B49CF2

Process Modelling
Process Control

Topic 1

Elements for Process Control

© Heriot-Watt University, B49CF2, Process

August 2015 v1 1
Modelling
 Learning Objectives
After studying this “Elements for Process Control” topic you
should be able to:
• Review and outline the need for Process Control, as well as
the need to model, not only individual items of equipment,
but also overall processes.
• List the individual elements needed to implement a control
loop.
• Differentiate between different control strategies.
• Review measuring devices, final control elements,
controllers and discuss their functionality within the
framework of a control loop.
• Differentiate how control loops can be represented using
different types of diagram.

© Heriot-Watt University, B49CF2, Process

August 2015 v1 2
Modelling
 1.1 Introduction
Controlling chemical and physical processes is a topic of key
importance for process engineers.
• The subject underpins not only plant safety and cost, but
also product yield and quality as well as environmental
considerations.
• Without an effective control system, no plant would be
licensed to operate in any modern state anywhere in the
world.
• An effective control scheme depends on understanding the
dynamic behaviour and physical characteristics of a process.
• Thus, being able to model a process is an essential
requirement of Process Control.
This module aims to demonstrate how mathematical “models”
can be developed to model the physical behaviour of processes;
once a process can be modelled, it can be controlled.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 3
Modelling
 The syllabus for Process Modelling and Control is as follows:
• Introduction to Process Control.
• Elements for Process Control: transmitters & sensors;
controllers; control valves.
• Control configurations.
• Piping & Instrumentation Diagrams (P&IDs). >TOPIC 1
• Laplace transforms and their use in solving ODEs. >TOPIC 2
• Process modelling and the use of ODEs to represent
chemical and physical processes.
• Definition of transfer functions . Review of typical first and
second order responses. >TOPIC 3
• Non-interacting and interacting dynamic systems.
• Block diagrams and block diagram algebra. >TOPIC 4
• Feedback control.
• P, P-I and PID control.
• Introduction to tuning. >TOPIC 5
© Heriot-Watt University, B49CF2, Process
August 2015 v1 4
Modelling
Recommended Textbooks
C.A. Smith, A. Corripio, Principles and Practice of Automatic
Process Control, Wiley, 3rd edition, 2006.

D.E. Seborg, T.F. Edgar, D.A. Mellichamp, F.J. Doyle, Process

Dynamics and Control, Wiley, 3rd edition, 2011.

G. Stephanopoulos , Chemical Process Control, An Introduction

to Theory and Practice, Prentice Hall, Englewood Cliffs, New
Jersey, 1984.

© Heriot-Watt University, B49CF2, Process

August 2015 v1 5
Modelling
 1.2 Introduction to Process Control
A chemical plant must simultaneously satisfy a number of
requirements including:
1. Safe operation.
2. Adherence to production specifications.
3. Adherence to environmental regulations.
4. Operation within specified constraints.
5. Operation at minimum cost.
To do this, Process Control is used to:
1. Ensure the stability of the process.
2. Minimise the effect of disturbances and track setpoint
changes.
3. Optimise the performance of the process.
Because of their dynamic nature, automatic process control is
normally required – manual control is rarely adequate.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 6
Modelling
 1.2.1 Process Variables
The variables that are associated with a process can be divided
into two groups:
1. Input variables: these are variables that affect the process.
2. Output variables: these are variables that are affected by
the process.
Processes may be classified as either SISO or MIMO systems:

Single-Input-Single-Output
(SISO) systems have only one
input/output pair.

Multiple-Input-Multiple- Output
(MIMO) systems have many
input/output pairs.

© Heriot-Watt University, B49CF2, Process

August 2015 v1 7
Modelling
Variables can also be divided into controlled variables,
disturbance variables and manipulated variables.
S.P.
TIC
1
The setpoint “SP” is the
Steam RA
desired value of the CV.
ATO
TT
The Controlled Variable “CV”
1
is the variable whose value
HX - 114 has to be controlled at a
T (t )
particular value.
T
The Manipulated Variable
“MV” is an input variable Identifying CV, MV, inputs and
whose value is changed to disturbances is a key step in the
keep the CV at the SP. design of a reliable control system.

Disturbances variables “DV” are input variables that fluctuate and cause the
CV to move away from its SP value – for instance, inlet temperature or flow
entering the heat exchanger.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 8
Modelling
 1.2.2 System Architecture
Typical control system architecture is shown below:
Complex/Extensive Boundaries/High Dependency/Low Integrity

Simple/Narrow Boundaries/Low Dependency/High Integrity

© Heriot-Watt University, B49CF2, Process

August 2015 v1 9
Modelling
 The term “regulatory control” is used here to cover the lowest
level of control. These systems are typically used to ensure
that the CV stays at or near its SP value.
• Regulatory control systems should be simple with minimal
instrument dependency. Each system must have a clear
function and a well-defined failure mode.
• Safety systems should take over from regulatory systems in
the event of an emergency. It must be possible to do a
HAZOP analysis of the control system.
Above the basic regulatory control there is “supervisory
control”, where up-to-date information about the process is
displayed at operator stations.
• The operators can then react to this current information by
changing setpoints or opening/closing valves.
• More complex “enhanced regulatory control” can also be
implemented at this level.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 10
Modelling
 The term “advanced control” is used to describe more complex
multi-variable control systems which are used to optimise a
system so that it runs most efficiently:
• Advanced control systems are normally not critical to plant
operation and do not normally have to be part of the HAZOP
process.
• If any parts are critical to the operation and safety of the
process, then basic regulatory control is used.
• Advanced control systems deal with larger systems with
many interacting variables. Also, these systems often use
some kind of modelling, for example distillation columns.
• The term “advanced control” is also used to describe control
systems which rely on analysers, such as HPLCs. These
instruments are more complex but less reliable.

