Subject Exam:

Control and Instrumentation of Chemical Processes

June 2009

3. h.

Problem 1

In Figure 1, a device can be seen in which a certain product acquires a

viscosity V after being processed. It is known that the relationship between viscosity V,
The input temperature T of the product and the signal u to the motor M is given by:

dV (1)
V + VT= but− 2)
dt

The temperature T is imposed by the previous process, but the signal u can be
manipulated. It is also known that the system normally operates with a value of u in
turns to 30% of its range, a value for which, under steady conditions and with a
inlet temperature of 60ºC, the viscosity is 50, all in the system of
units in which it is expressed (1). (t time in minutes)

T V
M

Figure 1.
It is requested:

a) Propose a regulatory scheme with ISA nomenclature and draw it

corresponding block diagram indicating the transfer functions.
b) Tuning a viscosity output regulator with the criterion of obtaining a
response to temperature changes in jump without steady-state error and that minimizes
the major errors at the previously indicated work point.
c) In case it is convenient, design a feedforward compensator for the
temperature changes.
d) If the temperature changes continuously at a rate of 2ºC per hour, what will be the error?
stationary with the previously designed regulator?
e) Yes, when working in a closed loop, the viscosity setpoint will experience changes.
sinusoidal waves of amplitude 2 units and a period of 2 minutes, what will be the evolution of
the viscosity after a certain amount of time?
Problem 2

In the process of the figure, a certain amount of product B is to be processed that

it must be mixed with another A in a reactor with a filler that operates endothermically
to produce product C. In the reactor, the temperature must be maintained so that
at no point in it should be below a certain value despite possible
disturbances.

Product C from the reaction is subsequently concentrated in a type of evaporator.

Robert in which it is desired to maintain the density of the concentrate accurately. The steam
of food is shared with a heat exchanger that allows heating the
current A, and it is known that the total steam flow cannot be manipulated and experiences
significant changes. The capacitor in the figure may also experience
pressure changes.

It is requested: Draw a control and instrumentation diagram with ISA nomenclature that
meet the aforementioned requirements and explain its functioning.

Condenser

Vapor
Condensed

C
Subject Exam:

Control and Instrumentation of Chemical Processes

June 2009

1 h.

Issues

1. What is the Nyquist diagram of a system and what can it be used for?

2. What is meant by party range control? Give an example.

3. What can you say about the type of response to a step jump in the
input that would present a system whose transfer function would be:

(s− 2)e−5s
2
+ 4s+1

What is a Grafcet? Explain it briefly.

5. If, starting from conditions of equilibrium, a system to which is given a
jump in the manipulated variable responds as shown in the figure, what type of
Could the model apply for it?

Step Response
0.5

-0.5

-1

-1.5

-2

-2.5
0 5 10 15
Time (sec)
Solution to problem 1

It is known that:

dV VT= this 50 . 60 = b 30 b = 100

V + VT= this
dt

Linearizing the system at the operating point u = 30, V = 50, T = 60:

dV
V + VT − but− 0.5)= 0
dt
dΔV dΔV
V0 + (T0+ )ΔV+ V0 ΔT= 100Δu(t− 0.5
dt dt 0
dΔV
50 + 60ΔV= 100Δu(t− 0.5)− 50ΔT
dt
10e− 0.5s 5
V(s)= U(s)− T(s)
5 seconds+ 6 5s+ 6

T
D(s)

U V(s)
G(s)

Design by Lopez with MISE criteria => because I want to eliminate changes in the
disturbance T(s), as I do not want steady-state errors, a PI or a POD, and as I want
minimize large errors at the workplace: MISE criterion

d = 0.5
τ = 5/6 = 0.83
K = 10/6 = 1.6

MISE

b b

= a ⎛⎜ ⎞⎟ = a ⎛⎜ ⎞⎟
b τ d Td d
K p K= a ⎛⎜ ⎞⎟
d
Ti ⎝τ ⎠ τ ⎝τ ⎠
⎝τ ⎠
In order to apply thisτ in our case: 0.5/0.83 = 0.6 <1 => it holds

1.6 Kp = 1.305 * 0.6-0.959 => Kp= 1.33

Integral => a=0.492 ; b= -0.739

0.833 / Ti = 0.492 * 0.6-0.739=> Ti = 1.16

PI => U(s) = 1.33 ( 1 + 1/ 1.16s)

c) A feedforward is not convenient => because the output of the system affected by the
perturbations is much faster than the system transfer function G(S)
which has a delay of 0.5 m.

d) Closed-loop system:

T
D(s)

W U V(s)
PI G(s)

Steady error:
1 D(s)
e ss = lim sE(s) = lim W(s) - T (s))
s → 0 → s 0 1 + G(s)R(s) 1 + G(s)R(s)

− 5 / ( 5 s + 6) 0.9663 seconds
E(s)= − −0.5s
T(S)= T(s)
1 + 1.6e (0.83s+ 1)*1.33(1+ 1/1.16s 0.9663s + 1.16s+ 2.12e−05s + 2.4592e−0.5s
2
s

Ramp => T(s)= 2ºC / hour = 2ºC/ 60 minutes = 0.033

T(t)=0.033 t => T(s)= 0.0333 / s2

 0.9663s 0.033 0.09633*0.033
e ss = limsE(s)= lims = = 0.4558
s→0 s→0 0.9663s2+ 1.16 seconds+ 2.12e−05s+ 2.4592e−0.5s 2 2.12

e) T = 2 minutes => w = 2ππ

Amplitude = 2 * | F(jw) |

G(s)R(s) D(s)
Y(s)= W(s)+ V(s)
1+ G(s)R(s) 1+ G(s)R(s)

1.6e−0.5s1.33(1.16s+ 1)
(0.83s+ 1.16s
F(s)= W(s)
1.6e−0.5s1.33(1.16s+ 1)
1+
(0.83s+ 1) 1.16s
Solution Problem 2

FT FC FF
LC
FC FT

A TT
Capacitor
TTConditions
Terms and

PC
TC LS TT
PT

FY FY
TT

Vapor FT

Condensed DC

LT LC
C DT
Problems of the Subject 'Control and Instrumentation of Chemical Processes'
4th year of Chemical Engineering

Problem 1

In the system of the figure, two storage tanks, A and B, can be seen,
Aliquid arrives from certain supply sources.The output from deposit B passes through
a reheater powered by heating steam and pours into the tank C where also
pour the output from tankA. The liquids must be mixed in a certain proportion. The mixture
both in the agitated tank C is sent to another process that imposes the consumption of
certain variable amounts over time. This mixture must also be sent to
constant temperature maintained with precision despite possible disturbances. It is assumed
that the liquid that reaches B and A does so at a sensibly constant temperature, due to the
On the contrary, the supply pressure of the heating steam undergoes significant changes.

vapor

A B

It is requested:

1) design a regulatory system that meets the proposed objectives, placing the
transmitters, actuators and necessary controllers.
2)justifythedesignmadebyexplainingitsobjectivesandhowitworks.
 Solution:

It is necessary to implement a control system that covers several objectives:

Maintain the level of the deposits to avoid overflow or situations of lack of any.
fluidforthemixture
Maintain a proportion between the flows ofAand B that are poured into C
Maintain the temperature in C despite disturbances from vapor pressure and
incoming flow changes
For this, the structure of the figure has been designed:

FY vape

PC

LC
LC
A B LT
FT FT

FF FC TT TC

TC

C TT
LC LT

Since the outlet flow of C is predetermined, the level controls LC in the three
deposits should be implemented backwards, manipulating the input to the deposits. They could
Design flow-level cascades, but the level is not critical and does not seem necessary, therefore.

To maintain the proportion between the inflows to reservoir C, and considering

the output flow from A is determined by the level regulator of C, this flow is measured and
a FF ratio control is implemented on the output flow of B.

To maintain the temperature of C accurately, there is a cascade of regulators.

the temperature controller sets the setpoint of the temperature regulator
from current B, the only one that can be altered. Since currents A and B enter into a
certain relationship, this seems to be a good policy, as it is important to be able to maintain the
temperature of B as needed. To do this, an additional cascade is implemented with a
pressure regulator of the heating chamber of the exchanger to absorb changes
of vapor pressure and a feedforward with flow rate changes to proactively correct
its effect.
Problem 2

Atemperature control system for a certain chemical reactor is desired.

exothermic,usingarefrigerantcontrolvalveasanactuator,asshown
inthefigure.

Terms and Conditions

TT

In an open-loop test, the temporal response of the temperature in ºC has been measured.
when the control signal to the valve varied by -20%, resulting in that of the figure with the
timescaleinminutes.

It is requested:

1) Design a regulator that does not exhibit steady-state error in response to step changes in the
reference, that it does not present overshoot before them and that it stabilizes in a time not
longerthan3minutes.Justifythedesignmade.
2) If the regulator reference changes at a ramp of 5 °C/min, will it present the
Isthereasteady-stateerrorinthesystem?Ifso,calculateit.
3) Draw the root locus corresponding to variations in the gain of the regulator and
comment on it.
Solution:

1) The system, given the shape of its step response, can be approximated by a system of
first order without delay, since it is a stable response without inflection in the response nor
overpeak.

