You are on page 1of 51

NONLINEAR MECHANICAL SYSTEM

We always assumed that the force F(x) exerted by the spring on the mass is a
linear function of x:F(x)= -kx (Hookes law).

However every spring in nature is nonlinear ( even if only slightly so).

Moreover, springs in some automobile suspension system deliberately are
designed to be linear.

So now we allow the force function F(x) to be nonlinear. Because F(0)=0 at
the equilibrium position x=0, we may assume that F has a power series
expansion of the form

...... ) (
3 2
+ + + = bx ax kx x F
We take k>0 so that the reaction of the spring is directed opposite of the
displacement when x is sufficiently small.
Suppose we know that the mass on a spring is connected also to a dashpot that
provides a force of resistance proportional to the velocity
of the mass. If the spring is still assumed nonlinear, then the
equation of motion of the mass is
(1)

Where c >0 is the dashpot constant. If b >0, then the equivalent first order system

(2)

(3)

Has critical point (0,0) and if we take m=1 for convinience, then at
(0,0) the characteristic of equation of the corresponding linear system is
(4)

dx dy y / =
) 0 , ( | k
DUMPED NONLINEAR VIBRATIONS
3
' ' ' bx kx cx mx + =
3
bx cy kx
dt
dy
m
y
dt
dx
+ =
=
0 ) (
2
= + + = + + k c k c
if we follow that (0,0) is a stable node if , a stable spiral point if

EXAMPLE:
Suppose that m=1, k=5, and c=2. Then the non linear system in () is





In this example no explicit solution for the phase plane trajectories is available, so we proceed
to investigate the critical points (0,0) ,(2,0) and (-2,0)
At (0,0) the linearized system





Has characteristic equation with roots .Hence, (0,0) is a stable spiral
point of (18) and the liniearized position function of the mass is of the form



4 5 = |
k c 4
2
>
k c 4
2
<
3
4
5
2 5 x y x
dt
dy
y
dy
dx
+ =
=
y y
dt
dy
y
dt
dx
2 5 =
=
0 5 2
2
= +
i 2 1 =
) 2 sin 2 cos ( ) ( t B t A e t x
t
+ =

An exponential damped oscillation about x=0
at (2,0) the substitution u=x-2, v=y in (18) yields the system





with corresponding critical point (0,0). The liniearized sytem




Has characteristics equation with roots and
. It follows that (2,0) is an unstable saddle point of the original
system in (18). A similar analysis shows that (-2,0) is also unstable point.
The phase



3 2
4
5
2
15
2 10 u u v u
dt
dv
v
dy
dx
+ + =
=
v u
dt
dv
v
dt
du
2 10 =
=
0 10 2
2
= + 0 11 1
1
< =
0 11 1
2
> + =
THE NON LINEAR PENDULUM
In section 3.4 we derived the equation

(20)

For the undamped oscillations of the simple pendulum shown in Fig 6.4.6. there we used the
approximation for small to replace Eq 20 with the linear model


(21)

Where . The general solution
(22)

Of Eq 21 describes oscillations about the equilibrium position with circular frequency
and amplitude
The linear model does not adequately describe the possible motions of the pendulum for large
values of . For instance, the equilibrium solution
u u ~ sin
u
0 sin
2
2
= + u
u
L
g
dt
d
0
2
2
2
= + u e
u
dt
d
L g =
2
e
t B t A t e e u sin cos ) ( + =
0 = u
e
2 2
B A C + =
u
We also want to include the possibility of resistance proportional to velocity, so we
consider the general non linear pendulum equation
(23)

We examine first the undamped case, in which c=0. with and
the equivelent first order system is

(24)


Upon substuting the Taylor series for sin x we get

(25)


Thus (24) is an almost linear system of the form


0 sin
2
2
2
= + + u e
u u
dt
d
c
dt
d
) ( ) ( t t x u =
) ( ' ) ( t t y u =
x
dt
dy
y
dt
dx
sin
2
e =
=
.......
! 5 ! 3
5 2 3 2
2
+ + =
=
x x
x
dt
dy
y
dt
dx
e e
e
) , (
) , (
y x g dy cx
dt
dy
y x f by ax
dt
dx
+ + =
+ + =
With and
the critical points of the system in (24) are the points with n as integer.
The nature of the critical point depends upon whether n is even or odd.
If is even then, because

