Solutions for Exercises in Dynamic Systems and Control Engineering by Jalili and Candelino
Solutions for Exercises in Dynamic Systems and Control Engineering by Jalili and Candelino
dx(t )
cos(3t ) + x(t ) = sin(t ),
8@
dt
we may first notice that this system is described by an ordinary differential equation, where
time is the only dependent variable. Therefore, this equation must describe a lumped-
parameters model. Next, we notice that x (t ) multiplies the time-varying coefficient cos(3t ).
Therefore, the model is time-varying. Finally, we notice that all terms containing x (t ) or x (t )
have coefficients that are independent of x (t ) or its derivatives. Therefore, the system is linear
and is described by a linear time-varying lumped-parameters model.
gm
2 2 y ( x, t ) 2 y ( x, t )
EI ( x ) + A = 0,
x 2 x 2 t 2
ail
we may first notice that this system contains partial derivatives with respect to time and x.
Therefore, this equation must be a distributed-parameters model. Next, we notice that all
coefficients of the dependent variable y ( x, t ) and its derivatives are independent of time.
.co
Therefore, this model is time-invariant. Finally, all coefficients of y ( x, t ) and its derivatives
are independent of y ( x, t ). Therefore, the system is linear and is described by a linear time-
invariant distributed-parameters model.
d 3 x dx
x + + 5x = t ,
dt 3 dt
t 2 x + x = e−t .
we may first notice that this system is described by an ordinary differential equation, where
time is the only dependent variable. Therefore, this equation must describe a lumped-
parameters model. Next, we notice that the coefficient of x (t ) is an explicit function of time
t 2 . Therefore, this model is time-varying. Finally, we notice that all terms containing x (t ) or
x (t ) have coefficients that are independent of x (t ) or its derivatives. Therefore, the system is
linear and is described by a linear time-varying lumped-parameters model.
3
Assuming the pairs of inputs and outputs shown in Figure ES1.2 were produced by the same
system y = f ( x), that system cannot be linear. To demonstrate this fact, consider that the
second input x2 is twice the magnitude of x1. However, the second output y2 is 2 23 the
magnitude of y1. Therefore, f (ax) af ( x). The scaling property does not hold, and the
system must be nonlinear.
Alternatively, if we consider the response of f ( x) to the third input x3 , we notice that
x3 = 12 x1. However, y3 = f ( x3 ) = f ( 12 x1 ) 12 f ( x1 ). Again, the scaling property does not hold.
Moreover, the response to x3 is delayed for a brief period before the response y3 rises to 0.1.
Therefore, superposing inputs x1 = x3 + x3 gives outputs f ( x1 ) = f ( x3 + x3 ) f ( x3 ) + f ( x 3 ),
and we see that the additivity property does not hold. In fact, neither scaling nor additivity
holds between any of the input/output pairs shown.
mx + bx + kx = 0,
x → y1 = f ( x) = mx + bx + kx
ax → y2 = f (ax) = max + bax + kax
f (ax) = a ( mx + bx + kx )
= af ( x).
f ( x1 + x2 ) = m ( x1 + x2 ) + b ( x1 + x2 ) + k ( x1 + x2 )
= mx1 + bx1 + kx1 + mx2 + bx2 + kx2
= f ( x1 ) + f ( x2 ),
which shows that f ( x1 ) + f ( x2 ) = f ( x1 + x2 ), and the additivity property holds. Therefore, the
system is linear.
d3x
+ xx + kx = 0,
dt 3
x → y1 = f ( x) = x(3) + xx + kx
ax → y2 = f (ax) = ax(3) + ( ax )( ax ) + kax
5
Therefore, the scaling property does not hold, and the system is nonlinear.
Alternatively, to test for additivity, we can apply the inputs
x1 → y1 = f ( x1 ) = x1(3) + x1 x1 + kx1
x2 → y2 = f ( x2 ) = x2(3) + x2 x2 + kx2
d 3 ( x1 + x2 ) d ( x1 + x2 )
x1 + x2 → y12 = f ( x1 + x2 ) = 3
+ ( x1 + x2 ) + k ( x1 + x2 )
dt dt
f ( x1 + x2 ) = x1(3) + x2(3) + ( x1 + x2 )( x1 + x2 ) + k ( x1 + x2 )
= x1(3) + x1 x1 + kx1 + x2(3) + x2 x2 + kx2 + x2 x1 + x1 x2
= f ( x1 ) + f ( x2 ) + x2 x1 + x1 x2
f ( x1 ) + f ( x2 ),
which shows that f ( x1 ) + f ( x2 ) f ( x1 + x2 ), and the additivity property does not hold. Again,
the system is nonlinear.
d4x 2 dx
4
+ (1 + b ) = 0,
dt dt
d4x 2 dx
x → y1 = f ( x) = + (1 + b )
dt 4 dt
d ( ax )
4
2 d ( ax )
ax → y2 = f (ax) = 4
+ (1 + b )
dt dt
f (ax) = ax (4) + (1 + b ) ax
2
(
= a x (4) + (1 + b ) x
2
)
= af ( x).