© Heriot-Watt University, B49CF2, Process

August 2015 v1 11
Modelling
 The term “regulatory control” refers generally to control
systems designed to keep the CV at its required SP value. The
two basic requirements of a regulatory system are:
1. To track changes in SP as inputted by operators in
conjunction with Standard Operating Procedures (SOP).
2. To compensated for process input variable fluctuations
which could cause the CV to move away from the SP value;
these are called “disturbances” or “load changes”.
The term “servo-control” refers to systems in which the
setpoint changes as a function of time and the control system
is then set-up to optimally reach the latest setpoint.
• In chemical industries regulatory control (such as level,
pressure, temperature & flow control) is more common.
• Servocontrol is more associated with purely mechanical
systems, such as control of robotic welding equipment.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 12
Modelling
 The term “business-wide optimisation” is used to describe
complex business models which are used to optimise the
operation of the whole plant or several sections plants.
• The objective of these optimisation models is to run the
plant (or plants) in the most economical way possible.
• This may be achieved by inputting key plant data into the
model and then altering production rates, determining
feed rates, changing recycle rates, etc.
• The idea might be to alter the plant operation in such a
way that (say) total production costs are minimised.

In this part of the course only basic regulatory control systems

will be discussed. When a control system is mentioned, it may
be taken to be a regulatory control system.

© Heriot-Watt University, B49CF2, Process

August 2015 v1 13
Modelling
 1.3 Basic Regulatory Control Structure
The basic components of a regulatory control system are as
follows – consider the heat exchanger example again:
S.P.
TIC

2. The Controller
1
RA
Steam
(effectively its “brains”).
ATO
TT
1 1. The Measurement
HX - 114 system (the sensor-
T (t )
transmitter).
3. The Final Control
T
Element (valve, pump,
heating element, etc.)
The roles of the above components, which must occur in the
same order are as follows:
1. Measurement (M).
2. Decision (D).
3. Action (A)
© Heriot-Watt University, B49CF2, Process
August 2015 v1 14
Modelling
 It is imperative in a control system that these three operations
“M”, “D” and “A” are in this order:
• The measurement “M” must lead to a decision “D”.
• This decision “D” must then result in some action “A” being
taken.
Even then, there are many ways in which a control engineer
can set up such a “closed-loop” system. Typically these include:
• Feedback control.
• Feedforward control.
• Cascade control.

..............but , there are actually many more possibilities !

© Heriot-Watt University, B49CF2, Process

August 2015 v1 15
Modelling
 No Control:
Cold liquid is flowing into a stirred tank which is fitted with a
steam heating coil………hot liquid is flowing out:
• Notice there is a measurement “M”, but there is no decision
element “D” and no action element “A”.
No Control ……Thus, no
Liquid …… Temperature
flow in
Here control……..Here
Steam …… Liquid flow out.
There is an
Or, Here indicator but no
transmitter

>> With no transmitter, no controller and no control valve.

>>>> The temperature leaving the tank will fluctuate up and
down in response to load changes…….
© Heriot-Watt University, B49CF2, Process
August 2015 v1 16
Modelling
 Manual Control:
There is now a sensor (the manual measurement element
“M”), an operator (the manual decision element “D”) and a
valve (the manual action element “A”):
Manual Control
Liquid
flow in Temperature
sensor

Liquid
Steam flow out
Valve

• The flow of steam can now be controlled by opening or

closing a manual valve located at the steam coil inlet.
>>> Open the valve a little & temperature goes up; close
the valve a little and temperature goes down.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 17
Modelling
 By adjusting the manual steam valve an operator could bring
the outlet CV to its SP value.
• At this point the “error” (CV-SP) would be zero.
• If the inlet liquid temperature to the heated tank were to
decrease (a disturbance or load change), then (CV-SP)<0
and an error would emerge.
• The operator would then need to open the manual valve
(by an amount that depends on the size of the error), until
the error is reduced back to zero, i.e. (CV-SP) 0 .
• It is clear that such an system requires “manual control”
which would be very expensive in terms of labour cost.
>>> It would also be impossible on a large plant where many
hundreds of CV need to be kept at their respective SP value; in
addition, many of these variables can interact.

