Process Control in the Chemical Industries 64

FEEDBACK CONTROL

1. Feedback closed-loop system

The simplest and most widely used method for process control is the feedback
control loop shown in Figure (1). A measurement of the controlled process variable
(CV) is compared to a set point (SP) to create an error. This error is used to derive the
correction action of the final control element (FCE) via the controller. The controller
output changes the manipulated variable (MV). The action of the controller may be
aggressive or sluggish; it depends on the internal control law and tuning that is used. In
this lecture we will discuss feedback controller types and tuning them in order to keep
the CV at the desired SP in presence of process disturbances. Disturbances typically
come in three types: input disturbances, load disturbances, and set point disturbances.
An input disturbance is the change in the mass or energy of the supply to the process
which could change the conditions of the process variables. A load disturbance is any
other upset, except for an input mass or energy changes. In this work we use the letter D
for the previous type of disturbances. A set point disturbance (SP) occurs when the
desired state of the CV changes, and the process must adjust to a new state.

All elements of the feedback loop can affect control performance. The controller
output consists of the feedback signal which has a range usually expressed as 0 to
100%. When the signal is transmitted electronically, it usually converted to a standard
signal range of 4 to 20 milliamperes (mA) and can be transmitted long distances (over
one mile). When the signal is transmitted pneumatically, it has a range of 3 to 15 psig
and can only be transmitted over a shorter distance. Pneumatic transmission is not
common with modern equipment since it requires longer time (several seconds) than
electronic transmission.

At the process unit, the transmitted signal is used to adjust the final control
element. The FCE as in over 90% of process control applications, is a valve. The valve
percent opening could be set by an electrical motor, but this is not usually done because
of the danger of explosion with high-amperage power supply a motor would require.
The alternative power supply typically used is a compressed air. The signal is converted
from electrical to pneumatic with 3 to 15 psig range. The pneumatic signal is
transmitted a short distance to the control valve. Control valves respond relatively
quickly, with typical time constant ranging from 1 to 4 seconds.

After the FCE has been adjusted, the process responds to the change. The
process dynamics vary greatly for the wide range of equipment in the process industries,
with typically dead times and time constants ranging from a few seconds (or faster) to
hours. When the process is by far the slowest element in the control loop, the dynamics
of the other elements are negligible. This situation is common in chemical engineering
applications.

The sensor responds to the change in the plant conditions. Usually, the sensor is
not in direct contact with the potentially corrosive process materials; therefore, the
protective equipment or sample system must be included in the dynamic response. For
example, a thin thermocouple wire respond quickly to a change in temperature, but the

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 65

metal sleeve around the thermocouple, the thermowell, can have a time constant of 5 to
20 seconds. Most sensor systems for the flow, pressure, and level have a time constants
of a few seconds. Analyzers that perform complex physiochemical analysis can have
much slower responses, on the order of 5 to 30 minutes or longer.

The sensor signal is transmitted to the controller, which is considered to be

located in a remote control room. The transmission could be pneumatic (3-15 psig) or
electrical (4-20 mA). The controller receives the signal and performs its control
calculations. The controller can be an analog system (electrical analog controller
consists of an electrical circuit that obeys the same equations as the desired controller
calculations) or a digital system where the controller is a continuous electronic
controller that performs its calculations instantaneously (usually a very fast digital
computer is used as in most modern control equipment).

It is worth recalling that the empirical methods for determining the process dynamics
presented in the previous lectures involve changes to the manipulated signal and
monitoring the response of the sensor signal as reported to the control system. Thus, the
resulting model includes all elements in the loop including instrumentation and
transmission.

Controller
Transmittor
Set Point (SP)
Controller algorithm

MV Controlled variable
(CV)

Transmitter
Process variable
Mass or energy (PV)
Process
Final control Sensor
element (FCE)

Disturbance (D)

Figure(1): Feedback Control System

2. Feedback Block Diagram

Consider a general block diagram for a feed back control system as shown in
Figure (2). The transfer functions for the control system relate the controlled variable to
the external disturbance (D) and to the desired set point (SP) can be defined as

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 66

CV ( s ) G p Gv Gc
=
SP ( s ) 1 + G p Gv Gc G s

CV ( s ) Gd
=
D( s ) 1 + G p Gv Gc G s

The block diagram provides a visual “picture of the equations” showing the
feedback loop. The general closed-loop transfer function model can be applied to any
specific system by substituting the transfer function models for the loop elements.

