Vous êtes sur la page 1sur 234

The ECOSSE Control HyperCourse

By Members of the ECOSSE Team and Staff of the Department of Chemical Engineering,
University of Edinburgh, Scotland

Welcome Who is it for?

to the ECOSSE Control HyperCourse. I Primarily the material is to be used by


hope that you will find the material students at Edinburgh as a follow up to
presented here both interesting and their lectures. However, it could also be
stimulating! used by other chemical engineering
departments or process engineers in
industry as a refresher course for those
What is it? involved in the development or use of
control systems.
Basically it is a course on control
available on the WWW.
Features of the Course
At Edinburgh the students have access
to the WWW whenever they want. All  The course is written for process
chemical engineering subjects have engineers in a language they will
taken advantage of this and most understand.
teaching material is available online. The  It is about controlling complete,
control hypercourse takes this concept a real process, not idealised parts
step further and includes interactive of imaginary processes.
simulations.  It aims to simplify a complex task
by providing a systematic
Why was it written? approach and explicit guidelines.
 It is not mathematical. Where
First and foremost it was written to
mathematics is necessary, a
benefit the students at Edinburgh
straightforward numerical and
University studying chemical
computational approach will be
engineering. Now if a lecture is not
used, rather than complicated
understood they can go to the
algebra.
computer where the material may be
 It will be useful both to students
presented in a different manner. Also
and to practising engineers.
the students can work at their own pace
and review or preview material at any
time. This is an excellent resource,
especially at exam time. What does it cover?
At present the course covers three years The course has been written in HTML, a
of undergraduate study. This work can HyperText Markup Language which
be found in modules 1 to 3. These are allows links from page to page by simply
intended to be a self contained clicking on an highlighted words. It
introductory course and could be incorporates images, sounds, movies
subtitled All You Really Need to Know and interactive sections to make the
about Process Control. All process subject matter more interesting and
engineers should become familiar with therefore hopefully easier to
the concepts and techniques discussed understand.
here, because they are an integral part
of understanding and designing The teaching material is presented to
processes, rather than specialist topics the students in a number of different
which may only occasionally be of ways:
importance to the working engineer.
 Firstly there are theory sections
Modules 4, 5, 6 and 7, which are in the which duplicate material from
process of being written will lectures and may also contain
discuss advanced control techniques. additional work for more able
This will include a brief resumé of students.
classical methods in process control.  Then there are case study
They will be written very much with sections which include worked
process engineers in mind, but some of examples and tutorial questions.
the contributors have a classical or non- In some cases the worked
process control background. This section examples may be interactive. The
is intended to allow process engineers student is given hints and
to use the knowledge gained from comments as they work through
modules 1 to 3 to understand state-of- the exercise, in response to
the-art techniques, judge their suitability answers to specific questions.
for a given process application, and then  Finally there is The Virtual Control
to supervise their implementation by Laboratory which is a set of
specialist control engineers. interactive simulations designed
to convey specific points in
process control. These are based
on experiments in our own
How does it Work? laboratory.

Contents

 Introduction
 Part 1: Basic Concepts
o Module 1.1: The Very Basics
 Introduction
 Why Control?
 Control Objectives
 Safety
 Operability
 Profitability
 Techniques of Control
 Basic Concepts of Feedback Control
 Disturbances and Setpoint Changes
 Advantages of Feedback Control
 Disadvantages of Feedback Control
o
o Module 1.2: The Control System
o Module 1.3: Practical Control Examples
o Module 1.4: Sensors and Tranducers
o Module 1.5: Control Actions
o Module 1.6: Variations on Basic Feedback Control
o Case Study Section
 Part 2: Further Developments
o Module 2.1: Controlling Simple Processes
o Module 2.2: More on Degrees of Freedom
o Module 2.3: Controlling Separation Processes
o Module 2.4: Controller Design and System Modelling
o Case Study Section
 Part 3: Control Systems For Complex Processes
o Module 3.1: Control of Distillation Columns
o Module 3.2: Extension of Hierarchical Decomposition
o Case Study Section
 Part 4: More Advanced Concepts
o Module 4: Further Controller Design
o Module 5: Development of Mechanistic Models
o Module 6: Multivariable Systems
o Module 7: Frequency Response Techniques
o Case Study Section
 Tutorial Exercises
 The Virtual Control Laboratory
Contents
1 Basic Concepts ..................................................................................................................................... 6
1.1 The Very Basics .......................................................................................................................... 6
1.1.1 Introduction .................................................................................................................. 7
1.1.2 Why Control? ................................................................................................................ 7
1.1.3 Control Objectives ...................................................................................................... 8
1.1.4 Techniques of Control ............................................................................................... 9
1.2 Control Systems ...................................................................................................................... 13
1.2.1 Introduction ................................................................................................................ 13
1.2.2 Manual Control System .......................................................................................... 13
1.2.3 Functions of a Control System ............................................................................. 14
1.2.4 Automatic Control System ..................................................................................... 15
1.2.5 Hardware of a Control System ............................................................................. 16
1.2.6 Software of a Control System ............................................................................... 16
1.2.7 Example - Hot Water Tank..................................................................................... 17
1.3 Module 1.3 - Practical Control Examples ....................................................................... 17
1.3.1 Introduction ................................................................................................................ 17
1.3.2 Flow Control Systems .............................................................................................. 19
1.3.3 Inventory Control Systems..................................................................................... 20
1.3.4 Temperature Control Systems .............................................................................. 23
1.3.5 Composition Control Systems .............................................................................. 24
1.3.6 Pressure Control in Two Phase Systems ........................................................... 26
1.4 Sensors and Transducers ..................................................................................................... 27
1.4.1 Basic Requirements .................................................................................................. 27
1.4.2 Pressure Transducers ............................................................................................... 28
1.4.3 Level Measurement .................................................................................................. 30
1.4.4 Flow Measurement ................................................................................................... 31
1.4.5 Temperature Measurement................................................................................... 32
1.5 Control Actions ........................................................................................................................ 34
1.5.1 The Controller............................................................................................................. 34
1.5.2 On-off or two-position action .............................................................................. 35
Advantage ................................................................................................................................. 35
Disadvantage............................................................................................................................ 35
1.5.3 Proportional Control Action .................................................................................. 35
1.5.4 Integral and Derivative Control............................................................................ 39
1.5.5 Examples ...................................................................................................................... 41
1.6 Variations on Basic Feedback Control ............................................................................. 41
1.6.1 Feedforward Control ................................................................................................ 41
Complex Computation.......................................................................................................... 42
Knowledge of Process ........................................................................................................... 43
Limitations ................................................................................................................................. 43
Specific Controller Required ............................................................................................... 43
1.6.2 Cascade Control ........................................................................................................ 43
1.6.3 Split Range Control .................................................................................................. 46
1.7 Case Study Section ................................................................................................................. 48
1.7.1 Laboratory Session ................................................................................................... 48
1.7.2 Hot Water Tank .......................................................................................................... 49
1.7.3 Level Control Experiment ....................................................................................... 49
1.7.4 Tutorial Questions and Answers .......................................................................... 49
1.7.5 Exercise 1: Simple Control Loops ........................................................................ 49
1.7.6 Exercise 2: Simple Processes ................................................................................. 49
2 Appendices........................................................................................................................................230
2.1 Hot Water Tank Experiment..............................................................................................230
2.1.1 Aims .............................................................................................................................230
2.1.2 Theory .........................................................................................................................230
2.1.3 Pre-Experimental Questions................................................................................231
2.1.4 Experimental Apparatus .......................................................................................232
2.1.5 Procedure ...................................................................................................................232
2.1.6 Answers.......................................................................................................................233
2.1.7 Summary ....................................................................................................................234

1 Basic Concepts

The six modules in this part of the course deal with the very basics of process control.
The concept of control systems are introduced with the aid of practical examples. The
hardware associated with control is discussed and finally various control actions are
examined. The Case Study Section contains extra material to reinforce what was covered
in the Modules. This includes experiments and tutorial questions and answers.

Module 1.1: The Very Basics

Module 1.2: The Control System

Module 1.3: Practical Control Examples

Module 1.4: Sensors and Transducers

Module 1.5: Control Actions

Module 1.6: Variations on Basic Feedback Control

Case Study Section

1.1 The Very Basics

This first module sets the basis of what is to follow. It discusses why we use control and
introduces the basic concept of simple feedback control.
1.1.1 Introduction

This part of the course starts with an outline and overview of basic control concepts.
Questions which process engineers routinely have to answer about process control
include the following:

 I have this process. What should I control?


 Where on the process do I put my control loops?
 As I proceed with the design of a process, what aspects of control should I
consider at which stages?

Most books with the words `process control' in the title do little to answer these
questions. Classical linear control theory, which forms the basis of most books on
control, is much concerned with how to design controllers and is less helpful on how to
design complete control systems. Other problems with this classical approach, for most
process engineers wishing to design control systems for real chemical processes, are the
restriction of most of its methods to idealised process models, and the extensive use of
rather specialised mathematics.

Satisfactory answers to questions such as the above frequently require little


conventional mathematics. What they do require, however, is a good understanding of
what a process is intended to do and how it works.

In this book we will approach process control from the standpoint of a chemical or
process engineer, and address these questions and others like them. We will consider
the process and its control system in the language of process engineering. We will use
mathematics, as such, only when necessary, and the language of classical control
engineering only when it is unavoidable, or will add very significantly to the process
engineer's understanding.

1.1.2 Why Control?


Chemical plants are intended to be operated under known and specified conditions.
There are several reasons why this is so:

Safety:
Formal safety and environmental constraints must not be violated.

Operability:
Certain conditions are required by chemistry and physics for the desired reactions
or other operations to take place. It must be possible for the plant to be arranged
to achieve them.
Economic:
Plants are expensive and intended to make money. Final products must meet
market requirements of purity, otherwise they will be unsaleable. Conversely the
manufacture of an excessively pure product will involve unnecessary cost.

