INDUSTRIAL INSTRUMETATION AND

CONTROL SYSTEMS

Lecture 14. Process control

Assoc. Prof. Dr. BUI Dang Thanh

School of Electronic & Electrical Engineering,
Hanoi University of Science and Technology
Dai Co Viet road, Ha Noi, Viet Nam

1/2/2025 Buidangthanh.3i@gmail.com 1
Presentation outline

❖ Introduction
❖ Control Modes
❖ Role of PID controller in process control
❖ Applications of cascade/feedforward controller in
process control
❖ Applications of inferential control
❖ Illustrate the applications of split-range control
❖ Control of MIMO processes in process control
applications

1/2/2025 Buidangthanh.3i@gmail.com 2
Introduction

• Control systems vary extensively in complexity and

industrial application.

• Ex: petrochemical industry, automotive industry, soda

processing industry

• The control loops can be very simple such as: controlling

water temperature or heating, ventilation, and air-
conditioning (HVAC)

Buidangthanh.3i@gmail.com
1/2/2025 3
Introduction
• To develop effective control solutions, it is essential for a process control
engineer to understand the underlying process.
• Understand the interconnections of the process, rank order priorities for the
objectives of the process, various constraints, and major disturbances
affecting the process is important.
• Control solutions design must effectively solve the real process problems and
function properly with regards to the entire plant.
• It is important to have a good understanding of the process and fundamental
characteristics of various control options.

Example of control loop tuning criterion Continuous-flow stirred tank reactor

1/2/2025 Buidangthanh.3i@gmail.com 4
Control Loop Hardware
Final Control Element

The final control element comprises of the control valve, I/P converter, and the
instrument air system. The signal from the control system like DCS is fed to the
I/P converter of the control valve, which in turn sends the pneumatic signal to the
actuator thereby forcing the valve to reach a certain movement/position as
desired by the operator or the control system.

1/2/2025 Buidangthanh.3i@gmail.com 5
Control Loop Hardware
Sensor Systems
A sensor system consists of a sensor, a transmitter, and the associated signal
processing treatment. Flow, pressure, temperature, level, and composition
sensors are most commonly used sensors in the chemical processing industry.
Most sensors in the chemical processing industry actually measure a quantity
that is directly related to the desired measurement.
The performance of a sensor is characterized by span, accuracy, and
repeatability:
• Span is the difference between the largest measurement value made by the
sensor/transmitter and the lowest measurement value.
• Accuracy is the difference between the value of the measured variable and its
true value.
• Repeatability is related to the difference between the sensor’s readings while
the process conditions remain constant, indicative of the noise on the sensor
reading.

1/2/2025 Buidangthanh.3i@gmail.com 6
Control Loop Hardware
Sensor Systems
❖ Flow Sensors: It is based on differential pressure measurement across an
orifice and may be called a differential pressure orifice meter.
❖ Pressure Sensors: A differential pressure cell uses a balance bar that is
deflected based on the pressure differential between two compartments that
are in contact with opposite sides of the balance bar. The balance bar is
maintained at a specific position using a precision forcing motor. The
measurement of the pressure is directly related to the force used by the
forcing motor to balance the bar.
❖ Temperature Sensors: The temperature of a process fluid is measured using
thermocouples (TCs) and resistance temperature devices (RTDs) by inserting
either the thermocouple or the RTD into a thermowell in thermal contact with
the process fluid. Thermocouples are more rugged and less expensive than
RTDs, but RTDs have a repeatability of ±0.1°C, while thermocouples have a
repeatability of ±1.0°C. The use of RTDs is recommended when an accurate
temperature measurement is required.

1/2/2025 Buidangthanh.3i@gmail.com 7
Control Loop Hardware

Sensor Systems
❖ Level Sensors: The most common type of level sensors measure the
hydrostatic head in a vessel using differential pressure. This approach typically
works well, as long as the difference between the density of the light and
heavy phases is large. Since the level sensor is based on pressure
measurement, it usually has relatively fast measurement dynamics.
❖ Composition Analyzers : The gas chromatograph (GC) is the most common
composition analyzer used in the process industries. It is extensively
employed for measuring the composition of light hydrocarbon streams. GCs
pass a vaporized sample of the stream to be analyzed through a small-
diameter tube. The tube contains a packed bed of various materials that
exhibit different tendencies to absorb each of the components in the sample.

