESS MATRIX
{R} is the GLOBAL LOAD VECTOR
Here
D U1 U 2 U3 U4
59 59 0 0
EAo 59 116 57 0
K
20L 0 57 112 55
0 0 55 55
1/ 6
1/ 3
R po L
1/ 3
1/ 6
2.25
MEEM 5150
FEA of axially loaded bar
his point, we apply the PMPE by choosing the nodal displacements which minim
approximate total potential energy of the structure :
˜ TOT
˜ TOT
˜ TOT
˜ TOT
0
U 1 U 2 U 3 U 4
in practice, one does not need to compute the total potential energy in (*):
y to (**) by assembling [k] and {r} into [K] and {R}, respectively.
2.26
MEEM 5150
FEA of axially loaded bar
Step 4 : Apply load and displacement boundary conditions
impose the nodal (concentrated) loads. In this case, we only have the reac
ll (node 1)
Ro
•1 •2 •3 •4
the nodal loads are added to the global load vector {R}
2.27
MEEM 5150
FEA of axially loaded bar
The final form of the linear system is thus :
116 57 0 U2 1/ 3
EAo
57 112 55
U p L 1/ 3
20L 3 o
1/ 6
0 55 55
U4
: - When the displacement b.c. are applied, the unknown reactions (here Ro)
from the system
- Is our approximate solution admissible? Why/why not?
Find
U1 0
U2 po L2 0.28249
U
3 EA o 0.45492
U4
0.51853
2.28
MEEM 5150
FEA of axially loaded bar
Step 6 : Post-processing
n this step, we
• plot the deformed shape of the structure
• obtain the stresses and strains in each element
• compute the nodal reactions at the supports
1) Deformed shape
2.29
MEEM 5150
FEA of axially loaded bar
2) Axial stress and strain in each element
•a E, A, l •b
have used a linear interpolation between Ua and Ub, the approximate axial st
ant in the element
du˜ s dNa dNb 1 1 1
˜ xx Ua Ub Ua U b Ub U a
ds ds ds l l l
E
˜ xx E
˜ xx U b Ua
l
2.30
MEEM 5150
ew the elements one by one and use the nodal displacements we just obtained i
Element #2 : l=L/3 a=2 b=3
po L2 po L2 E pL
Ua U2 0.28249 Ub U3 0.45792
˜ xx(2) U3 U2 0.52632 o
EAo EAo L /3 Ao
Element #3 : l=L/3 a=3 b=4
E pL
Ua U3 0.45792
po L2
Ub U4 0.51853
po L2
˜ xx(3) U4 U3 0.18182 o
EAo EAo L /3 Ao
2.31
MEEM 5150
FEA of axially loaded bar
2.32
MEEM 5150
FEA of axially loaded bar
3) Reactions at supports
consider the equation we deleted from the original system in Step 4 when we
lacement boundary conditions:
at we know the nodal displacements, we can use that equation to solve for the
on Ro. We find
1 59EAo po L2
Ro po L 0.28249 po L as expected!
6 20L EA
2.33
MEEM 5150
3.3 Notes
Conclusions :
nite element method
is a special form of the RRM
is based on the interpolation (using shape functions) of nodal displacement
contains 6 basic steps
1) discretization : mesh, nodes, elements
2) property of a generic element : local stiffness matrix [k] and load vec
3) assembling : global load matrix [K] and load vector {R}
4) apply the load and displacement boundary conditions
5) solve the linear system [K] {D} = {R}
6) post-processing : deformed shape, element stresses and support reaction
Important notes :
1) What if we use more elements? (N of them)
The property of a generic 2-node element remains unchanged :
k
E A 1 1
l 1 1
1/ 2
r pol 1/ 2
2.35
MEEM 5150
N = 10
2.36
MEEM 5150
Notes (Cont’d)
tand the physical significance of [k], let’s examine the behavior of a gener
to nodal forces
Ua Ub
Fa
•a E, A, l •b F b
’s fix Ua>0 and Ub=0, and compute the nodal forces Fa and Fb to achieve these
: U Ua U
xx b a
l l
U
xx E xx E a (compressive)
l
We have thus
Fa E AU a / l
Ua
Fa>0 Fb<0
Fb E AUa /l
•a •b
2.37
MEEM 5150
Notes (Cont’d)
w let’s fix Ua=0 and Ub>0, and compute the corresponding nodal forces Fa and F
U Ua Ub
xx b
l l
U
xx E xx E b
l
We have thus Ub
Fa<0
Fa E AUb /l
Fb E AU b / l
•a •b F b>0
n the general case, Ua and Ub≠0. We can combine (1) and (2) to get:
2.38
MEEM 5150
Notes (Cont’d)
3) How to incorporate a concentrated load?
Concentrated (nodal) loads must be added in Step 4.
E.g., Ro po (N/m)
P (N) x
0 L
2.39
MEEM 5150
Notes (Cont’d)
4) Higher-order elements
d of increasing the number of elements, we could increase the order of the s
ons from order 1 (linear) to order 2 (quadratic).
parabola
Ub ̃us Ua N a s Ub Nb s U c Nc s
Ua
Uc
• • •c 4
a l /2 b l /2
N (s) = ?
N b s 2 s s
b
where
2
aN (s) = ?