The substitution in (24) yields the system


(26)


Having (0,0) as the corresponding critical point. Just as in (25), the
liniearization of (26) is the system


(27)
0 ) , ( = y x f
...... ! 5 31 / ( ) , (
5 3 2
+ = x x y x g e
) 0 , ( t n
) 0 , ( t n
m n 2 =
u m u sin ) 2 sin( = + t
y v m x u = = , 2 t
u
dt
dv
v
dt
du
sin
2
e =
=
u
dt
dv
v
dt
du
2
e =
=
For which (0,0) is the familiar stable center with elliptical trajectories




If n= 2m+1 is odd then, because


The substitution in (24) yields the system


(28)

Having (0,0) as the corresponding critical point. The liniearization


(29)
u m u sin ) ] 1 2 [ sin( = + + t
y v m x u = + = , ) 1 2 ( t
u
dt
dv
v
dt
du
sin
2
e + =
=
u
dt
dv
v
dt
du
2
e + =
=
Of (28) has general solution


So we see that (0,0)is the unstable saddle point




In the undamped case we can see how these centers and saddle points fit together
by solving explicity for the trajectories of (24). If we write


and separate the variables,

then the integration from x=0 to x=x yields
(30)

We write E for the arbitary constant of integration because,if physical units are so
chosen that m=L=1, then the first term on the left is the kinetic energy and




t A t B t v
t B t A t u
e e e e
e e
sinh cosh ) (
sinh cosh ) (
+ =
+ =
y
x
dt dx
dt dy
dt
dy sin
2
e
= =
0 sin
2
= + xdx ydy e
E x y = + ) cos 1 (
2
1
2 2
e
The second term the potential energgy of the masss on the end of the pendulum. Then
Eis the total mechanical energy;Eq(300 thus expresses conservation of mechanical
energy for the undamped pendulum.
If we solve Eq 30 for y and use a half identity, we get the equatition
(31)

For the trajectories.
THE SEPARTICES

x E y
2
1
sin 4 2
2 2
e =
If the pendulum is released from rest with initial conditions
(32)

The Eq 30 with t=0 reduces to

(33)

Hence if , so a periodic oscillations of the pendulum ensues. To determine
the period of this oscillation, we substract Eq 33 from Eq 30 and write the result (with
and ) in the form

(34)

the period T of time required for one complete oscillation is four times the amount of time
required for to decrease from to one-fourth of an oscillation.
Hence we solve (34) for and integrate to get


(35)


o u = = ) 0 ( ) 0 ( x
0 ) 0 ( ' ) 0 ( = =u y
E = ) cos 1 (
2
o e
2
2e < E
t o < < 0
u = x
dt d y u =
) cos (cos ) (
2
1
2 2
o u e
u
=
dt
d
u
o u =
0 = u
u d dt
}

=
o
o u
u
e
0
cos cos 2
4 d
T
To attempt to evaluate this integral we first use the identity
And get
(36)


Where
(37)

Next the substitution yields

(38)

Finally the substitution

(39)



The integral in (39) is the eliptic integral of the first kind that is often denoted by
. It can be evaluated numerically as follows. First we use the binonial series


(40)
) 2 ( sin 2 1 cos
2
u u =
}

=
o
u
u
e
0
2 2
) 2 ( sin
4
k
d
T
2
sin
o
= k
) 2 sin( ) 1 ( u k u =
}

=
1
0
2 2 2
1 )( 1 (
4
u k u
du
T
e
) 2 , ( t k F
n
n
x
n
n
x

=


+ =
1
) 2 ( 4 2
) 1 2 ( 3 1
1
1
1
}

=
2
0
2 2
sin 1
4
t
|
|
e
k
d
T
| sin = u
With to expand the integrand in (39). Then we integrate termwise using
the tabulated integral formula

(41)
The final result is the formula




(42)
for the period T of the nonlinear pendulum released from rest with initial angle
in terms of the liniearized period and
1 sin
2 2
< = | k x
) 2 ( 4 2
) 1 2 ( 3 1
2
sin
2
0
2
n
n
d
n