1/2/2025 Buidangthanh.3i@gmail.com 8
Basic Terms

• The terms are redefined:

• Lag time is the time required for a control system to return
a measured variable to its set point after there is a change
in the measured variable.
• Dead time is the elapse time between the instant an error
occurs and when
• Dead-band is a set hysteresis between detection points of
the measured variable when it is going in a positive or a
negative direction.

Buidangthanh.3i@gmail.com
1/2/2025 9
Basic Terms

• The terms are redefined:

• Set point is the desired amplitude of an outpoint
variable from a process

• Error signal is the difference between a set reference

point and the amplitude of the measured variable.

• Transient is a temporary variation of a load parameter

after which the parameter returns to its nominal level

Buidangthanh.3i@gmail.com
1/2/2025 10
Basic Terms

• The terms are redefined:

• Measured variable is an output process variable that
must be held within given limits
• Controlled variable is an input variable to a process that
is varied by a valve to keep the output variable (measured
variable) within its set limits.
• Offset is the difference between the measured variable
and the set point after a new controlled variable level has
been reached

Buidangthanh.3i@gmail.com
1/2/2025 11
Control Modes

The two basic modes of process control:

❑ ON/OFF action
❑“Continuous control” action

Buidangthanh.3i@gmail.com
1/2/2025 12
ON/OFF action
• Simple ON/OFF action
• The simplest form of control in a closed loop system is ON/OFF
action
- When the variable is above the reference the system is turned ON
- When the variable is below the reference the system is turned OFF

Ex: a room heating system

- The top graph shows the room temperature
- The lower graph shows the actuator signal

Buidangthanh.3i@gmail.com
1/2/2025 13
ON/OFF action
• Differential ON/OFF action
• Differential or delayed ON/OFF action is a mode of operation where
the simple ON/OFF action has hysteresis or a dead-band built in

Ex: a room heating system

- The top graph shows the room temperature
- The lower graph shows the actuator signal

Buidangthanh.3i@gmail.com
1/2/2025 14
Continuous control action

The “continuous control” action has 4 types action:

❑ Proportional action:
❑ Derivative action:
❑ Integral action:
❑ PID action

Buidangthanh.3i@gmail.com
1/2/2025 15
Direct-acting and Reverse-acting Controllers

In some cases, the controller output is expected to increase with the increase in
the error from set point, while in some other cases the controller output must
decrease.

TWO TYPES OF CONTROLLER ACTION

Direct Action Reverse Action

CO increase => PV increase CO increase => PV decrease

CO decrease => PV decrease CO decrease => PV increase

• Set point - PV
• Control output - CO

1/2/2025 Buidangthanh.3i@gmail.com 16
PID Design

P-only Control

P-only control is used in applications that are slow and is acceptable to have a
degree of offset. Here the slow process means the processes that respond very
slowly for a change in the manipulated value.

PI Control

PI controllers, as discussed earlier, are most widely used. The controllers are
used for the control of the processes which are not sluggish and needs no offset.
The typical applications are flow control, level control, pressure control and
composition control, etc.

1/2/2025 Buidangthanh.3i@gmail.com 17
PID Design

PID Control
The PID control is applied in sluggish processes. The typical industrial sluggish
processes are temperature and composition control, etc. If we apply a PI control
to a sluggish process, there will be oscillations due to the lag in the process. In
order to reduce the oscillations, the derivative action/constant is added to the
control that acts counter to the oscillations and thereby allowing more of the
proportional action on the response. This helps in improving the control
performance and hence process becomes more stable. The key point here is to
know if the process is sluggish enough and is in need of a PID controller.

1/2/2025 Buidangthanh.3i@gmail.com 18
Analysis of Typical Control Loops

The five control loops commonly encountered in the chemical process industries
are flow control loop, level control loop, pressure control loop, temperature
control loop and composition control loop.
Flow Controllers
The dynamics of the process and sensor are quite fast compared with the
dynamics of the control valve. Since overall procedure is fast and calls for
accurate control to set point, a PI controller is the proper choice for most flow
control applications.
Level Controllers
The dynamics of the sensor are quite fast, and the dynamics of the actuator are
faster than the dynamics of the process.
Process dynamics typically control the response of the actuator/process/sensor
system.
In general, for the process of level filling is not a slow one, a P-only control is
enough if the offset is acceptable. In case if offset is not acceptable in
applications such as reactor level control, a PI control is used.