The first major reason for the feedback system is to maintain the controlled
variables at the desired set point which is also refereed as the servo problem. This can
be achieved by choosing the suitable controller (Gc) which make the ratio of
CV(s)/SP(s) equal one or

G p G v Gc ≈ 1 + G p G v G c G s

The second reason is to reject the effect of the disturbance on the controlled
variable. For this case, the controller transfer function (Gc) should be selected to make
the ratio of CV(s)/D(s) approaches zero

Gd <<< 1 + G p Gv Gc G s

D(s)
Gd(s)

SP(s)
e(s) MV(s)
+ Gc (s) Gv (s) Gp(s) CV(s)
+

CV m(s)
Gs (s)

Transfer Functions: Variables

Gc(s) = Controller CV(s)=Controlled variable
Gv(s)=Transmission, transdeucer and valve Cvm(s)=Measured value of controlled variable
Gp(s)=Process D(s)= Disturbance
Gs(s)=Sensor, transdeucerand transmission e(s)=Error
Gd(s)=Disturbance MV(s)=Manipulated variable
SP(s)=Set point

Figure (2) Feedback System Block Diagram

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 67

2.1 Closed loop stability

The essential information on stability is in the denominator of the closed loop

transfer functions, called the characteristic polynomial:

1 + G p G v Gc G s

A closed-loop control system is locally stable at the steady state point if all roots
of the characteristic polynomial have negative real parts. If one or more roots with
positive or zero parts exist, the system is locally unstable. Entries in the overall transfer
function denominator demonstrate that only the elements in the feedback loop affect the
system stability; neither the disturbance nor the set point change affects stability.
Essentially all real process control systems can be made unstable simply by using
unsuitable controllers. The most commonly used stability analysis methods are
summarized in Table (1).

Table (1) Stability analysis methods

Method Plant model Stability Results
results display
Root locus Polynomial in s Relative Graphical
Routh Polynomial in s Yes or No Tabular
Bode Open loop stable Relative Graphical
Monotonic decreasing amplitude
ratio and phase angle as frequency
increases
Nyquist Linear Relative Graphical

3. Controllers Modes

3.1 On-off control

The simplest form of control law is on-off control. This type of controller is
primarily intended for use with final control elements (FCE) that are non-throttling in
nature. An example is a home heating system, whenever a temperature goes above the
set point, the heating process shuts off. For this example, the behavior of this control is
given by:

MV = 0% for CV > SP
MV = 100% for CV < SP

This type of control can be used successfully for processes with large
capacitance where tight control is not important (capacitance represents system ability
to absorb or store mass or energy and it could be defined as the resistance of a system to
the change of mass or energy stored in it). A good example of this process is a surge
tank. One disadvantage of this type is the continual opening and closing of the
controller, the FCE quickly becomes worn and must be replaced. The primary
characteristic of on-off controller is that the process variable is always cycling about the

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 68

set point. The rate of CV cycling and its deviation from the set point are a function of
the dead time and capacitance in the system. Most on-off controllers are built with an
adjustable differential gap, inside which no control action takes place. This is to
minimize the controller cycling and extend the operational life of the FCE.

3.2 Proportional Control (P)

Proportional control is the simplest continuous control mode that can damp out
oscillations in the feedback control loop. This mode stops the CV from cycling but does
not necessarily return it to the set point. The output of the P controller is proportional to
the error:

MV = Kc e

the transfer function for the proportional controller has the form

Gc ( s ) = K c

where K c is the controller gain and e is the error defined by the difference between the
controlled variable and the set point.

e = SP − CV for reverse acting

e = CV − SP for direct acting

Increasing K c can decrease the error, but we should be aware not to increase K c such
that it makes the closed loop unstable. Because of this loop gain limit, there is another
approach to reducing the error to zero. Another term can be added to the proportional
controller:
mv = K c e + b

this additional term is called the bias, and is simply defined as the output of the
controller when the error is zero.