=
}
t
| |
t
] )
) 2 ( 4 2
) 1 2 ( 3 1
( 1 [
2
2 2
1
n
n
k
n
n
T

=


+ =
e
t
.......] )
6 4 2
5 3 1
( )
4 2
3 1
( )
2
1
( 1 [
6 2 4 2 2 2
0
+


+

+ + = k k k T
o u = ) 0 (
e t 2
0
= T ) 2 sin(o = k
In this example, we investigate what happens when the populations x and y still have to
compete for the same finite resources, but each grows according to a logistic growth law so that



Where a, b, c, d and k are positive constants. We are concerned only and
This system of equations (10.18) is nonlinear. In order to avoid unnecessary algebra, we will
consider the case where k=5, a=b, and d=c, so that (10.18) becomes



Where c>0. this means that in the absence of competition, the carrying capacity for both x and y
is 5/c. why?
Our first step is to find the equilibrium points, which in the case of (10.19) occur when



That is when x=0 or y=5-cx, and y=0 or y=(5-x)/c. if , we have four equilibrium points,
namely (0,0), (0,5/c),(5/c,0) and (5/(c+1), 5/(c+1). If c=1, we have an infinite number of
POPULATION MODEL

=
=
), ( '
), ( '
bx dy k y y
ay cx k x x
0 > x 0 > y

=
=
) 5 ( '
) 5 ( '
x cy y y
y cx x x
0 ) 5 (
0 ) 5 (
=
=
x cy y
y cx x
1 = c
equilibrium points, namely (0,0) and all points on the line y=5-x. in the two following examples
we will consider c<1 and c>1.

Example of Population Model
We illustrate the case c<1 by considering c = 5/8, so
that (10.19) becomes

10.20

with equilibrium points at (0,0), (0,8) and
(40/13,40/13).



If we use a nullcline analysis on (10.20), we find that
there are four lines to consider: x = 0 and y = 5- 5x/8
(x-nullclines, which have vertical arrows in the phase
plane), and y = 0 and y = 8(5-x)/5 (y nullclines, which
have horizontal arrows).
The directions of the arrows are determined from
(10.20) and are shown in Figure 10.21
We notice that the equilibrium points occur only
where a nullcline with vertical arrows intersects one
with horizontal arrows; namely, at (0,0), (0,8), (8,0)
and (40/13,40,13). The points (5,0) and (0,5) are not
equilibrium points.
In order to facilitate the discussion, we have labeled
the regions I, II, III and IV. We notice that if an orbit
enters region I, it cannot escape and appears to be
attracted to the equilibrium point (0,8) as t increases.

Similarly, orbits entering region III cannot escape and
appear to be attracted to the equilibrium point (8,0) as
t increases.
Orbits in region II are either drawn to the equilibrium
point (40/13, 40/13) or cross a nullcline into regions I or
III. Thus, the equilibrium point (0,0) behaves like an
unstable node, the equlibrium points (0,8) and (8,0)
behaves like stable nodes, and the equilibrium point
(40/13, 40/13) behaves like a saddle point.
This behavior can be confirm by linearizing the
nonlinear system about each of its equilibrium points.
The Jacobian matrix is


At the equilibrium point (0,0) this gives

which has determined D = 25 and trace T = 10 so
D = This is a rare case , so the Linearization
Theorem guarantees only that (0,0) is unstable
because T > 0.
graph
We now turn to the equilibrium point (40/13,
40/13), where


Here the determinant is -65. and the
trace is -50/13, so (40/13,40/13) behaves like a
saddle
point.
We now turn to the equilibrium point (0,8), where


Here the determinant is 15 and the trace is -8,
so the equilibrium point (0,8) behaves like a
stable node. By symmetry, a similar analysis
applies to the equilibrium point (8,0).
We can also evaluate those orbits numerically, and
these are shown in Figure 10.22 along with the
direction field.
We notice that there appear to be straight-line orbits
approaching (40/13,40/13) from both the left and the
right. If we substitute into (10.20), we find
m = 1, so y = x is a straight-line solution. The curve y =
x is a separatrix.
So, what is the ultimate fate of these populations?
If initially the x population exceeds the y population,
then the y population become extinct, and as t
increases. If initially the y population, then the x
population becomes extinct, and as t increases.

graph

Method of variation of parameter which- in principle (that
is, if the integrals that appear can be evaluated)-can always be
used to find a particular solution of the nonhomogeneous
linear differential equation
(18)
provided that we already know the general equation
(19)
of the associated homogeneous equation
(20)

Variation of parameters
Suppose that we replace the constant, or parameters,
in the complementary function in (19) variables:
functions of . We ask whether it is possible to
choose these function in such a way that the combination

(21)

is a particular solution of the nonhomogeneous Eq.(18). It
turns out that this is always possible.