1/2/2025 Buidangthanh.3i@gmail.com 19
Analysis of Typical Control Loops

Pressure Controllers
The pressure sensors have the highest response times where the valves remain
slow compared to other elements in the process loop. The changes in the
pressure and associated control expected from the pressure control is high and in
such situations a P-only control is enough if the offset is acceptable. However, if
the offset is not acceptable a PI control is recommended

Composition Controllers
A composition controller may be used for the overhead product of a distillation
column that uses the reflux flow rate as the manipulated variable. In this case, the
changes in the actuator from the flow controller are fast, where the sensor inputs
are slow. If the process and analyzer delay make the actuator/process/sensor
system sluggish, then a PID controller may be preferred. Once again, the
guidelines presented in the previous section should be used to determine if the
process is sluggish enough to warrant using PID control.

1/2/2025 Buidangthanh.3i@gmail.com 20
Analysis of Typical Control Loops

Temperature Controllers
The temperature has both a fast responding and slow responding loop for control
in any process applications.
Since the heater is likely to behave as a sluggish process, a PID controller would
usually be the controller mode of choice in this example. Derivative action
becomes ineffective in the presence of excessive sensor noise. A PI controller is
preferable for a less sluggish process. The guidelines provided in the previous
section must be used to determine if the process is sufficiently sluggish to
warrant using PID control.

1/2/2025 Buidangthanh.3i@gmail.com 21
Continuous control action
• Proportional action
• The most common of all continuous industrial process control
action is proportional control action.
– The amplitude of the output variable from a process is measured and
converted to an electrical signal
– This signal is compared to a set reference point..
– Any difference in amplitude between the two (error signal) is amplified and
fed to a control valve (actuator) as a correction signal.

Water heater showing a feedback loop Effect of load changes on the temperature
for constant temperature output of the water from the water heater.

Buidangthanh.3i@gmail.com
1/2/2025 22
Continuous control action
• Derivative action
• Proportional plus derivative (PD) action was developed in an attempt to
reduce the correction time
• Derivative action senses the rate of change of the measured variable and
applies a correction signal that is proportional to the rate of change

Proportional and derivative action Effect of proportional and derivative action

variable change with resulting slope on a variable

Buidangthanh.3i@gmail.com
1/2/2025 23
Continuous control action
• Integral action
• Proportional plus Integral (PI) action also known as reset action, was
developed to correct for long-term loads and applies a correction
proportional to the area under the change in the variable curve

Proportional and integral action variable Effect of proportional and integral action
change with area under the graph on a variable

Buidangthanh.3i@gmail.com
1/2/2025 24
Continuous control action
• PID action
• A combination of all three of the actions described above is more
commonly referred to as PID action.

Waveforms for proportional plus integral action and

waveforms for proportional plus derivative and integral action
Buidangthanh.3i@gmail.com
1/2/2025 25
Implementation of Control Loops

• Implementation of the control loops can be achieved

using:
❑ Pneumatic,
❑ Analog electronics
❑ Digital electronics.

Buidangthanh.3i@gmail.com
1/2/2025 26
Implementation of Control Loops
• ON/OFF action pneumatic controller
- When the temperature in the furnace reaches its set point the sensor moves the
flapper toward the nozzle to stop the air flow and allow pressure to build up
- The operates an air control relay that shuts OFF the air flowing to the control
valve turning OFF the fuel to the furnace
- When the temperature in the furnace drops below a set level the flapper is
opened, reducing the air pressure, opens the air control valve allowing the air
pressure to drop and the control valve to open, turning ON the fuel to the furnace

Pneumatic ON/OFF furnace controller

Buidangthanh.3i@gmail.com
1/2/2025 27
Implementation of Control Loops
• ON/OFF action electrical controller
- As the temperature decreases the bimetallic element tilts the mercury switch down
causing the mercury to flow to the end of the glass envelope and in so doing shorts
the two contacts together. The contact closure operates a low voltage relay turning
ON the blower motor and the heating element
- When the room temperature rises to a predetermined set point the bimetallic strip
tilts the mercury switch back causing the mercury to flow away from the contacts.
The low voltage electrical circuit is turned OFF the relay opens, and the power to
the heater and the blower motor is disconnected.

Simple ON/OFF room heating controller

Buidangthanh.3i@gmail.com
1/2/2025 28
Implementation of Control Loops
• PID action pneumatic controller

Pneumatic PID controller.