N a s 2 s s
2
1
1
a l /2 l /2 l /2 l /2
• •
b
•c a • •b •c
s s
2.40
MEEM 5150
Two applications
the finite element method (direct and PMPE approaches) to solve the followin
Q k6 6
P
• • • • •
k1 k2 k3 k4
1 2 2 3 3 4 4 5
1
5 k5
2) Define a suitable mesh and derive the 2- and 3-node finite
element formulation ([k] and {r}) for the following linearly
elastic beam torsion problem:
mo
x
J
x=0 x=L
x = L/2
2.41
MEEM 5150
Notes (Cont’d)
5) Other methods to derive the FE formulation
we have seen 2 ways to derive the FE formulation (i.e., to derive [k] and {r
ly loaded bar problem :
s indicated earlier, there are two other ways to obtain the same results:
2.42
MEEM 5150
3.4 PVW approach
he axially loaded bar problem examined earlier, the PVW is written as
L L
du du
W Win Wex E A dx p o u dx 0 u
0
dx dx 0
ux u˜ x N x D
•
1
•
2
•
3
•
4
•
5
•6 •
7
2.43
MEEM 5150
du du˜ dN x
Then, we have D
dx dx dx
The PVW becomes
L du dN x L
W EA dx D po u dx 0 u
0 dx dx
0
we did before, we construct the global stiffness matrix [K] and global load
ng [k] and {r} at the element level and then assembling into [K] and {R}.
2.44
MEEM 5150
3.5 Galerkin Weighted Residual approach
• For physical problems which do not have a functional form,
from which the differential equations and non-essential (i.e.,
natural) boundary conditions can be derived (e.g., when the DEQ
contains odd-order terms).
• We need a different approach which starts from the DEQ
itself, or, more precisely, from an integral formulation (or
weak form) of the problem. This approach is called the weighted
residual method (WRM).
• There
3.5.1. are many of
Presentation types
the of
WRMWRM.
Let us write the DEQ and natural BC as
Du f 0 in V
Bu g 0 on S V
where u is the unknown function of x, D and B are differential
operators and f and g are known 4 functions of x.
E.g., for beam bending problems: d
DEI 4 f qx
dx
d2
B E I 2 g M
dx
d3
B E I 3 g V
dx
2.45
MEEM 5150
u˜ x,a
Look for approximation satisfying essential
ai iand
boundary conditions such that
1, nchoose theu˜ x, a is "close
unknown to" ux
coefficients
ctive is thus to find the coefficients ai such that the residues are small.
* Collocation method
RD a, xi 0 i 1, 2,..., j 1
RB a, xi 0
i j, j 1,..., n
2.46
MEEM 5150
* (Continuous) least squares method
I
0 with I RD a, x dV RB a, x dS
2 2
ai V S
where is a scalar multiplier for dimensional homogeneity
and to stress the relative importance of RD vs. RB.
* Least squares collocation method
j1
I m
0 k 1,...,n with I RD a, x i RB a, xi
2 2
ak i1 i j
• Simple collocation if m = n.
• Overdetermined collocation if m > n.
* Galerkin method
Let vi x denote some weight functions (chosen as u˜ a i - see example) .
Then find the coefficients ai such that
Ri v i x RD a, x dV 0 i 1, 2,...,n
V
I.e., RD is made orthogonal to the shape functions v i x .
Before solving (*), use integration by parts to balance
the continuity requirements and to introduce the natural
bc (through RB).
Note: if a variational principle is available, the
Galerkin and RR methods yield identicalu˜ solutions with the
same approximating function 2.47
MEEM 5150
WRM: application
Problem description:
p(x) = po * x/L
P
E, A, L
2.48
MEEM 5150
WRM: application
Solution:
u(x)/cL^3 Residual methods for axially loaded bar with b/cL^2=1
2
exact
point collocation at x=L/3
1.8 subdomain and continuous least squares
Galerkin
1.6
b P
1.4
2 1
cL po L
1.2
0.8
0.6
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
x/L
2.49
MEEM 5150
WRM: application
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
x/L
2.50
MEEM 5150
3.5.2. Galerkin finite element method
idea of the GWRM is as follows : instead of satisfying the GDE at every poi
≤ x ≤ L), we will only satisfy it in a “weighted average sense” (weak form)
L
E A u x po vx dx
0
is any weight function defined on the domain vanishing where essential boun
are applied on u (here, the essential BC is u(0) = 0 and thus we must have
L L
E A u x vx dx po v x dx (*)
0 0
continuity requirements for u(x) and v(x) are “uneven”: u(x) must be at lea
be ?? only.
2.51
MEEM 5150
ance these requirements, let us integrate the left-hand side by parts :
L L
E A u v dx E A u vL
0
E A u v dx
0 0
L
E A u L v L E A u 0 v0 E A u v dx
0
= 0 = 0
Thus (*) becomes
L L
E A u x v x dx po v x dx (**)
0 0
hoose our weight functions to be the shape functions : v(x) = Ni(x) (i=1,
L dN x dN x L
E A
dx
dx
dx D po N x dx
0 0
K R
2.52
MEEM 5150
Or, in component form :
L
dNi x dN j x
Kij E A dx
0 dx dx
L
Ri po N i x dx
0
2.53
MEEM 5150