Week 4 Assignment 1 Solution

1. Choose the correct option to complete the sentence below.

“An anti-aliasing filter”

Solution: (c) “is an analog filter realised before sampling to remove low frequency errors due to
a too low rate of sampling”

“An anti-aliasing filter” is used before sampling an analog signal as it attenuates the signals with
frequencies more than the desired bandwidth. It is done to remove the effect of aliasing which is
caused by sampling at lower rate than the theoretical Nyquist rate. When aliasing occurs
frequencies towards the lower end of bandwidth get more effected.

2. The block diagram of a P-I controller is given in Fig. 1. How much time (in seconds) the
controller output (u) will take to reach 12 units when it is driven by a ramp error signal of slope
1 as shown in the figure? Assume initial condition of the integrator to be zero.

ee e 4
u
+
0 1t
u
+
2

Fig. 1
Solution: 2

1
Given e  t   tu  t  Its Laplace Transform will be: E  s  
s2
2
Controller transfer function: C  s   4 
s
1 2 4 2
Controller output: U  s   E  s  C  s   2  4    2  3
s  s s s
Taking inverse Laplace: u t   4 t u t   t u t 
2

Solving for the time taken (t) by the controller to reach 12 units we get:

4t  t 2  12
t 2  4t 12  0
4  42  4.12
t
2
4  16  48

2
4  8
 2
2
3. A controller of a closed loop system can be designed by optimising the performance
specification of the error. The choice of this specification used in designing the optimal
controller affects the nature of the error performance of the overall closed loop system. Match
the types of error specification given in the left column with the corresponding natures of their
error performances in the right column.

I. Suppresses large and small

A. Integral Square error
errors with equal weights
II. Strongly suppresses large
B. Integral Absolute error
errors
III. Suppresses long persistent
C. Integral Time Absolute error
errors

Solution: (b) A: II, B: I, C: III

Refer lecture videos

Ketd s
4. The open loop transfer function of a system is given by, G ( s)  . Unit step response
1 s
(input applied at t=0 second) of this system is shown in Fig. 2. Compute the initial guess of
Proportional Band (PB) and Integral Gain (KI) of a PI controller to be used with the given
open loop plant by Ziegler Nichols method of tuning based on the open loop step response. (Select
the closest option)

Fig. 2
 Solution: (d) PB=74.07, KI =0.405

Ketd s
Given G  s   where td  1sec as seen from Fig. 2
1 s
K  2 as the steady state value of G(s) for a unit step input is 2
One time constant is calculated as the time taken to reach 63.2 % of the steady state value
which is   3sec (4 – 1 = 3 sec, since 1 sec is the delay time)
Applying Ziegler Nichols method of tuning based on the open loop step response
K 2
M 
 3
0.9 0.9
Kp    1.35
M  td 2 3 1
Ti  3.33  td  3.33 1  3.33
100
PB   74.07
Kp
K p 1.35
KI    0.4054
Ti 3.33

1
5. Consider a closed loop unity feedback system with plant G p ( s)  , controlled
s( s  2)(s  6)
using a PID controller tuned by Ziegler-Nichols closed loop continuous cycling method. The transfer
 1 
function of the PID controller is given by G c (s)  K P 1   TDs  . Initial guess of controller
 TI s 
parameters (KP, TI, TD) using Ziegler- Nichols are: (Select the closest option)

Hint: Critical gain (corresponding to critical proportional band) = KC

Period of sustained oscillation = PC
Use the following relations: K P =0.6KC , TI = 0.5PC , TD =0.125PC

Solution: (c) KP = 57.6, TI = 0.907, TD = 0.227

K
Characteristic equation: 1   0 where K is the proportional gain of the PID
s  s  2 s  6
controller assuming TD = 0 and TI = infinity. This is done to calculate critical gain KC.
Characteristic equation  s3  8s2 12s  K  0
Applying Routh Hurwitz criterion
1 12
8 K
96  K P  0
96  K p
0 K P  96
8
Kp
Critical gain  Kc  96