Therefore, the scaling property holds.
Now, to test for additivity, we can apply the inputs
d 4 x1 2 dx
x1 → y1 = f ( x1 ) = 4
+ (1 + b ) 1
dt dt
4
d x 2 dx2
x2 → y2 = f ( x2 ) = 42 + (1 + b )
dt dt
d ( x1 + x2 )
4
2 d ( x1 + x2 )
x1 + x2 → y12 = f ( x1 + x2 ) = 4
+ (1 + b )
dt dt
f ( x1 + x2 ) = x1(4) + x2(4) + (1 + b ) ( x1 + x2 )
2
= x1(4) + (1 + b ) x1 + x2(4) + (1 + b ) x2
2 2
= f ( x1 ) + f ( x2 ),
which shows that f ( x1 ) + f ( x2 ) = f ( x1 + x2 ), and the additivity property holds. Therefore, the
system is linear.
x + kx (1 + x2 ) = 0,
x → y1 = f ( x) = x + kx (1 + x 2 )
ax → y2 = f (ax) = ax + k ( ax ) 1 + ( ax ) ( 2
)
and relating the outputs as
(
f (ax) = ax + k ( ax ) 1 + ( ax )
2
)
= ax + kax (1 + a x 2 2
)
(
= a x + kx (1 + x 2
) ) + ka x 3 2
− kax 2
= af ( x) + ka 3 x 2 − kax 2
af ( x).
Therefore, the scaling property does not hold, and the system is nonlinear.
Alternatively, to test for additivity, we can apply the inputs
7
x1 → y1 = f ( x1 ) = x1 + kx1 (1 + x12 )
x2 → y2 = f ( x2 ) = x2 + kx2 (1 + x2 2 )
d 2 ( x1 + x2 )
x1 + x2 → y12 = f ( x1 + x2 ) =
dt 2 (
+ k ( x1 + x2 ) 1 + ( x1 + x2 )
2
)
and compare the outputs as
f ( x1 + x2 ) = x1 + x2 + k ( x1 + x2 ) (1 + x12 + 2 x1 x2 + x2 2 )
= x1 + kx1 (1 + x12 + 2 x1 x2 + x2 2 ) + x2 + kx2 (1 + x12 + 2 x1 x2 + x2 2 )
= f ( x1 ) + f ( x2 ) + kx1 ( 2 x1 x2 + x2 2 ) + kx2 ( x12 + 2 x1 x2 )
f ( x1 ) + f ( x2 ),
which shows that f ( x1 ) + f ( x2 ) f ( x1 + x2 ), and the additivity property does not hold. Again,
the system is nonlinear.
Considering a linear time-invariant system y = f ( x ) that produces the input–output pair x1
and y1 shown in Figure ES1.4a, the expected system response y2 for the more complex input
x2 , shown in Figure ES1.4b, can be found by applying the superposition principle as
demonstrated in Figure ES1.4c.
Notice, when the complex input is applied, changes to the input occur quickly enough that
the response does not finish its transient portion before the next step begins to take effect (e.g.,
the input reaches a peak amplitude of 2, but the output reaches a peak of only about 1.8).
To determine the effect of changing the spring material from steel to aluminum, we can
begin by recognizing that the spring constant for the cantilever beam is given by
Ebh3
K= ,
4 L3
where the moduli of elasticity for steel and aluminum can be taken, respectively, as
Esteel = 3 107 and EAl =1107. Therefore, the stiffness of geometrically identical
cantilever beams of steel and aluminum can be found as
Esteel 3 107
= = 3 → Esteel = 3E Al ,
E Al 1 107
Therefore, the cantilever beam becomes three times softer when the spring material is
changed from steel to aluminum. Similarly, torsion bar springs and tension rod springs also
become three times softer when the material is changed from steel to aluminum.
To plot variations in the spring constants of cantilever beams, torsion bars, and tension rods
as a function of spring length L, we can write a MATLAB program like the one below:
% Parameter definitions
E_steel = 3e7; % Modulus of Elasticity in lb-f/inch^2
b = 1; % Width in inches
h = 1; % Height in inches
r = 1; % Radius in inches