A chemical plant might be thought of as a collection of tanks in which materials are


heated, cooled and reacted, and of pipes through which they flow. Such a system will
not, in general, naturally maintain itself in a state such that precisely the temperature
required by a reaction is achieved, a pressure in excess of the safe limits of all vessels be
avoided, or a flowrate just sufficient to achieve the economically optimum product
composition arise.

1.1.3 Control Objectives

Control systems in chemical plants have, as noted, three functions.

 Safety.
 Operability, i.e. to ensure that particular flows and holdup are maintained at
chosen values within operating ranges.
 To control product quality, process energy consumption etc.

To a large extent these are quite separate objectives. Indeed, in the case of safety
systems separate equipment is generally used. The aims of control for operability are
secondary to those of strategic control for quality etc., which directly affect process
profitability.

Control for Safety


Concern for safety is paramount in designing a chemical plant and its control systems.
Ideally a process design should be `intrinsically safe', that is, plant and equipment
should be such so that any deviation, such as an increase in reactor pressure, will itself
change operating conditions so that it is rapidly removed, for example by a fall in
reaction rate. For many perturbations this type of responsive, passive safety system will
not be possible and active systems will be required.

These active safety systems must be robust and of high integrity. Current processes
achieve this through simplicity. The ultimate safety system is in most cases the
mechanical relief valve which simply vents the plant to atmosphere, possibly through a
flare or scrubber.

We will not discuss control for safety explicitly in this book. Generally speaking a
complete and separate system is provided to handle emergency control action. The
need for this, and its design requirements, are established in hazard and
operability or hazop studies. These are typicaly carried out on the complete process with
its `normal' control systems in place.

A number of safety issues will be addressed in the course of developing the design of
the control systems for normal operation, but it must be emphasised that our treatment
of this vital issue will be relatively restricted.

Control for Operability


The operator of a process quite simply has to

 know what it is doing


 be able to make it do what he or she wants, rather than to follow its natural
inclinations.

The issue of making a plant behave in this way is called operability.

The majority of control loops in a plant control system are associated with operability.
Specific flow rates have to be set, levels in vessels maintained and chosen operating
temperatures for reactors and other equipment achieved.

Control for Profitability


There is no point in building a plant which is totally safe and can be made to take up
any (safe) conditions of flow, temperature etc., if the conditions under which it is
operated do not produce the correct amount of product to the correct specification,
thus allowing its operators to make a profit.

The top level of process control, what we will refer to as the strategic control level is thus
concerned with achieving the appropriate values principally of:

 Production rate,
 Product quality, and
 Energy economy.

1.1.4 Techniques of Control

Basic Concepts of Feedback Control


The task of maintaining these required conditions falls to one or, more usually
several, process control systems with which the plant will be equipped. The practical
aspects of these will be discussed more fully in the following module. The underlying
principle of most process control, however, is already understood by anyone who has
grasped the operation of the domestic hot water thermostat:

 The quantity whose value is to be maintained or regulated, e.g. the temperature


of the water in a cistern, is measured.
 Comparison of the measured and required values provides an error, e.g. `too hot'
or `too cold'.
 On the basis of the error, a control algorithm decides what to do.

Such an algorithm might be:

If the temperature is too high then turn the heater off. If it is too low then turn
the heater on.

 The adjustment chosen by the control algorithm is applied to some adjustable


variable, such as the power input to the water heater.

This summarises the basic operation of a feedback control system such as one would
expect to find carrying out nearly all control operations on chemical plants, and indeed
in most other circumstances where control is required. The diagram belows a feedback
control loop.

Notice that this extremely simple idea has a number of very convenient properties. The
feedback control system seeks to bring the measured quantity to its required value
or setpoint. The control system does not need to know why the measured value is not
currently what is required, only that this is so. There are two possible causes of such a
disparity:

 The system has been disturbed. This is the common situation for a chemical plant
subject to all sorts of external upsets. However, the control system does not need
to know what the source of the disturbance was.
 The setpoint has been changed. In the absence of external disturbance, a change
in setpoint will introduce an error. The control system will act until the measured
quantity reaches its new setpoint.

A control system of this sort should also handle simultaneous changes in setpoint and
disturbances.

Advantages of Feedback Control


Not only does the feedback control system require no knowledge of the source or
nature of disturbances, but it requires minimal detailed information about how the
process itself works.

Feedback control action is entirely empirical, so long as an adjustment is being made in


the correct `sense', e.g. more heat means increasing temperature and vice versa, then
the control system should remove the effect of an external disturbance.

As we will see, it helps to know more than this, but the minimum information required
to make a feedback control system work is whether the adjustment makes the
measurement go up or down.

Disadvantages of Feedback Control


The main disadvantage of feedback control is that the disturbance enters into the
process and upsets it. It is after the process output is different from the setpoint that the
controller takes some corrective actions. Although most processes allow some
fluctuation of controlled variable within a certain range, there are two process
conditions which can make the overall effectiveness of feedback control quite
unsatisfactory. One of these is the occurrence of disturbances of a large magnitude that
is strong enough to seriously affect or even damage the process. The other is the
occurrence of a large amount of lag (time delay) within the process. These are discussed
further below.

1.1.4.3.1 Large Magnitude Disturbance

An example where the occurrence of disturbances of large magnitude that are strong
enough to seriously effect the process, is temperature control of a catalyst reactor in
which strong exothermic reaction takes place. The reaction heat is very high, therefore
the reactant gas mixture is diluted by a inert gas to carry away most reaction heat,
although the temperature of the reactor is maintained by feedback control of a coolant
flowrate in coils inside. Assuming a large magnitude disturbance, the sudden large
increase in the reactant concentration in the feed, enters the reactor, a sudden increase
in the temperature is so large and so quick that the catalyst is burnt out before the
control system senses the change and takes any actions. A diagram of this situation is
shown below.

1.1.4.3.2 Large Time Delay

A simple example of a large time delay is the distillation column as outlined in the figure
below. If we use feedback control to regulate the purity of the top product, when the
feed composition changes (disturbance), the control system is not aware any takes no
action until the effects of the disturbance travels and arrives at the sensor position at the
top. When the controller takes the correction, the whole column may be far away from
the designed conditions.
The question of importance of either occurrence is defined in economic terms. In either
case, the principle concern is the existence of errors that have significant economic
consequences in the overall process operation. In these cases, feedforward control can
be used to deal with these disadvantages or inadequacies of feedback control.

1.2 Control Systems

1.2.1 Introduction
The key characteristic of control is to interfere, to influence or to modify the process.
This control function or the interference to the process is introduced by an organization
of parts (including operators in manual control) that, when connected together is called
the Control System. Depending on whether a human body (the operator) is physically
involved in the control system, they are divided into Manual Control and Automatic
Control. Due to its efficiency, accuracy and reliability, automatic control is widely used in
chemical processed.

The aim of this section is to introduce the concept of control systems, what their
function is and what hardware and software is required by them.

1.2.2 Manual Control System

First start with a simple manual control system, to examine how control is introduced,
how the control system is constructed and how it works.
A diagram of the system is shown below.

To begin with the shower is cold. To start the heating process the valve in the hot water
line is opened. The operator can then determine the effectiveness of the control process
by standing in the shower. If the water is too hot, the valve should be closed a little or
even turned off. If the water is not hot enough then the valve is left open or opened
wider.

1.2.3 Functions of a Control System

It can be seen that this control system, completed by the operator, possesses the
following functions:

 Measurement

This is essentially an estimate or appraisal of the process being controlled by the


system. In this example, this is achieved by the right hand of the operator.

 Comparison

This is an examination of the likeness of the measured values and the desired
values. This is carried out in the brain of the operator.

 Computation

This is a calculated judgment that indicates how much the measured value and
the desired values differ and what action and how much should be taken. In this
example, the operator will calculate the difference between the desired
temperature and the actual one. Accordingly the direction and amount of the
adjustment of the valve are worked out and the order for this adjustment is sent
to the left hand from the brain of the operator. If the outlet water temperature is
lower, then the brain of the operator will tell the left hand to open the steam
valve wider. If there is any disturbance, or variation of flow rate in water to the
shower inlet, some adjustment must be made to keep the outlet water
temperature at a desired value.

 Correction

This is ultimately the materilisation of the order for the adjustment. The left hand
of the operator takes the necessary actions following the order from brain.

Therefore, for a control system to operate satisfactorily, it must have the abilities of
measurement, comparison, computation and correction.

Of course, the manual operation has obvious disadvantages e.g. the accuracy and the
continuous involvement of operators. Although accuracy of the measurement could be
improved by using an indicator, automatic control must be used to replace the operator.
In industry, it is automatic control that is widely used.

1.2.4 Automatic Control System

Based on the above process, we can easily set up an automatic control system as shown
in the next figure.
 Firstly, we can use a temperature measurement device to measure the water
temperature, which replaces the right hand of the operator. This addition to the
system would have improved accuracy.
 Instead of manual valves, we use a special kind of valve, called a control valve,
which is driven by compressed air or electricity. This will replace the left hand of
the operator.
 We put a device called a controller, in this case a temperature controller, to
replace the brain of the operator. This has the functions of comparison and
computation and can give orders to the control valve.
 The signal and order connections between the measurement device, control valve
and controller are transfered through cables and wires, which replace the nerve
system in the operator.

1.2.5 Hardware of a Control System


Examining the automatic control system, it is found that it contains the following
hardware.

 Sensor - a piece of equipment to measure system variables. It serves as the


signal source in automatic control. These will be discussed at length in a later
module.
 Controller - a piece of equipment to perform the functions of comparison and
computation. The actions that a controller can take will be discussed at length in
a later module.
 Control Element - a piece of equipment to perform the control action or to exert
direct influence on the process. This element receives signals from the controller
and performs some type of operation on the process. Generally the control
element is simply a control valve.

1.2.6 Software of a Control System


Associated with a control system are a number of different types of variables.

First we have the Controlled Variable. This is the basic process value being regulated
by the system. It is the one variable that we are specially interested in - the outlet water
temperature in the example above. In feedback control the controlled variable is
usually the measured variable.

