Finite element method for structural dynamic

and stability analyses

Module-9

Structural stability analysis

Lecture-28 Nonlinear dynamical systems, fixed points, and

bifurcations

Prof C S Manohar
Department of Civil Engineering
IISc, Bangalore 560 012 India

1
 P P
q  x e e
P P

 
y0  x 
P P


 
y  x

These three problems are mathematically equivalent.

The
• transverse load
• eccentrically applied axial load
• initial imperfections
are manifestations of departures from an ideal situation.
How about the study of the ideal situation itself?
2
 y  x
P P

x

d2y
EI 2  Py  0; y  0   0; y  L   0
dx

d4y d2y
EI 4  P 2  0; y  0   0, y  l   0; y  0   0, y  l   0
dx dx

3
 P
      
3rd buckling load  P3




      
2nd buckling load P2





 P1
      
1st buckling load 

 

4
Nonlinear dynamical systems, fixed points, and stability
Consider free vibration of sdof, nonlinear dynamical system
governed by
x  2 x   2 x   g  x, x   0; x  0   x0 , x  0   x0
Define
y1  t   x  t 
y2  t   x  t 

y1  y2
y2  x  t   2 x   2 x   g  x, x   2 y2   2 y1   g  y1 , y2 

y1  y2
y2  2 y2   2 y1   g  y1 , y2  5
y1  y2
y2  2 y2   2 y1   g  y1 , y2 
In general we consider equations of the form
x  f  x, y 
y  h  x, y 
x
system state =X =  
 y
Fixed points: system states become time invariant 
x  f  x, y   0
y  h  x, y   0
Fixed points are thus roots of the equations
f  x, y   0
h  x, y   0 6
x   2 x  0; x  0   x0 , x  0   x0 x  2 x   2 x  0; x  0   x0 , x  0   x0
y1  y2 y1  y2
y2   2 y1 y2  2 y2   2 y1
Fixed point:  0, 0  Fixed point:  0, 0 

x   x   x 3  0; x  0   x0 , x  0   x0
y1  y2
y2   y1   y13  y1    y12 
 
Fixed points:  0, 0  ,  0,  
 

7
Questions
What are the fixed points of a dynamical system?
How many fixed points can a system have?
What happens if motion in the neighbourhood of a fixed point
is perturbed?
What happens if values of system parameters are varied?
Do the number of fixed points remain unaltered?
Do the nature of the motion in the neighbourhood of fixed points
change because of changes in values of the system parameters?

8
 15
R=1
R=2
10
x   x  0; x  0   x0 , x  0   x0
2

y1  y2 5

y2   2 y1

xdot(t)
0

Fixed point:  0, 0 
-5
x(t )  R cos(t   )
x  t    R sin(t   ) -10

x2
x2   R2 -15
-2 -1 0 1 2
 2
x(t)

Phase portrait

9
 4

x  2 x   2 x  0; 2

x  0   x0 , x  0   x0 1

xdot(t)
y1  y2 0

y2  2 y2   2 y1 -1

Fixed point:  0, 0  -2

-3

-4
-0.5 0 0.5
x(t)

x  t   exp  t  A cos d t  B sin d t 

x  t    x  t   exp  t   Ad sin d t  Bd cos d t 
A  x0
x0   x0
B
d 10
 x  2 x   2 x  f  t  ; x  0   x0 , x  0   x0
y1  y2
y2  2 y2   2 y1  f  t 
 y1 
y 
 y2 
 0 1   y1   0 
y 2      Ay  F
  2   y2   f  t  

x  2 x   2 x  g  x, x   f  t  ; x  0   x0 , x  0   x0
y1  y2
y2  2 y2   2 y1  g  y1 , y2   f  t 
 y1 
y 
 y2 
 y2 
y   A  y, t 
2 y2   y1  g  y1 , y2   f  t  
2 11
Autonomous system
y  A  y  ; y  0   y0
The system has autonomy to choose its frequency and amplitude of
oscillations.
Non-autonomous system
y  A  y , t  ; y  0   y0
External forcing influences the frequency and amplitude of
oscillations.

12
 Equillibrium points or fixed points
y  t   A  y  ; y  0   y0
Points at which the system state is at rest. That is, y  t   0.
These points are obtained as roots of the equation A  y   0

x   2 x  0; x  0   x0 , x  0   x0 x  2 x   2 x  0; x  0   x0 , x  0   x0
y1  y2 y1  y2
y2   2 y1 y2  2 y2   2 y1
Fixed point:  0, 0  Fixed point:  0, 0 

x   x   x 3  0; x  0   x0 , x  0   x0
y1  y2 Nonlinear systems have more
y2   y1   y13  y1    y12  than one fixed points.
 
