Final.1 (10 pts.

) Short Answer

Mnemonic for remembering since, cosine, and tangent?

Log-log and semi-log graphs that show the magnitude and phase
of the output of a linear ODE in response to harmonic forcing at
various frequencies.
Type of filter analogous to human hearing which passes an audible
range of frequencies from 20 Hz to 20 KHz
Symbol, name, and type (real, imaginary, or complex) for the two in-
dependent variables used in dynamic systems
Method used to analytically find the steady-state solution to a 3rd
order harmonically forced ODE
The name of the approach to removing vibration with a fixed fre-
quency from a system by adding a spring and a mass
The phase shift a 2nd-order low-pass filter at resonance
Definition of (equation for) the natural frequency of a 2nd-order
system
2
Location of the poles of a system of the form ddt 2y c1 dy
dt + c0 y = 0,
where c1 and c0 are positive constants
Form in which dividing two complex numbers is easier
The inclusion of a small amount of this term in a feedback control
system will drive the steady-state error to zero
 Final.2 (11 pts.) Inhomogeneous ODEs and Forcing Functions
(a) (5 pts.) Given the following ODE:

3ẋ + 2x = F(t)
If F(t) = 16 and x(0) = 2, find x(t).

x(t) =

(b) (3 pts.) Plot the steady state behavior, x(t)ss , for the system in part (a).
(c) (6 pts.) Transfer functions and system response
The transfer function for an ODE is
Y (s) 1
= 2
F (s) s + 8s + 7

Determine y(t) and sketch its solution for 0  t  5 when

y(0) = 3, ẏ(0) = 0, and f (t) = 0.

y(t) =

The dominant pole is p = .

(The dominant pole is the pole mostly responsible for y(t)’s behavior.)

(c) (3 pts.) For this system, if F(t) = 3sin(Wt + p/2), what value of W will produce the largest magnitude
steady-state response?

(d) (6 pts.) 1st -order transfer function and steady-state response

Determine the steady-state response of the system described by
the following transfer function when f (t) = 2 sin(3 t).

Y (s) 1
= ) y(t)ss =
F (s) s+1

Plot both f (t) = 2 sin(3 t) and y(t)ss for 0  t  6.

 Final.3 (12 pts.) Thinking in s and t
F.9 (10 pts.) Thinking in s and t
Y (s)
For each transfer function G(s) = in the following table:
F (s)
• Identify the poles of G(s), determine stability, and form a related ODE
• Sketch the unforced response for y(t) for initial values y(t = 0) = 1 and ẏ(t = 0) = 0.
• Use knowledge of inhomogeneous ODEs and/or the final value theorem in Section 16.7 to
calculate the “final value” (real number) lim y(t) when f (t) = 12 (a final value may not exist).
t

Transfer function and poles Related ODE and rough sketch of y(t) when f (t) = 0
G(s) = 1 ÿ + ẏ + y = f (t)
(s + 2 j) ⇤ (s 2 j)

Poles: s1 =
s2 =
stable/neutrally stable/unstable

When f (t) = 12, Final value =

G(s) = 1 ÿ + ẏ + y = f (t)
(s + 1 + 2 j) ⇤ (s + 1 2 j)

Poles: s1 =
s2 =
stable/neutrally stable/unstable

When f (t) = 12, Final value =

F.10 (3 pts.) Eigenvalues for a non-standard, non-generalized eigenvalue problem

Consider the following set of algebraic equations governing the unknowns u1 , u2 , and .

2
u1 + 5 u2 = 0 or equivalently u1 0
=
cos( ) 0.9 u1 + u2 = 0 u2 0

Find an equation which when solved produces “special values” of that allow for non-zero u1 and u2 .
Result: (These “special values” of are called eigenvalues.)
+ 4.5 = 0

Note: It is questionable whether this eigen-problem can be cast as a standard or generalized eigenvalue problem.
There turns out to be 3 eigenvalues, namely 1 = -1.7574, 2 = -0.5078, 3 = +0.4166.