© Heriot-Watt University, B49CF2, Process

August 2015 v1 18
Modelling
 Feedback Control
Whenever the SP is changed and/or a disturbance occurs
some automatic method is needed to bring the error to zero.
• The measured CV
Feedback Control
signal is “fed back” to
setpoint
the controller where
Temperature
Feedback
Controller
Liquid
flow in sensor
an error is generated.
Steam
Liquid • The controller then
Control
Valve
flow out generates an output
that automatically
adjusts the valve
position.

With feedback control the measurement is made at the output

and the difference “CV-SP” (the error) is used to change the
steam flow so that (CV-SP) 0.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 19
Modelling
 In summary this is what happens in the temperature control
feedback loop if a increase in process fluid inlet temperature
(a disturbance) occurs:
• The outlet fluid temperature (the CV) will increase – this is
the “M” step.
• The CV rises above the SP value and an error (positive error)
is generated. This is the first part of the “D” step where the
controller decides to decrease the steam flow.
• In the second part of the “D” step the controller decides by
how much the steam flow should be decreased. This
depends on controller set-up and the size of the error.
• The final “A” step is the action taken by the valve to reduce
the steam flow to heating coil – this brings (CV-SP) 0
>>> The controller action continues until the error returns to
zero (or very close to zero). Notice the controller works in
much the same way as a manual operator.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 20
Modelling
 Thus, the Measurement, Decision and Action sequence are in
a closed-loop and in this sequence.
• Also, once there is zero error the controller output remains
constant and the control valve position does not change
(until that is some new error is detected).
• When the error is zero the controller is doing nothing (it is
“idling”); it is taking NO action.
Therefore, for a feedback control system to take action, it is
necessary for the OUTPUT to be disturbed from its SP value.
• Feedback control systems work by measuring a change in
the OUTPUT and generating an error.
• Once the error has been generated the controller makes a
decision “D” which leads to an increase or a decrease in the
MV (this is the action “A”).
>>> It is the error that actually “drives” the control valve to
some new position and this new value of MV brings CV SP.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 21
Modelling
 Feedforward Control
In this configuration any disturbance is measured (see arrows)
and the controller is then programmed to compensate for it.
Feedforward control
Flow Temperature
measurement measurement • Notice INPUTS to
Liquid the model are
flow in
being measured.
Feedforward
Controller • Decision “D” and
Steam
Liquid action “A” are
Control
setpoint
Valve
flow out
based on INPUT
measurements.

The feedforward controller (a model of the process) then

decides whether to open/close the valve and by how much.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 22
Modelling
 By measuring the temperature directly at the inlet, the amount
of steam going to the heated tank can be increased or
decreased BEFORE an error in CV has time to emerge:
• This strategy, when properly implemented, could (in
theory) be used to prevent any major change in the CV.
With a feedback scheme, the steam flow rate is changed only
AFTER a change in CV has been measured (remember, it is the
error drives the valve to the new position):
• For a large heated tank system, there could be long time
delay between the disturbance and a change in outlet CV.
In summary, feedforward schemes act on measured INPUT
changes as they enter the system, while feedback schemes act
on measured OUTPUT changes leaving the system.
>>> Feedforward control is proactive, feedback control is
reactive. In practice if feedforward is needed, it is used in
conjunction with feedback control.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 23
Modelling
 Cascade Control
Consider what happens to the same feedback system when
steam supply pressure fluctuations occur (a common load
change in most plants):
Feedback Control
SP Liquid
flow in
Temperature
Feedback sensor …….leads to an upset
Controller
here.
Steam
Liquid
A load change Control
Valve
flow out

here……….

>>> If the steam pressure decreases (maybe another heater is

brought online elsewhere in the plant), then the steam flow to
the coil will drop......and the outlet temperature will drop.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 24
Modelling
 In order to restore the flow to it previous value, the valve
needs to open a little more (to compensate for the drop is
steam supply pressure):
• For feedback control nothing happens until the tank outlet
temperature, the CV, goes below its SP value (up until then
the controller is “idling”).
• Once a change is detected an error is generated at the
controller and this will eventually drive the valve to correct
position (it needs to open it a bit more).
• After, the loop settles down the steam supply pressure will
be lower, the valve will be open a little more, the steam
flow will be restored and the CV is at its old SP value.
>>> Notice, however, the tank temperature must first be upset
before the error can be generated and, remember, the prime
objective is to keep this temperature constant!!
© Heriot-Watt University, B49CF2, Process
August 2015 v1 25
Modelling
 Cascade Control
There are now two feedback loops as shown below:
Cascade control • The temperature
loop is called the
Temperature SP
controller primary or master
Flow
SP in Temperature loop.
Flow sensor • The flow loop is
controller
called the secondary
Steam
Flow or slave loop.
Flow out
sensor
Valve • The output of the
primary loop is
“cascaded to the SP
of the secondary
loop.
>> Once again the steam pressure decreases, the steam flow to
the coil will drop and the valve needs to open a little more, but
there are now two places to generate an error signal……..
© Heriot-Watt University, B49CF2, Process
August 2015 v1 26
Modelling
 As shown, there are now two places to generate an error:
1. The inner flow control loop, the setpoint comes from the
outer temperature controller (this is secondary CV).
2. The outer temperature control loop (primary CV).
If the steam supply pressure drops, then the decreased flow is
picked up (in seconds) by the steam flow sensor.
• An error is generated by the inner flow controller, the error
then opens the steam valve a little more and this quickly
returns the steam flow to its original SP value.
• Thus, after the disturbance is eliminated, the steam
pressure will be lower, the valve will be open a little more
and the steam flow should be at its original SP value.
>>> This all happens quickly and means that the disturbance
will have been corrected before it can significantly affect the
outlet tank temperature.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 27
Modelling
 In the event of disturbances to the inlet stream, the outer
temperature loop will return the primary CV (outlet
temperature) back to its SP value.
• Thus, the outer loop (slower temperature loop) corrects for
primary load changes associated with the inlet stream.
• And the inner loop (faster flow loop) corrects for secondary
load changes associated only with the steam supply
pressure.
>> For cascade control to have any chance of working the
inner loop must detect and correct errors MUCH FASTER than
the outer loop.
>>> Many (although not all) cascade systems therefore have
flow (fast) as the inner loop and temperature (slow) as the
outer loop.