Auxiliary eqn.
8s2  Kc  0
8s2  96  0
s2  12
s  12
 jc  12 j  c  12  3.464 rad/s

2
Period, Pc  c  2 f c
c
2 1 2
 Pc  
3.464 fc c
 1.814sec

K P  0.6  96  57.6
TI  0.5 1.814  0.907
TD  0.125 1.814  0.22675

6. Consider the water level control system of a boiler shown in Fig. 3.

Here,
q1 is the inflow rate of water;
q2 is the outflow rate of steam;
h denotes the level of water in the boiler which is to be controlled;
p1 and p2 are the water pressures at the inlet and outlet pipe, respectively;
x1 is the position of inlet valve;
x2 is the stem position of the outlet valve which is controlled to meet the steam demand.
The inlet water is drawn from a feed-water pipe fed by a pump and the steam flowing out is
going to the turbine. The power demand is met by supplying appropriate steam to the turbine
which is done by operating the outlet valve from time to time.
 Steam Control Valve
Steam

Steam Flow
Level FT Transmitter
Water Level Transmitter

Feedwater h
Feedwater
Control Valve

Feedwater Flow
FT
Transmitter
Differential Relay
Flow
Controller ID => Input Disturbance
Set Point PI Controller OD => Output Disturbance

Auto/Manual
Selector Switch

Fig. 3
Choose the correct option

Solution: (c) p1 is ID and x2 is OD

As explained in the question, here p1 is ID as the water entering the inlet is fed by a pump which
is far away from the tank. Therefore, this pressure may vary due to irregularities in the input
flow. x1 is not ID because it is being controlled by the actuator which is quite precise in its
operation. x2 is OD because this valve is operated from time to time i.e., not on a regular basis.
This valve may be operated manually also. p2 is not OD because this pressure depends on the
water level of the tank which is controlled precisely in spite of the ID p1.

7. A command pre-filter is needed in a dynamic reference feed-forward controller to avoid

Solution: (a) the effect of step change in set-point

For a reference feedforward controller, control action is generated by considering both

reference input and disturbance. Set points can change randomly. This change can be an
increase/decrease of high step input. Due to this sudden change, the controller output can also
rise/drop suddenly which can have adverse effect on the actuator. To avoid this, a command
pre-filter is required in a dynamic reference feed-forward controller.

8. The diagram shown in Fig. 4 is a typical structure a closed loop control system along with a
feed-forward controller. Gc is a PI controller and H is the feedback transfer function of the
 Kd Kp
system. Suppose G d = , Gp= , G t =K t , Gv =Kv . The dynamic transfer function model
τds+1 τ ps+1
of the feed forward controller (Gf) which will mitigate the transient effect of disturbance from
D
the output response is:
Gf Gt
FF controller Disturbance
PFF sensor/trans Gd
mitter
Yd
PFB + Disturbance
Ysp E P U Yu + Y
K +- Gc + Gv Gp + Model

FB controller Control valve

Ym
H
Sensor/transmitter Plant
Fig. 4

K d  τps+1 
Solution: (b) -  
K t K v K p  τds+1 
D
Gf Gt
PFF
PFB +
+ Gd

+ Y
Gv Gp +

Kp
G v =K v Gp =
τps+1
Kd
Gd = G t =K t
τds+1

Output Y when there is No disturbance (D=0):

Y=PFBGvGpPFB

Output Y when there is disturbance ( D  0 ):

Y   PFB  PFF  GvGp   DGd

  PFB  Gf Gt DGvGp   DGd

If the effect of disturbance in the output is nullified by the feedforward controller,

Output Y when there is No disturbance = Output Y when there is disturbance D.
 
PFB GvGp  PFBGvGp  G f Gt GvGp D  DGd
DGd  Gf Gt GvGp D
 Kd
Gd  d s 1
 Gf   
Gt GvGp Kp
Kt K v
 ps 1
Kd  s 1
  p
Kt K v K p  d s  1

9. Consider a liquid level control system (modelled as a first order system) of an industrial
process shown in Fig. 5.1. The effect of disturbance in this process will take a lot of time to
reflect in the output. So a feed-forward controller is used to reduce this time taken. Let the
1
nominal transfer function of the process be . But the actual plant transfer function is
5s  2
1
. The effect of output disturbance on the process output is modelled by the transfer
4.9s  2.1
1
function (Assume actual and nominal models are same for this). A PI controller is used as
s 1
a feedback controller to achieve a desired performance by taking care of the modelling error.
The parameters of the PI controller are: KP=10 and TI=1 unit.

