Professional Documents
Culture Documents
Performance Parameters - Module II Reading III-2
Performance Parameters - Module II Reading III-2
Performance Parameters - Module II Reading III-2
________________________________________________________________________________
SYSTEM RESPONSE
1. Introduction
2. Response Analysis of First-Order Systems
3. Second-Order Systems
4. Sinusoidal Response of the System
5. Bode Diagrams
6. Basic Facts About Engineering Systems
1. Introduction
The order of a system is defined as being the highest power of derivative in the differential
equation, or being the highest power of s in the denominator of the transfer function. A first-order
system only has s to the power one in the denominator, while a second-order system has the highest
power of s in the denominator being two.
Types of the input functions (or test input signals) commonly used are:
Impulse function: In the time domain, u(t) = c(t). In the s domain, U(s) = c.
Step function: In the time domain, u(t) = c. In the s domain, U(s) = c/s.
Ramp function: In the time domain, u(t) = ct. In the s domain, U(s) = c/s2.
Sinusoidal function: In the time domain, u(t) = csin(t). In the s domain, U(s) = c/(s2+2).
where c is a constant in all the above.
With these test signals, mathematical and experimental analyses of control systems can be carried
out easily since the signals are very simple functions of time.
Which of these typical signals to use for analysing system characteristics may be determined by the
form of the input that the system will be subjected to most frequently under normal operation. If
the inputs to a control system are gradually changing functions of time, then a ramp function of time
may be a good test signal. Similarly, if a system is subjected to sudden disturbances, a step function
of time may be a good test signal, and for a system subjected to a shock input, a pulse or an impulse
function may be best.
Exercise: What are the orders of the systems described by the following transfer functions:
1
a) G (s) =
2
ms + bs + k
1
b) G (s) =
RCs + 1
1
c) G (s) =
2
LCs + RCs + 1
________________________________________________________________________________
The time response of a control system consists of two parts: the transient response and the steadystate response. The transient response is defined as the part of the time response which goes from
the initial state to the final state and reduces to zero as time becomes very large. The steady-state
response is defined as the behaviour of the system as t approaches infinity after the transients have
died out. Thus the system response y(t) may be written as:
y(t) = yt(t) + yss(t)
where yt(t) denotes the transient response, and yss(t) denotes the steady-state response.
2. Response Analysis of First-Order Systems
Many systems are approximately first-order. The important feature is that the storage of mass,
momentum and energy can be captured by one parameter. Examples of first-order systems are
velocity of a car on the road, control of the velocity of a rotating system, electric systems where
energy storage is essentially in one capacitor or one inductor, incompressible fluid flow in a pipe,
level control of a tank, pressure control in a gas tank, temperature in a body with essentially
uniform temperature distribution (e.g. steam filled vessel). Next we will present several examples
to show how to obtain the dynamic equations of first-order systems.
Example 1: Mechanical system
m is the mass, u(t) is the external force, y(t) is the velocity
and b is the friction coefficient. By Newtons law, we
have the following differential equation:
m
y(t)
u(t)
dy( t )
+ by( t ) = u ( t )
dt
by(t)
y(t)
Thus
RC
dy( t )
+ y( t ) = u ( t )
dt
R(s)
Y(s)
+ _
R(s)
1/Ts
1
Ts + 1
Y(s)
________________________________________________________________________________
Tasks: Write the system outputs or responses to inputs such as the unit-step, unit-ramp, and unitimpulse functions, respectively. The initial conditions are assumed to be zero. Draw the response
curves. T is the time constant of the system.
2.1. Unit-step response of first-order systems
1
s(Ts + 1)
1
T
1
1
=
Expanding Y(s) into partial fractions: Y(s) =
s Ts + 1 s s + 1 / T
R(s) = 1/s, and therefore the unit-step response is: Y(s) =
The solution has two parts: a steady-state response: yss(t) = 1, and a transient response:
y t ( t ) = e t / T , which decays to zero as t .