K
G(s)
ts+1

According to the design criteria, and considering the type of process model,
we can choose the tuning method called d-tuning in which the design criterion
is to obtain a system that in closed loop has a response similar to that of the function of
transfer
1
ls+1

which is overdamped. In our case, if it is desired that the closed-loop system

stabilize in less than 3 minutes, it is enough that the time constant meets the relationship
4l=3 so we can choose l= 0.75 min.

The regulator, since a first-order system like this process does not have
integrators should be of the PI or PID type. The tuning table of Rivera-Morari for a PI
Improved provides the values of the regulator parameters according to:

2t + d d
Kp= T=
i t +
2Kl 2

Where K is the gain, tau is the time constant, and d is the delay of a first-order model.
order with delay that can represent the open loop process. In our case d=0, of
so that the validity condition for using the table is fulfilled:
l
1.7
d
The first-order model can be obtained graphically as can be seen in the figure.
Gain is the change in output divided by the change in input:

2.5 °C
K= -0.125
-20 %

and the time constant can be calculated, either using the maximum slope method,
Well searching at the moment in time when the final change of 63.2% is reached. Both
methods give very similar values = 1.2 min., so that:

K -0.125
G(s)= =
ts+1 1.2s +1
andtheregulatorparametersareasfollows:

Kp-1.2/(0.75 0.125) = -12.8 % / ºC Ti1.2min

1.58
2.5

1.2

The steady-state error due to changes in the reference is given by:

1
E(s) = W(s)
1+G(s) R(s)

Where R(s) is the transfer function of the regulator:

1 Ti s+1 1.2s + 1
R(s)=K p(1+)=Kp( )= -12.8
Ti s Ti s 1.2s
so that:
 1
E(s)= W(s)
1 + G(s)R(s)
1 5
=
(-0.125) (-12.8)(1.2s+1) s 2
1+
1.2s + 1 1.2 seconds
5
e ss=lim sE(s) = 2.14
s®0 1 + 1.33

and the system will present a steady-state error of 2.14 ºC

3) The place of the roots is a diagram of the solutions of the equation:

1+ Kp G(s)R(s) = 0

In our case:

-0.125 K p (1.2s +1) 0.1042Kp

1+G(s)R(s)=1+ =1- 0
1.2s + 1 1.2s s
s=0.1042Kp

andthediagramis:
s
Subject Examination:

Control and Instrumentation of Chemical Processes

June 2007

3.5 hours.

Problem 1

In the process of figure 1, a closed container thermally isolated from is observed.

10 liters of volume that receives a flow q of a fluid, which exits by overflow, and is heated.
from a temperature Tiuntil another T thanks to an electrical resistance R powered
by a direct current I that comes from an amplifier. This amplifier has
an input that accepts signals in the range of 4-20 mA and provides an output of
variable current in the range of 0–8A. A transmitter is also available.
calibrated temperature in the range of 10-60ºC and a commercial PID that operates in
units of %.

The properties of the fluid are largely constant in the temperature range of
operation and it is known that when the system is in steady state, the flow of
The input is 2 l/min, the amplifier current is 4A and the input temperature
If it is 20ºC, then the exit temperature is 28ºC.

TI T

R
q

I
Amplifier

u
It is requested:

a) Calculate a mathematical model of the process that relates the main variables.
It is assumed that the amplifier has a linear and fast response.
b) Propose a regulation scheme with ISA nomenclature and draw it
corresponding block diagram indicating the transfer functions.
c) Tune an output temperature regulator with the criterion of obtaining a
response to reference changes in jump without steady-state error and with a time
of closed loop settling time of 12min. in an operating point environment
as previously indicated.
d) If the inflow rate experiences sinusoidal changes like those in figure 2,
(Time in min.) How will the outlet temperature evolve in closed loop?
 2.5

2.4

2.3

2.2

2.1

1.9

1.8

1.7

1.6

1.5
0 2 4 6 8 10 12 14 16 18 20

e) If the gain of the regulator is increased from its design value, how
the closed-loop response of the system will evolve in response to step changes in the
input temperature?

Problem 2

In the scheme of the figure, a certain amounts of liquid product A are to be processed.
which requires heating it to a temperature of 90ºC in a gas-powered oven and
keep the product in a tank for a residence time of 20 min. It is known that the
Gas supply pressure experiences significant changes.

The oven simultaneously heats a non-manipulable flow of another product B that

it must always be kept at a temperature above 80ºC and is used for
heat a third current C to 40ºC, a temperature that must be maintained with
precision despite possible disturbances in the heating current.

It is requested:

Draw a process control diagram, with the appropriate instrumentation and

ISA nomenclature, and explain its functioning.

Air

gas
Subject Exam:

Control and Instrumentation of Chemical Processes

June 2007

1 h.

Questions

a) Could you apply PID controllers in the control of a multivariable system whose
RGA out:? Justify the answer.
[ y1 y2 y3 ]
⎡ u1 ⎤⎡
0.8− 1.2 1.4 ⎤
⎢ u2− ⎥⎢ 1.2 0.9 1.3 ⎥
⎢ ⎥⎢ ⎥
⎥⎢1.4
⎣⎢ u3 ⎦⎣ 1.3− 1.7 ⎦⎥
If so, what would be the best combination of input/output bindings?

b) What is the place of the roots and what can it be used for?

c) What can you say about the type of response to a step change in the input
what would a system whose transfer function would be:
(s+ 2)e−3s ?
s 2+ s+1

d) What is meant by the robustness of a control loop? What measures of robustness

do you know?

e) Under what conditions can an advance compensation be used or

feedforward?
Solution Problem 1

a) Calculate a mathematical model of the process that relates the main variables.
It will be assumed that the amplifier has a linear and fast response.

An energy balance leads to:

dT I 2R
V = q(T−T)+
I
with V, R,ρ,ce= cte.
dt ρce
8-0
I= (um−4)= 0.5(um−with umenmA
20-4
8-0 16
I= u= 0.08 ucon u in %
20-4100

In steady state:
42R R
0 = 2(20−28)+ ⇒ =16/16=1
ρce ρce
8−0
4= (u− ⇒ u0=12mA= 50%
20−4

therefore:

dT
10 = q(T−T)+
i
0.0064u2 T in °C

b) Propose a regulation scheme with ISA nomenclature and draw it.

corresponding block diagram indicating the transfer functions.

Ti T
TT
R

I
u
Amplifier TC

To calculate the transfer functions, we must first linearize the model:

 dT
V = q(T−T)+6410
I
u
−4 2

dt
f(T,T,T,q,I)=
I
0
∂f ∂f ∂f ∂f ∂f
(T−T0) + (T−T0) + −T)+ (q−q0) + (u−u0) = 0
∂T 0 ∂T 0 ∂Ti 0 i i0 ∂q0 ∂u0

d ΔT
V = −q 0 ΔT+ q 0 ΔTI+ I0− T0)Δq+ 0.0128 micro0 Δu
dt
V dΔT (T− T)
i0 0 0.0128u0
+ ΔT= ΔT+I Δq+ Δu
q0dt q0 q0
dΔT
τ + ΔT= K1ΔT+ i
K 2 Δq+ K 3 Δu
dt
dΔT
5 + ΔT= ΔT−i 4Δq+ 0.32Δu
dt
and taking Laplace transforms, the transfer functions result:

K1 K2 K3
T(s)= TI(s)+ Q(s)+ U(s)
τs+ 1 τs+ 1 τs + 1
T in °C
1 −4 0.32
T(s)= Ti(s)+ Q(s)+ U(s)
5s+ 1 5s+ 1 5s+ 1

and since the span is 100/(60-10) = 2,

2 −8 0.64
T(s)= Ti(s)+ Q(s)+ U(s) T in % of the span, u in %
5s+ 1 5s+ 1 5s+ 1

The block diagram is then:

Q(s) Ti(s)

−8 2
5s+ 1 5s+ 1

w E(s) U(s)
0.64 T(s)
R(s)
5s+ 1
c) Tuning an output temperature regulator with the criterion of achieving a
response to reference changes in jump without steady-state error and with a time of
closed-loop settling time of 12 minutes in an operating point environment before
indicated.

Depending on the tuning objective, we can use the criterion ofλ-tuning of Rivera-
Morari IMC, with a PI or a PID. Theλ desired would have to be 12/3 = 4 min. which
meets the minimumλ 0.2τ = 0.2 * 5 = 1

In our case, the delay is zero, then the tuning criterion gives:

2τ + d 5
Kp= = = 1.953 % / %
2Kλ 0.64*4
Ti= τ + d/ 2= 5 min

1 5s+ 1
being the same for a PID it results in Td= 0. R(s)= k p(1+ ) = 1.953
Tis 5s

d) If the incoming flow experiences sinusoidal changes like those in figure 2,

(Time in min.) How will the outlet temperature evolve in closed loop?
2.5

2.4

2.3

2.2

2.1

1.9

1.8

1.7

1.6

1.5
0 2 4 6 8 10 12 14 16 18 20

The closed-loop transfer function is:

G(s)R(s) D1(s) D2(s)

Y(s)= W(s)+ V1(s)+ V 2(s)
1 + G(s)R(s) 1 + G(s)R(s) 1 + G(s)R(s)
0.64 5s+ 1 2 −8
1.953
T ( s ) = 5s+ 1 5s W(s)+ 5s+ 1 Ti(s)+ 5s+ 1 Q(s)
0.64 5s+ 1 0.64 5s+ 1 0.64 5s+ 1
1+ 1.953 1+ 1.953 1+ 1.953
5s+ 1 5s 5s+ 1 5s 5s+ 1 5s

− 40s
T(s)= Q(s)
(5s+ 1)(5s+ 1.235)

The oscillation frequency is:

2*pi/5 = 1.256 rad/min
The response will be a sinusoid with a frequency of 1.256 rad/min, oscillation amplitude
|G(jω)| y phase shift arg G(jω):
 − 40jω
T ( j ω) = Q(jω)
(5jω + 1)(5jω + 1.235
− 40jω − 40 1.256
= = 1.2344
(5jω + 1)(5jω + 1.235) (5 j1.256+ 1)(5j1.256+ 1.235
− 40jω 1 1.235
arg = −π / 2− arctan − arctg = −1.9294 rad
5jω + 1)(5jω + 1.235 5ω 5ω

f) If the gain of the regulator is increased from its design value, how
the closed-loop response of the system will evolve in response to step changes in the
input temperature?