The controller gain is the ratio of the change of controller output to the change in
error. Since there is a one-to-one relationship between CV and e the controller gain can
be written as:

ΔMV
Kc =
ΔCV

The proportional band is defined as the change in CV that will cause the output of the
controller to change by 100%. The controller gain can be related to the proportional
band by:

100%
Kc =
PB%

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 69

Virtually all modern controllers use a gain adjustment, however a few older controllers
exist that still use a proportional band adjustment.

In general, a proportional controller provides a fast response compared to other

controllers but a sustained error occurs where the CV does not return to the set point
even when steady-state is reached. This sustained error is called offset and is
undesirable in most cases.

3.3 Integral control (I-only)

The action of integral control is to remove any error that may exist. As long as
there is an error present, the output of this controller mode continues to move the FCE
in a direction to eliminate the error. The equation for this control mode is:

1
Ti ∫
MV = edt + mvo

mvo is defined as either the controller output before integration, or the initial condition
at time zero. The integral time Ti , is defined as the amount of time it takes the controller
output to change by an amount equal to the error. Thus, it is measured in minutes per
repeat. As a result of the reciprocal relationship of the integral time, some manufactures
adjust their controllers in repeat per minute (1/ Ti ). Then increasing such adjustment
gives less integral action.

Although an integral only controller provides the advantage of eliminating the

offset, there is a significant difference in its response time when compared to a P only
controller. The output of the P controller changes as quickly as the measurement
changes, so if the error measurement change as a step, the controller output also changes
as a step by an amount depending on the controller gain. For a step input to an integral
controller, the output does not change instantaneously but rather by a rate that is
affected by Ti and e.

Hence, an integral only control, due to the additional lag introduced by this
mode, has an overall response that is much slower that that for proportional controller.
If no offset is required then a slower period of response must be tolerated. If the
requirement is a return to the set point with no offset and a faster response time is
necessary, then the controller must be composed of both P and I actions.

3.4 PI Control
The majority (>90%) of controllers found in plants are PI controllers. The
equation for PI is given as:

1
Ti ∫
MV = K c (e + edt )

and the PI controller transfer function is:

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 70

1
Gc ( s ) = K c (1 + )
Ti s

The PI controller gain has an effect not only on the error, but also on the integral action.
The bias term for PI control when compared with the P controller equation is:

1
Ti ∫
b = Kc edt

Therefore, the integral action provides a bias hat is automatically adjusted to eliminate
any error. The PI control is faster in response than the I-only controller because of the
addition of the proportional action. The gain of the PI controller, K PI can be defined as:

Kc
K PI = K c +
Ti

The PB and Ti are used to adjust the PI controller to give the loop a desired response.
When setting Ti to a very large number in min/rep, the integral action is minimized and
the controller response would be very close to the P-controller. While, if we set Ti to a
very small value, the controller gain would approach that of integral-only controller and
the controller action will return the CV to the set point but a with a long response
period.

The response of the PI controller will be slower than the P controller. Thus, the
response period of the a loop under PI control is 50% longer than that for a loop under
P-only control. In order to increase the speed of the response it may be necessary to add
an additional control mode.

3.5 Derivative control

The purpose of the derivative controller is to provide lead to overcome lags in
the loop. It anticipates where the process is going by looking at the rate of change of
error. The output of the controller is defined as:

de
output = Td
dt

Figure (3) shows how the output from a derivative block would vary for different inputs
given a fixed value of Td . As the rate of change of the input gets larger, the output gets
larger. When the slop of the input approaches infinity (very close to a step function), the
output would be a pulse of infinite amplitude and zero time length. This output is
unrealizable since a perfect step with zero rise time is physically impossible, but signals
that have short rise and fall time do occur and these are referred as noise. Thus, the
output from the derivative block would be a series of positive and negative pulses,
which would try to derive the FCE either full open or full close. This would result in
accelerated wear on the FCE and no useful control.

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 71

For processes, where the measured variable has high frequency noise (even with
small amplitude), the derivative control will notice them and the control outputs would
be a series of large amplitude pulses. For such case, the noise must be filtered or
eliminated by modifying the installation of the primary sensor.

It is important to note that derivative control would never be the sole control
mode used in a controller. The derivative action does not know what the set actually is
and hence can not control to a desired set point.

