M2794.

001200 Dynamics
PROBLEM SET 11 SOLUTIONS

Problem 1
(a)
det(λI − A) = λ2 − 3λ + 2 = 0 (1)
λ1,2 = 1, 2 (2)
(b)
det(λI − A) = λ2 − λ − 2 = 0 (3)
λ1,2 = −1, 2 (4)
(c)
det(λI − A) = λ2 − 3 = 0 (5)
√
λ1,2 = ± 3 (6)

(a) plot for 1(a) (b) plot for 1(b) (c) plot for 1(c)

1
Problem 2
(a)
−x1 + x21
   
ẋ1
= (7)
ẋ2 x1 + x2
   
ẋ 0
The equilibrium points ( 1 = ) are
ẋ2 0

(x1 , x2 ) = (0, 0), (1, −1) (8)

Then Jacobian at equilibrium points are

     
∂f 2x1 − 1 0 −1 0 1 0
= ⇒ , at equilibrium (9)
∂x 1 1 1 1 1 1

Obtain eigenvalues of each Jacobian then,

Figure 2: plot for 2(a)

(b) Let x1 = x, x2 = ẋ, then    

ẋ1 x2
= (10)
ẋ2 −x1 + (1 − x21 )ϵx2
The equilibrium point is (x1 , x2 ) = (0, 0) and the Jacobian is,
   
∂f 0 1 0 1
= ⇒ at equilibrium (11)
∂x −1 − 2ϵx1 x2 (1 − x21 )ϵ −1 ϵ

The eigenvalues of the Jacobian are

√
ϵ± ϵ2 − 4
λ1,2 = (12)
2
If |ϵ| < 2, the plots are figure 3.

If |ϵ| = 2, saddle point(one positive and one negative real eigenvalues). If ϵ < −2, sink(two negative
real eigenvalues). If ϵ > 2, source(two positive real eigenvalues).

2
 Figure 3: plot for 2(b)

Problem 3
(a) The equilibrium points are obtained by setting ẋ = 0. For K ̸= −2, the origin is the unique
equilibrium point. When K = −2, the line x1 = 2x2 is an equilibrium set.
(b) The Jacobian is given by  
∂f −1 −K
= (13)
∂x 1 −2
with eigenvalues r
3 1
λ=− ± −K (14)
2 4
Thus, the closed loop system is asymptotically stable about the origin for K > −2. Depending on
the value of K, we can origin has the following character
1
<K decaying oscillation (stable)
4
1
−2<K < stable node
4
K < −2 saddle (unstable)

Problem 4
The equation of motion for mass m1 is

m1 ẍ1 = c1 (ẋ2 − ẋ1 ) + k1 (x2 − x1 ) (15)

For mass m2 ,
m2 ẍ2 = −c1 (ẋ2 − ẋ1 ) − k1 (x2 − x1 ) + k2 (y − x2 ) (16)
Let z1 = x1 , z2 = x˙1 , z3 = x2 , z4 = x˙2 . Then, we can convert the equation of motion into the
vector differential equation form.
0 1 0 0
      
z˙1 z1 0
z˙2  − m k1 c1 k1 c1   
 = − m1 m1 m1  z2  +  0  y
 
1 (17)
z˙3   0 0 0 1  z3   0 
k1 c1 k
z˙4 m2 m2 − k1m+k
2
2
−mc1
2
z4 2
m2

3
Problem 5
(a) The fluid gives dissipative torque to two shafts which is proportional to relative angular velocity.
Then the equations of motion for two shafts are given as follows:

I1 ẇ1 = −T1 − cT (w1 − wd ),

Id ẇd = Td − cT (wd − w1 ).

(b) Assume the external torque T1 and Td are zero. Then the equations of motion becomes:
       
I1 0 ẇ1 cT −cT w1 0
+ = ,
0 Id ẇd −cT cT wd 0
 1
− I11
   
ẇ1 w1
= −cT I11 1 .
ẇd − Id Id wd
Then by solving the matrix exponential using eigenvalues and eigenvectors, the solution for w1 (t)
and wd (t) is given as follows:
 I +I
  I +I

1 −c 1 d t Id −c 1 d t
w1 (t) = I1 + Id e T I1 Id w1 (0) + 1 − e T I1 Id wd (0),
I1 + Id I1 + Id
 I +I
  I +I

I1 −cT I1 I d t 1 −cT I1 I d t
wd (t) = 1−e 1 d w1 (0) + Id + I1 e 1 d w1 (0).
I1 + Id I1 + Id

4
Problem 6
(a) The free body diagram for this system is shown in Figure 4. Since the problem assume angle ϕ
to be small, we can approximate sin ϕ, cos ϕ as ϕ, 1 respectively. Frome the diagram we can write
the equations of motion as
mL2 ϕ̈ = mgLϕ − kL21 ϕ − cL21 ϕ̇
or more specifically,
mL2 ϕ̈ + cL21 ϕ̇ + (kL21 − mgL)ϕ = 0