An important concept related to the controlled variable is the Setpoint. This is the
predetermined desired value for the controlled variable. The objective of the control
system is to regulate the controlled variable at its setpoint.
To achieve the control objective there must be one or more variables we can alter or
adjust. These are called the Manipulated Variables. In the above example this was the
input hotwater flow rate.

Conclusively, in the control system we adjust the manipulated variable to maintain


the controlled variable at its setpoint. This meets the requirement of keeping the
stability of the process and suppressing the influence of disturbances.

1.2.7 Example - Hot Water Tank

In the Virtual Control Laboratory there is an experiment which shows a control system in
action. This example is based on the hot water thermostat mentioned in Module 1.1 -
The Very Basics. It describes how the control system is put together and then how it
works. In this simple example feedback on/off control is used to regulate the
temperature of the hot water at its setpoint. Note that a link to this experiment can also
be found in the Case Study Section for this part of the HyperCourse.

Go to Hot Water Tank Experiment

1.3 Module 1.3 - Practical Control Examples

1.3.1 Introduction

Many different operations take place in a chemical plant. The classical approach of Unit
Operations might thus be extended to process control, and we could consider in turn
the control of heat exchangers, chemical reactors, distillation columns etc.

This turns out not to be a useful approach in most cases. The reason for this is that we
are in the end concerned with the control of processes which consist
of several operations, and these cannot be considered in isolation. This makes the
engineer's task of designing a control system a difficult one, since it is hard to find just
where to start!

The starting point we shall choose here is to consider how we regulate each of the basic
quantities we may wish to keep constant in a process.

These quantities are

 Flow
 Inventory - level or pressure
 Temperature
 Composition
 Pressure - two phases

The following sections discuss simple, but real, examples of how feedback control is
applied to these basic quantities in a chemical plant. They are primarily examples of
control for operability, and most of them will refer to single items of equipment or very
simple combinations. A number of safety issues will be identified.

Strategic control for profitability will be dealt with in a later section in the context of
control of complete plants and processes.

A number of fundamental concepts will be illustrated in the course of these examples.


They are `graded' in the sense that the simplest examples come first; the reader is
advised to follow the sequence we have presented. Even apparently trivial examples may
be used to introduce important ideas.

Terminology
At this early stage there are three terms that should be used correctly. They are:

 Control
 Regulate
 Adjust

The word `control' is sometimes loosely used to mean either regulation or adjustment.
We have not actually seen the sentence 'In a control system the controller controls the
controlled quantity by controlling a control valve position' However the term is regularly
misused in this way. Control should refer only either to the actions of the controller
element itself or to the function of the complete system. We are not just being pedantic,
it is possible to misunderstand what is happening through misuse of terminology. We
may also slip up ourselves, particularly in talking loosely about e.g. 'flow control' when
we strictly mean 'flow regulation'.

Control Hardware
In all the examples which follow there will be various types of measurement sensors and
transducers mentioned. These are simply considered here as black boxes and will be
discussed in greater depth at alater stage.
1.3.2 Flow Control Systems
The most basic requirement in any chemical plant is to be able to make the flow
through a pipe take a particular value. Consider first therefore the simplest item of plant
equipment, namely a pipe, as shown below.

The basic pipe has had the following parts added to it, to make a control system.

 A flow measuring device or Flowmeter. This consists of two parts


o Firstly an Orifice Meter. This is shown in the diagram by two parallel lines.
o This is connected to a sensor or Flow Transducer labelled FT in the figure.
 An adjustable valve or Control Valve which alters the flowrate. This is shown by
its conventional flowsheet symbol.
 Finally these are connected by the Controller itself identified by the element FC.

This completes a control system to regulate the measured quantity, here the flow,
by adjustment of the valve position. Compare this with the block diagram which we used
earlier to introduce the feedback control system.

Positioning of Elements
One of the problems with designing control systems is that, as in any design problem,
we are faced with alternatives. We have an alternative here in the positioning of the
elements.

 Should the measurement element be placed upstream of the valve as shown in


the diagram?
 Or should it be downstream?

Consideration of the properties of flowmeters and valves suggests that we were correct
in our first choice. If the valve were upstream of the flowmeter then there are a number
of ways in which it might affect the flowmeter calibration.
Control Algorithms
In the simple illustrative example of the water heater the rule for making the adjustment
was:

 If the temperature is too high then turn the heater off.


 If it is too low then turn the heater on.

This is an example of an on-off control algorithm. The heater is either on (full) or off
(completely). What will happen if we try to use such an algorithm here, where the
objective is to maintain a particular flow?

 If the flow is too high then shut the valve off.


 If it is too low then open it.

Clearly, this is unlikely to serve, as rather than maintaining a specified flow the
conditions will switch between zero and some maximum value. To achieve a specified
steady flow we require something like:

 If the flow is too high then shut the valve some more.
 If it is too low then open it more.

This is a proportional control algorithm; the larger the error in the measured quatity, the
larger will be the adjustment. This arrangement should result in the system settling at or
near the required flow.

In practice, on-off control is seldom used. Most adjustment elements are valves, or
occasionally other mechanical elements. These do not take kindly to being regularly or
rapidly swung accross their full range of adjustment; they very quickly wear out or break
down.

In most cases, therefore, proportional control or some variant is used. More detailed
investigation of control algorithms requires quantitative information about the process.
This aspect will be dealt with in a later section.

1.3.3 Inventory Control Systems


The next most basic requirement in a plant is a control system to regulate the amount of
material or inventory in an item of equipment or over part of the process.

Inventory may be measured in a number of ways. Mass holdup may sometimes be


determined directly, but usually volume is measured. In liquid systems volume is
measured by level. In gas or vapour systems pressure is used as a measure of inventory.
Level Control Systems
Here we will consider simple feedback control of the level in a tank. This being the case
it is necessary to measure the level directly and adjust the flow into or out of the tank to
keep it constant.

Alternative Control Systems

Below is a diagram showing the two alternative control systems available for feedback
control of the level. Both are equally valid and the decision as to which to use is based
on

 What is upstream or downstream of the tank?


 Which streams are already being controlled?
 The relative sizes of the flowrates, if there are several input or output flows.
 Etc.

As can be seen the control system consists of

 A Level Transducer denoted by LT in the diagram.


 A Control Valve.
 A Level Controller denoted by LC.

The level can be regulated by altering the flow via the adjustment of the valve position.

Control Algorithms
It is possible to control the level in a tank using

 On/Off control.
 Proportional control.
 Extensions to proportional control.
The theory behind the algorithms will be found in a later section. There is also a level
control experiment based on an actual experiment carried out by the undergraduates in
the laboratory. Note that this can be found in the Case Study Section and in the Virtual
Control Laboratory.

Pressure Control Systems


In gas or vapour systems we regulate inventory as pressure. A typical system is shown
below. Both the inlet and outlet are gas or vapour. Therefore if the control valve is shut
then the pressure in the tank will rise and vice versa.

In principle we might, like the level control system, have the valve either upstream or
downstream of the tank. In practice in gas systems it is more likely to be downstream for
the following reason.

Raising the pressure of a gas requires energy, and normally this energy is imparted by
some mechanical device, such as a compressor. Both the compressor itself, and the
energy to drive it, are expensive. To minimise the first cost we try to minimise the
number of compressors in a process. Where possible we would use only one, locate it at
the front of the process, and perform any subsequent manipulations to obtain the
required pressure by downstream valves.

The energy used in compression is expensive, and throttling through a control valve
throws this energy away. Therefore in proesses where compressor costs are very
significant we may sometimes avoid such valves and manipulate the compressor speed
in order to maintain the system at the required pressure. This control system is shown in
the diagram below.
When we have vapour we usually also have liquids. Regulating pressure in two phase
systems can be somewhat different. This is dealt with later.

1.3.4 Temperature Control Systems


In this section the control of temperature is to be discussed. Again only simple feedback
loops are considered.

To change the temperature of something it is necessary to add or take away energy.


This can be achieved in one of two ways.

 Transfer energy indirectly, using a second stream, through coils, tubes, jackets
etc. The second stream could be, for example, steam, cooling water, another
process stream or even a source of power as in an electric element.
 Mix in a second stream directly. This stream will have a different energy content
from the original.

There are advantages and disadvantages for both methods. With the first there is the
problem of transferring heat through the walls of the 'coil'. In the second the energy is
absorbed directly but with the additional problem of increased flowrate/volume.

Diagrams of these alternative schemes can be found below.


1.3.5 Composition Control Systems
The control of composition is probably the most important objective in the chemical
industry due to the requirement for specification on products. It is thus
a strategic rather than an operational control problem and can only be considered
sensibly in the context of whole process control. This will be discussed fully later but it is
relevent to include a short example here.

Simple Composition Control Problem

To illustrate composition control consider the simplest process in which composition


can be changed, namely blending. Here two streams of different compositions are
mixed together e.g. a concentrate and a diluent as shown in the diagram below.
Either of the above schemes could be used although the first is preferred. The reasons
why are discussed in a later section.

It is worth mentioning that the composition of a stream is rarely measured directly.

Typical composition analysers include

 Gas chromatagraphs
 Spectroscopic analysers

Features of this type of hardware which make them ineffective for control purposes are

 Large time delay in their response


 Low operational reliability
 Relatively high cost

Thus an alternative method has to be sought to control the composition. This could be
via the

 Temperature of the mixture


 Pressure of the mixture
1.3.6 Pressure Control in Two Phase Systems
An example of a process which contains both a vapour and a liquid is distillation. We
generally wish to regulate the pressure in the column, which contains mainly vapour.
This could be done by placing a valve in the vapour line leaving the column, exactly as
we did with the simple tank.

There are several disadvantages to this system. One is that the control valve is on a
vapour line. These are generally much bigger than liquid lines and hence require a much
bigger valve i.e. of a much increased cost.