Slope = 1/T
Unit-step response, T = 1
1
0.9
0.8
0.7
0.632
0.6
y(t) 0.5
0.4
0.3
0.2
0.1
0
Tr
t/T
Ts
At t = T, y(T) = 1 e-1 = 0.632. T is called the time constant, and it is the time it takes for the
step response to rise to 63.2% of its final value.
y(2T) = 0.865; y(3T) = 0.95; y(4T) = 0.982; y(5T) = 0.993 ... It can be seen that for t 4T, the
response y(t) remains within 2% of the final value; this time is known as the settling time, Ts.
________________________________________________________________________________
The rise time, Tr, is defined as the time for the waveform to go from 10% to 90% of its final
value.
The steady-state error is the error after the transient response has decayed leaving only the
continuous response. The error signal:
e(t) = r(t) - y(t) = 1 - 1 + e-t/T = e-t/T
As t approaches infinity, e-t/T approaches zero and the steady-state error is:
e ss = e() = lim[r ( t ) y( t )] = 0
t
The larger the time constant T is, the slower the system response is.
It is noted that the transient response dominates the response of the system at times
immediately after the input is applied and can make significant contribution to the system
response when the time constant is large.
Y(s)
2
=
R (s) 10s + 4
For a step input r(t) = 2V, what is the time taken for the output of the RC circuit to reach 95% of its
steady-state response?
Y(s)
50
=
R (s) s + 50
Find the time constant, T, the settling time, Ts, and the rise time, Tr for a unit-step input.
Exercise: A system has transfer function:
________________________________________________________________________________
2.2. Unit-ramp response of first-order systems
Unit-ramp response, T = 1
6
1 1
s 2 Ts + 1
Steady-state error:
e ss = e() = lim[r ( t ) y( t )] = T
Steady-state error
4
r(t) = t
y(t) 3
t/T
Unit-impulse response, T = 1
1.2
1/T
0.8
1
Ts + 1
y(t)
0.6
0.4
0.2
t/T
________________________________________________________________________________
3. Second-Order Systems
Example 1: Mechanical system
For the mechanical system shown in the figure, m is the
mass, k is the spring constant, b is the friction
coefficient, u(t) is the external force and y(t) is the
displacement. From Newtons second law force = ma:
y(t)
u(t)
d 2 y( t )
dy( t )
m
+b
+ ky( t ) = u ( t )
2
dt
dt
di( t )
+ y( t )
dt
u(t)
dy( t )
where i( t ) = L
dt
Hence:
y(t)
C
LC
d 2 y( t )
dy( t )
+ RC
+ y( t ) = u ( t )
2
dt
dt
Transfer function:
k 2n
Y(s)
= 2
R (s) s + 2 n s + 2n
k: the gain of the system
: the damping ratio of the system
n: the (undamped) natural frequency of the system
and
Three cases:
= 1, critically damped case
> 1, overdamped case
0 < < 1, underdamped case
s 2 = n + n 2 1
________________________________________________________________________________
3.1. Step Response of Second-Order Systems
3.1.1. Critically damped case ( = 1)
Two equal poles: s1 = s2 = -n
2n
For a unit-step input R(s) = 1/s, the output is: Y(s) =
s(s + n ) 2
n
1
1
nt
(1 +
t)
n=3
0.8
n=2
0.7
s-plane
0.6
y(t)
-n
0.5
0.4
0.3
0.2
0.1
0
t (sec)
Exercise: A system has the following transfer function:
Y(s)
1
= 2
R (s) s + 8s + 16
What is the state of damping of the system when it is subjected to a unit-step input? Determine the
natural frequency of the system.