Fixed points:  0, 0  ,  0,  
  13
Stability of equillibrium points
Consider
x  f  x, y 
y  g  x, y 
The equillibrium points are given by the conditions
f  x*, y *  0 
  x*, y * : a set of equillibrium points.
g  x*, y *  0 
Let us examine the nature of motion around each of the
equillibrium points by perturbing the equillibrium states by
small amounts as:
x  t   x *   t 
y  t   y *   t 
x    t   f  x *   t  , y *   t  
y    t   g  x *   t  , y *   t   14
 f f
  t   f  x *   t  , y *   t    f  x*, y *   t    t 
x x  x* y x  x*
y  y* y  y*

g g
  t   g  x *   t  , y *   t    g  x*, y *   t    t 
x x  x* y x  x*
y  y* y  y*

f f
 t    t    t 
x x  x* y x  x*
y  y* y  y*

g g
 t    t    t 
x x  x* y x  x*
y  y* y  y*

 f f 
  t    x y    t  
      Linear homogeneous set of ODE-s 
  t    g g    t  
 x y  x  x*
 y  y*
15
  f f 
  t    x y    t  
     Omit subscripts 
  t    g g    t  
 x y 
 
Seek the solution in the form
  t    
     exp  st 
  t     
 f f 
   x y   
 s   exp  st       exp  st 
   g g    
 x y 

 f f 
 x y     
     s    Eigenvalue problem 
 g g      
 x y 
 16
 f f 
 x y     
     s     s1 , s2  : eigenvalues; complex, in general.
 g g      
 x y 

s  a  ib

  t    exp  at  ibt    exp  at  cos bt  i sin bt 
  t    exp  at  ibt    exp  at  cos bt  i sin bt 
  t  
If a  0, lim 
t    t 
    the fixed point  x*, y * is unstable
 
  t  
If a  0, lim    0  the fixed point  x*, y * is stable
t    t 
 
17
Classification of fixed points
Node
both eigenvalues are real and are of the same sign
the fixed point can be stable (if roots are <0) or unstable (if roots are >0)
Saddle
both roots are real and of different signs
the fixed point is unstable
Focus
the roots are complex conjugates (but not pure imaginary)
the fixed point could be stable or unstable
Center
roots are pure imaginary
linearized stability analysis is inadequate to
answer the question on whether the fixed point is
stable or unstable 18
x  2 x   2 x  0; x  0   x0 , x  0   x0
y1  y2  f  y1 , y2 
y2  2 y2   2 y1  g  y1 , y2 
Fixed point:  0, 0 
f f g 2 g
 0;  1;   ;  2
y1 y2 y1 y2
Eigenvalues
 1
 0    2      2  0
 2 2  
 2  2   2  0       2 2   2
  0    i  origin is a center
0    1      i 1   2  origin is a stable focus
  1        2  1  origin is a stable node
 <0  origin is unstable (node/focus?) 19
x  x  x  x 3  0; x  0   x0 , x  0   x0
y1  y2  f  y1 , y2 
y2   y2  y1  y13   y2  y1 1  y12   g  y1 , y2 
Fixed points:  0, 0  ,  1, 0 
f f
 0;  1;
y1 y2
g g
 1  3 y1  ;
2
 1
y1 y2
Consider  0, 0 
Eigenvalues
0 1
  1     1  0
1 1  
1 5
   Origin is unstable and is a saddle 20
2 2
f f g g
 0;  1;  1  3 y1  ;
2
 1
y1 y2 y1 y2
Consider  1, 0 
Eigenvalues
0 1
  1     2  0
2 1  
1 7
   i  Origin is stable focus
2 2

21
 x  2 x   x  0
2 x  x  x  x3  0
 2 x2 1 1
V V   x2  x4
2 2 4
4.5 0.15

4 0.1

3.5
0.05
3
0
2.5

U(x)
U(x)

-0.05
2
-0.1
1.5
-0.15
1

0.5 -0.2

0 -0.25
-3 -2 -1 0 1 2 3 -1.5 -1 -0.5 0 0.5 1 1.5
x x

22
 g
  sin   0
l
1   2  f 1 , 2 
g
 2   sin 1  h 1 , 2 
l
Fixed points:  n , 0  , n  0, 1, 2, , 
Eigenvalues
0 0
g
g  0  2  cos n  0
 cos n  l
l
g
n  0, 2, 4,   i  Fixed points are centers
l
g
n=1,3,5,   Fixed points are saddles
l 23
    

Phase portrait

24
 P

A 
AB: rigid rod with mass m per
La unit length
k C C: sleeve that can slide along AB

without friction
Direction of P remians unchanged
a

B 
Question
Will the rigid rod remain in the upright
position under quasistatically applied load
P? 25
 a
cos  
L

L
a
a 

a
1  a 
1
 cos   cos  
L L

26
 y P
A  L 1  cos  

x   sin  ; y   cos 

dT  md   x 2  y 2 
1
L cos 2

  x, y
  a
1
  
2
 md   cos    sin 
2

2

1
 m 2 2 d
B
x
 2
L 3
1 mL
T   m 2 2 d   2
0
2 6
1 2
V  ka tan 2   PL 1  cos  
2
 mL3 2 1 2
L   ka tan 2   PL 1  cos  
6 2
d  L  L
  0
dt    
ml 3  ka 2 
    3  PL  sin   0
3  cos  
Consider damped system
ml 3  ka 2 
  c   3  PL  sin   0
3  cos  
1   2
3c 3  ka 2 
 2   3  2  3  3  PL  sin 1
ml ml  cos 1 