However, we remember that in a two phase system temperature and pressure are not
independent. We can thus change the pressure of a vapour which is in equillibrium with
a liquid by changing the temperature of the system. Raising the temperature raises
the vapour pressure of the liquid which must equal the equillibrium pressure of the
system.

Hence we can manipulate the temperature in the condenser by means of a small valve
on the cooling water line, thus changing the pressure in both condenser and column.
There are a large number of different ways of manipulating pressure in two phase
systems. These are discussed in a later section under Control of Distillation Process.

1.4 Sensors and Transducers

1.4.1 Basic Requirements

In order to control the process performance, we need a control system, which consists
of a sensor, a controller and a final control element. Obviously, the sensor is a very
important part of the control system. It monitors the process and serves as a signal
source for the control system. In our previous discussion, we always assumed, there was
some suitable measuring device available, but not all measuring devices can be used in
automatic control. The basic requirements for a sensor used in a control loop are the
abilities :

 to indicate the values of measured variables


 to transmit the signals to the controller

The signals could be transmitted through either a electric circuit or a pneumatic


pipeline, therefore, in order to transmit the signal, the sensors must have the ability to
convert the measured variable values into either electric signals or pneumatic signals.

In this section of the course, we will very briefly discuss some kinds of sensors used in
process control. However it is not intended to examine their working mechanism in any
detail. Common variables in Chemical Engineering covered in the following discussion
are pressure, temperature, flowrate, and liquid level. Analytical instruments for chemical
composition measurement are usually specially designed for the specific purpose and
hence are not included.
1.4.2 Pressure Transducers

Many kinds of pressure transducers are widely used in industry for pressure
measurement. Although devices like the manometer and Bourdon tube etc are quite
common, they are not suitable for control purposes due to the difficulties for signal
transmission.

In most pressure transducers, there is a diaphragm to contact the fluids and protect the
measuring setup isolated from the measured fluids, most of which may be corrosive.
Due to the existence of this diaphragm, most pressure transducers can be used as
pressure differential transducers, as long as the second (lower) fluid is introduced into
the other side of the diaphragm. Actually, when measuring the pressure, it is measuring
the pressure difference between the measured pressure and the pressure of atmosphere
guage pressure.

Strain meter (resistance):


In this sensor, under the diaphragm there is a pressure sensitive electric
resistance. It is a spring type of resistance wire. Under pressure, some of the
spring is pressed together causing a short circut and reducing the resistance. The
resistance will be inversely proportional to the strain on it, or the pressure. This
changing resistance may be measured by being included into a electric circuit.
The electric signal, current or voltage, is easily transmitted.
Piezo - electric sensor:
In this sensor, under the diaphragm is a quartz crystal. The working principle is
based on the characteristics of quartz crystal. If the crystal is cut in a special way
and placed between two plates, then the electromotivated force (e.m.f) set up
between the plates will be a measure of the pressure applied to the crystal. This
property of crystal is called piezo-electric effect. By measuring this e.m.f. setup,
the applied pressure can be indicated and transmitted. This technique is mainly
used for higher pressure measurements.

Traditional transducer (air pressure transmitted):


The principle of this system is that the pressure on a diaphragm is arranged to
control the flow of air into, or out of, a chamber on the opposite side of the
diaphragm, until a balance is obtained. The balancing pressure is an indication of
the measured pressure. In this case, the measured signal is transmitted in a
pneumatic circuit through the air pipeline.
1.4.3 Level Measurement

Pressure operated sensor:


As there exists a unique quantitative relationship between the liquid level (head)
and the static pressure at the bottom of the tank, the latter is widely used as an
indication of the liquid level. Thus this is another case of pressure measurement
and pressure transducers discussed above could be used for level measurement.

Float operated sensor:


In this kind of system, there is a float on the surface of liquid. The change in
liquid level will cause movements of this float. By monitoring this movement a
signal of the level is generated and transmitted.

Capacity bridge sensor:


This equipment consists of an electrode, an electronic unit and an indicator (or
transmitter). The electrode is in the form of a long metal rod which reaches from
the top to the bottom of the vessel. The electrode is bare when the liquid is
electrically non conducting, but is sheathed in polymer like polyethlene etc when
the liquid is conducting. The electronic unit is merely a power supply and a highly
stabilized capacitance measuring bridge. One arm of the bridge is formed by the
capacitance between the level sensing electrode and the earth (the vessel wall). A
change in the capacitance owing to the rise and fall of the material around the
electrode produces an out-of-balance current flow from the bridge which is
measured and transmitted.

Optical sensor:
This kind of sensor is based on the difference in the reflecting and transparent
properties of liquid and the gas above it. It takes the form of a light source and a
receiver to the reflected light. When there is no liquid around, the receiver can
detect the reflected light, and this light signal is converted to a electric signal
which can be transmitted. When the sensor is surrounded by liquid, the receiver
can't get the same amount of light reflected. The change in the light received is
then converted into the change in the electric signal which is indicative to the
amount of liquid. Thus to monitor level, you need a number of these kinds of
sensors in series, spread over the whole height of the tank. This signal can also be
transmitted.

1.4.4 Flow Measurement

Differential Pressure Method:


This is still the most commonly used method. Whatever the construction of the
meter, the principle involved is same. The net cross-section area of the stream is
reduced, causing an increase in the velocity, and hence an increase in kinetic
energy. This increase in kinetic energy is obtained at the expense of pressure
energy, so that the pressure of fluid is reduced. By measuring this pressure
reduction, or the pressure differential, the velocity of the fluid can be calculated.
Examples of this kind are orifice plates and Venturi tube. The nature of this
measurement is to measure the pressure differential and then to use pressure
transducers.

Rotating vane meter:


Liquid passing through the meter is directed on to the rings of the vane, and
rotates it at a rate which depends on the velocity of the liquid. This rotation can
be arranged to drive some electric transducers to give out electric signal, like
frequency.

1.4.5 Temperature Measurement

Resistance thermometer and thermistor:


The electrical resistance of metals depends on temperature. By measuring the
changing resistance, the temperature can be determined. The change in
resistance can easily be converted to a electrical signal transmittable. Commonly
used thermometers are made of Platinum or Nickel because they have a stable
and preferable resistance-temperature coefficients.

A thermistor is made of semiconductor, a mixture of metal oxide. Unlike metals,


the semiconductors have a negative resistance coefficient. This is the main
difference between a thermometer and a thermistor.
Thermocouple:
If an electric circuit consists of all metallic conductors and all parts of the circuit
are at the same temperature, there will be no electric force in the circuit, and
hence there is no current. However, if the junctions between two metals are at
different temperature, then there will be an e.m.f. and a current will flow. This
e.m.f is called the thermoelectric e.m.f. , and the junction between the two metals
is a thermocouple. The e.m.f will depend on the temperature difference between
the two junctions. Therefore, when one junction (cold end) is kept at zero
degrees, the e.m.f will indicate the temperature of the heated junction (measured
temperature).

It has been shown the electric circuit for transmitting this signal does not alter the
signal itself, which is indicated by the law of intermediate metals which states: In
a thermoelectric circuit composed of two metals A and B, with junctions at
temperature T1 and T2, the e.m.f is not altered if one or both the junctions are
opened and one or more metals are interposed between A and B, provided that
all the junctions by which the single junction at temperature T1 may be replaced
are kept at T1, and all those by which the junction at temperature T2 may be
replaced are kept at T2.

Silicon semiconductor:
Diodes have an important parameter called pass-required voltage. Below this
voltage, there is no current through the diode. Above this voltage, the diode
allows current to pass through.
The pass required voltage of silicon diodes depend uniquely on temperature, and
thus this voltage signal can be used to indicate temperature. The main advantage
of silicon semiconductor thermometers is that this pass-required voltage has a
temperature coefficient which is essentially the same for all silicon devices of -
2mV/ degC, and this linear change feature is a great advantage for control
porposes.

1.5 Control Actions

1.5.1 The Controller


The controller plays an essential role in the control system. Of the four basic functions of
a control system, (measurement, comparison, computation, and correction) comparison
and computation are solely achieved by the controller. The correction is materialized by
the final control element, but this is done according to the controller's calculation.

The control mechanism in the controller may be considered as consisting of two


sections

 The Comparator
 The Controller

The purpose of the first is to compare the measured and the desired values of the
controlled variable and then compute the difference between them as the error. If there
is no error, i.e. the controlled variable is at the setpoint, then no action is taken.

If an error is detected, the second section of the controller operates to alter the setting
of the final control element in such a way as to minimize the error in the least possible
time with the minimum disturbance to the system. To achieve this objective, different
actions could be taken by the controller and hence different signals are sent to the final
control element.

The remainder of this section is concerned with the different control actions that the
feedback controller could take.
1.5.2 On-off or two-position action

This is the simplest and most commonly experienced type of control. A typical example
is the thermostatically controlled domestic immersion heater. Depending on the
temperature of the water in the tank, the power supply to the heater is either on or off.
In The Virtual Control Laboratory and Case Study Section there is an experiment which
shows on-off control of the temperature in a hot water tank.

 Advantage

Advantage of this type of control action is that it is inexpensive and extremely


simple.

 Disadvantage

The disadvantage mainly lies in the oscillatory nature of the control, which makes
it suitable only for those applications where it can be used alone and close
control is not essential.

Note that it is impractical to use on-off control when trying to regulate a flowrate.
It is necessary to have some sort of capacity available. This can be illustrated by
means of an experiment which uses as a scenario someone taking a shower. Once
again the experiment can be found in The Virtual Control Laboratory and Case
Study Section

Thus the application of on/off control in industry is severely limited. In most


industrial applications continuous control modes, such as those described below
are widely adopted.

1.5.3 Proportional Control Action

Proportional action is the simplest and most commonly encountered of all continuous
control modes. In this type of action, the controller produces an output signal which is
proportional to the error. Hence, the greater the magnitude of the error, the larger is the
corrective action applied.