Since
2
5s+ 1 10s
T(s)= Ti(s)= Ti(s)
0.64 5s+ 1 (5s+ 1)(5s+ 0.64kp)
1+ kp
5s+ 1 5s
The system has a pole at -1/5, another at -0.64kp/5 that will become more negative, and
it will always be real, as the profitpincrease, just like a zero in 0.

Note that a zero at s=0 produces a response that is the derivative of the same.
response without the zero. The response if there were no zero would be that of a second system
overdamped order increasingly faster that would bring its response closer to that of a first
first-order system with a time constant of 5, and therefore, with a settling time of 15 minutes.
So in response to a jump in the input temperature, the response will be:

Without zero
With zero
Problem 2 Solution

B TT

FT
FT FC + TC

TT TY
A
T TC Tm
TT
FT FC
Air TC

FF τ/S
FC HS
FT FC

FT LC LT

Gas
FF
Problems of "Control and Instrumentation of Chemical Processes"
4th Chemical Engineering

Problem 3

U
Ti T
Changer
q

The system in the figure represents a heat exchanger with an internal heating system.
non-manipulable that heats a flow q of water from a temperature T to a temperature T. I
For this system, it is known that the relationship between the control signal to the inlet valve U in
% and the output temperatureTin ºC (time in minutes) is given by:

dT
3 -6T + 8.8U2+2TI
dt

and that when the system is in steady state and the inlet temperature is 10 ºC,
The outlet is 40ºC.

It is requested:

1) Draw a temperature control system on the diagram

2) Assume that the regulator is of the PI type, draw a block diagram and calculate the
transfer function in closed loop neglecting the dynamics of the transmitter.

3) If the regulator gain is 0.1 and the integral time is 1 minute, study the stability of
system and the form of the response to a step change in the reference.

4) Calculate the steady-state error in response to a jump of 0.2 ºC in the input temperature.
liquid.
1) To control the temperature, the only option is to act on the flow of liquid to
through the valve, then the control system should be as in the figure, where it has been
installedatemperaturetransmitterandacontrollerthatmanipulatesthevalve:

TC TT

U TI
Changer
q T

2) In order to establish a block diagram and calculate the transfer function in

In a closed loop, the transfer function of the process must be calculated. This has to be done
linearizing the mathematical model that relates temperature T with the opening of the
valveandthetemperatureT: i

dT
3 -6T + 8.8U2+2Ti
dt

To do this, we will start by calculating a stationary linearization point. According to the

data of the problem it is known that when the input temperature is 10 ºC, the output temperature
it is at 40ºC in equilibrium, so that:

0 = -6*40 + 8.8U02+2*10

U 0(240-20)/8.8 = 5%
T040ºC
Ti 0 10 degrees Celsius

Developing in Taylor series:

dT
3 = -6T + 8.8U2+2TI
dt
dDT
3 = -6DT+(2*8.8U0)DU+2DT I
dt
-6DT + 88DU + 2DTi
conDT=T - TDU=U-U
0 0DT=T
i iTi0

So taking Fourier transforms on both sides of the equal sign, it results in:
 3sDT(s) = -6DT(s) + 88DU(s) + 2DT(s)i
3sDT(s) + 6DT(s) = 88DU(s) + 2DT(s)I
88 2
DT(s) DU(s)+ DTi (s)
3s+6 3s+6

And now, taking into account that the transfer function of a PI regulator is:

1
Kp(1+ )
Tis
Is it possible to draw the closed-loop diagram:
DTI
2
3s + 6

w error YOU DT
1 88
K p(1+ )
Ti s 3s + 6

andcalculatetheclosed-looptransferfunction:

88 2 88 T s+1 2
DT(s) = DU(s)+ DT(s)=
i K p( i (W(s)-DT(s))+ DT(s)
i
3s + 6 3s+6 3s+6 Tis 3s + 6
é 88 Tis+1 ù 88 T s+1 2
ê1 + 3s + 6 Kp( T s )DT(s)= ú Kp( I )W(s)+ DT(s)
i
ë i û 3s+6 Ti 3s+6
88 T s+1 2
Kp ( i )
3s+6 Ti 3s + 6
DT(s)= W(s)+ DTi
88 TIs+1 88 Tis+1
1+ K p( ) 1+ K p( )
3s+6 Tis 3s+6 Tis
K p 88(Tis+1) 2Tis
= W(s)+ DT(s)
i
(3s+6)Ts+K
i p 88(Ti s+1) (3s+6)T is+Kp88(Tis+1)
3) Assigning values to the parameters of the PI according to the statement:

8.8(s+1) 2s
DT(s)= W(s)+ DT(s)=
i
(3s+6)s+8.8(s+1) (3s + 6)s + 8.8(s + 1)

8.8(s+1) 2s
= 2 W(s)+ 2
DT(s)
i
3s +14.8s+8.8 3s +14.8s+8.8

and the stability in closed loop can be calculated using the roots of:

3s2+14.8s + 8.8 = 0
-14.8±14.82-4 3 8.8
s 1,2 = -0.69,-4.24
6

Since both are real and negative, the response to a change in the reference would be stable and
withoutoscillations,withapointofinflection,althoughnotverypronouncedduetothepresenceofapole
(-4.24) quite a bit faster than the other (-0.69). The settling time will be imposed by the
the slowest pole will be on the order of 3/0.69 = 4.34 min for a 5% criterion (4/0.69 = 5.79
minimum for a criterion of 1%. However, since the system also has a zero in the loop
closed at s = -1, a lead will occur. The gain is 1. The
Total response is shown in the figure.

Step Response
From:U(1)
1

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
0 1.4 2.8 4.2 5.6 7

Time (sec.)

4) To calculate the steady-state error in response to a change of 0.2 ºC in

I T:
 -D(s) -2s 0.2
E(s)= DT(s)=
i
1+G(s)R(s) 3s +14.8s+8.8 s
2

-2s 0.2
e ss =lim sE(s) =lim s =0
s®0 s®0 2
3s +14.8s+8.8 s

and there will be no steady-state error.

Problem 4

A B

vapor Reactor

Changer

The figure shows an endothermic chemical reactor in which two products A and B
react at a certain temperature to form product C, which, after being heated in the
heat exchanger, is temporarily stored in a tank. The liquid exits through
rebose of the reaction chamber so it is always full of products, while the
heating chamber receives a steam whose pressure experiences strong changes and exits as
condensed. The flow of product B is not manipulable.

Design a control system that is capable of accurately maintaining the final temperature
of product C, as well as its concentration. Install the instrumentation, transmitters and
actuators, necessary and explain their operation.
Solution:

FT FC FF FT

A
B
TC TT

TC TT

vapor
Reactor
AC

AT

TT

Changer
PC PT LT

TC LC

The objectives that must be met are:

∙ Maintain the level in the tank, as it should be done in all of them.
∙ Since it is a chemical reaction, maintain the proportions in the input of
products to the reactor.
∙ Maintain the temperature of the exothermic reactor to ensure proper operation at
despite the changes in the heating vapor pressure
∙ Maintain the required temperature in product C
∙ Maintain the concentration of product C
For this purpose, a control scheme has been designed as shown in the figure. Given that the flow rate of the
Product B is not manipulable, we will implement the control of proportions of productsAand
B as a ratio over B, managing the flow ofA.

This means that the output flow of the reactor is imposed, and the only alternative for
installing the tank level control is done by acting on its output.

The reactor temperature control is implemented as a cascade with an internal loop of

temperature control of the shirt, in order to absorb more quickly in the latter the
variations in supply vapor pressure.

The temperature control of C after the heat exchanger is also implemented as another cascade.
whoseinternalbondisoneofvaporpressureforthesamereason.

The composition control of C is carried out using an analyzer for its measurement and acting
about the proportion of both products, which can thus be corrected if the
the composition of C is not suitable.

Note that both the temperature and the composition of C could be measured in the tank.
this would include in the control loop the possible disturbances that occurred in it and
it would act as a filter for quick changes, although it would also introduce an additional element
slowthatwouldmakethesystem'sresponselessagile.
Problems of 'Control and Instrumentation of Chemical Processes'
4th Chemical Engineering Course

Problem 5

The figure shows a diagram of an endothermic reactor where two are introduced.
products A and B, which react to form another C, which are maintained notably
constants.