________________________________________________________________________________
3.1.2. Overdamped case ( > 1)
We can write the transfer function of a second-order system by factoring the denominator as:
2n
2n
Y(s)
=
=
R (s) (s + n + n 2 1)(s + n n 2 1) (s s1 )(s s 2 )
Taking the inverse Laplace transform yields the time response (prove this time response as an
exercise):
e s1t e s 2 t
n
y(t) = 1
2
s
s
2
1 1
2
When is much greater than unity, i.e. >> 1, then |s1| >> |s2| and the term involving s1 in the
time response will decay faster than the term involving s2. The term involving s1 can therefore
be neglected and the system becomes first-order decided mainly by the pole s2:
|s1 | >> |s 2 |
2n
2n / s1
2n / s1
Y(s)
=
=
2n
s(s s1 )(s s 2 )
s2
Y(s)
=
R (s) s s 2
( 2 1 )t
, t0
y(t) = 1 es 2 t
0.8
0.7
s-plane
0.6
y(t)
n ( 1)
2
e s1t es 2 t
y(t) = 1
s 2
2 2 1 s1
n
0.5
0.4
n ( + 1)
0.3
0.2
0.1
0
10
15
t (sec)
>>1, s1 can be neglected.
20
25
________________________________________________________________________________
Unit-step responses of 2nd order system (overdamped case), =1.1, n=1
1
0.9
0.8
y(t) = 1 es 2 t
0.7
0.6
y(t)
0.5
es1t e s 2 t
2
s
s
1 1
2
y(t) = 1
0.4
2
0.3
0.2
0.1
0
t (sec)
1, s1 cannot be neglected.
3.1.3. Underdamped case (0 < < 1)
Transfer function:
2n
Y(s)
=
R (s) (s + n + j d )(s + n j d )
j
s-plane
s1 = n j n 1
and s 2 = n + j n 1
1 2
= tan
-n+jd
= cos()
-n
-n-j d
-d
s + 2 n
s + n
n
1
1
Y(s) = 2
=
2
2
2
s s + 2 n s + n s (s + n ) + d (s + n ) 2 + d2
Taking the inverse Laplace transform using the table of Laplace transforms yields:
s + n
= e n t cos(d t )
2
2
(s + n ) + d
n
= e n t sin( d t )
2
2
(s + n ) + d
-1
-1
________________________________________________________________________________
y( t ) = 1 e n t cos(d t ) +
sin( d t )
1 2
n t
e
= 1
sin( d t + )
1 2
1 2
where = tan
When = 0, the response becomes undamped and oscillations continue indefinitely at frequency
n. The time response in this case becomes:
y(t) = 1 cos(nt)
The natural undamped frequency, n, is the frequency of oscillation of the system without
damping.
Unit-step responses of 2nd order system (underdamped case), n=3
2
=0
1.8
=0.1
1.6
=0.5
1.4
=0.9
1.2
y(t)
1
0.8
0.6
0.4
0.2
0
0
t (sec)
10
________________________________________________________________________________
3.2. Transient system specifications
Maximum overshoot
1.2
0.02
1
0.8
y(t)
0.6
0.4
0.2
Tr
t (sec)
Tp
Ts
3.2.1. Maximum overshoot
The maximum amount by which the system output response proceeds beyond the desired response.
Let ymax denotes the maximum value of y(t), and yss = y() the steady-state value of y(t), then the
maximum overshoot of y(t) is defined as:
maximum overshoot = ymax - yss
The maximum overshoot is often represented by a percentage of the final value of the step response:
y y ss
y ss
1
3.2.2. Peak time, Tp
The time required for the response to reach the first peak of the overshoot:
Tp =
n 1 2
3.2.3. Rise time, Tr
The time required for the step response to rise from 10% to 90% of its final value for critical and
overdamped cases, and from 0% to 100% for underdamped cases. For the underdamped case:
Tr =
1
2
tan ( 1 / )
=
d
n 1 2
11
________________________________________________________________________________
3.2.4. Settling time, Ts
The time required for the step response to settle within a certain percentage of its final value. A
frequently used figure is 2% in which case the settling time is approximately:
4
Ts
n
3
For varying within 5% of the final value, the setting time is: Ts
n
Note that NOT all these specifications necessarily apply to any given case. For example, for an
overdamped system, the terms peak time and maximum overshoot do not apply.