Buidangthanh.3i@gmail.com
1/2/2025 29
Implementation of Control Loops
• PID action control circuits

Circuits used in PID action error generating circuit and proportional circuit.

Buidangthanh.3i@gmail.com
1/2/2025 30
Implementation of Control Loops
• PID action control circuits

Derivative amplifier (a) circuit and (b) waveforms

Buidangthanh.3i@gmail.com
1/2/2025 31
Implementation of Control Loops
• PID action control circuits

Integrating amplifier (a) circuit and (b) waveforms

Buidangthanh.3i@gmail.com
1/2/2025 32
Implementation of Control Loops
• PID electronic controller

Block schematic of a PID electronic controller

Buidangthanh.3i@gmail.com
1/2/2025 33
Implementation of Control Loops
• PID electronic controller

Circuit of a PID action electronic controller.

Buidangthanh.3i@gmail.com
1/2/2025 34
Implementation of Control Loops
• PID electronic controller

Circuit of a PID electronic controller with feed back from the actuator position

Buidangthanh.3i@gmail.com
1/2/2025 35
Implementation of Control Loops
• Digital Controllers

Computer based digital controlled process

Buidangthanh.3i@gmail.com
1/2/2025 36
Process and Control Diagrams
Process and Instrumentation Diagrams

Simplified P&ID of the section of a column

1/2/2025 Buidangthanh.3i@gmail.com 37
Control Loop Hardware

Schematic of a feedback controller Schematic of the hardware for a feedback

on a CST thermal mixing tank controller for the mixer

1/2/2025 Buidangthanh.3i@gmail.com 38
Characterizing Dynamic Behavior
First-order System
A first-order process is a self-regulating process. The other characteristic of a
first-order process is the steady-state gain of the process, Kp. The gain indicates
the change in the output of the process for a unit-input change.

Dynamic response of first-order process to

step input change
1/2/2025 Buidangthanh.3i@gmail.com 39
Characterizing Dynamic Behavior
Second-order Process
The dynamic response of a second-order process also known as “over-damped
response” involves an asymptotic approach to a new steady-state condition or an
oscillatory approach.

Dynamic response of second-order Key characteristics of an under damped

processes to a step input change second-order response

1/2/2025 Buidangthanh.3i@gmail.com 40
Application of Cascade/
Feedforward Control in
Process Control

1/2/2025 Buidangthanh.3i@gmail.com 41
Cascade Control
❖ Cascade control is used to effectively reject specific disturbances. Cascade
control uses two control loops that are connected together. The primary or
master controller determines the set point for the secondary or slave controller.
The secondary controller responds to disturbances more quickly than the
primary controller, thereby rendering the secondary controller more efficiency
than the primary controller in rejecting certain disturbances.
❖ The secondary control loop must be at least three times faster than the
primary loop for a cascade controller to be effective.
❖ The cascade arrangement’s effectiveness with respect to improving the
rejection of disturbances depends on how well the secondary control loop
rejects a specific disturbance.

1/2/2025 Buidangthanh.3i@gmail.com 42
Cascade Control
We should use Cascade control in cases:
• When the single registration ring does not meet the quality
• Easily measure and control the second variable
• There is a causal relationship between the control variable and the second
controlled variable
• The second variable acts on the joystick variable faster than the first one

Temperature controllers applied to a steam-heated exchanger

(a) without cascade control and (b) with cascade control

1/2/2025 Buidangthanh.3i@gmail.com 43
Feedforward Controller

Feedforward control is applicable to process control loops that are significantly

affected by online measurable disturbances. A feedback controller reacts to
deviations from set point caused by the disturbance until the process is returned
to set point.
In this control, the proportional and derivatives constants are kept zero at the
steady state. The integral constant in the PID drives the compensation for the
process upsets. The feedforward controller can generate a signal based on the
upset to compensate the effect of upset. In other words, the feedforward control
generates an output signal that otherwise might have been generated by the
feedback controller after the effect or upset.