Mathematical Description

Mathematically, proportional control could be expressed as:


Where

o V is the adjustment or signal for the adjustment from the controller.


o is the error.
o =S-L
o L is the measured value of the controlled variable.
o S is the setpoint.
o K is the proportional constant, named as the gain which shows the
sensitivity of the control.
o Vo is the signal output when no error exists.

The gain is often replaced with another parameter, called the proportional
band, PB. This quantity is defined as the error required to move the final control
element over its whole range and is expressed as a percentage of the total range
of the measured variable. What is the relationship between K and PB.

According to this definition we can see that the whole range of the final control
element adjustment should be Vmin to Vmax.

At point Vmin

At point Vmax

The error required to move from Vmin to Vmax will be

Therefore

Recall that the proportional band, PB, is defined as the error required to move
the final control element over it's whole range expressed as a %. So for the
controlled variable, L, with its total rangeLmin to Lmax the definition for the
proportional band is
or

Therefore we have the relationship between gain K and proportional band PB as

With proportional band, the relationship between the adjustment and the error
can be expressed as

It can be seen both from the expression above and by running the experiments in
the Virtual Laboratory that the larger the gain K, or equivalently the smaller the
proportional band PB, the higher the sensitivity of the controller's actuating
signal to deviations will be.

Dynamic Response

Now let's examine the dynamic response of the proportional control. Assume the
process is at steady state and the level is at the setpoint. At time = 0, an increase
in the inlet flowrate, regarded as a disturbance, enters into the process. If no
control action is taken, i.e. the outlet flowrate is not altered, the level (controlled
variable) will increase.
With proportional control, the level is brought back and maintained in a certain
range near the setpoint. The history curve could typically be like that shown
below. Different responses are obtained depending on the proportional band, B,
of the controller.

As can be seen the smaller the proportional band the closer to the setpoint the
controlled variable becomes but the more oscillatory the response.

Advantages

The advantages of this type of controller are

o It is relativly simple and easy to design and tune


o It provides good stability
o It responds very rapidly
o Dynamically it is relatively stable

Disadvantages

From the response curve to a step change in the input two features should be
noted. These are two points which make proportional control unsatisfactory.

o Offset

For a sustained change of load, the controlled variable is not returned to


the original or desired value, but attains a new equilibrium value termed
control point. The difference between the control point and the desired
value (set point)is referred to as offset.

The reason for this offset with proportional action is that the control action
is proportional to the error. Consider the above simple level control
system. For the step increase in the flow of liquid into the tank, in order to
maintain the level, the valve on the outlet must be opened wider. This will
only occur if there is a continuous output from the controller. The output
itself can only exist if there is an error signal supplied to the controller. In
order to maintain this error, the level will rise above the desired level at the
new control point, hence create an offset.

o Overshoot

There is a significant time of oscillation, or in other words, overshoot.


Although the period of this oscillation is moderate, this, in some cases,
could be highly undesirable.

1.5.4 Integral and Derivative Control

Two other common control techniques are used to eliminate the problems found
when using proportional only control. These are known as integral and derivative
control. Sometimes they are used individually but more often they are combined
with proportional control. A more mathematical definition can be found in a later
section.
Integral Action

Integral action gives an output which is proportional to the time integral


of the error. It is also called reset control. It is possible to use integral
action itself, but this is not a common situation.

 P and I Control

Integral action is generally applied with proportional control, yielding so-


called proportional and integral control (P+I). This combination is
favorable in that some of the advantages of both types of control action
are available.

The main advantage of P+I is that it can eliminate the offset in


proportional control.

The disadvantages of P+I are that it gives rise to a higher maximum


deviation, a longer response time and a longer period of oscillation than
with proportional action alone. This type of control action is therefore
used where the above can be tolerated and offset is undesirable.

Derivative Action

This kind of action gives an output which is proportional to the derivative


or the rate of change of the error. It is also known as rate control. This kind
of action could not be used alone in practice. This is because its output is
only related to the rate of change of the error. The error could be huge,
but if it were unchanging, the controller would not give any output. Thus
although it is theoretically possible, it is practically impossible.

 P and D Control

Derivative control is usually found in combination with proportional


control, to form so-called P+D. By adding the derivative action, lead is
added in the controller to compensate for lag around the loop, and so
P+D can eliminate excessive oscillations. A disadvantage is that it can not
eliminate the offset although somehow it makes it smaller.
Proportional, Integral and Derivative Action

In applications, sometimes the above three action are combined together


to set up the proportional plus integral plus derivative action, i.e, P+I+D.

This combined action is able to:

 eliminate the offset due to the existence of integral action


 reduce the maximum deviation and time of oscillation, which is a
compromise between the advantage and disadvantage of P+I and
P+D

1.5.5 Examples

The aim of this module was to introduce the various types of control action. In
the Case Study Section and Virtual Control Laboratory there is a level control
experiment which allows the user to practice what they have learned in this
section. i.e. it is possible to vary the deadzone on an on-off controller, see the
effect of altering the proportional band etc.

1.6 Variations on Basic Feedback Control

1.6.1 Feedforward Control


In this configuration, a sensor or measuring device is used to directly measure the
disturbance as it enters the process and the sensor transmits this information to the
feedforward controller. The feedforward controller determines the needed change in the
manipulated variable, so that, when the effect of the disturbance is combined with the
effect of the change in the manipulated variable, there will be no change in the
controlled variable at all. The controlled variable is always kept at its setpoint and hence
disturbances have no efffect on the process. This perfect compensation is a difficult goal
to obtain. It is , however, the objective for which feedforward control is structural. A
typical feedforward control loop is shown in the figure below.
Another name for feedforward control is open loop control. The reason is that the
measured signal goes to the controller parallely to the process. This can be seen in the
next figure. This is in contrast to feedback or closed loop control.

As mentioned previously the main advantage of feedforward control is that it works to


prevent errors from occuring and disturbances have no effect on the process at all.
However, there are some significant difficulties.

 Complex Computation

The feedforward control computation involves determining exactly how much


change in manipulated variable is required for a specific change in disturbance.
To be able to make this computation accurately requires significant quantitative
understanding of the process and its operation. There is also a tremendous
escalation of the theoretical know-how required in the feedforward controller's
computation activities.
 Knowledge of Process

The structure of feed forward control assumes that

1. The disturbances are known in advance.


2. The disturbance will have sensors associated with them (measurable).
3. There will not be significant unmeasured disturbances.

These limitations on the disturbances constrains the application of feedforward


control, simply as most disturbances in the industrial processes are unpredictable
and unmeasurable.

 Limitations

In pure feedforward control, there is no monitoring on the controlled variable. If


the controlled variable strays from its setpoint there is no corrective action to
eliminate the error. This makes pure feedforward control somewhat impractical
and a rarity in typical process application.

 Specific Controller Required

The feedforward controller must be specifically and uniquely designed for the
one particular control application involved, because of the necessity of accurate
and quantitative calculations.

It can be seen that feedforward control requires a significant increase in technical skills
and capabilities. As a result, feedforward control of specific variables is limited to the
most economically significant cases. In practical industrial application, only few cases are
handle with feedforward control. While the number of application is small, their
importance is quite significant.

1.6.2 Cascade Control


The second alternative to simple feedback control is cascade control. In this setup there
is

 one manipulated variable


 more than one measured variable

An inner and outer control loop are formed each with an individual feedback controller.
The outer loop controller is also known as the master or primary controller.
 The input to this controller is the measured value of the variable to be controlled.
 The setpoint is supplied by the operator.
 It passes its output signal to the inner control loop.

The inner loop controller is known as the slave or secondary controller.

 It measures a second variable whose value affects the controlled variable.


 The setpoint is supplied by the output from the outer loop.
 Its output signal is used as the signal to the manipulated variable.

The above points can be shown clearly in a diagram.

The major benefit from using cascade control is that disturbances arising within the
secondary loop are corrected by the secondary controller before they can affect the
value of the primary controlled output. Cascade control is especially effective if the inner
loop is much faster than the outer loop and if the main disturbances affect the inner
loop first.

Below are described examples of cascade control in practise. It should be noted that in
two of the three examples, the secondary loop is used to compensate for flowrate
changes. In process systems this is generally the case.
Example 1 - Reactor Temperature Control

In this example the aim is to keep T2 at its setpoint. The primary control loop detects
and eliminates changes in T1, the temperature of the reactants. The secondary control
loop detects changes in the temperature of the cooling water. Hence it can adjust the
flow accordingly before the effects are detected by the primary control loop. If there
was no second controller the effect of the cooling water would take a long time to
materalise and hence eliminated.

Example 2 - Distillation Bottoms Temperature Control


In this example the primary loop detects changes in the temperature brought about by
changes in composition, pressure, etc. The secondary loop detects changes in the steam
flowrate and hence eliminates anticipated effects on the temperature.

Example 3 - Heat Exchanger Temperature Control

This is similar to example 2. The aim is to keep T2 constant. Again the secondary loop is
used to compensate for flowrate changes.

1.6.3 Split Range Control


The final alternative to simple feedback control to be discussed in this section is Split-
Range Control. This is distinguished by the fact that it has

 one measurement only (the controlled variable)


 more than one manipulated variable

The control signal is split into several parts each associated with one of the manipulated
variables. A single process is controlled by coordinating the actions of several
manipulated variables, all of which have the same effect on the controlled output.

Below are described two situations where split-range control is used in chemical
processes.
Example 1 - Control of Pressure in a Reactor

The aim of this loop is to control the pressure in the reactor. It may be possible to
operate this system with only one of the valves but the second valve is added to provide
additional safety and operational optimality.

In this case the action of the two valves should be coordinated. Thus for example if the
operating pressure is between 0.5 and 1.5 bar then the control algorithm could be

 If the pressure is below 0.5 bar then valve 1 is completely open and 2 is
completely closed.
 If the pressure is between 0.5 and 1 bar then valve 1 is completely open while 2 is
opened continuously as the pressure rises. Note that both these actions lead to a
reduction in pressure.
 If there is a large increase in pressure and it rises to above 1 bar then valve 2 is
completely open while 1 is closed continuously.
 If the pressure reaches 1.5 bar then valve 1 is shut and 2 is open.