5 /13
F.12 (10 pts.) State-space control with PID feedback control
Consider the following coupled linear constant-coefficient 2nd -order ODEs written in terms of the:
Final.4 (15 pts.) State-space control with PID feedback
F.12 (10
F.12 (10• pts.)
pts.) State-space
constant
n⇥n State-space control
matrices M , B,
control with
with PIDfeedback
K PID feedbackcontrolcontrol
constant matrix M Ẍ nd + B Ẋ + K X = G Fc
Consider the
•
Consider
n⇥p thefollowing
followingcoupled
coupledlinear
G linearconstant-coefficient
constant-coefficient
n n 2 2nd -order
-order
n ODEs
n ODEs nwritten
n written inninterms
pterms ofofthe:
the:
• n⇥1 variable matrix X(t) t
constantmatrices
n⇥n constant
•• n⇥n matricesMM, ,B,
B,KK
• p⇥1 control matrix Fc - - -
constantmatrix
n⇥p constant matrixGG FcMẌẌ ++
M = Kp BBẊẊ ++
X + Kd KKXX ==
Ẋ + K i GGX FcFdct̄
•• n⇥p pn n1 n n p n n n n n p n n nn n p nn np p
• p⇥n constant matrices Kp , Kd , Ki .
variablematrix
n⇥1 variable
•• n⇥1 matrixX(t)
X(t) t̄ = 0
t t
p⇥1 control
p⇥1 control
(a)••Complete matrixFFcequation
the matrix
following that multiplies F ==-FK-cK ++- K -K -K
++-matrix.
c cand pto produce dan
c GF pXX dẊẊ
n⇥1 K i i XX dt̄dt̄
••Next, constant
constantmatrices
p⇥n substitute
p⇥n matrices KKp p,the
G Fc into ,d ,KKi .i .
,KKdright-hand side pofp 1the
1 p p n n(above) pand
ODE pn n p pn n
rearrange. t̄ =t̄ 0= 0
Result t
(a)
(a) Complete
CompleteG Fthe
thefollowing
-G Kequation
c =following X + that
pequation thatmultiplies
multiplies
Ẋ G+ Gand c ctotoproduce
andFF produce X dt anann⇥1
n⇥1matrix.
matrix.
Next,
Next, substitute
substitute GGFFc c into
intothe
theright-hand
right-handside
sideofofthe
theODE (above)and
ODE(above) andrearrange.
rearrange.
t̄ = 0 t
Result
Result t t
M Ẍ + Ẋ + X + X dt = [ 0 ]
GFFcc == -G
nG n -GKKnp p n XX ++ nẊn++
Ẋ nXX ndtdt n 1
t̄ = 0
t̄ =t̄ =
00 t t

M ẌẌ ++
M
nn nn
ẊẊ ++ XX ++ dtdt== [ 0[]0 ]
XX
nn nn nn nn n nn n n n1 1
t̄ =t̄ =
0 0

(b) (2 pts.) After assuming a solution X = U ep t where p is a constant (to-be-determined) and U

is a non-zero n⇥1 matrix of constants (to-be-determined), form a matrix equation governing p
and U .
(b) Next, form a scalar equation that suffices to find
p t p (without U ).
(b) (2 pts.) After
(2 pts.) Afterassuming
assumingaasolutionsolution XX==UUe ep t where wherep pisisa a-constant
constant(to-be-determined)
(to-be-determined)andandU U
Result:
isis aa non-zero
(in termsn⇥1of p,matrix of
U , M , B, constants
K, G, Kp , K(to-be-determined)
d , Ki – but without, form
M 1 )a matrix equation governing p
non-zero n⇥1 matrix of constants (to-be-determined), form a matrix equation governing p
and
and UU..
Matrix equation for p and U : = [0]
Next,
Next, formformaascalar
scalarequation
equationthat thatsuffices
sufficestotofindfindp p(without ). ).
(withoutU U n 1
Result: (in terms of p, U , M , B, K, G, K , K , K – but without M -1
-
)
Result:
Scalar (inpterms
equation for (no Uof
p d i
): p, U , M , B, K, G, Kp , Kd , Ki – but without M )
1
= 0
Matrix equation for p and
Matrix equation for p and U : U : ==[ 0[ ]0 ]
n 1
n 1

Scalar equation for p (no U ):