The minimum controller configuration containing derivative action is a

combination of proportional plus derivative action:

de
MV = K c (e + Td )+b
dt

The addition of the derivative action results in a faster response for the measurement
process variable and a smaller offset than the loop under P-only control. In the previous
PI case, Ti can be set to a large number to eliminate the integral action, while in the PD
controller, even by setting Td to a very small value, there is still the possibility of a
sizable derivative action if there is a noisy input (if dCV/dt is large). In electronic
controllers and distributed control systems (DCS) the derivative action can be
eliminated by setting Td to zero. In a pneumatic controller the derivative action can not
be eliminated but can be reduced to a minimum value of approximately 0.01 min. If a
PD controller is installed on a flow loop there will still be considerable derivative action
due to the noisy flow measurement. It is therefore important when applying a pneumatic
controller to a noisy loop to make certain the controller does not contain a derivative
block.
e

4 1
2
3
Derivative output

Time

1
2
3
4
∞

Time
Figure (3) Derivative action

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 72

3.6 Proportional integral derivative (PID) control

The primary purpose of a PID is to provide a fast response that is much the same as
with P-only controller but which has no offset. The derivative action adds the additional
response speed required to overcome the lag in the response from the integral action.
The control algorithm for the PID controller is given by:

100 1 de
MV = (e + ∫ edt + Td )
PB Ti dt
100 1 dCV
MV = (e + ∫ edt − Td )
PB Ti dt

The PID transfer function is given by:

1
Gc ( s ) = K c (1 + + Td s )
Ti s

Figure (4) presents a comparison of the responses for P, PI and PID controllers to a step
change in the load. Therefore, a PID controller provides a tight dynamic response, but
since it contains a derivative term, it cannot be used in any process which noise is
anticipated.

The flow chart given in Figure (5) summarizes a procedure for controller selection.

Disturbance

P only
CV

PI
CV

PID
CV

Time
Figure 4: P-only, PI and PID responses to a load disturbance

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 73

Start

Yes
Can offset be telorated? Use P only
No
Yes
Is there noise presented? Use PI only
No
Yes
Is dead time excessive?
No
Yes
Is capacity extremely small?
No

Use PID
Figure (5) Flowchart for Controller selection

4. Tuning Feedback Controllers

Controller tuning can be defined as an optimization process that involves a

performance criterion to the form of the controller response and to the error between the
process variable and the set point. Depending on the process to be controlled, the first
consideration is to decide what type of response is optimal. Typical process responses to
load changes are shown in Figure (6). There are three general possible responses:

1. Over-damped - slow response with no oscillations

2. Critically damped – fastest response without oscillations
3. Under-damped – fast return to set point with considerable oscillations.

The selection of good control is a trade off between the speed of the response and
the deviation from the set point. A highly tuned controller may become unstable if large
disturbances occur, whereas a sluggish tuned controller provides poor performance but
is very robust. What is typically required for most process control loops is a
compromise between performance and robustness.

There are several common performance criteria that can be used for controller
tuning, based on the closed loop response. In the following we review the most
important criteria:

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 74

Cyclic radian frequency

ω = 2πf

And
1
f =
period

Underdamped
Process Variable

Overamped

Time
Figure (6): Typical responses to a set point change

The cyclic radian frequency can be related to the under-damped natural frequency
ωn and the damping coefficient ζ as follows:

ω = ωn 1 − ξ 2

Overshoot
Overshoot is the amount by which the response exceeds the steady-state final value.
Referring to Figure (7), the overshoot is defined as:

B 2
= e −πξ (1−ξ )
A

Decay ratio
Decay ratio is the amplitude of an oscillation to the amplitude of the preceding
oscillation, (C/B in Figure 7). The quarter decay ratio (QDR) which lies between critical
damping and under-damping:

C 1
QDR = =
B 4

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 75

The QDR is often used to establish whether the controller is providing a satisfactory
response. It has been shown through experience that the QDR provides a good trade off
between minimum deviation from the set point and fastest return to the set point. For a
second order system it can be shown that:

C
= exp(−2π(1 − ξ 2 ))
B

Rise Time
The rise time is the time required by the transient response to reach the final steady-state
value.

Response time
Is the time required for the response to settle within the specified arbitrary limits. These
limits are typically set as ± 3 − 5% of the process variable steady state value.