A graph of these valve positions with respect to pressure is shown below.


Example 2 - Control of Pressure in a Steam Header

The aim of this control loop is to maintain a constant pressure in the steam
header subject to differing demands for steam further downstream. In this case
the signal is split and the steam flow from every boiler is manipulated. An
alternative manipulated variable could be the steam production rate at each
boiler via the firing rate. A similar control scheme to the above could be
developed for the pressure control of a common discharge or suction header for
N parallel compressors.

1.7 Case Study Section

The aim of this case study section is to reinforce material taught in the lecture
notes. There are two sections as outlined below. The first is a Laboratory Session.
There are links to various relevent experiments found in the Virtual Control
Laboratory. The second includes tutorial questions and answers covering control
loops and simple processes.

1.7.1 Laboratory Session

There are a number of experiments in the Virtual Control Laboratory which


involves work covered in this section. In particular please work through the
following
1.7.2 Hot Water Tank

1.7.3 Level Control Experiment

1.7.4 Tutorial Questions and Answers

1.7.5 Exercise 1: Simple Control Loops

1.7.6 Exercise 2: Simple Processes

2 Further Developments

The four modules in this part of the course cover qualitative issues in the development
of control system structures for simple processes, quantitative issues in the design of
simple single loop feedback controllers and writing models for chemical processes.

 Module 2.1: Controlling Simple Processes


 Module 2.2: Degrees of Freedom
 Module 2.3: Controlling Separation Processes
 Module 2.4: Controller Design and System Modelling
 Case Study Section

2.1 Controlling Simple Processes

This first module introduces a hierarchical method of placing control loops onto real
processes.

 Introduction
 Degrees of Freedom
o Example 1: Vapouriser Problem (1)
o Example 2: Mixing of Two Streams
 Hierarchical Decomposition
o Controlling a Real Process
o Vapouriser Problem (2)
2.1.1 Introduction

In all the examples so far discussed it has been assumed that we know at the outset
what quantity is to be measured and thus regulated, and what will be the
corresponding adjustment. This information will in fact be readily available only for
the simplest of cases.

As soon as we consider the control of any sort of process, or even a modestly complex
piece of equipment, we are faced with the need to provide severalcontrol loops. The
result of this is to create a range of choices. In this section, we introduce ideas to help
resolve this choice systematically. These ideas will enable us to design complete
control systems for large and complex processes, although they will first be
introduced in the context of simple examples.

2.1.2 Degrees of Freedom


When faced with the task of devising a control scheme for a process it is necessary to
know how many of the process variables am I entitled to attempt to regulate. By
process variables we mean temperatures, pressures, compositions, flowrates or
component flowrates. The answer arrived at is known as thenumber of degrees of
freedom. The degrees of freedom of a process are here defined as the number of
process variables which can be set by the designer, operator or control system ie

 Temperatures
 Pressures
 Compositions
 Flowrates (component or total)

In our case we are concerned with the Control Degrees of Freedom which will be the
number of the above types of process variable which may be set once non-adjustable
design variables, such as vessel dimensions or number of trays, have been fixed.

In this context the number of degrees of freedom thus corresponds strictly to the
number of manipulated variables which may be used in control loops. Note that this is
also the number of single-input-single-output control loops and of regulated variables
in the loops.
2.1.3 Example 1: Vapouriser Problem (1)

To illustrate the nature of the problem, consider how a process unit which vaporises a
liquid feed stream might be controlled.

In this device the quantities which we might choose to regulate include:

 Feed rate
 Product rate
 Operating pressure
 Operating temperature
 Liquid level

The first question to be resolved is which and how many of these can legitimately be
regulated independently?

Similarly, what adjustments may be made in order to regulate the chosen quantities?
There appear to be three candidates for streams on which control valves might be
located, namely:

 Liquid feed
 Vapour product
 Steam supply

Suppose that liquid level is chosen as one of the regulated quantities. Which of these
three possible adjustments should be paired with this measurement to complete the
control loop?

We will return to this particular example after addressing individually the problems
noted above. In summary these are:

 How many, and which, quantities can be measured and regulated?


 How many, and which, quantities can be adjusted?
 Which measurement should be paired with which adjustment?
Before proceeding to this it is worth noting the following points about the vaporiser
example, to illustrate that these questions can indeed be answered using our
knowledge of the process.

 Since there are only three potential adjustments, there cannot be more than
three control loops, and hence no more than three regulated quantities.
 Physics and thermodynamics tell us that certain variables in this problem
cannot be set independently, and thus cannot be regulated in separate control
loops. Unacceptable combinations here are:
o Inlet and outlet flows, which must be the same by conservation, and
o Temperature and pressure, which are related in a single component
two phase system.
 This leaves us with three possible control loops, in which the three
adjustments regulate:
o Temperature or pressure,
o Feed rate or product rate,
o Liquid level or holdup.
 The choice of pairing remains, to some extent, as a genuine choice between
alternatives which must each be evaluated. Here any of the three possible
adustments can be seen to affect both temperature (or pressure) and holdup,
two (vapour and steam valves) can affect vapour rate, and only one can affect
feed rate.

2.1.4 Example 2: Mixing of Two Streams

Consider the exceptionally simple process shown below in figure (a), where two feed
streams are mixed together to produce a single product stream. Suppose that what is
required is that the two feed streams shall have individually specified flowrates. A
suitable control system for this would be as shown in figure (b).
It should be immediately apparent that this control system is complete, i.e. we
cannot put any more control loops on it. Having fixed two of the three streams which
are connected together, conservation requires that the third must be the sum of
these two.

This suggest the validity of the following rule.

Rule 1: `(n-1) out of n'. If n streams join together in a process or part of a process
over which mass must be conserved (normally any process), then the flows of only
(n-1) of these may be set by flow controllers.

It is possible to prove this formally. As will be seen, further generalisation is also


possible.

Conservation of mass requires that the sum of the inflow and outflows shall match
over an extended period, but is it necessary to take steps to ensure that this happens
from minute to minute? The flows would not always match if there were any
possibility of material accumulating within the junction. This will not occur if the
fluids are incompressible. This implies that it is necessary to have some mechanism to
ensure that mass balances do actually balance. In this example, the design of the
process, i.e. simple closed junction and incompressible fluid, ensures that this will be
so.

However, consider what happens if we replace the closed junction by an open tank,
see below figure (a). Here there is nothing to stop the tank from running dry or
overflowing, unless, as in figure (b), we provide the tank with a level controller.
Rule 2: Mass balances must balance. To ensure that mass balances do balance,
there must either be an inplicit mechanism in the process, or an inventory controller
must be supplied. The valve for the holdup control loop goes on the remaining
stream.

Inventory or holdup is measured either by level or some equivalent measurement in a


liquid system, or by pressure in a gas system. A simple junction with a liquid system
is rather special in being `self regulating' with respect to holdup.

If we wished to regulate the total product rate from a simple mixing process and one
only of the feeds, then we may require explicit holdup control for either compressible
or incompressible fluids in either of the arrangements below. The vessel in the right
hand figure is a closed vessel, which, in the case of a liquid system, would be run full.
Whether or not a control loop for pressure is required will depend on the specific
process conditions, in particular the source and sink pressures for the flows and the
type of device, pump, compressor etc., if any, driving the flow.
Finally, there is a further rule implied by all these examples which really belongs
before any of the others:

Rule 3: Strategic aims. The primary control objectives of a process are set by the
strategic aims of the process. These define the basic control stucture.

Thus the fact that flow controllers were placed on the two feed streams in the first
example and on the product and one feed in the last was a consequence of a decision
by the process designer that these were the streams whose flows were to be fixed. It
should thus be clear that the design of a process and of its control system cannot really
be separated.

2.1.5 Hierarchical Decomposition


In the above examples it was simple enough to look at the process or single piece of
equipment and see how many degrees of freedom there were and how the control
loops would interact. What happens, however, when it is a complete process that is
to be controlled with many loops.

In this section of the course we shall introduce a hierarchical procedure for


developing the design of a process control system. This approach will be seen to have
a number of advantages. Firstly, it provides a systematic approach to resolving what
can otherwise seem to be a complex and unstructured problem. Secondly, it enables us
to concentrate on individual parts of the problem, rather than trying to do several
things at once. Finally, it corresponds to a standard systematic approach to
designing processes, enabling us to evolve the design of the process and its control
system together.
The process and control scheme can be looked in to at differing levels of complexity:

 Input-Output Stage: Strategic Decisions


 Functional Level Stage: Further Details
 Separation Stage: Final Details

There now follows two examples of this hierarchical approach in use. Firstly there is a
simple process and then we return to the vapouriser problem introduced earlier.

2.1.6 Controlling a Real Process

Consider a rather more sophisticated mixing process.

Process description
(i): The aim of the process is to deliver a fixed amount of product, made by blending
together two streams of two constituents, a concentrate and a diluent, and to supply
this product to a specified composition.

In this example we shall introduce a hierarchical procedure for developing the design of
a process control system. This approach will be seen to have a number of advantages.
Firstly, it provides a systematic approach to resolving what can otherwise seem to be a
complex and unstructured problem. Secondly, it enables us to concentrate on individual
parts of the problem, rather than trying to do several things at once. Finally, it
corresponds to a standard systematic approach to designing processes, enabling us to
evolve the design of the process and its control system together.

Input-Output Stage: Strategic Decisions

Starting from the above statement of the process requirements, viz specified product
rate and composition, without reference to any detail of the process itself, other than
the input and output streams, we can define immediately a part of the control system
structure, as shown below. Here the, unspecified, process is shown as a box. It is clear
that the product stream will require flow control and that that can be implemented as
shown. It is also clear that composition measurement and some sort of composition or
quality control loop will be required. Half of this loop can be immediately defined, and is
also shown.
Still without any detailed knowledge of the contents of the PROCESS box, consider how
the composition control loop might be implemented. What can be adjusted to cause the
composition of the product stream to change? Clearly, it will be necessary to
manipulate either the amount of diluent or the amount of concentrate. These lead to
two alternative structures shown below.