28
1   2  f 1 , 2 
3c 3  ka 2 
 2   3  2  3  3  PL  sin 1  g 1 , 2 
ml ml  cos 1 
Fixed points
 ka 
1/3
1  PL 
2
 2  0,  3  PL  sin 1  0   2  0, 1  0,sec  2 
 cos 1   ka 
f f
 0;  1;
1  2
g 3  ka 2  3  ka 2 
 3  PL  cos 1  3 sin 1  3cos 1 sin 1 
2

1 ml  cos 1
3
 ml  cos 1
6

g 3c
 3
 2 ml

29
Consider the fixed point  0, 0 
f f
 0;  1;
1  2
g 3  ka 2  3  ka 2 
 3  PL  cos 1  3 sin 1  3cos 1 sin 1 
2

1 ml  cos 1
3
 ml  cos 1
6

  3  ka 2  PL 
3
ml
g 3c
 3
 2 ml
Eigenvalues
0 1
0
  ka  PL  c  
2

30
 c  4  ka 2  PL 
c 1 2
 
2 2
Suppose c  0
  PL   0  Re     0  fixed point is unstable
ka 2

 ka 2
 PL   0  Re     0  fixed point is a center
ka 2
Condition for stability of  0, 0  is P 
L
 1  PL 1/3 
Consider  sec  2  , 0 
  ka  
 
 ka 2  g 3ka 2 sin 2 1
When   PL   0,     2

 cos 3
1  1 cos 4
1
Other gradients remain the same.
31
 0 1
Eigenvalues: 0
 2 c  
  2  c   2  0
c 1 2
  c  4 2  Fixed points are always stable
2 2
P
  unstable
Large deformation
 
 
dynamic analysis
  
 
2
ka
P* 
 l
 stable 
 

a 1a
1
 cos   cos  
L L 32
Linear dynamic analysis
ml 3
  c   ka 2  PL   0
3
1   2  Fixed point:  0, 0 


 2   3  2  3  ka  PL 1  Stable if  ka  PL   0  P* 
3c 3 2 2 ka 2
ml ml  l
P
  unstable
Linear 
dynamic analysis 
       P* 
2
ka
 l
 stable 
 

a a
1
1
 cos   cos   33
L L
Does the linearized stability analysis give qualitatively
correct picture of the phase portrait?
Yes, provided the fixed point for linearized system is not a
border line case (as in center)
If the linearized system predicts a saddle, node, or a focus, then
the fixed point is really is a saddle, node, or focus.

Example to follow:
H S Strogatz, 1994, Nonlinear dynamical systems and chaos,
Westview, Cambridge, MA.

34
x   y  ax  x 2  y 2 
y  x  ay  x 2  y 2 
Origin is a fixed point
Eigenvalue problem
0 1
 0   2  1  0    i
1 0
Origin is a center.
Let us analyse the governing equation and see how the
solutions behave near the origin.
x  r cos  , y  r sin 
x 2  y 2  r 2  xx  yy  rr
rr  x   y  ax  x 2  y 2    y  x  ay  x 2  y 2  

 ax  y
2

2 2
 ar 4  r  ar 3 35
x   y  ax  x 2  y 2  ; y  x  ay  x 2  y 2 
x  r cos  , y  r sin 
r  ar 3
xy  yx
 2
 1
r
r  ar 3
 1
a  0  lim r  t   0  origin is a stable focus 
t 

a  0  r  t   r0  origin is a center 
a  0  lim r  t     origin is an unstable focus 
t 

Linearized system: origin is center for all values of a.

Exact solution: does not support this.
36
Bifurcations
As the parameters in a nonlinear dynamical system are changed
one observes
 Number of fixed points can change
The nature of the fixed points can change
The stability of the fixed points can change

The subject of bifurcation theory deals with these changes due to

changes in values of system parameters.

The occurrence of bifurcations is accompanied by qualitative changes

in the nature of the system behavior.
The system paramter values at which system undergoes bifurcations
are called the bifurcation points. 37
x   x   x3  0

  0,   0  origin is a center
  0,   0  center+saddle
  0,   0  2 centers and one saddle

38
Limit cycles
x   x 1  x 2   x  0; x  0   x0 , x  0   x0

Isolated periodic free vibration solutions of nonlinear systems.

Stable
Unstable
 1 
r   r  r  1 sin 
2
2 
 1  r   this system has infinite number of limit cycles
  1 


39
40
41
42
Energy methods for stability analysis
Axiom - 1
A stationary value of the total potential energy with respect to the
generalized coordinates is necessary and sufficient condition for the
equillibrium state of the system.
Axiom - 2
A complete relative minimum of the total potential energy with respect
to the generalized coordinates is necessary and sufficient for the stability
of an equillibrium state of the system.

J M T Thompson and G W Hunt, 1973, A general

theory of elastic stability, John Wiley, London

43