© Heriot-Watt University, B49CF2, Process

August 2015 v1 28
Modelling
 1.4 Elements of a Feedback Loop
The elements of a control system are as follows:
• The measuring elements “M” .
• The decision elements “D” .
• The action elements “A” (very often, although not always,
these are valves).

1.4.1 Sensor-Transmitters; the “M” element

A sensor is a device capable of detecting and measuring
certain variables, such as (but not restricted to) level,
pressure, temperature, flow and composition.
• The transmitter then turns this measurement into a signal
that can be easily transmitted from one location to another.
• This can be an analogue signal (pneumatic, voltage,
hydraulic, current signal, etc.).........or a digital signal.

© Heriot-Watt University, B49CF2, Process

August 2015 v1 29
Modelling
 The sensor-transmitter unit produces a signal that provides
information about the value of the CV (also called the process
variable “PV”). The transmitter should have following features:
• It should have a fast response time. However, some
measurements experience significant time delays between
same and measurement, e.g. Gas Chromatograph.
• The measuring technique should be sensitive.
• The measuring technique should also be accurate and
precise.
• Preferably this relationship should be linear; i.e. the signal
should be proportional to the value of the CV.
• For non-linear devices, it may be possible to linearise the
signal……for an orifice plate this is known as a “square root
extractor”.
>>> Common sensor/transmitters are level, pressure,
temperature, flow and composition….but, there are others !!
© Heriot-Watt University, B49CF2, Process
August 2015 v1 30
Modelling
Typical measuring devices for Process Control
• The range is set by the
lowest and highest values of
the CV it can measure.
• The span is the difference
between the highest and
lowest CV it can measure.
• The lowest CV is often
called the zero (notice it
doesn’t have to have a value
of zero).
• Each value of a PV has an
output signal value. Thus,
the output signal also has a
range, span and zero.

>> Note, the terms Controlled Variable “CV”, Process Variable

“PV” and Controlled Output “CO” are all used interchangeably.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 31
Modelling
 For example, a pressure sensor-transmitter system might have
a CV range of 100 to 300 psig (the span is 200 psig, the zero is
100 psig); the output signal range is 4 to 20 mA (DC current).
• The transmitter could then be calibrated to give a linear
change in output signal over the entire input range

20
Transmitter • Notice the “live zero” of 4 mA.
Output Signal
(mA) • This feature helps to identify
12
transmitter faults.
• The output range is 4 mA to 20
4
mA, the output span is 16 mA
100 200 300 and the output zero is 4 mA.
Pressure
(psig)

>>> If the input-output calibration is non-linear it should be

linearised in some way.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 32
Modelling
 The slope of this line is known as the “gain” of the sensor –
transmitter KT :
(20  4)mA 16mA mA
KT   0.08
(300  100)psig 200psig psig

• The gain may also be expressed in terms of “percent of

transmitter output” (%TO) as follows:
(100  0)%TO 100%TO %TO This is a better
KT   0.5
(300  100)psig 200psig psig approach

© Heriot-Watt University, B49CF2, Process

August 2015 v1 33
Modelling
 1.4.2 Tutorial Questions - Transmitters
1. A level sensor is linear and has an input range of 0 – 0.5 m
and gives an output of 0 – 10 V. What is the overall gain KT ?
a) In SI units
b) In percent transmitter output?

2. The pH of a solution is defined as –log [H3O+]. It is usually

measured using potentiometry (by measuring the voltage
drop across a thin pH sensitive glass membrane). A change
in 50mV corresponds to a change of 1 pH unit.
This potentiometric pH probe, according to the
manufacturer’s instructions, can measure pH values
between 1 and 13.

© Heriot-Watt University, B49CF2, Process

August 2015 v1 34
Modelling
 During calibration it is found that at pH 6 the voltage across
the probe is 100 mV, while at pH 4 the voltage across the
probe is 0 mV.
a) Determine the spans of the sensor and the transmitter
and the gain of the probe (don’t forget to specify the
units!).
b) A step changes occurs in the pH from 5.5 to 6.5. What
are the corresponding voltages?