Which of these is the better structure? Without detailed and quantitative information
about the process in the box, it is not possible to decide. This is a common situation in
engineering design. Ultimately, it may well be necessary to explore both alternatives,
and make a decision on the basis of some measure of overall system performance. The
designer could proceed with both alternatives in parallel. Unfortunately, this is almost
certainly only the first of many points where alternatives arise, and very soon the `tree'
of possible designs will become intractably large. Unless the whole design procedure is
automated and carried out by a very powerful computer, this is not a realistic approach.

The following heuristic approach, based on the ideas of Douglas is recommended.

 Based on information currently available, choose the more promising alternative


using rules-of-thumb. This might mean making an arbitrary choice.
 Note this point in the design as on at which a choice was made.
 Proceed with the chosen alternative until either:
o it is found to be unsatisfactory, or
o the design is complete.
 If the design was unsatisfactory, backtrack to the decision point and explore the
other branch.
 If the design was completed, then evaluate it and:
o if satisfied that the design is the best available, or good enough, finish, or
o if not entirely satisfied, backtrack and explore the other branch.

This algorithm is a rather general one for any kind of design.

Here we can apply the following heuristic or rule-of-thumb.

Heuristic : Small streams. Manipulate small streams rather than large ones in
important control loops.

This has a number of justifications. Firstly, small valves are cheaper than large ones, so it
may be possible to save money. Secondly, small valves can be manipulated more quickly
and precisely than large valves, and so a control loop with a smaller valve will often work
better.

Clearly, the concentrate stream will be a smaller one than the diluent, and so we will
choose to follow up the right hand alternative where this is the manipulated variable for
the composition loop.

Examination of the flowsheet shows that we have control valves on two out of the three
streams associated with the process. Since these flows have been set, one to a specific
flow and another to ensure that a particular product composition is achieved, the flow
of the third stream cannot now be chosen independently, it must match these two flows
to ensure that the mass balance is maintained. In fact our `(n-1)' rule, and its corrollary,
can be generalised to cover any type of controller with valves on (n-1) out of n streams.

Rule 1a : Generalised `(n-1) out of n'. If n streams join together in a process or part of
a process over which mass must be conserved (normallyany process), then the flows of
only (n-1) of these may be set by control loops other than one regulating inventory
within the process or part process.

Without a knowledge of precisely the type of process element on the box we cannot
completely define any control loop associated with inventory or holdup regulation, but
we do know that we cannot put a control valve on the remaining stream for any
purpose other than inventory regulation. We will indicate this on the flowsheet as shown
below. The shaded `valve' implies that no other valve may be put on this line. The
square, rather than round `controller' indicates that some mechanism, not necessarily an
actual controller, will regulate inventory, which might, if measured, be a level, mass
holdup or pressure.

Functional level: further details

The steps which were followed above illustrate that it is sometimes possible to design a
significant part of the control system for a process by reference to:

 The objectives of the process with respect to amount and composition of


products.
 The basic structure of the process.

It will not in general be possible to determine the whole control system with just this
information. Further steps in developing this involving `opening up' the box labelled
PROCESS in this example into succesive levels of increasing detail. This approach will be
explored in later examples. For more complex processes there will be several levels. The
purpose of this hierarchical approach is to help the designer to concentrate on the
decisions that can be taken at each stage, by presenting details of the process in
sequence rather than all at once. This makes the design task easier both by reducing the
amount of new information presented at one time, and by allowing some earlier
decisions to be finalised and thus removed from the list of tasks still to be tackled. A
glance ahead at a complete process flowsheet will enable the reader to appreciate how
daunting a task placing the control loops on a flowsheet might be if this approach
is not adopted.
This example may however be completed in one further level by providing some more
details of the process.

Process description, (ii): The process equipment consists of an open mixing tank,
followed by an in-line static mixer (a section of pipe with internal vanes or baffles) and a
second mixing tank.

The PROCESS box to include this, and with the control system so far defined, is shown
below.

Looking at this more detailed structure, the questions to ask, in order, are:

1. Can any partial control loops now be completed?


2. Look at `new' streams created by expanding the structure: what stategic control
requirements arise?
3. having identified strategic controls, can any inventory controls be identified using
the `(n-1)' and mass balance rules?

In answering question 2, it is important to realise that in any process, streams do not


simply happen! They are their for a reason, and that reason usually defines what will
determines their flow.

Refering to process, the incomplete inventory loop cannot be unambiguously


completed. Intuition suggests that it should probably regulate the level in T1, and this
will indeed prove to be the case, but we cannot be certain of this yet.

The two new streams, from T1 to M1 and from M1 to T2, have no obvious strategic
requirements for regulation of flow or composition.
However, T2 has two streams, one of which has its flow set by the product flow control
loop. The tank is not self regulating, and so an level controller must be placed with a
valve on the feed stream as shown.

It now becomes clear that mixer M1 requires inventory regulation, having one of its two
streams now set, which would imply a valve on its input stream from M1. However, it is
self regulating, being essentially just a piece of pipe, but this still means that no valve
can be placed there for any other reason. It is now clear that the measurement end of
the original level control loop must be in T1.

All streams being accounted for by having explicit control valves or implicit regulatory
mechanisms determining their flows, the control scheme is now complete.

2.1.7 Review: What did we do?

This was a very simple process. However, there were potentially a significant number of
alternative process structure, not all of which would have worked. We applied a logical
procedure, each of whose steps could be justified with reference either to a knowledge
of the process or the rule which have been proposed, and ended up with a complete,
and workable, control system with four loops.

Before proceeding to more complex examples, let us review how the three questions set
out at the beginning of the section were answered.

How many and which quantities to measure?

The question of `how many?' was not posed or answered explicitly in this example. As
will be seen later, it is sometimes convenient to do so. However in this case it was
subsumed in the question of `which?'.
The question of `which quantities to measure' was answered in two ways. Firstly by
reference to the Strategic aims of the process rule. This established the outlet flow and
concentration as regulated quantities. Further strategic regulated variables, other than
inventories can often be established by by reference to identified adjustable variables,
but in this process there were no others.

Secondary regulated variables are usually inventories and are identified by the (n-1) and
mass balance rules.

How many and which quantities to adjust?

It is clear that there cannot be more adjusted variables than there are streams whose
flows can be manipulated independently. These were all identified in this process by the
requirement for inventory regulation. In general we can use the following Rule both to
identify adjustments and to check the final control system structure.

Rule: flows do not just happen. Stream flows in a process do not just happen. Either a
valve or a mechanism (such as continuity) must set the flow of every stream.

Which measurements and adjustments are paired?

Here there will almost invariably be alternatives. To identify these and help choose
between them we have one firm Rule, and some guiding heuristics.

Rule: Cause-and-effect. An adjustment chosen to pair with a measurement must have


an effect on that measurement.

This is rather obvious. It is nonetheless given a name in the control literature where it is
called structural controllability.

A number of heuristics serve to aid choice. One has been given, the Small streams
heuristic. Here are two more.

Heuristic: immediate response. Prefer pairings in which the measurement responds


immediately, rapdicly and unambiguosly to the adjustment.

Their are a number of important quantitative elements embodied in this heuristic. These
are dealt with in detail elsewhere.

Heuristic: noninteraction. An ideal adjustment should affect its paired measurment


and no other measurements.
2.1.8 Vapouriser Problem (2)

We now have sufficient understanding of how to develop whole process control


schemes to tackle the vaporiser problem.

Input-Output Stage: Stategic Decisions

Redrawing this `process' as an input-output block yields the structure shown. In the
block we have distinguished two sub-blocks, noticing that while both feed and steam
enter the process they are subject to separate material balances.

The objectives of this process will be taken to be:

1. deliver a specified quantity of vapour,


2. at specified temperature and pressure.

Objective 1. suggests that we should place a flow controller on the vapour outlet as
shown in figure (a). Noting that the process fluid side of the block has only one input
and one output, the feed stream may be `blocked' for any purpose other than inventory
regulation as shown by the shaded valve.
Figure (a) fulfills the stated objective of delivering a specified rate of vapour product.
However, because there is only one input and one output to the process side, so would
the structure of figure (b). The first structure is prefered according to the following
Heuristic.

Heuristic: close adjustment. It is usually better to make an adjustment as close as


possible to the measurement with which it is paired.

Also because the sub-block has only two streams, the scheme in figure (c), which
regulates the feed flow would also serve to maintain the product rate once inventory
regulation was provided. This is clearly a less direct way of performing the specified task,
and depends of the satisfactory operation of the inventory control system. It is therefore
avoided on the basis of a further Heuristic.

Heuristic: direct action. Prefer the the structure which manipulates the regulated
quantity most directly. In particular, avoid arrangements which depend on the
satisfactory operation of additional control loops.

The final structure, figure (d) violates both of the above guidelines.

Objective 2. explicitly states that the vapour temperature shall be regulated, as shown
below, (a). Since all streams on the process side are set when inventory regulation is
added, only the steam rate can be adjusted. If this is done on the steam supply, then the
condensate outflow must be adjusted to maintain the steam side material balance,
figure (b).

This control system must now be complete, as there are no more possible adjustments.
The designer cannot therfore be tempted to try and regulate the pressure of the system.
The Phase Rule could also have been invoked to check this.

N=C-P+2

N is the nuber of extensive thermodynamic variables which may be chosen or set, C is


the number of components and P the number of phases. Here P = 2, there being both
vapour and liquid present and so:

n=C

However, the feed composition is fixed and because of the requirements for material
balance, the vapour will have this same composition. Specifying the composition on
a C component stream fixes (C-1) concentrations, leaving one intensive variable which
can be fixed by a control system. This can be eithertemperature or pressure, but clearly
not both independently.