A unit step is applied at reference (R) of the closed loop system assuming all initial conditions
to be zero. At t=10 secs, through an actuator a unit step disturbance (D) affects the process.
Output responses of the above mentioned closed loop system using different control schemes
like only feedback, only feedforward and feedback-feedforward are shown in Fig. 5.2. Match each
response with its corresponding control scheme:

Feedforward Controller

+ - D
Eff Process

Yd
Feedback Controller Uff
R Efb Ufb + U Yr + Y
+- + +

Fig. 5.1
 Fig. 5.2
I: Using only feedback controller
II: Using only feedforward controller
III: Using feedback-feedforward controller

Solution: (b) A – II; B – III; C – I

Reference input to the system=u(t)

Controller is of PI type and the system is modelled as a first order type-0 system.

Using only feedback controller, the system will take time to settle when a disturbance is applied
to it. In this case it is observed that the response has a high overshoot. Steady state error of the
system is 0 as PI controller is used for this type 0 system. These can be observed by calculating
settling time, maximum peak overshoot and steady state error of the following system with
Y (s)
transfer function ]
D( s)

+ - D
Eff Process

Yd
Feedback Controller
R Efb Ufb + U Yr + Y
+- + +
It can be seen from Plot C that the time taken to nullify the effect of disturbance is high and
overshoot is also high.

Using only feedforward controller, the effect of disturbance to the system can be nullified quickly.
But there will always be some steady state error due to modelling inaccuracy. It can be observed
Y ( s)
by calculating the steady state error ( y(t )  r(t ) , at t   or R( s)  lim s R( s) ) of the following
s 0 R ( s )

system.

Feedforward Controller

+ - D
Eff Process

Yd
Uff
R Efb Ufb + U Yr + Y
+- + +

It can be seen from Plot A that there exists a steady state error.

Using both feedback and feedforward controller, the effect of disturbance will be taken care of
quickly by the feedforward controller and the effect of modelling inaccuracies will be taken care
of by the feedback controller. It can observed from Plot B.

10. A ratio control scheme as shown in Fig. 6 is to be used to maintain a stoichiometric ratio of
H2 and N2 as feed to ammonia synthesis reactor. Individual Flow Controllers (FC) will be
used for both H2 and N2 streams. The electronic Flow Transmitters (FT) have built-in-square
root extractors. The spans of the flow transmitters are 45 L/min for H2 and 20 L/min for N2.
The control valves have pneumatic actuators. Each current-to-pressure (I/P) transducer has
a gain of 0.75 psi/mA. The Ratio Station (RS) is an electronic instrument with 4-20 mA input
and output signals. The ratio station is a device with an adjustable gain. The input signal to
the ratio station is dm i.e., the measured H2 flow rate. Its output signal usp serves as the set
point for the N2 flow control loop. The gain setting of the ratio station takes into account the
spans of the two flow transmitters.

The stoichiometric equation for the ammonia synthesis reaction is: 3H2 + N2=2NH3

The appropriate gain for the ratio station is:

Note: For the sake of simplicity, assume that the ratio of the molar flow rates is equal to the ratio
of the volumetric flow rates. But in general, the volumetric flow rates also depend on the
temperature and pressure of each stream (from ideal gas law).

d
(H2)
FT

I/P FC dm

Set-point, dsp NH3

Ratio RS NH3
synthesis
station Set-point, usp reactor

I/P FC

um
FT

u
(N2)
Fig. 6

Solution: (b) 3/4

The stoichiometric equation for the ammonia synthesis reaction is

3H2 + N2=2NH3
In order to introduce the feed mixture in stoichiometric proportions, the ratio of the molar
flow rates (H 2/N 2) should be 3:1.
We assumed that the ratio of the molar flow rates is equal to the ratio of the volumetric flow
rates. But in general, the volumetric flow rates also depend on the temperature and pressure
of each stream (cf., the ideal gas law).
The H2 flow rate is considered to be the disturbance variable, although this choice is arbitrary
because both the H 2 and N 2 flow rates are controlled. Note that the ratio station is merely a
device with an adjustable gain. The input signal to the ratio station is dm, the measured H 2
flow rate. Its output signal usp serves as the set point for the N 2 flow control loop. It is
calculated as usp = KR dm.
From the stoichiometric equation, it follows that the desired ratio is Rd = u/d = 1/3.

KR = Rd × ( Sd/Su)
where R d is the desired ratio, Su and Sd are the spans of the flow transmitters for the
manipulated and disturbance streams, respectively.
The spans of the flow transmitters H2 , Sd = 45 L/min
The spans of the flow transmitters N2 , Su = 20 L/min
KR = (1/3) × [(45 L/ min)/ (20 L/ min)]=3/4