3. A temperature sensor has a range of -50 to 150 oC. The

temperature controller (full range 0-100%) is set at 70%.
What is the temperature?

© Heriot-Watt University, B49CF2, Process

August 2015 v1 35
Modelling
 4. A pressure sensor is used to measure the level of fluid in a
tank filled with oil. The specific gravity of the oil
is G oil  oil /  water 0.85
a) If the measured pressure is 1.2 barg, what is the oil
level in the tank?
b) What would the level be if the fluid in the tank were
water?
5. The absorbance of light by a liquid is often used to measure
concentrations of (coloured or UV absorbent) compounds.
Absorbance is easily measured by measuring the intensity
of light passing through a liquid (the light intensity
decreases when an absorbent compound is present).
The intensity of the light when it entered the sample is I in ,
it then decreases when it has passed through the sample to
I out .
© Heriot-Watt University, B49CF2, Process
August 2015 v1 36
Modelling
 The Transmittance of a sample is the ratio of the intensity of
the light that has passed through the sample to the intensity
of the light entering the sample. Thus, Transmittance = I out / I in
Transmittance can be converted to absorbance A using
the equation:
A  log(%Transmitt ance/100)
The absorbance A follows the Lambert-Beer law:
A  bc
Where  is proportionality constant, b is the pathlength
of the sample, and c is the concentration of the
compound in the solution.
The measurement of the absorbance of a compound X
dissolved in methanol gave the following results:
© Heriot-Watt University, B49CF2, Process
August 2015 v1 37
Modelling
 The measurement of the absorbance of a compound X
dissolved in methanol gave the following results:

a) Plot the absorbance as a function of the concentration of

compound X, and determine the value of  . Make sure
to use only the linear range!

© Heriot-Watt University, B49CF2, Process

August 2015 v1 38
Modelling
 b) A sensor is constructed from the same materials as
before with a pathlength of 5 cm for use in a reactor.
The sensor gives the following results:

Estimate the unknown concentration.

© Heriot-Watt University, B49CF2, Process

August 2015 v1 39
Modelling
 1.4.3 Final Control Element – The “A” Element
The final control element takes the signal from the controller
and converts into an “action”:
• The purpose of the action is to change the value of some
Manipulated Variable “MV”.
• Very often the MV is the flow of some fluid. If the MV is
flow, then the final control element will be a control valve.
However, the MV need not always be flow. Say that the stirred
tank heater had been fitted with an electrical heater (rather
than a steam coil):
• The MV would now be current passing through the heating
element.
• The final control element would now be a heating element
(and not a control valve), this is because the MV is current
and not flow.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 40
Modelling
 Control Valves
However, very often the final control element in chemical
engineering is a control valve. Valves are used to isolate,
switch, and control fluid flow in piping systems.
• They also play an important role in plant safety.
• Valves may be operated manually (using levers or gear
operators) or remotely (using either electric, pneumatic, or
hydraulically powered actuators).
• Manual valves are only used if the valve is operated
infrequently or if no power source is available.
>>> This discussion here will be limited to control valves. Most
of these valves are operated pneumatically – that is using
compressed air as the power source.

© Heriot-Watt University, B49CF2, Process

August 2015 v1 41
Modelling
 Types of Valve
Control valves can be divided into valves with rotating stems
and valves with reciprocating stems, see below:

>>> The different symbols used for these valves are given in
the hand-out – see Appendix B for further details.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 42
Modelling
 For control purposes the pneumatic control valve is usually
the preferred choice because it has excellent response
characteristics.

Air • The air pressure acting

on top of the diaphragm
produces a force F 
Actuator F
F • The spring acting below
Spring the diaphragm produces
Valve plug
Stem a force F 
Seat
• At equilibrium the two
flow forces are in balance and
the plug is stationary.
Valve body
Pneumatic Globe Valve (Fail-Open)
>>> The flowrate is then fixed somewhere between 0 and Qmax.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 43
Modelling
 Much less detail is needed to represent a control valve
symbolically – again this is diaphragm actuated globe valve:

Instrument Symbol
for a Control Valve

• The symbol above is fairly typical of many control valves

found in industry.
>>> As stated before, a full list of symbols for different types of
control valve is attached in Appendix B.

© Heriot-Watt University, B49CF2, Process

August 2015 v1 44
Modelling
 Control Valve Actuator
When the final control element is a valve, the controller sends
a signal m(t ) to the valve actuator:
• The valve actuator/spring assembly then converts this
signal into a valve position, which in turn determines the
flow through the valve.
• Valve position is usually defined in terms of the “fractional
valve position”, denoted by vp :
• A vp of 0 means that the valve is fully closed.
• A vp of 1 means that the valve is fully open.
>>> The compressed spring contains stored energy and this
stored energy is used to drive the valve to either the fully
open or the fully closed position on instrument air failure.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 45
Modelling
 Control Valve Action
When choosing a control valve it is important to consider
what would happen if the air supply were to fail:
• This is called the its “fail position” or “failure mode”.
• The failure mode of the valve is largely determined by a
combination of safety and process requirements. For
instance what is the best failure mode for valve below?
Steam • On air failure, the steam
FC should stop heating the
process fluid.
HX - 114
T (t ) • This is safest option.
T
• The valve should Fail
Closed “FC”.