A
C
PV

time
Rise time
Figure (7) Second or higher order typical response to a set point change

4.1 Error Performance criteria

The previous simple criteria use only few points in the response and therefore are
simple to use. The following criteria are based on the entire response of the process.

Integrated error (IE)

The integrated error is defined as:
∞
IE = ∫ edt
0

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 76

The controller parameters are selected to minimize the IE criteria. But, the designer
should bare certain situations where the positive and negative deviations from the set
point cancel each other.

Integrated absolute error (IAE)

Both negative and positive errors will be considered. The IAE is defined as:
∞
IAE = ∫ | e | dt
0

Integrated square error (ISE)

This error uses the square of the error, thereby penalizing large errors more than small
errors. This gives more conservative response (faster return to set point).

∞
ISE = ∫ e 2 dt
0

Integrated time absolute error (ITAE)

This criterion is based on the integral of the absolute value of the error multiplied by
time. It results in errors existing over time being penalized even though may be small,
which results in a more heavily damped response.

∞
ITAE = ∫ t | e | dt
0

Figure (8) shows the various responses of a loop that is tuned to the above criteria.

CV Type damping
1 ITAE Least
3 2 IAE More

2 3 ISE Most

Time
Figure (8) Responses to various error criteria

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 77

5. Tuning methods

5.1 Trial and error method

Guidelines:
1. Proportional action is the main control. Integral and derivative actions
are used to trim the response.
2. The starting point is always with the controller gain, integral action and
derivative action all at a minimum.
3. Make adjustments in the controller gain by using a factor of two.
4. Use the QDR criteria to determine the optimal response.
5. When in trouble (for example, unstable behavior) decrease the integral
and derivative actions to a minimum and adjust the controller gain for
stability.

It is important to know that controller parameters are strongly dependent on the

individual process. When dealing with a flow control loop, P-only control can be used
with a low control gain. For accuracy, PI controller can be used with a low gain and
high integral action. Derivative actions can not be used for flow loops because such
process has a very fast dynamics and flow measure met is inherently noisy.

Levels represent material inventory that can be used as a surge capacity to

dampen disturbances. Hence, loosely tuned P-only control is sometimes used. However,
some operators don not like offset, so PI controllers are typically used. Temperature
dynamic responses are usually fairly slow, so PID control is used.

5.2 Process reaction curve method

In this method a process reaction curve is generated in a response to a disturbance.
This curve is then used to calculate the controller gain, integral time and derivative time.
These methods are performed in open loop so no control action occurs. To generate this
a curve, the process is allowed to reach steady state. Then, a small disturbance (step
input) is introduced and the reaction of the process variable is recorded. Figure (9)
shows a typical process reaction curve for a generic self-regulating process. The process
parameters than may be obtained from this curve are as follows:

θ = lag time (min)

τ = time constant estimate (min)
P = initial step disturbance (%)
ΔCp = change in process variable (PV) to step disturbance
N = ΔCp/τ = reaction rate (%)
R = (N θ)/ ΔCp = lag ratio

Based on the process reaction curve parameters, three tuning methods are presented:

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 78

Ziegler-Nichols (ZN) open loop rules

The ZN open loop recommended controller parameters for quarter decay ratio are as
follows:
P-only P
kc =
Nθ
PI ⎛ P ⎞
kc = 0.9⎜ ⎟, τ I = 3.33θ
⎝ Nθ ⎠
PID ⎛ P ⎞
kc = 1.2⎜ ⎟, τ I = 2θ, τd = 0.5θ
⎝ Nθ ⎠

These setting should be taken as recommendations only and tested thoroughly in closed
loop. Note that the ratio ΔCp/P is the process gain, i.e. kp. Note also that the integral
time Ti and derivative time Td is replaced (in the above and following tables) by τI and
τd, respectively.

Cohen-Coon open loop

The Cohen-Coon (C-C) recommendations correct for one deficiency in the ZN open
loop rules. This is the sluggish closed loop response given by the ZN in the relatively
rare occasions when process with dead time is large relative to the dominant time
constant. The C-C settings are as follows:

P-only P R
kc = (1 + )
Nθ 3
PI P R ⎛ 30 + 3R ⎞
kc = (0.9 + ) τ I = θ⎜ ⎟
Nθ 12 ⎝ 9 + 20 R ⎠
PID P R ⎛ 32 + 6 R ⎞ ⎛ 4 ⎞
kc = (0.33 + ) τ I = θ⎜ ⎟ τd = θ⎜ ⎟
Nθ 4 ⎝ 13 + 8 R ⎠ ⎝ 11 + 2 R ⎠

As with the Z-N open loop method, the C-C recommendations should be tested in
closed loops and adjust accordingly to the QDR.