A B

valve

An experiment has been carried out consisting of increasing the opening of the intake valve
of vapor by 10% and record the change in the temperature of the reactor when the others
variables were constant, which can be seen in the graph, where it is supposed that the
The reactor temperature was initially at the value marked as 0. The scale of
the temperature is in ºC and the duration in minutes. The temperature transmitter was calibrated
in the range of 0-80 ºC.
 r
60

50

40

30

20

10

0
0 5 10 15

It is requested:

1) Calculate a dynamic model that relates the reactor temperature to the opening of
thevalve,usingunits%/%forthegain.
2) Calculate the settling time and the oscillation frequency and compare them with the
values that can be estimated from the model.
3) Design the simplest temperature controller that provides the following
characteristics: no steady-state error, without or with very little overshoot and with a time of
settlementofabout5minutes.

Solution:

1) Since the only information is the answer to a jumping essay, we will use that.
graph to deduce an approximate linear model. From the shape of the response, with
overshoot and oscillation and without delay, it is deduced that we can choose a model of
second order of the type:

Kwn2
s 2+2dw +n wn2

r
6
5%
8.5
5

4
The gain K is calculated by the quotient of the change in temperature in state
stationary, 50ºC, and the change in the valve opening 10%. As units are requested in
To convert 50ºC to a percentage scale, we need to consider that 80ºC is 100%.
ofthetransmitter.

K = (50 100 / 80) / 10 = 6.25 % / %

To calculate the value of damping we will use the overshoot measurement. The value of the
The peak is 8.5 ºC and in % over the final value: 8.5 100 /50 = 17 % and it is known that the
the relationship of this value with damping is given by:

- dp
1- d2
100e
therefore:
-dp
1- d2
-dp d2 p 2
e -1.771956 ; (-1.772)2=
1- d 2 1- d 2
1.318d2=0.318;d =0.491

A similar value can be obtained from the %M /d graphs.

To calculate the undamped natural frequencynwe will use peak time. The time
the peak is 3.53 min and it is known to be related to the parameters of the function of
transfer by:
p
tp =
wn1- d 2
therefore:
p
3.53= ;wn1 . 0 2
wn1 - 0 . 4 9 1 2
and the resulting transfer function between changes in temperature and position of the
valveturnsouttobein%/%:

6.5
s +s+1.04
2

The oscillation period obtained from the graph is approximately 2(7.5 - 3.53) =
7.94 min. Thus, the oscillation frequency will be: 2p/7.94 = 0.79 rad/min. Being difficult
make the measurements accurately on the graph due to the flat shape of the descending oscillation.

Similarly, the settling time of 5% obtained from the graph is

approximately5.5min.

The calculated values of the model are:

wr= wn1- d 20.53120.89 rad / min

dw n=0.491.02=0.5;3/ 0.5=6min

What are values close to those measured. In particular, please note that the calculation of
settlement time through 3/dwnin second order systems it is just an expression
approximate. It should also be noted that the response of the graph is being
approximating it for a second order system, but the statement does not say it is
exactly a second-order system.
3) To design the requested regulator, we must first choose its type. Having
It should be noted that the process has no integrators; to eliminate the steady-state error, it will be necessary to ...
a PI or PID regulator. On the other hand, given the type of transfer function and the
design specifications, it is not possible to apply the Ziegler-Nichols rules or tables
Rovira, Lopez or Morari. An applicable design procedure is to obtain a phase margin.
(related to the overshoot) at a given frequency (related to the speed of
response). If you want the closed-loop system to have a slight or no overshoot
A phase margin of about 55º can be taken. To estimate the frequency at which one wants
obtain this phase margin for the process-regulator set, given that the time of
the required settlement is slightly lower than open loop, we will take a frequency
slightly superior to that of open-loop control.

Given that in our case the transfer function is:

6.5 6.25 1.04

=
s +s+1.04 s +s+1.04
2 2

and we know that for a transfer function:

w2n
2
+2dw + w 2
n n

the cutoff frequency is approximately at the frequencynwe can do the design

regarding this transfer function for which we can estimate this frequency, and
then divide the resulting regulator's profit by 6.25. However, given that the value of
the profit from the process only intervenes in the denominator of the calculation formula of the
profit Kp of the regulator, (as can be seen in the formulas below) is equal
use for them the transfer function of the process that includes the gain 6.25 and not
divide afterwards by this factor. In the design, therefore we will choose a frequency
slightlysuperiorawne.g. 1.02f1.2 rad/min.

In the case of a PI, it is about obtaining certain values of the gain and the integral time of the
a regulator such that the phase margin of G(jw)R(jw) is 55º at the frequency of 1.2

f
-
w
G(jw)R(jw)

The solution, according to the theory, is given by the expressions:

q = p - f + arg [G(jw f )]
1
Ti=
w f tgq
cosq
Kp=
G( jw f )
Where
6.5 6.5
G(jw f)= = =5.14
(jw f) 2+jwf +1.04 (j1.2)2+j1.2+1.04
[
arg [G(jw f ) ] = -arg 2+j1.2+1.04 ] -1.89 rad
55p
q=p- -1.89=0.29 rad
180
1 cos0.29
Ti= 2.8 minutes K p= =0.19%/ %
1.2tg(0.29) 5.14

Once the PI regulator parameters have been obtained, and since there is no exact relationship
for the resulting third-order system with a zero, the settling time would be
It is convenient to verify in simulation that the closed-loop system obtained meets the
specifications.
Another design alternative is a direct synthesis procedure. If we take as
desired behavior in closed loop a first-order transfer function
(overdamped) with unit gain and time constant l = 5/3 = 1.66, we can
calculate the regulator R(s) that gives that function M(s) in closed loop:

1 1 Kwn2 6.5
M(s)= = G(s)= =
ls+1 1.66s +1 s +2dwns+ wns + s + 1 . 0 4
2 2 2

1
M(s) ls+1 +2dws+nwn2
2
R(s)= = = =
G(s)(1-M(s)) Kwn2 1 Kwn2 (ls+1-1)
(1- )
2
+2dws+nwn2 ls+1
s 2+2dws+n w22
n /wn+(2d/wn )s+1
2
= = =
Kwls
2
Kls
2d(2d/wn) (1/2 dw) s2+(2d/w n)s+1
= n

wn Kl (2d/wn)s

2D 2d 1
Kp= Ti= Td =
w nKl wn 2dw

K p (TT
i d s +Ts+1)
2
i
PID ideal=
Tsi
MatchingtermstoanidealPID

andsubstituting:Kp= 0.093, T = 0.96 Ti d0.998

Problem 6

The process in the figure represents a heating furnace for a material, which enters into
temperature T and iit must come out at temperature T. In the oven, the flow F can be manipulated.
heating element to make the final temperature T of the material reach the values
desired.

Ti Oven T

It is known that the relationship between the temperature T, the input temperature of the material T and the i
flow F is given by:
dT
(5+3F) +2T2=3FT+T i
dt
withTandTi in ºC, F in Kg/min and time in minutes, and when the system is in state
stationary at 40ºC the inlet temperature is 10ºC. The request is:

1) Obtain the transfer function of the system at that operating point.

2) A PI controller with a gain of 0.1 %/°C and time is used for the regulation of the system.
Integral 20 seconds. It is known that the relationship between the control signal U of the regulator
in % and the flow F in Kg/min is F=U. Draw the control diagram of the process and the
closed loop block diagram corresponding to it, and study the stability of
system.
3) Calculate the regulator gain so that the system reaches the limit of
stability.
4) For a ramp change in temperature T of 2ºC/min, I calculate the steady-state error.
with the tuning parameters from section 2).
5) If the changes in Ti are significant, how would you design a lead compensator?
What will improve the functioning of the system? Calculate it and draw the diagram.
corresponding.

Solution:

1) Given that the system model is nonlinear, to obtain the transfer function between
the outputT, the input F and the possible disturbanceTwe must linearize
i this model.The point
Thelinearization,accordingtothestatement,istheequilibriumpointthatverifies:
 2T2=3FT+Ti; 2 402=3F040+10; F0=3190/120=26.6 Kg/min
withwhatsaidpointoflinearizationturnsouttobe:

T0= 40°C, F0= 26.6 Kg/min, T = 10°C, i0 T&00

dT
where we have used the notation T= &
dt

The model equation is a function of F, T, T and its derivative

i T, and can be linearized
& using
aTaylorexpansion:

dT
(5+3F) +2T2-3FT-T i=0 f (T,T,F,T
& i=0
dt
¶f ¶f ¶f
Linearization is an expression of the type: DT+
& DT+ DF+ DT=0
I
¶T&0 ¶T 0 ¶F 0 ¶T i 0
whereDT=T& - &0T
;DT=T
& - 0T;DF=F - 0F;DTI=TiTi0with what:
(5+3F0)DT+(4T
& 0-3F )DT+(3T
0 0-3T&
0 )DF-DT i=0

dDT
84.8 +80.2DT=120DF+DTi
dt

Taking Laplace transforms on both sides of this linearized equation, and

consideringthat,ifattheinitialmomenttheprocessisinequilibrium,theinitialvalues
theincreaseswillbenull:

ì dDT ü
84.8Lí ý +80.2L {DT} =120L {DF} +L {DTI }
î dt þ
84.8T(s) + 80.2T(s) = 120F(s) + T(s) where
i
T(s) = L {DT}, F(s) =L {DF}, Ti (s)=L {DTi }
(84.8s+80.2)T(s)=120F(s)+T(s) I

andthetransferfunctionturnsouttobe:

120 1 1.5 0.0125

T(s)= F(s)+ Ti(s)= F(s)+ Ti(s)
84.8s + 80.2 84.8s + 80.2 1.06s + 1 1.06s + 1

Theregulationdiagramis:

T
TI

Process

TT
F
U
TC
 and the block diagram:
Ti
0.0125
1.06s + 1

W E T
U F
1 1.5
0.1(1+ ) 1
(20/60)s 1.06s + 1

The closed-loop transfer function can be calculated as:

G(s)R(s) D(s)
T(s) = W(s)+ Ti (s)=
1 + G(s)R(s) 1+G(s)R(s)
1.5 0.1s +0.3 0.0125
= 1.06s + 1 W(s)+ 1.06s + 1 T (s)=
1.5 0.1s +0.3 1.5 0.1s +0.3i
1+ 1+
1.06s + 1 s 1.06s + 1 s
0.15s + 0.45 0.0125s
= W(s)+ Ti (s)
1.06s2+1.15s+0.45 1.06s2+1.15s+0.45

and the stability in closed loop can be studied by calculating the poles in closed loop, this
yes, the roots of:

-1.15±1.152-4 1.06 0.45

1.06 seconds2+1.15s + 0.45 = 0; -0.54±j0.36
2 1.06
Since the real part of both poles is negative, the system will be stable in closed loop.