>>> See abbreviation “FC” below the valve symbol.

© Heriot-Watt University, B49CF2, Process
August 2015 v1 46
Modelling
 Valves can either be Fail-Closed “FC”, or Fail-Open “FO” when
the energy supply (usually compressed air) fails:
• On failure a “FC” valve adopts the fully closed position – no
flow through the control valve.
• On failure a “FO” valve adopts the fully open position – that
is maximum flow through the control valve.
• Valves that control heating fluids usually “FC”, those that
control cooling fluids usually “FO”....... beware, process
requirements can occasionally override these rules !!
Pneumatic control valves need a supply of pressurised air to
be energised:
• “FO” valves need an air supply to close, thus they are also
called Air-To-Close “ATC” valves.
• “FC” valves need an air supply to open them, thus they are
also called Air-To-Open “ATO” valves.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 47
Modelling
 A FC control valve is shown below:

Air • The diaphragm force

acts to open the valve.
• The spring force acts to
close the valve.
flow
• The failure mode
depends on the spring;
Fail Close (FC) Control Valve whether it opens or
closes the valve.

>>> Remember, the spring is a stored energy source. On failure

of the main energy source (the air), the stored energy source
(the spring) CLOSES the valve.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 48
Modelling
 A FO control valve is shown below:
Air
• The diaphragm force
acts to close the valve.

• The spring force acts to

open the valve.
flow
• This means that a FO
valve will OPEN on air
Fail Open (FO) Control Valve supply failure.

>>> On failure of the main energy source (the air), the stored
energy source (the spring) OPENS the valve.

© Heriot-Watt University, B49CF2, Process

August 2015 v1 49
Modelling
 If the controller output signal m(t ) is expressed in terms of
percent controller output (%CO), then the valve position,
expressed as a fraction, is given by
FC: vp m(t ) / 100 If m(t ) 0, FC  vp 0

FO: vp 1  m(t ) / 100 If m(t ) 0, FO  vp 1

Valve Positioner
Knowing the controller output m(t ) the factional valve
position vp is fixed as shown above.
• However, it is important to ensure that the control valve
stem actually adopts this position and does not get stuck.
• This is the job of the “Valve Positioner”.
>>> The positioner must overcome dynamic fluid and static
friction effects; valve HYSTERSIS must be minimised.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 50
Modelling
 Valve positioners should be considered for all process control
applications.
• The positioner measures the actual position of the valve
stem and compares it to the required valve position.
• If there is an error, then the positioner will drive the valve
quickly to the correct position.
• The positioner must have a very high sensitivity to small
changes (0.125% or less) in the controller output signal,
otherwise it cannot respond to minor CV errors.
• It must supply large amounts of air to the actuator so that
the stem responds quickly.
• Typically the positioner receives a 4-20 mA DC signal and
an “I/P converter” converts this into a 0.2-1.0 bar signal;
the I/P converter now usually supplied with the valve.
>>The latest trend is towards microprocessor-based positioners
which have an inbuilt fault-finding/diagnostic capability.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 51
Modelling
 Control Valve Characteristics
Flow of liquid through a control valve in imperial units is
described by the equation:
Pv
f C v
Gf
• In which f is the flow (US gallons per minute), Pv is the
pressure drop across the valve (in psi) and G f is the specific
gravity “sg” of the fluid, which is the density of a fluid
relative to that of water.
• Cv is called the “flow coefficient” or often simply the
“valve Cv” – in this case it is defined in imperial units.
>>> It can be understood as the volume flow (gpm) of water,
at a temperature of 60° Fahrenheit, passed by the valve, when
the pressure drop across the valve is 1 psi.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 52
Modelling
 The metric equivalent of Cv is K v . It can be understood as
the volume flow (cubic metres per hour) of water passed by
the valve when the pressure drop across the valve is 1 bar.

Pv
f K v
Gf
• In which f has units (m3/h), Pv has units (bar).
Conversion between K v and Cv is straightforward:
K v 0.86 C v
• For compressible gases different equations are used (Smith
and Corripio, 2006).
>> If all the other factors can be measured (when the valve is
tested on a test bench) then the valve Cv can be calculated.
Larger valves will have a larger value than smaller valves.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 53
Modelling
 Valve Characteristics
The value of the coefficient Cv depends not only on the valve
position but also on the construction of the valve:
100
Cv, % of
• Some valves approach
maximum
80
Quick
opening maximum Cv at low vp
these are quick opening”
60
valves.
40
linear
• Linear valves have a
Equal
percentage linear characteristic.
20
• Equal percentage valves
0
0 0.2 0.6 0.8 1.0
pass more flow as the
valve opens.
0.4

Valve position

>>> The Cv versus valve position is known as the “inherent

characteristic” (above). It is measured when the valve is on a
test bench subjected to a constant pressure drop.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 54
Modelling
 The way Cv varies with valve position is called the valve
characteristic. Valve manufacturers can alter this by changing
the shape of the valve plug, see below:
The wetted components (that controls
Valve Plug Shape
flow) is called the “valve trim”.
>>> This shape of plug produces a
“quick opening” inherent
Quick opening
characteristic

