Professional Documents
Culture Documents
Module-5
Prof C S Manohar
Department of Civil Engineering
IISc, Bangalore 560 012 India
1
Numerical integration of equations of equillibrium
Remarks:
This equation constitutes a set of semi-discretized system of
coupled second order ode-s. That is, these equations have been
obtained after discretizing the spatial variables.
2
The time variable, t , however, is still continuous.
We consider now the problem of discretizing in time.
We consider solution of the above equation at a set of
discrete time instants t0 t1 t2 tn with
tn tn 1 tn .
3
Equations in standard form for a N -dof system
MU CU KU R U t ,U t , t F t ;U t0 U 0 ,U t0 =U 0
U M 1CU M 1 KU M 1R U t ,U t , t M 1F t
Define X I t U t & X II t U t and
X I t
x t system state vector
X II t
X I X II
X II M 1 F t M 1CX II M 1 KX I M 1R X I , X II , t
X I t 0 I X I t 0 0
1 1
X II t M K M C X II t
1
M 1
R I II
X , X , t M F
t
Equation in state space form given by
x a x t , t ; x t0 x0 ; x t is 2 N 1.
4
The set of N coupled 2nd order ODE-s
MU CU KU R U t ,U t , t F t ;U t0 U 0 ,U t0 =U 0
have been thus recast as the following set of 2 N first order coupled ODE-s
x a x t , t ; x t0 x0 ; x t is 2 N 1.
Denote by x t , x0 , t0 solution of x a x t , t with x t0 x0
Discrete time approximation
time: t0 t1 t2 tn ; tn tn 1 tn
Denote yn x tn , x0 , t0
How to approximate x in terms of yn ?
5
Use finite difference approximations. For example,
x t t x t yn 1 yn
x t yn
t tn
yn 1 yn tn a yn , tn ; n 0,1, 2, with y0 x0
and yn a yn , tn given that x t a x t , t
Here tn ; n 0,1, 2, algorthmic parameters.
Accuracy depends upon the choice of tn ; n 0,1, 2,
6
Errors
7
How to formulate the integration algorithms to achieve
acceptable accuracy? How to understand the nature of the errors?
0?
lim en c ?
n ?
u0
0
10
Outline
• Mathematical preliminaries
• The integration schemes
– Explicit/implicit
– Single step/multi step
• Consistency
• Stability
• Accuracy
• Energy conservation
11
Mathematical preliminaries-1
f x O g x as x a
M 0 & 0 f x M g x x a
12
Examples
ax 7 bx3 cx d
ax bx cx d is O x
7 3 7
as x lim 7
a
x x
ax 7 bx3 cx d
ax 7 bx3 cx d is O x 0 as x 0 lim 0
d
x 0 x
ax 7
ax 7 is O x 7 as x 0 lim 7 a
x 0 x
ax 7
ax 7 is not O x8 as x 0 lim 8
x 0 x
13
Examples
sin x
sin( x) is O x as x 0 lim 1
x 0 x
sin x 2
sin( x 2 ) is O x 2 as x 0 lim 2
1
x 0 x
cos x
cos x is O x 0
as x 0 lim 0
1
x 0 x
sin x
sin( x) is o x o 1 as x 0 lim
0
0
0
x 0 x
12 cos x
cos x is o x as x 0 lim 1
lim x cos x 0
x 0 x 0
x
2
14
Mathematical preliminaries-2
Mean value theorem
Consider a function f ( x) to be continuous for a x b and differentiable
for a x b. According to the mean value theorem, there exists at least one
c satisfying a c b such that
df f b f a f x
c
dx ba
f b f a b a
df f a
c f b
dx
a b
x
c1 c2
15
Mathematical preliminaries-3
Taylor's series
Let f be a function of x with derivatives of all orders throughout
an interval containing the point a. The Taylor series generated by f
at x a is given by
f
k
a f a
x a f a f a x a x a
k 2
k 0 k! 2!
f
n
a
x a
n
n!
16
Mathematical preliminaries-3
Taylor's polynomial
Pn x f a f a x a
x a x a
2 n
.
2! n!
17
Corollary to Taylor's theorem
f ( x) f (a ) f a x a x a x a Rn x
2 n
2! n!
f c
n 1
n 1!
f x
f
n
x
x a
n
n 1 n!
18
Mathematical preliminaries-4
Forward difference approximation to a derivative
h2 h3
f x h f x hf x f x f x
2! 3!
f x h f x h h2
f x f x f x
h 2! 3!
f x h f x
f x O h as h 0
h
Backward difference approximation to a derivative
h2 h3
f x h f x hf x f x f x
2! 3!
f x f x h h h2
f x f x f x
h 2! 3!
f x f x h
f x
O h as h 0
h 19
Central difference approximation to a derivative
h2 h3
f x h f x hf x f x f x
2! 3!
h2 h3
f x h f x hf x f x f x
2! 3!
h3
f x h f x h 2hf x 2 f x
3!
f x h f x h
f x O h 2 as h 0
2h
Intutively
Error is of O h and we half step size error gets halved.
Error is of O h 2 and we half step size error gets quartered.
Standard form
MU t CU t KU t R U t ,U t ,U t , t F t
U 0 , U 0 specified
x a x t , t ; x 0 x0 ;0 t t f
Strategy: replace derivatives by finite difference approximations.