From the design requirement, the swifter and closer, the better. However, when n is fixed,
small Tr and Tp require a small , while small Ts and PO require a large . Note that
Exercise: A second-order system is underdamped with a damping ratio of 0.4 and a natural
frequency of 10Hz. Find:
a) the transfer function
b) the time response when it is subjected to a unit-step input
c) the percentage overshoot with such an input
d) the rise time
12
________________________________________________________________________________
Exercise: Given the transfer function:
G (s) =
100
s + 15s + 100
2
k 2n
s 2 (s 2 + 2 n s + 2n )
13
________________________________________________________________________________
3.4. Impulse response of a second-order system
The output transform is therefore equal to the transfer function of the system, i.e.:
k 2n
Y(s) = 2
s + 2 n s + 2n
Fact: the unit-impulse function is the time derivative of the unit-step function.
Therefore, the impulse response of a LTI system can be found from the time derivative of the
step response for a given damping.
Taking the example of a critically damped system where =1, the unit-step response is given by:
y(t) = 1 e n t (1 + n t)
dy(t)
= 2n te n t
dt
Note, again, that the impulse input gives the transfer function of the system.
14
________________________________________________________________________________
4. Sinusoidal Response of the System
Although step responses are commonly used in both simulation and experimental tests, it is also
common to undertake frequency response tests on the system.
The frequency response of a system is defined as the steady-state response of the system to a
sinusoidal input signal.
u(t) = Uosin(t)
yss(t) = Uo|G(j)|sin(t+)
Uo|G(j)|
Uo
The linear, time-invariant system G(s) subjected to a sinusoidal input of amplitude Uo and
frequency describe by:
u(t) = Uosin(t)
will, at steady state, have a sinusoidal output of the same frequency as the input but, generally, with
different amplitude and phase given by:
yss(t) = Uo|G(j)|sin(t + ())
where Uo|G(j)| is the amplitude of the output sine wave:
| G ( j) |= {Re[G ( j)]}2 + {Im[G ( j)]}2
The sinusoidal transfer function of any linear system is obtained by substituting j for s in the
transfer function of the system.
Proof:
Consider a system described by:
Y(s)
= G (s)
U(s)
u(t)
U(s)
The input u(t) is a sine wave with and amplitude Uo and frequency :
u(t) = Uosin(t)
The Laplace transform of u(t) is:
U(s) =
Uo
s + 2
2
15
y(s)
G(s)
Y(s)
________________________________________________________________________________
With zero initial conditions, the Laplace transform of the output is:
U
Y(s) = G (s) 2 o 2
s +
A partial fraction expansion of a general system (assuming the poles of G(s) are distinct) yields:
c1
c2
cn
c0
c *0
Y(s) =
+
+L+
+
+
s + p1 s + p 2
s + p n s + j s j
14444
4244444
3
Partial fraction terms from G(s)
where cn are constants and c0 and c *0 are a complex conjugate pair that can be obtained using the
cover-up rule:
G ( j) U o
G ( j) U o
c0 =
c *0 =
2j
2j
Since G(j) is a complex quantity, it can be written in the form:
G(j) = |G(j)|ej
where |G(j)| is the magnitude and is the phase given respectively by:
Im[G ( j)]
= G ( j)
| G ( j) |= {Re[G ( j)]}2 + {Im[G ( j)]}2
() = tan 1
Re[G ( j)]
Similarly:
Therefore:
| G ( j) | e j U o
c0 =
2j
| G ( j) | e j U o
c =
2j
*
0
Substituting for c0 and c *0 and noting that sin(x) = (ejx e-jx)/(2j), gives the steady-state output:
yss(t) = Uo|G(j)|sin(t + )
16
________________________________________________________________________________
The amplitude and phase of the frequency response can be used to predict both time-domain
transient and steady-state system performances.
Systems may be designed to achieve transient and steady-state requirements using frequency
response analysis, and such analysis and design may be extended to certain nonlinear control
systems.
Exercise: For the sinusoidal input u(t) = sin(10t) applied to the system: G (s) =
Amplitude
0.1
-0.1
0
4
6
Time (sec.)
17
10
1
,
s+2
________________________________________________________________________________
There are two commonly used representations of sinusoidal transfer functions:
1) Nyquist or polar plot, and
2) Bode diagram
We shall focus on the more popular Bode analysis and show how we can use MATLAB to produce
these plots.