>>> This shape of plug produces a

“linear” inherent characteristic
linear

>>> This shape of plug produces an

“equal percentage” inherent
Equal percentage characteristic
© Heriot-Watt University, B49CF2, Process
August 2015 v1 55
Modelling
 Equal percentage valves get their name from the fact that if
the controller output changes from 30 -31%, then the flow will
also change by 1 % of the flow at 30%.
• If the controller output changes from 70 -71%, then the
flow will also change by 1 % of the flow, but now 1% of the
flow at 70%.
• Thus, this type of inherent characteristic produces a rising
flow as the valve opens, because a 1% change in flow at
70% is a much larger increment than 1% in flow at 30%.
• With equal percentage valves, the algebraic relationship
between Cv and valve position vp is simply
C v C v ,max  vp  1
In which  is a constant – usually 50, but sometimes 100.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 56
Modelling
 In the case of valves with a linear responses, the algebraic
relationship between Cv and valve position vp is simply
Cv Cv ,max vp
• The “inherent” valve characteristic is obtained on a test rig
where the valve is subject to a constant pressure drop. The
valve is gradually opened and Cv is measured.
• However, when the valve is installed in the process, the
pressure drop across the valve is often varies with flow.
• This means that the “installed” characteristic can often be
very different from the “inherent” characteristic.
• For this reason a linear characteristic may produce a non-
linear flow response when installed in the process.
>>> The idea of using equal percentage trims (for these cases)
is to approximately linearise the flow response.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 57
Modelling
 Valve Rangeability
When a valve is nearly fully closed, or nearly fully open, it
becomes difficult to properly control the flow.
• The valve rangeability gives a measure of the range of
operating flows a valve can actually control.
• Rangeability is defined at the ratio of the maximum
controllable flow to the minimum controllable flow, usually
taken to correspond to 95% and 5% of valve position.

Flow at 95% valve position

Rangeability 
Flow at 5% valve position

>>> A control valve should always be modulating flow.

Whenever the control valve is either fully open or fully closed
it is not by definition controlling flow.

© Heriot-Watt University, B49CF2, Process

August 2015 v1 58
Modelling
 Control Valve Sizing
To be able to select the correct valve for a given system, the
valve Cv needs to be calculated. This may be found by re-
arranging the valve flow capacity expression as follows:
Gf
Cv  f
Pv

• In which f is the nominal flow through the system, i.e. the

flow under normal steady-state circumstances.
• In practice the flow must be varied up and down which
means that a larger flow coefficient Cv ,max is needed.
• The ratio of the larger Cv ,max to the nominal Cv is the
overcapacity factor of the valve, defined as follows:
C v , max
Overcapacity Factor  Usually fixed at around 1.5 - 2
Cv
© Heriot-Watt University, B49CF2, Process
August 2015 v1 59
Modelling
 Another important parameter when choosing a valve is the
pressure drop across the valve
• Frictional loss across a valve represents an energy loss and
therefore a cost to a process. The pressure loss has to be
kept to a minimum, but with following considerations:
• It must be large enough to effectively control flow.
• If too small a change in valve position will not affect the
flowrate.
• Typically 15 psi is used. The pressure drop across the valve
is seldom less than 5 psi.
• A rule of thumb used by many engineers is that the
pressure drop across a control valve Pv should be roughly
1/3 of the total frictional losses of the pipework.
>>> Cavitation will occur in the valve whenever the pressure
(within the valve) drops below the fluid’s vapour pressure.
This can often be avoided by locating the valve correctly.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 60
Modelling
 Control Valve Gain
The pressure drop across any control valve may be calculated
from a re-arrangement of the valve capacity expression
2
f
p v G f 2
Cv
• The pressure drop across the pipework pL is in most
cases dependent on the square of the flow rate. This is
because p L are dependent on the velocity head u / 2 g .
2

• Thus, pipework pressure losses are given by

2
p L k L G f f

The total pressure drop p0 for any type of system/valve

arrangement is, therefore, just the sum of these two drops.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 61
Modelling
 Adding the two component parts together leads to:

p 0 p v  p L
2
f 2
 p 0 G f 2
 k LG f f
Cv

• After some re-arrangement the flow through any system

and the valve arrangement can be expressed as

Cv p 0
f 
1  k L Cv
2 Gf

>>> The overall pressure drop “PD” is assumed to be fixed,

even though the line and valve PD will vary.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 62
Modelling
 The “gain” of a valve K v is the steady-state change in flow f
divided by the change in controller output signal m that is
needed to produce this change in flow, or mathematically
df
Kv 
dm
• The units of K v are (gpm)/(%CO). Of course, units other
then gallons per minute (gpm) may be used.
• If the pressure drop across the valve is constant (usually the
case when line losses are negligible), then for valve with
linear characteristics:
f max
K v 
100
>>> A FC valve has a positive valve gain, while a FO valve has a
negative valve gain.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 63
Modelling
 If the pressure drop across the valve is constant, then for a
valve with equal percentage characteristics the valve gain K v
is given by
 ln  
K v   f
 100 
• In which f is the nominal flow. For a valve with linear
characteristics, if the pressure drop across the valve is
variable, then the valve gain K v is given by