3) If the regulator's gain were variable instead of 0.1, the transfer function in
closed loop would be:
 1.5 s+3 0.0125
Kp
T(s)= 1.06s + 1 s W(s)+ 1.06s + 1 T (s)=
1.5 s+3 1.5 s+3 i
1+ Kp 1+ Kp
1.06s + 1 s 1.06s + 1 s
1.5Kps+4.5Kp 0.0125s
= W(s)+ TI(s)
1.06 seconds (1+1.5Kp)s+4.5Kp
2
1.06 seconds2+(1+1.5Kp)s+4.5Kp
and stability would be given in terms of the roots of:

-1-1.5K p±(1+1.5Kp) 2-4 1.06 4.5Kp

1.06s +(1+1.5Kp)s+4.5Kp=0; s =
2
2 1.06

the roots will be real if it is verified:

(1+1.5Kp ) 2-4 1.06 4.5Kp³0; that is: 1+2.25K2p-16.08Kp=0

16.08±16.082-4 2.25 1 16.08±15.79
Kp = = =7.08; 0.0627
4.5 4.5

that is between Kp= 0.0627 and 7.08 the roots s will be imaginary and the stability condition
Itshouldbethattherealpartofthemisnegative,inotherwords,itshouldbefulfilled:

-1-1.5K p£0; Kp³ - 2/3 so that in this range of values of Kpthesystemwouldbe

always stable. Out of the K rangep[0.0627, 7.08], the roots in the closed loop are real
and the stability condition is:

-1-1.5K p±(1+1.5Kp) 2-4 1.06 4.5Kp£0

for which the absolute value of the radicand must be less in absolute value than 1 + 1.5Kpthe
which holds for any Kppositive, so that the system will be stable for any
Kp0

Alternatively could to have beenused the criterion of Root for

2
1.06s (1+1.5Kp)s+4.5Kp=0 that leads to:

1.06 4.5Kp
4.5Kp>0 ; 1+1.5KpSo Kpgreater than zero
1+1.5Kp

4)The steady-state error in response to a ramp change inTof 2ºC/min can

i be calculated using
the corresponding transfer function:
 1 D(s)
E(s)= W(s)- Ti (s)=
1+G(s)R(s) 1 + G(s)R(s)
0.0125
-
1.06s + 1 -0.0125s 2
= Ti (s)=
1.5 0.1s +0.3 2
(1.06s +1.15s+0.45)s 2
1+
1.06s + 1 s
-0.0125s 2 -0.0125 2
e ss=lim s = -0.055
s®0 (1.06s +1.15s+0.45)s
2 2
0.45

5) A lead compensator can be used, since the dynamics of the output in response to
changes in the disturbance T are noti faster than changes in the manipulated variable
U. The transfer function of the compensator will be given by:

0.0125
-D(s)
G F(s)= = 1.06s + 1 -0.008
G(s) 1.5
1.06s+1

andwouldcorrespondtothefollowingscheme:

T
TI

Process

TT
TT
F
U
TC

TY

TI
0.0125
1.06s + 1

-0.008

W E T
1 U
1.5
0.1(1+ )
(20/60)s 1.06s + 1
Examination of the Subject "Control and Instrumentation of Chemical Processes"
4th year of Chemical Engineering

Problem 7

In a drying process, a certain constant flow of material is introduced into the dryer, along with
with a flow of hot gas, whose flow rate can be manipulated through the signal to
afansystem.Inanexperimentconsistingofreducingthesignaltothefanby10%
from a position of equilibrium, a change in the moisture of the material has been observed
like the one in the figure. The transmitter is calibrated with a span of 5 units. Units of
timeinminutes.

It is requested:

1) To control humidity, a P-type controller is used. Draw a block diagram.

of the resulting control system, specifying the transfer functions.
Calculate the gain value of the regulator in %/%, to achieve a peak increase of 20%
in the closed-loop response to a two-unit step in the command.
3) Calculate the value of the regulator gain to achieve a steady-state error lower than
1% in the closed-loop response to a 2% step change in the setpoint. What will be the value?
stationary of the manipulated variable in this case?
4) If the reference oscillates sinusoidally with an amplitude of 10% and a period of 0.1 min.And
the value of the gain from section 4) How will the system's output be after
a pretty long time?
5)Calculateandexplaintherootlocusdiagramofthesystem.
Solution

1) Given that the only information is the response to an essay in jump, we will use it.
graph to deduce an approximate linear model. In the shape of the response, with overshoot
and oscillation and without delay, it is deduced that we can choose a second-order model of
type:

Kwn2
2
+2dw + w 2
n n

0.1 0.83

1.1

The gain K is calculated by the quotient of the change in moisture in state

stationary, 0.83, and the change in the valve opening -10%.As units are requested in
%/%, we must convert the humidity change to a % scale taking into account that 5
units are 100% of the transmitter.

K = (0.83 100 / 5) / (-10) = -1.66 % / %

To calculate the damping value, we will use the overshoot measurement. The value of the
The overshoot is 0.1 units and in % of the final value: 0.1 100 / 0.83 = 12% and it is known.
that the relationship of this value with the damping is given by:

- dp
1- d2
100e
therefore:
 -dp
1- d2
-dp d2 p 2
e =0.12; ln( 0.12)= ; (-2.12)2=
1- d 2 1- d 2
14.36d2=4.495;d =0.56

A similar value can be obtained from the peak percentage graphs /d

To calculate the undamped natural frequencynwe will use peak time. The time
The peak is 1.1 min and it is known to be related to the parameters of the function of
transfer by:
p
t p=
wn1- d 2
therefore:
p
1.1= ;wn 3.44
wn1-0.562

andtheresultingtransferfunctionbetweenthechangesintemperatureandpositionofthe
valveturnsouttobein%/%:

-19.64
s 2+3.86s+11.8

and the block diagram in closed loop is:

W E Y
U -19.64
Kp 2
s +3.86s+11.8

2)The transfer function in closed loop turns out to be:

-19.64Kp
G(s)R(s) s 2+3.86s+11.8 -19.64Kp
Y(s)= W(s)= W(s)= W(s)
1 + G(s)R(s) -19.64Kp s 2+3.86s+11.8-19.64Kp
1+ 2
+3.86s+11.8

To achieve a 20% peak at a jump, regardless of its value, the

closed-loop damping must be
 - dcp
2
1- dc
20=100e Th dc0.45

thatistosay: 3.86 = 2dcw=2 0.45 11.8-19.64KpTHKp-0.34

3) The expression of the closed-loop error is:

1 1 s 2+3.86s+11.8 20
E(s)= W(s)=
1+G(s)R(s) -19.64Kp 2
+3.86s+11.8-19.64Kp
1+
s 2+3.86s+11.8
s 2+3.86s+11.8 20 11.8 20
e ss=lim s 2 = = 1 Þ K p-11.41
s®0 s +3.86s +11.8 -19.64Kp 11.8-19.64Kp

Given that the transfer function between U andWis:

U(s)=R(s)(W(s)-Y(s))=R(s)(W(s)-G(s)U(s))
R(s) Kp Kp(s2+3.86s+11.8)
U(s)= W(s)= W(s)= W(s)
1+G(s)R(s) -19.64Kp 2
+3.86s+11.8-19.64Kp
1+
s 2+3.86s+11.8
K p(s2+3.86s+11.8) 20 20Kp 11.8
u =lim sU(s)=lim s = -11.41
s®0 s®0 s +3.86s+11.8-19.64Kp
2
s 11.8-19.64Kp

that is, 11.41% below its equilibrium value.