1/2/2025 Buidangthanh.3i@gmail.com 44
Feedforward Controller

Steam boiler level control; (a) Feedback control; (b) Feedforward

control; (c) Combined feedforward and feedback control

1/2/2025 Buidangthanh.3i@gmail.com 45
Feedforward Controller

Generalized Feedforward Controller

A general formulation of a feedforward controller can be derived as:
−Gd (s)
Gff s =
Gp (s)
Where Gff s is the transfer function for the feed-forward controller, Gd (s) is the
transfer function for the effect of the measured disturbance on the controlled
variable, Gp (s) is the transfer function for the effect of the manipulated variable on
the controlled variable.
If Gd (s) and Gp (s) are described using FOPDT models, the feedforward controller
would be represented as a lead-lag element plus dead time with a gain, a lead, a
lag, and a dead time. Note that the gain of the feedforward controller is the
negative of the ratio of the disturbance gain and the process gain, the lead is the
process time constant, the lag is the disturbance time constant, and the dead
time is the dead time for the disturbance minus the process dead time.

1/2/2025 Buidangthanh.3i@gmail.com 46
Feedforward Controller
Tuning a Feedforward Controller
The steps for tuning a feedforward controller are as follows:
1. Estimate the lead-lag plus dead time parameters based on process knowledge.
2. Under open-loop conditions, adjust K ff while maintaining the rest of the tuning
parameters at their initial levels so as to minimize the steady-state deviation from set
point.
3. Adjust qff after analyzing the dynamic mismatch. The direction of the deviation causes
dynamic mismatch indicating whether the feedforward correction is applied too soon
or too late.
4. Finally, adjust either t ld or t lg until you attain approximately equal areas above
and below the set point.

Tuning feedforward controller

1/2/2025 Buidangthanh.3i@gmail.com 47
Feedforward Controller

Combined Feedforward and Feedback Control

❖ A feedback-only controller efficiently absorbs disturbances in processes such
that, the advantage provided by feedforward control for fast-responding
processes is insignificant.
❖ The controller starts acting on the system after the error is seen, but by the
time the corrective action has the effect on the process, the process is already
upset. If there is a feedforward controller added to the same process with
more dead time, it reduces the deviation from the controlled variable from set
point leading to smaller disturbances to the process.
❖ Feedforward provides a better feedback control performance when the
performance compensates for a major disturbance to the process. Typically,
feedforward control is useful when:
• Feedback control by itself is not satisfactory, i.e. for a process having
substantial dead time or for a process having significant dead time.
• The major disturbance to a process is measured on line.

1/2/2025 Buidangthanh.3i@gmail.com 48
Feedforward Controller
Feedback Control
Advantages Disadvantages
Takes action only after the disturbance has
Disturbance measurement not required.
affected the process.
Can effectively reject disturbances for a fast Susceptible to disturbances if the process is
responding process. slow or in the presence of significant dead time
Can make the closed-loop system unstable
Simple to implement.
because of nonlinearity
Feedforward Control
Advantages Disadvantages
Compensates for disturbances before they
Requires the measurement of the disturbance.
affect the process.
Can improve the reliability of the feedback
Does not compensate for unmeasured
controller by reducing the deviation from set
disturbances.
point.
Offers noticeable advantages for slow
Since it is a linear-based correction, its
processes or processes that have significant
performance deteriorates with nonlinearity
dead time.

1/2/2025 Buidangthanh.3i@gmail.com 49
Feedforward Controller
Ratio Control
Ratio control is another way of feedforward control where two loads are
measured, and the ratio is maintained at constant rate. In general, this will be
with respect to different flow rates, where the flow rate ratio is maintained at a
constant rate, commonly one of the streams is not controllable and referred as
wild stream.

Schematic diagram of ratio control Schematic of ratio control applied

for feed rate changes in a for pH control of an acid
distillation column neutralization process

1/2/2025 Buidangthanh.3i@gmail.com 50
Check point

1. What are the limitations of a PID Controller?

2. Name the controllers that form the cascade control.

3. Which controller in the cascade form responds fast for disturbances?

1/2/2025 Buidangthanh.3i@gmail.com 51
 Applications of
Inferential Control

1/2/2025 Buidangthanh.3i@gmail.com 52
Inferential Control
❖ Generally, the widespread assumption is that sensor measures the process
variable by direct means. The above is true, but there are various methods in
which the measurement is made for some other variable and the actual
measurement is inferred.
❖ Parameters like pressures, temperatures, and flow rates that can be readily
measured can be effectively used to infer difficult to measure quantities, such as
molecular weight, compositions, and extent of reaction. In all these cases, the
inferred value is treated as actual process measurement and is used in the
controller to generate the output.
❖ Following are some of the needs in the industry where the inferential
measurement is needed as controlled variable:
• Inferential measurements are the industrial chosen method for counteracting large
measurement delays for controlled variables.
• Inferential measurements can greatly reduce the measurement dead time
because they use measurements (e.g. temperatures, pressures, and flows) that
have relatively low levels of measurement dead time.
• The installation and maintenance cost of an online analyzer can be expensive.
Whereas the installation and maintenance cost of an inferential measurement is
less expensive as they are based on pressure, temperature, and flow
measurements.