1 Cv ,max P0
K v 
100 1  k L Cv2 3 / 2 Gf

© Heriot-Watt University, B49CF2, Process

August 2015 v1 64
Modelling
 For a valve with equal percentage characteristics, if the
pressure drop across the valve is variable, then the valve gain
K v is given by

ln  f
K v 

100 1  k L Cv2 

© Heriot-Watt University, B49CF2, Process

August 2015 v1 65
Modelling
 1.4.4 Tutorial Questions - Valves
1. Explain the reason and specify safe actions (fail-close or
fail-open) for the valves in the following services:
a) A highly flammable solvent is heated in an indirect
contact vessel by hot oil. The valve controls the flow of
oil into the heater.
b) A hazardous liquid is stored in a tank. The valve controls
the flow of liquid to downstream equipment.
c) A tank is used to store high pressure gas from upstream
equipment. The pressure in the tank is controlled by the
valve at the exit. The equipment downstream of the
valve is capable of dealing with surge in the flow.
d) A valve controls the flow of gas fuel to a burner whilst
another valve controls the flow of combustion air.
e) A valve controlling the flow of reactant into a reactor.
The reaction is highly exothermic.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 66
Modelling
 2. Calculate the valve rangeability of:
a) A linear valve.
b) A valve with equal percentage characteristics (  50 ).
c) A valve with equal percentage characteristics (  10 ).
d) A valve with square root characteristics C v C v ,max vp .

3. You are asked to size a control valve for regulating the flow
of benzene in a line as in Figure 3.1 (over the page). The
flow rate is 600 (gpm). The relative density of benzene is
0.730. The pressure drop caused by friction between points
“1” and “2” is 100 kPa. The pressure at point “1” is 970 kPa,
at point “2” 630 kPa; assume that the pressures do not
change with flow.
a) Determine where you would place the valve.
b) Determine the pressure drop and the flow coefficient Cv
for 100% overcapacity…….Pressure conversion: 1
atm = 101.3 kPa = 14.7 psi
© Heriot-Watt University, B49CF2, Process
August 2015 v1 67
Modelling
 Fig. 3.1: Benzene Flow System

25 m

© Heriot-Watt University, B49CF2, Process

August 2015 v1 68
Modelling
 1.4.5 Controllers – The “D” Element
The controller is the “brain” of the control loop. A feedback
controller receives a CV signal, from the sensor/ transmitter,
and compares it with the SP.
• It then generates an error signal “e = SP-CV” internally
within the controller.
• The controller then sends a signal to the valve (or other
final control element) in order to keep the CV at its SP
valve. Thus, the controller seeks to bring the error to zero.
• The signal that the controller sends to the valve is based on
a mathematical equation, which is determined by the
engineer as a result of his/her experience of the process.
>>> Later it will be shown how to derive these equations and
how to set up a controller that will operate efficiently in
conjunction with the process that it is controlling.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 69
Modelling
Controller Action
A controller may either be “direct-acting” or “reverse-acting”.
• If an increase in signal from the sensor/transmitter requires
an increase in output, then controller is DIRECT-ACTING. A
direct-acting controller has a negative controller gain K C .
• If an increase in signal from the sensor/transmitter requires
an decrease in output, then controller is REVERSE-ACTING. A
reverse-acting controller has a positive controller gain K C .
• A direct-acting controller is also called an “increase-
increase” controller.
• A reverse-acting controller is also called an “increase-
decrease” controller.
>>> Selecting the correct controller action is very important. If
the wrong controller action is selected, then in all likelihood
the control loop will be unstable.
© Heriot-Watt University, B49CF2, Process
August 2015 v1 70
Modelling
 1.4.6 Tutorial Questions – Controllers
1. Determine the action of all the controllers shown below:
a) A SISO heat exchanger control system (a single loop)
SP
FC TC

Heat Exchanger
Control Loop
Steam
TT
Process Fluid

Ti(t) T(t)

T
Condensate

© Heriot-Watt University, B49CF2, Process

August 2015 v1 71
Modelling
 a) A MIMO flash drum (three separate loops)
FO
Flash
Drum Vapour PT
product
PC
FC

Feed
FT LC
LT
Steam

FC

Condensate

T FC
Liquid
product

© Heriot-Watt University, B49CF2, Process

August 2015 v1 72
Modelling
 Diagrams: see Appendix A for……
• Block Flow Diagrams or “BFD” diagrams
• Process Flow Diagrams or “PFD” diagrams
• Piping and Instrumentation Diagrams or “P&ID” diagrams
• Control system “Block Diagrams” (not to be confused with
BFDs.

Symbols: see Appendix B for……

• P&ID symbols. These are the symbols used when producing
P&ID diagrams.

© Heriot-Watt University, B49CF2, Process

August 2015 v1 73
Modelling