4) The response of the closed-loop system to sinusoidal changes in the reference will also be
sine wave of the same period 0.1 min, but with an amplitude and phase shift that will be given in
magnitudeandphasefunctionoftheclosedlooptransferfunctionatthefrequency2π/0.1
= 20p. To calculate them, we will substitute s with j20p in the same, we will assign Kp-11.41
andwewillcalculatethemodulusandargumentoftheresultingcomplexnumber:

G(jw)R(jw) -19.64Kp
= 2 =
1 + G(jw)R(jw)
s=20pj
s +3.86s +11.8 -19.64Kp
s=20pj

-19.64(-11.41) 224.2
=
2j(0)+3.86j20+11.8-19.64(-11.41)-164+77.2j
2

224.2 224.2 224.2

= =1.14; arg( )= -154.8º
-164 + 77.2j 164 2+77.22 -164 + 77.2j

The amplitude of the oscillation will be 1.14 . 10% = 11.4% y

It will be delayed 154.8º = 0.1 154.8/360 = 0.043 minutes with respect to the reference.
5) The root locus diagram corresponds to the positions in the s-plane of the roots.
from the characteristic equation of the closed-loop system for different values of the gain
Kpthat is, the roots of:

s 2+3.86s+11.8-19.64Kp=0
-3.86±3.862-4(11.8-19.64K p) -3.86± - 32.3 + 78.56K
p
s= =
2 2

-3.86±5.68 j
For Kp= 0 result:
2

Since the process has a negative gain, the controller must also have it. If this were not the case,
When humidity increases, the flow of hot gas would decrease, operating in the opposite manner.
What is expected to correct the deviations. When K takes negative p values each time.
greater, the radicand will be negative, so we will have the same negative real part -
3.86/2 and an imaginary part that will become increasingly larger. The resulting diagram is:

This means that for any negative value of the gain, the response will be of type
subamortigued, since the roots are complex conjugates. Furthermore, the time of
settlementwillbesimilar,havingallthesamerealpart.Whenitdecreasesnegatively
profit also comes from the imaginary part, so that the frequency of the oscillations
it will be increasingly greater, just like the peak.
Problem 8

It is known that the transfer functions relating the two inputs and two outputs of a
the process are:

2 2 s-1 2
G 11= G 12 = G 21= G 22 =
s+1 2
s +3s+1 2
+4s+2 3s+1

u1 y1

u2 y2

It is desired to know:

1) To install two SISO controllers, what will be the best pairing of inputs and
outputs?
2) Would that form of regulation be advisable?
3) Assuming that unit gain proportional controllers are placed, what will be the
transfer function that relates output 1 with the inputs of both regulators?

Solution

1) The measure of interaction and the best way to match inputs and outputs can
study oneself with the relative profitability matrix of Bristol. For this, it must be found
previouslytheprofitmatrix.

The profits between the inflows and outflows can be obtained from the functions of
transfer

2 1 2 1
m
il s m
il s
k 11=
s®0 s + 1s =2; k 12 =
s®0 s +3s+1 s =2;
2

1 1

s-1 1 2 1
m
il s m
il s
k 21 =
s®0 s 2
+4 seconds +2 seconds
-0.5 k 22 =
s®0 3s+1s =2
1 1

é 2 2ù
so that: K= ê-0.5 2 ú
ë û
To calculate the RGA, we will calculate one of its elements, for example11andthenestimatethe
others using the properties of the RGA:

k 11k 22 2 2
l 11 = = 0.8 l 12 =1- l11=0.2;
k 11k 22-k12k 212 2 -2(-0.5)
l 21=1- l11=0.2;=1- l =0.8;l 22 21

é0.8 0.2ù
then: RGA= ê0.2 0.8ú
ë û

Now, clearly the best way to associate inputs and outputs is to control output 1 with the
input 1 and output 2 should be regulated with input 2, since the corresponding term of the
RGA (0.8) is closer to 1, so the interaction between the resulting loops will be
minor. The other association, output 1 regulated with input 2 and output 2 regulated with the
Entry 1 has a value of 0.2 in the RGA, which means that, since 0.2 = 2/10, there is
a 500% gain change in a bond of that association when the other switches from
manualtoautomaticandviceversa,whichisnotacceptable.

u1 y1
R1
Process
R2
u2
y2

2) In this case, the change in gain in one loop when the other switches between automatic and
it is manual, taking into account that 0.8 = 8/10, in the same proportion as going from 8 to 10,
Whatisanadmissiblechangeandamultivariablecontrollerwouldprobablynotbenecessary.

3) To calculate this transfer function, we will start from the block diagram:

y1
G11
w1 u1
R1 G21

R2
w2 u2 G12

G22
y2
And operating:

y1=G11u 1+G12 u 2=
=G11R 1 1-y1 ) + G12R 2(w2-y2)
y 2=G21u 1+G22 u 2=
=G21R 1 (w 1-y1 ) + G22R 2(w2-y2)

G11R1 G12R 2
y1= w 1+ (w 2-y2)
1 + G 11R1 1 + G 11R1
G21R1 G 22R 2
y 2= (w 1-y1)+ w2
1+G 22R 2 1+G 22R 2

G11R 1 G 12R 2 G 21R1 G 22R 2

y1= w1+ (w 2- (w 1-y1)- w 2)
1 + G 11R1 1 + G 11R1 1 + G 22R 2 1 + G 22R 2
G 11R1(1+G22R 212R 2G 21R1 G12R 2(1+G22R 2) - G 12R 2G 22R 2
y1= w 1+ w2
(1+G11R1) (1 + G22R 2) - G 12R 2G 21R1 (1+G11R1)(1+G22R 2) -G12R 2G 21R1

Now, replacing R and R1 with 2Kp1y Kp2as well as assigning values to the functions of
transferG:

2 2 s-1 2
G 11= G 12 = G 21= G 22 =
s+1 2
s +3s+1 2
s +4s+2 3s+1

we will obtain the requested transfer function.

Problems of 'Control and Instrumentation of Chemical Processes'
4th Chemical Engineering

Problem 9

The pressure of a storage system of a certain gas is regulated by means of a

systemliketheonereflectedinthefigurebymanipulatingtheoutputline.

PT PC

gas

The system normally operates at a pressure of 3 bar in the storage tank and with
a 30% valve signal. It is known that the system, in open loop, in the face of a change in the
signal to the valve from 30% to 35% gives a temporary response in pressure similar to that of the
Figure, where time is shown in seconds, and pressure in bars, with the transmitter calibrated.
in the range of 1-5 bar.

It is requested:
1) Design a regulator that does not exhibit steady-state error in response to step changes in the
reference, and to minimize the deviation in time on it by correcting possible
perturbations.

2) Calculate the steady-state error that would be obtained with the designed regulator if the reference of
Pressurevariesaccordingtotheequation3+0.4t.

3)Iftheregulatorisplacedinmanualpositionandsinusoidalchangesof2%amplitudearegiven
and period pi/2 sg, to the manual control signal around the value of 30%, what would be the evolution
What is the pressure temporal when a steady state situation is reached?

It is known that the relationship between the temperature of the product that arrives at the device
storage at ºC and the pressure in bars, for a valve opening of 30%,
is given by:
dp
1800 =(-3p2+30)T(t-0.4)-30
dt
How would we modify the previous control scheme to eliminate the effect of the possible
temperature changes of the product concerning pressure? Calculate the parameters of the
new control scheme to meet the specifications of 1) and 4).

Solution

The system, based on the form of the response, shows that it does not have open-loop integrators.
so that a regulator does not present steady-state error in response to changes in the reference
It must be of type PI or PID. On the other hand, the design criterion of the statement is to minimize
the integral of the error modulus (MIAE) in the face of disturbances.This is the criterion used in the
tuning tables of Lopez et al. that apply to underdamped response processes in
open loop, like that of the problem, as shown in the figure. The method of the tables of
Lopez is based on the knowledge of a first-order model with delay of the process of
tipo

Ke - ds
ts+1

Where K is the gain, d is the delay, and t is the time constant. The response of our
the process is not of this type, but it can be approximated by a model of this class.

The profit is calculated in the usual way, by the quotient between the change in output in
steady state and the change in the input. To determine the time constant and the
We can follow several methods, the most common one is based on drawing the tangent line to
the response curve of greatest slope, then determining the intersection points of the
samewithparallelsatthestartandendpoints.Theresultcanbeseeninthefigure:
 0.3 1.9

2.8125

2.8125-3 bar 100% %

Therefore: K= -0.0375 -0.0375 -0.94
35-30 % 5-1 % %

d = 0.3 sg. t = 1.9 - 0.3 = 1.6 sg. and the estimated model is:

-0.94e- 0.3 seconds

1.6s + 1

Another procedure is to calculate the moments in time when 28.3% is reached and the
63.2%ofthefinalvalueandapplythecorrespondingestimationformulas:

0.75 1.4

2.9469=
28.3%

2.8815=
63.2%

2.8125
In this case t2= 1.4 sg, t1= 0.75 sg. so that:

t = 1.5(t2–t1) = 0.975 sg; d = t2-t = 0.425 sg. and the model obtained with this approximation

-0.94e- 0.42s
0.98s + 1

Both are valid, considering that they are first-order type approximations with
delay of a higher-order overdamped system.