1/2/2025 Buidangthanh.3i@gmail.com 53
Inferential Control
Inferential Temperature Control for Distillation
❖ In distillation columns, the tray temperatures at different levels can be used to
infer the quality and compositions of the liquid inside. Some experts can tell
the quality by measuring the temperature alone at different heights. Therefore,
inferential control is widely used for distillation product composition control.

Schematic for inferential control of the bottom

product composition control of a distillation column
1/2/2025 Buidangthanh.3i@gmail.com 54
Inferential Control

Inferential Temperature Control for Distillation

❖ In multicomponent columns, the correct tray temperature for a specified

product composition changes as the feed composition to the column changes.
In these situations, a composition controller can be cascaded to the tray
temperature controller. Alternatively, the composition controller can be
administered by an operator by adjusting the set point for the tray temperature
controller as laboratory analysis results on the product become known.

1/2/2025 Buidangthanh.3i@gmail.com 55
Check point

1. What is inferential measurement?

2. List two examples of inferential control.

1/2/2025 Buidangthanh.3i@gmail.com 56
Illustrate the applications
of Split-range Control

1/2/2025 Buidangthanh.3i@gmail.com 57
Split-range Control

This control is used when a single output variable is to be produced from several
manipulated variables. This control ensures that a proper coordination among
various manipulated variables is present.

Split-range control

1/2/2025 Buidangthanh.3i@gmail.com 58
Override Control
The strategy of this control involves a selection between two or more controller
output signals, while a single controller exerts control over a process at any given
time. All other “de-selected” controllers are thus overridden by the selected
controller.
Most of the applications of override control are seen with variable speed motors
and drives as the final control element than the traditional control valves. In the
low-flow conditions, by reducing pump speed a lot of energy can be saved, over a
period of time.

Override control

1/2/2025 Buidangthanh.3i@gmail.com 59
Override Control

Running the pump “dry” for a long period will result in damaging the pump. This
may render the pump unusable, right at the time when the customer needs it
most. One of the solutions to this problem is installing a level switch in the well.

Override control with switch protection

1/2/2025 Buidangthanh.3i@gmail.com 60
Override Control
One way to address this issue would be to allow the pump reduce the water
output when water levels are low. In this mode of control, the drive/pump
mechanism keeps driving the pump with minimum amount of pressure
maintained and at the same time reducing the demand on the well along with
maintaining water needs, while protecting the system from damage. This is a soft
override control achieved through control systems such as DCS.

Override control with soft control

1/2/2025 Buidangthanh.3i@gmail.com 61
Override Control
The level control signal overrides the pressure control signal to protect the
motor/pump for their longevity. The second level protection provided is also
considered where the level switch installed in the lower side of the tank shuts off
the pump. These two layers of protection, one is a soft means by the controllers
and other by the hard means through a switch is a common practice for all the
critical controls. The soft control tends to be moderate in action and tries to
throttle, where the hard control will aggressively shutoff.

Override control with control strategy and switch

1/2/2025 Buidangthanh.3i@gmail.com 62
Scheduled Tuning

❖ Nonlinear process behavior injects instability to a controller in certain

situations and renders it extremely sluggish in others.
❖ A PID controller tuned with high gain can reduce the oscillations but introduces
delay in performance. The magnitude of the disturbances in combination with
the inherent process nonlinearity determines the degree of observed process
nonlinearity. Certain measurements, for many processes, directly indicate
whether the process parameters have increased or decreased and by how
much. Therefore, an effective way to compensate for process nonlinearity is by
scheduling the controller tuning based on process measurements. The feed
rate and the controlled variable are examples of key process measurements.
These are used, typically, to schedule the controller tuning.

1/2/2025 Buidangthanh.3i@gmail.com 63
Constraint Control
❖ Process control systems operate with constraints from the process. The process
always tends to move towards better yield and naturally tends to reach the limits. It
becomes necessary to apply different control loops from those previously used
once the higher and lower limits of the process variable in reached.
❖ It is important to install safeguards while implementing an industrial controller
effectively, while operating safely, economically and environmentally friendly using
the control strategies such as override control, etc.