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 79

τ
Output Process
variable

ΔCp

Time
θ
Figure (9) Process reaction curve
Internal model control (IMC) tuning rules
Many practitioners have found that the Z-N open loop and C-C rules are too
aggressive for most chemical processes since they give a large controller gain and
short integral time. Tuning rules based on the internal model of the process were
developed with robustness in mind. These rules were related directly to the closed
loop time constant and the robustness of the control loop. As a sequence, the closed
loop step load response exhibits no oscillations or overshoot. The following
simplified IMC rules were developed by Fruehauf et. al. (1993) for PID controller
tuning.

τ τ τ
>3 <3 < 0.5
θ θ θ
kc P P P
2 Nθ 2 Nθ Nθ
τI 5θ τ 4
τd ≤ 0.5θ ≤ 0.5θ ≤ 0.5θ

5.3 Constant Cycling Method

Ziegler-Nichols closed loop method
The closed-loop technique of Ziegler and Nichols is a technique that is
commonly used to determine the two important system constants: ultimate period
and ultimate gain. These are determined by disturbing the closed loop system and
using disturbance response to extract the values of these constants. The following is
a step-by-step approach to using Ziegler-Nichols closed loop method for the
controller tuning:

1. Attach a P-only controller with low gain.

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 80

2. Increase the controller gain until a constant amplitude limit cycles occurs
(Figure 10).
3. Determine the following parameters from the constant amplitude limit cycle:

Tu = ultimate period = period taken from the limit cycle

ku = ultimate gain = controller gain that produces the limit cycle

4. calculate the tuning parameters using the following equations:

P-only kc = ku/2
PI: kc = ku/2.2, τI = Tu/1.2
PID kc = ku/1.7, τI = Tu/2, τd = Tu/8

5. Fine-tune by adjusting the controller parameters.

Process variable
Tu

time
Figure 10: Continuous cycling response

6. Guidelines for common control loops

Flow Control
Flow and liquid pressure control loops (Figure 11) are characterized by fast
responses (seconds), with essentially no time delay. The process dynamics are due to
compressibility ( in a gas stream) or inertial effects (in a liquid). The sensors and signal
transmission line may introduce sufficient dynamic lag if pneumatic instruments are
used. Disturbances in flow-control system tend to be frequent but generally not of large
magnitude. Most of the disturbances are of high-frequency noise due to stream
turbulence, valve change, and pump vibration. PI flow controllers are generally used
with intermediate values of the controller gain Kc. The presence of high frequency
noise rules out the use of derivative action.

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 81

FC sp
FT

Valve

FC Sp

FT
Po

F Valve

Figure (11) Flow and liquid pressure control loops

Liquid Level
Figure (12) shows two control loops used to control the liquid level. Because of
the liquid level process integrating nature, a relatively high-gain controller can be used
with little concern about instability of the control system. Integral action is normally
used but is not necessary if small offsets in the liquid level can be tolerated. Derivative
action is not normally used, since the level measurements often contain noise due to
splashing and turbulence of the liquid entering the tank.

In many level control problems, the liquid storage tank is used as a surge tank to
damp out fluctuations in its inlet streams. If the exit flow rate from the tank is used as
the manipulated variable, then conservative controller setting should be applied to avoid
large, rapid fluctuations in the exit flow rate (this strategy is called as averaging
control). If the level also involves heat transfer such as in an evaporator, the process
model and controller design become much more complicated.

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 82

sp
Fi LC

LT

Fo
(a)

Fi
LT

LC sp

Fo
(b)
Figure (12) Level Control

Gas Pressure
Gas pressure is relatively easy to control, except when the gas is in equilibrium
with a liquid. A gas pressure process is self-regulating: the vessel or pipeline admits
more feed when the pressure is too low, and reduce the intake when the pressure
becomes too high. PI controllers are normally used with only a small amount of integral
action.