Figure 4: Free body diagram of Problem 6

(b) It is obvious that equilibrium points are ϕ = ϕ̇ = 0. In order to determine the condition for
stability, we have to analyze the equations of motions we derived above. Simplifying the equation,
we get
cL21 (kL21 − mgL)
ϕ̈ + aϕ̇ + bϕ = 0, where a = , b =
mL2 mL2
By defining x1 = ϕ and x2 = ϕ̇, the equation of motion can be expressed in vector differential
equation form.  
0 1
ẋ = x
−b −a
√
Thus, the characteristic equation is s2 + as + b = 0 and the eigenvalues are 12 (−a ± a2 − 4b). Note
that a ≥ 0, since all the parameters consisting a couldn’t be negative. If b > 14 a2 , the real part of
the eigenvalues are − 12 a, which is always negative. Thus, it is stable. If 41 a2 ≥ b > 0, eigenvalues
are real and negative. Thus, it is also stable in this case. If 0 > b, one of the eigenvalues is real and
positive. In this case, system is unstable. If b = 0, it is impossible to discuss about stability with
linearization theorem. In this case, nonlinear equation of motion considering the geometry should
be considered.
In conclusion, the system is stable when b > 0, or kL21 > mgL, which indicates that the stiffness of
the spring is big enough to hold on to the torque generated by the mass.

5
Problem 7
(a)

mL2 θ̈ = −mgL sin θ

g
θ̈ + sin θ = 0
L

(b) Set x1 = θ, x2 = θ̇ and x = [x1 x2 ]T , then

   
ẋ1 x2
ẋ = = = f (x)
ẋ2 −(g/L) sin x1

Equilibrium point is when ẋ = 0, or sin x1 = 0 and x2 = 0. That is, when x̄ = [0 0]T or [π 0]T are
equilibrium points.
Linearize the differential equation by setting x = x̄ + η as follow:
∂f
η̇ = (x̄)η
∂x
where  
∂f 0 1
(x̄) =
∂x −(g/L) cos x̄1 0
Therefore, the linearized differential equation is
 
0 1
η̇ = Aη where A =
−(g/L) cos x̄1 0

The solution of above equation is

η(t) = eAt η(0)
p
The eigenvalues of eAt is same with them of A, so they are ± −(g/L) cos x̄1 . Same as the eigen-
values, the eigenvectors of eAt is same with them of A, so they are
 
p 1
± −(g/L) cos x̄1
p
(i) When x̄ = [0 0]T , eigenvalules are ±i g/L, so the phase plot near the equilibrium point is
drawn as in Figure 5.
p h p iT
(ii) When x̄ = [π 0]T , eigenvalules are ± g/L and eigenvalues are 1 ± g/L . Therefore the
phase plot near the equilibrium point is drawn as in Figure 5.

6
 Figure 5: Phase portrait of two equilibrium points

Problem 8

1
T = m(Lϕ̇)2
2
1
V = k(ϕL1 )2 + mgL cos ϕ
2
1
R = c(Lϕ̇)2
2
Euler-Lagrange equation with Rayleigh’s dissipation function R becomes
d ∂L ∂L ∂R
− + =0
dt ∂ ϕ̇ ∂ϕ ∂ ϕ̇
where the Lagrangian L = T − V . Substituting T, V and R into Euler-Lagrange equation, we have

mL2 ϕ̈ + kL21 ϕ − mgL sin ϕ + cL2 ϕ̇ = 0

Approximating sin ϕ as ϕ, the equation of motion becomes

mL2 ϕ̈ + kL21 ϕ − mgLϕ + cL2 ϕ̇ = 0

which is same result as problem 6.

7
Problem 9
The root s1 , s2 are p
s1,2 = −ζωn ± ωn ζ 2 − 1.

• Overdamped: ζ > 1. The roots s1 and s2 are real and distinct, and the solution x(t) is

x(t) = c1 es1 t + c2 es2 t ,

which is the sum of two decaying exponentials. The initial conditions are x(0) = 1 and
ẋ(0) = 0, and c1 and c2 can be calculated as

1 ζ
c1,2 = ± p
2 2 ζ2 − 1

• Critically damped: ζ = 1. The roots s1,2 = −ωn are real and equal, and the solution is

x(t) = (c1 + c2 t)e−ωn t ,

which is a decaying exponential multiplied by a linear function of time. c1 and c2 can be

calculated as
c1 = 1 and c2 = ωn

• Underdamped:
p ζ < 1. The roots s1,2 are complex conjugates at s1,2 = −ζωn ± jωd , where
ωd = ωn 1 − ζ 2 is the damped natural frequency. The solution is

x(t) = (c1 cos wd t + c2 sin ωd t)e−ζωn t ,

which is a decaying exponential multiplied by a sinusoid. c1 and c2 can be calculated as

ζ
c1 = 1 and c2 = p
1 − ζ2

Figure 6: The plot of overdamped, critically damped, underdamped