Scalar equation for p (no U ):
==0 0

(c) The unknowns in the previous equations are p and U . Classify the previous matrix equation by
picking the relevant qualifiers from the following list. Next, classify the previous scalar equation.
Classify matrix equation for p and U Classify scalar equation for p
Uncoupled Linear Algebraic Uncoupled Linear Algebraic
(c) The unknownsNonlinear
Coupled in the previous equations are p and U . Classify
Di↵erential the previous matrix equation by
(c) The unknowns in the previous equations are p and U . Coupled
Classify the Nonlinear Di↵erential
previous matrix equation by
picking the relevant qualifiers from the following list. Next, classify the previous scalar equation.
picking the relevant qualifiers from the following list. Next, classify the previous scalar equation.
(d) InClassify
the context of matrix
matrix equationalgebra, the U“special values” of p Classify
for p and are calledscalar equation for
andp the
Classify matrix equation for p and U Classify scalar equation for p
Uncoupled Linear Algebraic Uncoupled Linear Algebraic
corresponding
Uncoupled U s are called
Linear Algebraic , whereas in the context of ODEs
Uncoupled and X(t)Algebraic
Linear = U ep t ,
Coupled Nonlinear Di↵erential Coupled Nonlinear Di↵erential
Coupled Nonlinear Di↵erential Coupled Nonlinear Di↵erential
these special values of p are called and the corresponding U s are called mode shapes.
(d) In the context of matrix algebra, the “special values” of p are called and the
(d) In the context of matrix algebra, the “special
7 values” of p are called and
/5 the
corresponding U s are called
. corresponding U s are called , whereas in the context of ODEs and X(t) = U ep t,
, whereas in the context of ODEs and X(t) = U ep t ,
these special values of p are called and the corresponding U s are called mode shapes.
these special values of p are called and the corresponding U s are called mode shapes.
7 /5
7 /5
(e) (2 pts.) After defining the state matrix Y as shown below determine the 3 n ⇥ 3 n matrices C
and D so the ODE in item (12a) can be cast into the 1st -order form C Ẏ = D Y .
Result: (in terms of M , B, K, G, Kp , Kd , Ki , zero matrix [ 0 ], and identity matrix I – but without M -1 )
(e) (2 pts.)2 After3 defining
2 the state matrix 3Y2 as shown3 2 below determine the 3 n ⇥ 3 n matrices 3C2 3
X 0 0 Ẋ X
and 6 1st -order form C Ẏ = D Y .
Y D=so
4 the ODE in4 item0 (12a) can0 be 5cast
4 Ẍinto 5 =the 74
Ẋ 5 4 5 Ẋ 5
Result: (in
3 n 1 ... - 1
Ẍ terms of M , B,
0 K, G,
0 Kp , Kd , Ki , zero
X matrix [ 0 ], and identity matrix I – but without M ) Ẍ
2 3 2 32 3 2 32 3
X 0 0 Ẋ X
6 74
Y = 4 Ẋ 5 4 0 0 54 Ẍ 5 = 4 5 Ẋ 5
3n 1 ...
Ẍ 0 0 X Ẍ

(f) After assuming a solution Y = U e t where is a constant (to-be-determined) and U is a non-

zero 3 n⇥1 matrix of constants (to-be-determined), form a matrix equation governing and U .
Next, form a scalar equation that suffices to find (without U ).
Result:
(f) After assuming a solution
(in terms of , C, DY– but C -1 , then
e t where
= Uwithout is aexpress
constant
C and D in terms of K,and
(to-be-determined) G, KUp , is
Kda, Ki)
non-
zero 3 n⇥1 matrix of constants (to-be-determined), form a matrix equation governing and U .
Matrix equation for and U : = [0]
Next, form a scalar equation that suffices to find (without U ). 3n 1
Result: (in termsScalar -1
of , C, D – but without C , then express C and D in terms of K, G, Kp , Kd , Ki )
equation for (no U ): = 0
Matrix equation for and U : 2 = [0] 3
2 3 3n 1
6 7
C = 4 5
Scalar equation for (no U ): D = 4 = 0 5

2 3 2 3
6 7
C = 4 5 D = 4 5

(g) (1 pt.) Since the previous generalized eigenvalue problem has the form D U = C U , the
“special values” of can be solved with the MATLAB R command eig(D,C).
After solving for , how does one know whether there is a stable response for X(t)?
Result:
(g) (1 pt.) Since the previous generalized eigenvalue problem has the form D U = C U , the
“special values” of can be solved with the MATLAB R command eig(D,C).
8 /5
After solving for , how does one know whether there is a stable response for X(t)?
Result:

8 /5
Final.5 (5 pts.)
Given the double pendulum shown below with Frame b attached to link 1 (connecting points N0 and P) and
with Frame c attached to the springy link 2 (connecting points P and Q).

If N w
~ B = q̇~bz , B w
~ C = ḟ~bz = ḟ~cz , and N w
~ C = (q̇ + ḟ )~cz ,
and the velocity of Q in frame n is: N~V Q = ẏ~cx + (q̇ + ḟ )(Ln1 + y)~cy + q̇ L1~bx
and the velocity of Q in frame b is: B~V Q = ẏ~cx + ḟ (Ln1 + y)~cy
Find the acceleration of Q with respect to the rigid link, frame b.

B~
aQ =
 Vi# L# Vo#
Final.6 (11 pts.)
Given this electrical system: R# C#

(a) (5 pts.) Find the equation of motion of the system describing the output voltage v0 as forced or driven
by the input voltage vi .

(b) (1 pts.) Given a different system that has the equation of motion below, find the transfer function of this
new system.