Now we can apply Lopez's tables for parallel PI controllers. First of all
we verify that they are applicable when d/t = 0.3 / 1.6 < 1 (and the same for the other
model). Now to calculate the gain Kpwewillusetheformulafromtheregulator:
b
æd ö
K p K =a ç ÷
èt ø

For which the Lopez tables, following the MIAE criteria, provide us with the values:
Criterion Proportional Integral

MIAE a=0.984 a=0.608

b=- b=-

SET a=1.305 a=0.492

b=- b=-

MITAE a=0.859 a=0.674

b=- b=-

a = 0.984, b = -0.986, with which:

-0.986
1 æ0.3 ö
Kp= 0.984ç ÷ -5.45 % / %
-0.94 è 1.6 ø

whileforthetotaltimeTtheformulaisused: I
b
t æd ö
=a ç ÷
Ti èt ø
for which Lopez's table gives the values: a = 0.608, b = -0.707, which leads to:

-0.707
1 1 æ0.3 ö
= 0.608ç ÷ THi0.8sg.
TI1.6 è 1.6 ø

In the same way, the other model could have been used, which would have led to: Kp= -
2.41; Ti = 0.89

2) If the reference pressure varies according to 3 + 0.4 t, since the equilibrium point is 3 bars, the
Change in reference is 3 + 0.4t - 3 = 0.4t bars = 0.4t 100/(5-1) = 10t in %.
the expression of the error is:

1 1 10
E(s)= W(s)= =
1+G(s)R(s) -0.94e- 0.3s K p(Ts I+1) s2
1+
1.6s+1 Tsi

(1.6s+1)0.8s 10
(1.6s+1)0.8s-0.94e- 0.3s (-5.45)(0.8s+1)s2
(1.6s + 1)0.8s 10 0.8 10
e ss =lim s = 1.56%
s®0 (1.6s + 1)0.8s - 0.94e (-5.45)(0.8s+1)s -0.94(-5.45)
- 0.3s 2

3) If sinusoidal changes occur in open loop around 30% of the signal to the valve, the
pressure will oscillate sinusoidally around the equilibrium point of 3 bars, after a
a certain time with a frequency equal to that of the control signal 2pi/(pi/2) = 4 rad/s. and with
anamplitudeandphaseshiftthatdependonthemodulusandargumentofthetransferfunctionat
that frequency:

-0.94e- 0.3 jw -0.94e- 0.3 jw 0.94

G( jw)= ; G(jw) = ; G(j4)= =0.145
1.6(jw)+1 1.6(jw)+1 1.624 2+ 1 2
arg(G(jw))=-p+arg(e- 0.3 jw-arg(1.6(jw)+1);
1.6 4
arg(G(j4))=-p-0.34-arctan -5.75rad
1

That is, it will oscillate with an amplitude of 2. 0.145 = 0.29%, a frequency of 4 rad/s and a phase shift.
regarding the command signal of -5.57 rad = 1.44 s.

4) Since the temperature of the input product acts as a measurable disturbance,

we can improve the control system by incorporating advance compensation
(feedforward). For this we need the transfer function between the pressure and the
input temperature at the operating point, which we can obtain from the equation of the
statement:
 dp
1800 =-3p2+30)T(t-0.4)-30
dt
At the considered operating point, p = 3 bars, so 0 that in steady state
2
we can write: (-3p0 + 30)T0= 30, where T = 10 ºC. Known 0 the point of
operation, we can linearize the previous equation around:
dp
T0= 10ªC, p0= 3 bar, p=0 where
&0 we have used the notation p= &
dt
The model equation is a function of T, p, and its derivative p, and can be linearized
& using a
Taylor expansion. In what follows, to simplify notation, we will use the abbreviation
T = T(t-0.4).

dp
1800+(3p2-30)T+30=0 f (p,p,T)=0
&
dt
¶f ¶f ¶f
linearization is an expression of the type: DT=0 Dp+
& Dp+
&0 0 0

where Dp& = &p &- 0 p; DT = 0 T - T; Dp0 = p - p; with which:

1800Dp+6p
& TDp+(3p
0 0
2
-30)DT=0
0

dDp
1800 +180Dp=3DT(t-0.4)

Taking Laplace transforms on both sides of this linearized equation, and

Consideringthat,ifattheinitialmomenttheprocessisinequilibrium,theinitialvalues
theincreaseswillbenull:

ì dDp ü
1800Lí ý +180L {Dp} =3L {DT(t-0.4) }
î dt þ
1800sP(s) + 180P(s) = 3e- 0.4s T(s) where T(s)=L {DT(t) },P(s)=L },
(1800s+180)P(s)=3e- 0.4s T(s)
andthetransferfunctionturnsouttobe:

3e- 0.4s 0.01661e- 0.4s 0.416e- 0.4s

P(s)= T(s) T(s)bar= T(s)%
1980 10s+1 10s + 1

Since the dynamics of the response to changes in the disturbance T is not faster (greater
time constant 10 versus 1.6 and greater delay, 0.4 versus 0.3) in response to changes in the
Manipulated variable U can use a feedforward compensator to eliminate the
disturbance

The transfer function of the compensator will be given by:

 -0.416e- 0.4s
-D(s) 10s+1 0.44e- 0.4s(1.6s+1)
GF(s)= = =
G(s) -0.94e- 0.3s 10s+1
1.6s + 1

andwouldcorrespondtothefollowingscheme:

T
0.416e- 0.4s
10s+1
0.44e- 0.1s(1.6s+1)
10s+1

W E P
U -0.94e- 0.3s
1
-5.45(1+ )
0.8s 1.6s + 1

Sincetheleadcompensatordoesnotalterthedynamicsofthecontrolloop,itwouldnotbe
Itisnecessarytomodifythetuningoftheregulatortomaintainthespecificationsof1)

The process scheme would be:

PC
TY

TT
gas
Problem 10

A tubular reactor like the one in the figure carries out the conversion of some reactants R that are
they are previously mixed with other S and are introduced from their lower part after passing through a
preheater to produce product A. The reactants have a concentration
constant. The flow of reagents R is imposed by other parts of the process. The reactor does not
it has manipulable variables, although one can significantly act on the
concentration of productAthrough the inlet temperature using the preheater.
This uses water vapor as a heating fluid that partially heats the reactants and
then it is used for other purposes with variable demand.Ascheme is desired to be designed for
control that allows maintaining the concentration of A as precisely as possible.
Justify the proposed scheme.

reactor
S

R
vapor

Solution

Thepointsthatmustbeconsideredinthedesignofthecontrolsystemarethefollowing:

The concentration ofAis influenced by the proper ratio between the flows of R and S. For this purpose
a control ratio must be established between both magnitudes. Since the flow of R is imposed,
the only alternative is to measure it and act with a ratio on the setpoint of a control loop
flow of S. Note that the concentration of R and S is constant according to the statement. Note
also that, in this way, the total flow that reaches the reactor is preset by R and the control
ratio.

To regulate the concentration of productA, the only manipulated variable is the temperature.
preheater outlet. In turn, to be able to set that temperature we need a system
of temperature control of the output of the preheater. Therefore, the control scheme
it will include a control loop for the concentration ofA, in cascade with a control loop
of the preheater temperature. The concentration ofAcan be measured with an analyzer
online.
And operating:

y1=G11u 1+G12 u 2=
=G11R 1 1-y1 ) + G12R 2(w2-y2)
y 2=G21u 1+G22 u 2=
=G21R 1 (w 1-y1 ) + G22R 2(w2-y2)

G11R1 G12R 2
y1= w 1+ (w 2-y2)
1 + G 11R1 1 + G 11R1
G21R1 G 22R 2
y 2= (w 1-y1)+ w2
1+G 22R 2 1+G 22R 2

G11R 1 G 12R 2 G 21R1 G 22R 2

y1= w1+ (w 2- (w 1-y1)- w 2)
1 + G 11R1 1 + G 11R1 1 + G 22R 2 1 + G 22R 2
G 11R1(1+G22R 212R 2G 21R1 G12R 2(1+G22R 2) - G 12R 2G 22R 2
y1= w 1+ w2
(1+G11R1) (1 + G22R 2) - G 12R 2G 21R1 (1+G11R1)(1+G22R 2) -G12R 2G 21R1

Now, replacing R and R1 with 2Kp1y Kp2as well as assigning values to the functions of
transferG:

2 2 s-1 2
G 11= G 12 = G 21= G 22 =
s+1 2
s +3s+1 2
s +4s+2 3s+1

we will obtain the requested transfer function.

Problems of 'Control and Instrumentation of Chemical Processes'
4th year of Chemical Engineering

Problem 23

In Fig. 1, a process can be seen where a controllable currentAenters and another current
non-manipulable B. CurrentAcan be regulated through a flow control system.
as seen in the figure. When the setpoint z of the regulator is increased by 2 l/min
flow of A, starting from a certain steady state of the entire system, the response that is
The density of the output product can be seen in Fig. 2. In this figure, the
Time axis units are minutes and density units are Kg/l, and the latter is measured with a
transmitter calibrated in the range of 0.5-3 kg/l to provide a signal of 4-20 mA. Similarly
TheflowtransmitterAiscalibratedintherangeof0-20l/min.

FT FC

r
process
A

Fig. 1
It is requested:

1) Estimate a dynamic mathematical model that relates changes in flow setpoint.

of product A with the changes in the output density of the product in that area of
work.
2) Propose a density control system and draw a diagram of it using
ISA nomenclature. Design the simplest density regulator that does not present error.
stationary against step changes in the reference and that minimizes the deviations of
the density over the reference value when B experiences changes.
3) Draw a block diagram of the closed-loop process and express the relationships that
link the changes in density with the changes in density reference and B, as well as
the unitsofthedifferentvariables.
4) In open loop, when flow B experiences a step change, the density varies from
continuous form without stabilizing, and when B experiences a positive unit impulse
(l/min) the density stabilizes, after a certain time, at a new value of 2 Kg/l for
below the initial steady value. Knowing this, we are asked to calculate the error
 stationary, if it exists, that the system in closed loop with the regulator will present
calculatedin2),againstaunitstepchange(l/min)inflowB.
5) What is the phase margin of the closed-loop system calculated in 2)?, What
What interpretation can you give to this value?