Introduce 0 t1 t2 t N t f & denote yn x tn , 0, x0
Forward difference approximation
x t t x t
At time=t , x t a x t , t
t
yn 1 yn tyn yn ta yn O t
Backward difference approximation
x t t x t
At time=t t , x t t a x t t , t t
t
yn 1 yn tyn 1 yn ta yn 1 O t 21
Central difference approximation
x t t x t t
At time=t , x t a x t , t
2t
yn 1 yn 1 2tyn yn 1 2ta yn O t 2
Trapezoidal rule
t tn1 tn tn1
x t x s ds x tn 1 x s ds x s ds x s ds
0 0 0 tn
yn 0.5t yn 1 yn
yn 1 yn 0.5t yn 1 yn
yn 0.5t a yn 1 a yn
22
Forward difference approximation
yn 1 yn tyn yn ta yn O t
Backward difference approximation
yn 1 yn tyn 1 yn ta yn 1 O t
Central difference approximation
yn 1 yn 1 2tyn yn 1 2ta yn O t 2
Trapezoidal rule
yn 1 yn 0.5t a yn 1 a yn
Generic form
yn 1 1 yn 2 yn 1 m yn 1 m t 0 yn 1 1 yn k yn 1 k
23
Generic form
yn 1 1 yn 2 yn 1 m yn 1 m t 0 yn 1 1 yn k yn 1 k
Definitions
Scheme is said to be self starting if yn for n 0 does not enter the calculations.
24
Forward difference approximation
yn 1 yn tyn yn ta yn O t
Backward difference approximation
yn 1 yn tyn 1 yn ta yn 1 O t
Central difference approximation
yn 1 yn 1 2tyn yn 1 2ta yn O t 2
Trapezoidal rule
yn 1 yn 0.5t a yn 1 a yn
Generic form
yn 1 1 yn 2 yn 1 m yn 1 m t 0 yn 1 1 yn k yn 1 k
Forward difference approximation: single step, explicit, self-starting.
Backward difference approximation: single step, implicit, self-starting.
Central difference approximation: multistep, explicit, not self-starting.
Trapezoidal rule: single step, implicit, self-starting.
25
Remarks
t0 0 t1 t2 tN t f
tn nt constant step size
System states: U t , U t , U t
Assume: System states at t tn are known. To find system states at t tn 1.
Integration algorithms essentially achieve this.
Computation effort is proportional to number of time steps to advance
solution from t 0 to t t f .The step size, therefore, need not be smaller than
what is essential. We need to be concerned about
Growth of errors (stability) and
Accuracy of the solution.
Undamped free vibration of a sdof system:
Check for amplitude
Frequency distortions
26
Remarks (continued)
The time integration method is the most generally applicable
method to solve equillibrium equations:
For linear systems, the size of the model (dof-s) depend upon the
details of mesh used in spatial discretization. These details need
to be chosen such that the system behavior is well represented over
the frequency range over which excitations have significant power.
If max is the highest frequency present in the excitations, then all the
modes present in the frequency range up to about 1.25 max needs to be
correctly captured. If the structure has, say, r modes in frequency range
up to 1.25 max , then the mesh details should be such that the model has
at least 10r dof-s.
28
Remarks (continued)
This invariably means that there would be a large number of higher order
modes that are present in the model (with frequencies beyond max ) which
however do not contribute significantly to the response.
29
Discussion on following methods
Explicit method with first order accuracy
Forward Euler
Implicit method with first order accuracy
Backward Euler
Explicit method with second order accuracy
Central difference
Implicit method with second order accuracy
Newmark's family of methods
HHT- method and generalization
Explicit-Implcit methods
HHT- with operator splitting
30
Strategy
Development of basic formulary
Pseudocode for implementation
Analysis of the method
Illustrative examples
Discussion on relative merits
31
Forward Euler Method
t0 tn tn 1 tf
MU t CU t KU t F t
U 0 U 0 ;U 0 U 0 ;0 t t f
U t t U (t )
U t U n 1 tU n U n
t
This is an explicit scheme
U t t U (t )
U t U n 1 tU n U n
t
Condtion for equillibrium at tn 1 is given by
MU n 1 C tU n U n K tU n U n Fn
33
Implementation of the forward Euler Method
x 2 x 2 x f t
U n 1 tU n U n t 2U n 2U n f n U n
U n 1 tU n U n
U n 1 1 t U n 0
2
U n 1 t 1 2t U n tf n
Yn 1 AYn Ln
Yn Yn n
Yn 1 n 1 AYn A n Ln
n 1 A n
35
Digression
Consider the scalar equation xn 1 axn with x0 0 sepecified.
x1 ax0
x2 ax1 a 2 x0
xn axn 1 a n x0
Clearly
0 if a 1
lim xn x0 if a 1
n
if a 1
36
Let us now consider the s 1 vector equation xn 1 Axn
with x0 0 sepecified. Here A is a s s matix.
x1 Ax0
x2 Ax1 A2 x0
xn Axn 1 An x0
Let us introduce the transformation xn zn
zn 1 Azn
Let be such that t I & t A Diag i
We can select by solving the eigen value problem
associated with matrix A.
37
xn zn
zn 1 Azn
t zn 1 t Azn zn 1 Diag i zn
znk1 k znk ; k 1, 2, ,s
znk1 kn z0k ; k 1, 2, ,s
s
xnj jk kn z0k with z0 t x0
k 1
Clearly,
max k 1 lim xnj 0 for j 1, 2, s
1 k s n
39
max k 1 lim xnj 0 for j 1, 2, s
1 k s n
Re
Asymptotically
stable Stable