1 R 1
V̈0 + V̇0 + LRV0 = Vin
CL L CL

(c) (5 pts.) If L = 4H, R = 2MW (1E6) , and C can vary from 1µF (1E-6) to 100µF, plot the root locus of
mber addition,
the systemmultiplication, division,
from part (b) as a function of C. and exponents
Imaginary
Axis
numbers in the complex plane, and
hase form. Then, make the calcu-
de results in Cartesian and magni-
phase reported in radians).
Real
Axis

ej

ej

) = + j
 x3
F.13 (19 pts. ) Earthquake analysis of a three-story building

The schematic to the right shows a three story building. The

motion of the building’s floors are mostly horizontal and their
horizontal displacements from equilibrium from a vertical line x2
fixed in ground are denoted x1 , x2 , x3 , respectively. Each floor
is supported by several columns and the combined sti↵ness and
material damping in each floor’s columns are modeled with a
linear horizontal spring with spring constant ki (i = 1, 2, 3) and a x1
linear horizontal damper with damping constant bi (i = 1, 2, 3).
When the ground’s horizontal motion is known to be x0 (t), the
di↵erential equations governing x1 (t), x2 (t), x3 (t) are x0

m1 (x¨0 + x¨1 ) + b1 x˙1 + b2 (x˙1 x˙2 ) + k1 x1 + k2 (x1 x2 ) = 0

m2 (x¨0 + x¨2 ) + b2 (x˙2 x˙1 ) + b3 (x˙2 x˙3 ) + k2 (x2 x1 ) + k3 (x2 x3 ) = 0
m3 (x¨0 + x¨3 ) + b3 (x˙3 x˙2 ) + k3 (x3 x2 ) = 0

(a) After defining the 3⇥1 matrix X as shown below, determine the M , B, K, and G matrices
when the previous set of ODEs are cast into the matrix form M Ẍ + B Ẋ + K X = G x¨0
Result: 2 3 2 3
2 3 m1
x1 6 7 6 7
6 7 6 7
X = 4 x2 5 M = 6 6 7 6
B = 6 7
7 7
x3 4 5 4 5

2 3 2 3
6 7 6 7
6 7 6 7
K = 6
6
7
7 G = 6 7
4 5 4 5

(b) The matrices M , B, and K are symmetric. True/False.

(c) (3 pts.) When the Earth is at rest (not shaking) - and for certain values of mi , bi , ki (i = 1, 2, 3),

the ODEs can be put into the matrix form Ẍ + A X = [ 0 ] as (you do not have to show this)
2 3 2 32 3 2 3
x¨1 a11 -h 0 x1 0 where a11 , a22 , a33 are
4 x¨2 5 + 4 -h a22 -h 54 x2 5 = 4 0 5
-h a33 positive real numbers
x¨3 0 x3 0
The eigenvalues of A are equal to the roots of the following cubic polynomial equation in .
3
+ c2 2
+ c1 + c0 = 0

Express the coefficients c2 , c1 , c0 of ’s polynomial equation in terms of a11 , a22 , a33 , and h.
Result: Note: The roots of the polynomial equation are 1 20
20, 2 155
155, 3 325
325.
c2 =

c1 =

c0 =

9 /5
(d) As described in Section 24.2, the solution for X(t) can be written in terms of the constants !1 ,
!2 , !3 (called the system’s natural frequencies), three 3⇥1 matrices of constants U1 , U2 , and U3 ,
(called the system’s mode shapes), and some yet-to-be-determined constants ci and i (i = 1, 2, 3), as
3
X = ci Ui cos(!i t + i)
i=1
Find numerical values for !i (i = 1, 2, 3) (or provide a clear way to calculate them).
Result: !1 = !2 = !3 =
(e) The solution for X(t) is stable/neutrally stable/unstable (circle one).
(f) How does one usually determine numerical values for the constants ci and i (i = 1, 2, 3)?
Result:
(g) Knowing the eigenvectors U1 , U2 , U3 , draw a picture of the building in a position that corre-
sponds to the building moving in each mode. Show the direction of motion with arrows.
2 3 2 3
2 3 0.74 0.59
0.33
U2 = c2 4 0.33 5 U3 = c3 4 -0.74 5
U1 = c1 4 0.60 5
-0.60 0.33
0.74
Mode 2 Mode 3
Mode 1

x0 x0
x0

(h) In view of your pictures and natural frequencies, answer the following questions.
Mode with the largest potential energy associated with column deflection? Mode
Mode with the largest kinetic energy due to floor motion? Mode
Mode that is the “easiest to excite” (takes the least energy to excite)? Mode
(i) One model of earthquake ground motion is harmonic motion of the form x0 = Amplitude sin(⌦ t),
where Amplitude relates to the earthquake’s magnitude and ⌦ is the frequency of ground motion.
The values of ⌦ that cause significant damage to this building are: ⌦ = .
(j) Assume unmodeled damping a↵ects each mode equally, i.e., each mode is damped with ⇣ = 0.02.
The value of ⌦ that probably shakes the building longest is ⌦ =
Explain:
(k) Is it possible for the building to move only in the second mode? Yes/No.
Explain:

Congratulations on completing dynamic systems.

Best wishes on your future endeavors.