THEOREM
Let and linearly independent solutions of . If
the functions and are chosen so that their derivatives satisfy




Then will be the solution of the non
homogeneous equations

Proof. Preparing to force to satisfy the differential equation
into the differential equation, we compute the derivative


Since we have two functions to determine, we can impose a second condition on
and namely, that


So that simplifies to Now compute from this
simplified expression for . We find


Substitution into and regrouping terms gives


Since and satisfy the associated homogeneous equation, the first two terms
equal zero. Thus for our second requirement.

Before illustrating the variation of parameter method with a more compelling
example well outline in general terms the steps that follow the application of
Theorem 3.5. The equations



are the result of applying the theorem. We can solve this system of equations
for and by elimination or by Cramers rule. Either way we find







The expression in the denominators is the same in both formulas
and we can write it as the two-by-two determinant



Called the Wronskian determinant of Since were
dealing with second order constant-coefficient equations, we can
check directly that is never zero:
(i) If if then
(ii) If then

If the solutions and not independent, one of them would
be a constant multiple order, so would be zero for all .
To complete the solution, integrate the formula for ,
to find and ,and then combine with , to get a
particular solution



Example:
Use variation of parameters to find a particular solution
for each of the following constant coefficient equations.
a)

The equation has homogeneous
solutions and Since
and the two conditions of theorem 3.5
translate into






Adding these two equations gives so
We integrate and take .Subtracting the second

equation from the first gives so

We integrate and take According to the

Theorem 3.5, well gets a solution of the nonhomogeneous
differential equation by writing
We recognize the term as a part of the general
homogeneous solution, so we write the general solution of the
nonhomogeneous equation as



b)

The complementary function is ,
so



Hence the equations are



we can easily solve these equations for





Hence we take


and


Thus our particular solution is




that is,


Integral Transform Formulas
Method
Assume is continuous on an interval
containing , and let be the solution of the
normalized constant coefficient equation

Satisfying = = 0. If , are the roots of the
characteristic equation
By that :
i) if , are real and unequal,


ii) If = ,


iii) If , = i, 0,


In each cases, we use an integral operator defined by



Where the variable is held fixed during the
integration. The operator acts as an inverse
integral operator to the differential operator
with initial condition .
The function g has one of the three forms







Those forms are example of an integral transform of
a type called a convolution integral that convert
into .

Since the variable x that pccur in the upper limit and in
the difference must be treated as a constant with
respect to integration in the other variable, some care
is necessary in doing the integration.
We can separate the x variable from the integration
process. For example, the addition formula for the sine
function enables us to write

EXAMPLE 1 OF INTEGRAL TRANSFORM
FORMULAS
Here is an equation in which the right-hand side is itself a
solution of the associated homogeneous equation :

The homogenous solution is
corresponding to a double root of the
equation The function
.
Since the differential equation is normalized, we can
compute the solution satisfying
from

Note that integration is with respect to t, so x is
temporarily fixed. We find





This is the solution with .By
Corollary 3.2 the general solution is


To satisfy initial conditions of the form
we use the flexibility inherent in the
general homogeneous part of the solution .
Since the particular solution given by the integral
transform formula satisfies , ,
we just need to determine values for and
in so that the independent solutions
and satisfy
and
For this example, we have , ,
and .
Hence we solve
and
To get and . Thus our solution is

EXAMPLE 2 OF INTEGRAL TRANSFORM
FORMULAS
Here is an example involving trigonometric functions

The characteristic roots of the associated homogenous
equation are , . The function is
then so the integrand in the
integral transform formula is then .
The integration with respect to t follows from using
formula


or more directly using the identity that proves formula
above :


Thus the particular solution satisfying
is

You might also like