Temperature
General guidelines for temperature control loops are difficult because of the
wide variety of processes and equipment involving heat transfer. For example, the
temperature control problems are quite different for heat exchangers, distillation
columns, chemical reactors, and evaporators. Due to the presence of time delays and/or
multiple thermal capacitances, there will be usually a stability limit in the controller
gain. PID controllers are commonly employed to provide more rapid responses than can
be obtained with PI controllers.

Temperature control loops can be divided into two main categories:

1.Endothermic – require heat energy

2.Exothermic – generating heat energy.

Both of these processes have similar characteristics in that they are typically
comprised of one large and many small capacities (valve, transmitter,…). The net result
is a response of a process with a dominant capacitance plus dead time. It is very
important to select the measuring device which adds a minimum lag to the process lag.

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 83

The exothermic reactor is perhaps the most difficult process to control due to its
instability and extreme nonlinear response. A control scheme for an exothermic
chemical reactor is shown in Figure (13). The degree of stability that can be achieved in
this temperature control loop depends on the rate at which the heat can be removed
from the reactor. The reactor can be stabilized if the reaction temperature changes fairly
slowly when compared to the rate at which the jacket temperature changes. The idea of
the control loop shown in Figure 13 is that once the feedstock and catalyst are added,
hot water in the jacket is used to initiate the reaction. As the reaction temperature
increases, the controller output decreases, closing the hot water valve, and opening the
cold water valve. Multiple water inputs to the jackets can be used to minimize dead time
and to change the jacket temperature as quickly as possible (to minimize the time
constant of the jacket). Typically a PID controller is used, but using a proportional only
controller may stabilize the reactor provided that the reactor is the dominant single
capacitance in the loop and there is no appreciable dead time.

A good example of an endothermic process is a heat exchanger being used to

heat a fluid from the inlet temperature Ti to an outlet temperature To. This heat
exchanger response will be that of single large dominant capacitance with dead time.
Typically either a PI or PID controller is used. Derivative action can be used since the
temperature measurement is not noisy.

Composition
Composition loops generally have characteristics similar to temperature loops,
but with several differences: (1) measurement noise is a more significant problem in
composition loops, and (2) time delays due to analyzers are significant factors. These
factors can limit the effectiveness of the derivative action. Due to their importance and
the difficulty to control, composition and temperature loops often are prime candidates
for the advanced control strategies discussed in the following lectures.

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 84

sp

TT TT

Cold water

Hot water

Figure (13) Control scheme for an exothermic reactor

References
Shinsky, F., Process Control Systems, McGraw Hill, New York, 1988.

Chemical Engineering Department

King Saud University, 2002
Process Control in the Chemical Industries 85

Luyben, M. and Luyben, W., Essentials of Process Control, McGraw Hill, New York,
1997.

Ogunnaike, B. and Ray, W., Process Dynamics, Modeling and Control, Oxford
University Press, UK, 1994.

Marlin, T., Process Control, McGraw Hill, New York, 1995.

Clair, D., Controller Tuning and Control Loop Performance, Straight-line Control Co.,
USA, 1993.

Svreck, W., Mahoney, D. and Young, B., A Real Time Approach to Process Control,
Wiley & sons, New York, 2000.

Astrom, K. and Hagglund, T., PID Controllers: Theory, Design and Tuning, ISA, NC,
USA, 1995.

Friedmann, P., and Stoltenberg, T., Continuous Process Control, ISA, NC, USA, 1996.

Murrill, P., Application Concept of Process Control, ISA, NC, USA, 1988.

Murrill, P., Fundamentals of Process Control Theory, ISA, NC, USA, 1991.

Luyben, W.,Process Modeling, Simulation and Control for Chemical Engineers,

McGraw Hill, New York, 1990.

Smith, C. and Corripio, A., Principles and Practice of Automatic Process Control,
Wiley & sons, New York, 1997.

Seborg, D., Edgar, T., and Mellichamp, D., Process Dynamics and Control, Wiley &
sons, New York, 1989.

Stephanopoulos, G., Chemical Process Control: An Introduction to Theory and

Practice, Prentice Hall, 1984.

Chemical Engineering Department

King Saud University, 2002