5. Bode Diagrams (Asymptotic Approximation)
plot of the logarithm of the magnitude of the a sinusoidal transfer function, |G(j)|
plot of the phase angle, ()
both are plotted against the frequency (rad/s) on a logarithmic (base 10) scale.
Logarithmic magnitude (also called gain) of G(j), M = 20log10|G(j)| (unit in decibels, dB)
X()
Phase: () = tan 1
R ()
Bode plots of systems in series simply add, which is quite convenient. For example,
consider the transfer function:
b (s z1 )(s z 2 ) L (s z m )
G (s) = m
(s p1 )(s p 2 ) L (s p n )
The magnitude of the frequency response of the system is given by:
b m (s z1 ) (s z 2 ) L (s z m )
G ( j) =
(s p1 ) (s p 2 ) L (s p n ) s j
Taking the logarithm yields:
20log10|G(j)| = 20log10bm + 20log10|(s - z1) + 20log10|(s z2)| +
- 20log10|(s - p1) - 20log10|(s p2)| - |sj
Knowing the respone of each term, the algebraic sum would give the total response in dB.
18
________________________________________________________________________________
Example: Find the Bode plots for the following RC filter
1
1
=
with: G ( j) =
RC( j) + 1 jT + 1
u(t)
Solution:
Magnitude: G ( j) =
1
1+ T
2
y(t)
1/ T
1 / T 2 + 2
1
1
j T
=
= R () + jX()
2 2
1 + j T 1 + T
1 + 2 T 2
X
Phase: () = tan 1 = tan 1 (T ) (rad)
R
At low frequencies ( 0): (0) tan 1 (0) = 0 rad
At high frequencies ( ): () tan 1 ( ) = / 2 rad
At corner frequency ( 1/T): (1 / T) tan 1 (1) = / 4 rad
19
________________________________________________________________________________
Asymptotic curve of |G(j)|
Asymptote
Corner frequency
Exact curve
0
Asymptote
-5
Asymptote
Exact curve
dB
-/4
-10
-15
-20
0.01/T
0.1/T
1/T
-/2
0.01/T
10/T
0.1/T
Exercise: Sketch the Bode plot for the following transfer function: G (s) =
20
20
4s + 1
1/T
10/T
________________________________________________________________________________
Bode Diagrams
Magnitude (dB)
30
25
20
15
10
Phase (deg)
0
-20
-40
-60
-80
10 -2
10-1
100
(rad/sec)
The MATLAB command bode(SYS) computes the logarithmic gain and phase angles of the
frequency response of the LTI SYS=tf(num,den), where num and den are the numerator and
denominator coefficients of the system, respectively. For example, to plot the Bode diagrams
shown for the transfer function of the previous exercise, we enter on the MATLAB command line:
num = [0 20];
den = [4 1];
SYS = tf(num,den);
bode(SYS) or bode(num,den)
21
________________________________________________________________________________
Bode Diagrams
30
25
20
15
10
0
-20
-40
-60
-80
10-2
10-1
100
Frequency (rad/sec)
22
________________________________________________________________________________
up to the plant. Even those controllers that are available commercially require an element of design
either off-line or on-line and hence an effective engineer requires an appreciation of the design
process and design tools currently available.
CONTROLLERS DEPEND ON THE DETAILED DYNAMICAL CHARACTERISTICS OF THE
PROCESS TO BE CONTROLLED If that were not the case then it would only be necessary to
have one control system on sale. Practical experience has shown that this is not feasible it is
found that it is necessary to have appropriate data and some understanding of the process to be
controlled. This typically takes the form of EITHER
1) a mathematical model expressed in differential equation, transfer function or state space
form, AND/OR
2) data on the behaviour of the plant output(s) in response to known inputs (such as steps or
sinusoids) from which
(a) the desired parameters for the model (obtained in (1)) can be estimated or
(b) if a physical model is not available a model can be constructed from a curve
fitting or a, so-called, identification procedure.