Schematic of a furnace fired heater Schematic of the stripping section of a distillation column with
with low select firing controls low select controls applied to prevent flooding of the column

1/2/2025 Buidangthanh.3i@gmail.com 64
Check point

1. What is split range control?

2. What is override control?

1/2/2025 Buidangthanh.3i@gmail.com 65
Control of MIMO processes in
Process Control applications

1/2/2025 Buidangthanh.3i@gmail.com 66
Selecting Control Configuration

❖ The process of pairing the manipulated and controlled variables consists of

selecting a control configuration. The selection of control configuration is
crucial for PID controllers, where each controlled variable must be paired with
a manipulated variable and control performance depends on control
configuration.
❖ It is essential to identify the manipulated variables and the controlled variables
for multivariable controllers also, which do not have to pair manipulated and
controlled variables due to the MIMO nature of multivariable controllers.
❖ The closed-loop performance of a decentralized control configuration is
affected by the following three factors:
• The coupling or interaction between SISO control loops
• The dynamic response of a controlled variable to a change in its manipulated
variables
• The sensitivity of disturbances.

1/2/2025 Buidangthanh.3i@gmail.com 67
Selecting Control Configuration
❖ Select a configuration such that each of the manipulated variables has a fast and
consistent dynamic effect on its controlled variable. It is recommended to avoid pairings
that result in a sluggish response or inverse action.
❖ Each control configuration has its own sensitivity to disturbances. A control configuration
with less sensitivity to a disturbance requires less feedback correction during regulatory
control than another configuration that is more sensitive to the disturbance. Therefore,
the sensitivity of a control configuration to the key disturbances is a major consideration
when selecting configuration.

Schematic of a two-input/two-output process with single-

loop controllers showing coupling between the control loops

1/2/2025 Buidangthanh.3i@gmail.com 68
Tuning Diagonal PID Controllers
❖ The tuning procedures discussed earlier for the single PID controllers can also be used
as a subset for decentralized control of MIMO. At first an ATV test is performed for each
loop in such a way that while one is tested, the other is maintained in open-loop
condition. The second step is to know if any of the loop responds faster than the other.
Compare the values of the ultimate periods, Pu, obtained in the ATV tests to achieve
this. If the smallest value of Pu is five times or more small than the next largest Pu,
implemented that loop first by itself before tuning the other loops. It can be tuned as a
single PID loop.
❖ The results from the ATV test are used for calculating the gain and reset time using
techniques such as Zeigler-Nichols methods, etc. After this the single tuning factor FT is
used for both control loops.

1/2/2025 Buidangthanh.3i@gmail.com 69
Model Predictive Control

❖ The most widely used form of multivariable control is the model predictive
control (MPC). There are more than five thousand industrial MPC applications
worldwide, and the number is rapidly growing. Dynamic matrix control (DMC)
is the most popular form of MPC and is based on having a step-response
model for each input-output pair of the process. A step response model is a
series of terms that represent the deviation response of the process to a unit
step input change.
❖ When MPC determines control action, it also calculates the future behavior of
the process. Even though it calculates a number of manipulated variable
moves into the future, only the first move is implemented. In the next control
cycle, another set of future moves is calculated, and, once again, only the first
move is implemented. Because of this methodology, MPC is known as a
moving horizon controller

1/2/2025 Buidangthanh.3i@gmail.com 70
Model Predictive Control

❖ The widespread industrial use of DMC and MPC, in general, is the result of
their ability to operate processes more profitably. The industry use of DMC
would have been drastically lower if DMC provided only reduced variability
operation. DMC improves the profitability of processes because it can operate
processes for more highly valued products at higher production rates.
❖ A linear program (LP) assesses the economics of the process and constraints
and specifies to the DMC controller the constraints against which it should
control the process. An LP determines the optimum values of the decision
variables for a linear economic objective function, subject to a set of linear
constraints.

1/2/2025 Buidangthanh.3i@gmail.com 71
Check point

1. What is MIMO?

2. List three factors on which the decentralized control configuration depends.

3. How coupling can be analyzed from steady state point of view.

4. What is MPC?

1/2/2025 Buidangthanh.3i@gmail.com 72
Thank you for your attention!

1/2/2025 Buidangthanh.3i@gmail.com 73