1.45

1.4

1.35

1.3

1.25

1.2

1.15

1.1

1.05

1
0 0.5 1 1.5 2 2.5 3

Time (min)

Fig.2

Solution

1) In view of the statement of the problem, we must obtain the requested mathematical model.
starting from the response curve of Fig.2. Since it is a process
overdamped, we can try to approximate the response to that of a first-order system.
order with delay:

Ke - ds
ts+1

To estimate this model, the profit is calculated in the usual way, by the quotient between
the change in steady-state output and the change in input.

With reference to Fig. 3, we have:

 1.417-1 Kg
K= =0.21
2 min
or well expressed in %/%:

100
(1.417-1)
2.5 18.8 %
K= = 1.88
100 10 %
2
20-0

To determine the time constant and delay, we can follow several methods, the most
common is based on drawing the tangent line to the curve of response with the greatest slope,
then determining the intersections of it with parallels at the starting points and
final.Theresultcanbeseeninthefigure:

1.374
1.45

1.417
1.4

1.35

1.3

1.25

1.2

1.15

1.1

1.05

1
0 0.5 1 1.5 2 2.5 3

Time (min)
0.22

Fig.3
From where it is deduced:

d = 0.22 sg. t = 1.374 - 0.22 = 1.15 sg. and the estimated model is:

1.88e- 0.22s
r(s)= z(s)
1.15s + 1
2) A schematic of a density control system can be seen in Fig. 4. It incorporates a
density transmitter and a regulator that acts in cascade on the flow reference
A.

w
DC
z

FT FC
DT
r
process
A

Fig.4

Since the process does not have integrators in its response to z, the regulator is more
simple that can eliminate steady-state errors in response to setpoint changes is a PI.
according to the requested specifications, we can use the Tables of Lopez et al. with the
MIAE criteria for tuning, as they correspond to the type of identified model, verify
theconditiond/t=022/1.15<1,andtheyconformtotherequestedcriterion.
Now to calculate the profit Kpfromtheregulatorwewillusetheformula:

b
æd ö
K p K =a ç ÷
èt ø

For which the Lopez tables, following the MIAE criteria, provide us with the values:
Criterion Proportional Integral

MIAE a=0.984 a=0.608

b=- b=-

PUT a=1.305 a=0.492

b=- b=-

MITAE a=0.859 a=0.674

b=- b=-
a = 0.984, b = -0.986, therefore:
-0.986
1 æ0.22 ö
Kp= 0.984ç ÷ 2.67% / %
1.88 è 1.15 ø

whileforthetotaltimeTtheformulaisused: i
b
t æd ö
=a ç ÷
Ti èt ø

for which the Lopez table gives the values: a = 0.608, b = -0.707, which leads to:

-0.707
1 1 æ0.22 ö
= 0.608ç ÷ THI=0.59sg.
Ti1.15 è 1.15 ø

3) A block diagram of the process can be seen in the figure:

B D(s)

%
W% E% r%
z%
1 1.88e - 0.22s

2.67(1+ )
0.59s 1.5s + 1

The output of block D, unknown, must be in %, while the input can be listed
in l/min since we do not know anything about the respective transmitter. From the diagram, it can be deduced that
expression of the transfer functions that relate the density to the reference and the
caudal B:
 1.88e- 0.22s 0.59s + 1
2.67
GR D D
r(s)= w+ B= 1.15s +- 0.22 1 0.59s w+ B=
1+GR 1+GR 1.88e seconds
0.59s + 1 1.88e - 0.22s
0.59s+1
1+ 2.67 1+ 2.67
1.15s + 1 0.59s 1.15s + 1 0.59s
1.88e 2.67(0.59s+1)
- 0.22s
0.64s(1.5s+1)D
= w+ B
(1.15s+1)0.59s+1.88e- 0.22s
2.67(0.59s + 1) (1.15s+1)0.59s+1.88e- 0.22s2.67(0.59s+1)

4) The expression of the error in closed loop for a change in B is given by:

D D(s) -0.59s(1.15s+1)D(s)
E(s)= - B= - - 0.22s
B= B
1+GR 1.88e 0.59s + 1 (1.15s + 1)0.59s + 1.88e- 0.22s2.67(0.59s+1)
1+ 2.67
1.15s + 1 0.59s

and the steady-state error in response to a step change in B can be calculated as:

-0.59s(1.15s+1)D(s) 1 -0.59sD(s)
e ss= l imsE(s) = l im s =l im
s®0 s®0 (1.15s + 1)0.59s + 1.88e- 0.22s
2.67(0.59s+1) s s®0 1.88 2.67

To calculate this limit, according to the statement, it is known that D(s) contains an integrator.
that responds monotonically without stabilizing to a step change in B, and
Moreover, it only has an integrator since it reaches a steady value in response to an impulse.
This will cancel the s of the previous expression. In reality, the only information
necessary to solve the problem is related to the steady-state value of the response to a
impulse in B in open loop.This is given by:
100
m
il sDs(1)=2-Kg=l/2- -80%
2.5

so that the previous error reaches the value:

-0.59sD(s) (-0.59)(-80)
e ssl imsE(s)=l im = 9.4%
s®0 s®0 1.88 1.88 2.67

5) To calculate the phase margin of the system, the value of the expression MF = must be evaluated.
p+ arg(G(jw)R(jw)) at a frequency value for which | G(jw)R(jw)|=1.

To estimate that frequency, we must first solve the equation:

1.88e- 0.22jw 0.59jw+1 1.88 0.592w2+ 1
2.67 =1; 2.67 =1
1.15 jw+1 0.59 jw 1.152w2+ 1 0.59w
72.4 0.592w2+ 1
=1; 72.4(0.592w2+ 1 ) = ( 1 . 1 5 2w2+ 1 ) w 2;
1.152w2+1 w2
24.2±24.22+4 1.15272.4 24.2±31.12
1.152w4-24.2w2-72.4=0; w2= 2
= 20.9
2 1.15 2 1.152
w = 4.57 rad / min
Now we can calculate the MF:
é1.88e -0.22jw 0.59jw +1 ù æ pö
MF = p + arg ê 2.67 ú = p + -0.22w
ç + arctan 0.59w - arctan 1.15w - ÷=
ë 1.15jw +1 0.59jw û è 2ø
p p
= -0.22(4.57)+arctan 0.59(4.57)-arctan1.15(4.57)= - 1+1.21-1.38=0.4 rad=
2 2
23 degrees

Based on the relationship between phase margin and damping in second order systems
order, the small phase margin found allows us to predict that the response of
the system will have a high overshoot and the degree of robustness will be low in the face of changes in the
processes that increase the system's lag, such as delays and time constants,
thereisadangerofdestabilizationiftheseincreaseforanyreason.
Problem 24

Inthediagramofthefigure,aprocesscanbeseeninwhichacurrentA,notmanipulable
and experiences significant changes, exchanges heat with another liquid C at high
temperature after passing through tank 1. At the supply point of liquid C, the
temperature is noticeably constant, but the pressure undergoes some variations
importance. Current A, after passing through the exchanger, goes to tank 2, where a
part is recycled to deposit 1 through a pumping system that imposes a flow rate
constant and another part, B, is the output of the process. It is desired to maintain the flow rate and temperature.
as constant as possible, as well as ensuring the safety of the process. There is a
overflow system in tank 2 that prevents overflow from being problematic in
said deposit.

It is requested:

Choose the appropriate instrumentation and control loops for it and draw the
corresponding control scheme according to ISAstandards, justifying it.

1 D C

Solution

The regulation objectives are to maintain the flow and temperature of B. Moreover, for reasons
For safety, the levels in both tanks must be regulated and the circulation must be ensured.
1 to 2 through the changer.
The flow rate of B can be maintained using a flow controller and a flow transmitter.
To act on the temperature of B, the simplest way that does not interfere with the others.
circuits, it is to vary the flow of C. However, as the supply pressure of C
significant changes, it is logical to use a cascading structure in which the
temperature regulator acts on the setpoint of a flow regulator of C, of this
this last regulator absorbs the disturbances of the feeding pressure that do not
theywillsignificantlytransmittothetemperature.

The level of tank 1, since entry A is unmanageable, must be regulated with the
output D. Note that, although A were zero, output D would not be stably null due to
to the recirculation flow and therefore, the heat exchanger would not be empty.Aconsequence
The secondary of this tie is that it modifies the output temperature, so if you want to
to maintain accurately, a feedforward correction must be incorporated to the flow of C for
compensate for these disturbances.

The level of tank 2 must also be regulated, but since there are no other manipulated variables left,
there is no other remedy than to establish an override control over the flow of B. Note that
here the important thing is to maintain the level above a minimum, since it must be protected
pumping equipment and overflow problems are not considered as such.

Temperature, level, and flow transmitters will be used to measure these variables.
conventional.The entire proposal can be seen in the figure:

FT FY

LC FC FT

LT
1

2 TT TC

B
FT
LT

FC LS LC Lmni