THE DEGREE OF CONTROL OBTAINED DEPENDES ON THE AVAILABILITY OF
SUITABLE MEASUREMENTS OF SYSTEM BEHAVIOUR and the more accurate and
extensive these measurements are, the better control will be.
YOU CANNOT ALWAYS MEASURE WHAT YOU WANT TO MEASURE if you cannot
measure the output variable of interest (due to extreme physical conditions of speed, temperature or
pressure .. etc, it is necessary to create an intelligent device which observes the (available)
measurement and uses them to create a useful estimate of the (unavailable) output. As an example,
how is it possible to control the temperature in the centre of a furnace when the temperature sensors
are placed on the external wall?
CONTROLLERS CAN DO AMAZING THINS As control requirements are as varied as the
applications and needs for new products, even the virtually impossible has been asked for in the
search for the intelligent controller.
This requires the development of an abstract way of thinking but has amazing consequences e.g.
(a) the development of control elements capable of observing and accurately estimating
variables that cannot be measured
(b) the development of control systems capable of adapting to new situations and learning
from experience.
23
________________________________________________________________________________
dy( t )
+ y( t ) = ku ( t )
dt
where T is the time constant, k is the gain, u(t) and y(t) are the input and output of the system
respectively. The unit-step response of a first-order system is shown in the following figure.
Determine the parameters k and T from this figure.
Step Response
0.5
0.45
0.4
Amplitude
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
6
Time (sec.)
24
10
12
________________________________________________________________________________
4. Consider the first-order system,
Y(s)
1
=
R (s) Ts + 1
Obtain the unit-step response curves for T = 0.1, 0.5, 1.0, 5.0 and 10.0 respectively, with
MATLAB.
5. Consider the first-order system,
Y(s)
k
=
R (s) s + 1
Obtain the unit-step response curves for k = 0.1, 0.5, 1.0, 5.0, and 10.0 respectively, with
MATLAB.
6. A general second-order system has the form:
k 2n
Y(s)
= 2
R (s) s + 2 n s + 2n
What are the values of k, , and n for the following system:
Y (s)
3
= 2
R (s) s + 2s + 9
7. A second-order system is described by the differential equation:
d 2 y( t )
dy( t )
+5
+ 25y( t ) = 25u ( t )
2
dt
dt
a) Write down the transfer function Y(s)/U(s) of the system, where U(s) and Y(s) are the Laplace
transforms of u(t) and y(t), respectively.
b) Obtain the damping ratio and the natural frequency n of the system.
c) Calculate the rise time and percent overshoot of the system.
d) Evaluate y(t) for a unit-step input u(t).
e) Check your answers of the above with MATLAB.
8. For the control system shown by the block diagram, the numerical value of J = 1 kg-m2 and B =
1 N-m/(rad/sec).
R(s)
+_
+_
1
Js + B
K1
Y(s)
1/s
K2
________________________________________________________________________________
9. When the second-order system
Y(s)
K
= 2
R (s) Ts + s + K
is subjected to a unit-step input, the system output responds as shown in the following figure.
Determine K and T from the response curve.
Step Response
1.4
1.31
1.2
Amplitude
1
0.8
0.6
0.4
0.2
0.55
0
0.5
1.5
2.5
Time (sec.)
10. A sinusoidal input u(t) = 2sin(2t) is applied to a system with transfer function:
Y(s)
2
=
U(s) s(s + 2)
KP
1
s2
KD s
26
Y(s)
________________________________________________________________________________
12. Consider the second-order system
Y(s)
1
= 2
R (s) s + 2s + 1
Obtain the unit-impulse response curves for = 0.1, 0.3, 0.5, 0.7, 1.0, and 4.0 respectively, with
MATLAB.
13. Consider the second-order system
k 2n
Y(s)
=
R (s) s 2 + 2 n s + 2n
Assuming that n = 2, k = 2, obtain the unit-impulse response curves for = 0.1, 0.3, 0.5, 0.7, 1.0,
and 4.0 respectively, with MATLAB.
27