Rule: Phase Rule. The Phase Rules still says:

N=C-P+2

Completing the Example

The control system structure has been completely defined at the input-output level. This
is rather unusual, but it is now very easy to open up the block and turn the conceptual
controllers into `real' ones, see below figure (a). Note the special form of inventory
regulation on the steam heating side using a steam trap, essentially a very small vessel
with an internal level control system to allow steam and condensate to be disengaged.
Just to show that no heuristics are totally reliable, figure (b) shows another version of
the control system which would be acceptable in many circumstances. This breaks the
Direct action heuristic by regulating the feed rather than the product. However, it saves
a control loop by having a `self regulating' material balance through the following
mechanism.

Heat transfer from the coil to the fluid happens only in the liquid phase. If too much
vapour leaves the vessel, the liquid level falls, uncovering some of the steam tubes. Thus
the heat transfer area falls, reducing the rate of heat transfer and the rate of
vaporisation. Normal operation will be with the tube bundle partly uncovered.

This arrangement will only be acceptable if this is allowable, e.g. with relatively low
temperature steam. Another problem could be that there will always be some liquid
boiling dry on the exposed tube surface. This could tend to degrade and build up
deposits.

Finally, to show that, in this system, temperature and pressure regulation are equivalent,
the temperature control loop on the steam has been replaced by a pressure controller.

2.1.9 Rigorous Identification of Number of Regulated Quantities

It is possible to determine the number of variables in a process which it is permisable to


regulate by a formal mathematical procedure. In the previous example it this was not
necessary, as identification of the number of available adjustments served to define this.
However, it will sometimes be the case that two or more potential adjustments have the
same effect and are alternatives in the sense that if one is used, the other cannot be. It is
also possible, in principle, to devise a process which is uncontrollable because it has
more variables to regulate than there are adjustments available.
The mathematical technique is called degrees of freedom analysis. It can be carried out in
a number of ways, but essentially it consists of counting the number of equations
required to describe the process or part process under consideration, and the number of
variables which appear in these equations. The excess of variables over equations
represents the number of quantities which may be set to arbitrary values, for example by
a control system.

2.2 More on Degrees of Freedom

This second module introduces the idea of degrees of freedom i.e. how to determine
the number of control loops to place on a process.

The approach used is a formal mathematical one which those who are only interested in
practical control can omit. Instead they can read this informal summary. (table below)

 Introduction
 Example 1 - Simple Blender
 Alternative Equation
 Example 2 - Adiabatic Flash
 Example 3 - Total Condenser
 Example 4 - Countercurrent Cascade
 Application to Complete Processes
 Inventory

2.2.1 Introduction
When faced with the task of devising a control scheme for a process it is necessary to
know how many of the process variables am I entitled to attempt to regulate. By process
variables we mean temperatures, pressures, compositions, flowrates or component
flowrates. The answer arrived at is known as the number of degrees of freedom.

What follows is a summary of a method for analysis of the number of degrees of


freedom which is described in:

Ponton JW, 1994, Degrees of Freedom Analysis in Process Control, Chemical Engineering
Science, Vol. 49, No. 13, pp 1089 - 1095.

In fact it is possible to understand the key concepts from a much simpler viewpoint. This
is described here. (in table below)

When faced with the task of devising a control scheme for a process it is necessary to
know how many of the process variables am I entitled to attempt to regulate. By process
variables we mean temperatures, pressures, compositions, flowrates or component
flowrates. The answer arrived at is known as the number of degrees of freedom.

This particular problem has exercised the minds of quite a few clever people over the
years. It is non-trivial if we start (as everyone did) by considering all
possible measurements. This is because not everything that can be measured can be
regulated independently. For example, if a piece of equipment has a single input input
and a single output, only one of these can be flow controlled, since a mass balance
about the unit must be maintained.

However, it is much easier to understand if we consider the number of quantities we


can adjust. Since in practice:

 all adjustments on a chemical plant are made by changing the flow of a stream
by means of a valve, and
 only one valve can be placed on any one pipe,

then the maximum number of adjustable quantities is the same as the number of
streams on the process flowsheet. Hence the following are all the same:

 Number of streams
 Maximum number of adjustments
 Maximum number of control loops
 Maximum number of meaurements

This is the maximum number since some material balances will balance automatically,
e.g. a simple mixing tee. In this case we do not need a control loop and valve are not
required on the stream and must not be put there for any other purpose.

However we must check that there is either an explicit (control loop) or implicit (self
regulating) mechanism to ensure that all mas balances balance about each processing
unit.

Where there are two or more phases in a unit, e.g. a liquid and a vapour, we must
maintain an interface between them, which is equivalent to maintaining a mass balance
on each phase. this will normally require an interface level control system, although
some devices, such as distillation trays, are designed so that a hydaulic balance can
maintain the interface over a limited range of operation.

When counting streams, it is convenient to count utilities such as cooling water and
steam, as a single stream, although they both enter and leave a unit. An incompressible
utility stream will have a self-regulating mass balance, and steam heating coils and
jackets are usually equipped with a steam trap, a self contained device which maintains
a steam-condensate interface independent of other control systems.

Counting Degrees of Freedom


The fuller treatment of this concept describes how to count the degrees of freedom for
both individual units and complete processes. This procedure is a means to an end, to
enable the designer to determine how many control systems must be provided.

In practice, it is better to work as follows. It is useful to determine both the total number
of control loops (total count T) and the number of loops associated with regulating
`strategic' quantities, i.e. other than inventories, interfaces and levels (strategic count, S).

 Count the total number of streams in the process; this is the upper limit on the
number of control systems required. This is the initial value of both T and S.
Now look at each unit in turn.
 Each will require at least one material balance:
- deduct 1 from S
 Note those which will have self-regulating mass balances:
- deduct one from T
 Note where an interface level has to be maintained:
- deduct one from S

Finally:

 S is the number of required non-inventory loops


 T - S is the number of explicit inventoru loops

Definitions
Before we start analysing the degrees of freedom it is perhaps wise to define a few
terms.

Degrees of Freedom of a process

The number of process variables which can be set by the designer, operator or
control system.
Control Degrees of Freedom

 The number of the above types of process variable which may be set once
non-adjustable design variables have been fixed.
 The number of manipulated variables which may be used in control loops.
 The number of SISO control loops.
 The number of regulated variables in the control loops.

The procedure which will be described is a counting process which identifies potential
manipulations, but does not directly identify variables which may be regulated.

Derivation
The technique can be derived by applying the Kwauk method. This was first developed
by Kwauk [1952] and later by Smith [1963]. The equation to be solved is

Degrees of Freedom = Unknowns - Equations

Unknowns

There are two types of streams - process streams and energy streams. If we take a
process stream containing C components then the unknowns are C flowrates,
temperature and pressure. An energy stream has 1 unknown associated with it.

Equations

The equations are best thought of as types of equations i.e. component balances,
energy balances, equilibrium equations, equivalent T and P etc.

Degrees of Freedom

Once the final number of D.O.F. have been found then it is necessary to decide which of
these are already fixed and which can be controlled.

2.2.2 Example 1 - Simple Blender


Let us take, for our first example, a simple blending process. Assume that the mixing
takes place in the vapour phase. There two inputs and one output as shown in the figure
below.
Figure 1 - Simple Blender

It is possible to apply the above equation to evaluate the number of degrees of freedom

Unknowns

3 streams with C+2 unknowns = 3C + 6 Unknowns

Equations

 C Mass Balance Equations


 1 Energy Balance Equation
 C+1 Equations

Degrees of Freedom

D.O.F = 2C + 5

However in this system we know the composition, temperature and pressure of the two
input streams. This adds on a further

2C + 2 Constraints

Hence there are 3 D.O.F and so we can fix the output composition, pressure and
throughput.

A typical control scheme is shown below.


Figure 2 - Simple Blender with Control

2.2.3 Alternative Equation


The question now is Is there a relationship between the control degrees of freedom, the
number of streams and the number of phases present. In the paper it is proved
theoretically that there is such a relationship.

It is possible to

 Number all the streams associated with the unit


 Count all the phases
 Determine the number of extra phases ie the (number of phases - 1)
 This number is also the number of interfaces
 Subtract interface count from stream count giving answer

Hence for the example discussed earlier

Blender

 No. of Streams = 3
 No. of Extra Phases (interfaces) = 0
 CDF = 3 - 0 = 3 - as before
This equation is extremely simple to use and discards the need to think about unknowns
and equations.

2.2.4 Example 2 - Adiabatic Flash


The next example we will look at is the adiabatic flash. Note that we will be returning to
this example in a later section. A diagram is shown below.

As you can see in this example there are:

 No. of Streams = 3
 No. of Extra Phases (interfaces) = 1
 CDF = 3 - 1 = 2

Therefore it is possible to control two strategic variables in an adiabatic flash eg feed


flowrate and pressure, leaving the third stream free to determine the inventory.

2.2.5 Example 3 - Total Condenser


The third example of this technique is a total condenser.
This time there are:

 No. of Streams = 3
 No. of Extra Phases (interfaces) = 1
 CDF = 3 - 1 = 2

Again it is possible to control two strategic variables.

2.2.6 Example 4 - Countercurrent Cascade

A more complex unit that we can look at is a countercurrent cascade.

Figure 3 - Countercurrent Cascade

First consider a single countercurrent equilibrium stage:


 No. of Streams = 4
 No. of Extra Phases (interfaces) = 1
 CDF = 4 - 1 = 3

A stack of N such units built into a cascade will have 2N + 2 streams, N two phase
elements and thus N + 2 apparant degrees of freedom. These would never all be used
in practise however, but it is, in principle, possible to maintain each stage at a different
pressure by some valve-like arrangement between the trays.

If there is a fixed rather than adjustable restriction between trays then a potential degree
of freedom is lost for each of N-1 vapour interstreams, so the practical degrees of
freedom for this device is [N + 2] - [N - 1] = 3.

This is the same number of degrees of freedom as would be calculated by simplistically


counting the streams to the complete cascade, 4, and subtracting 1 for the presence of
two phases in the device, again giving 3 Control Degrees of Freedom

2.2.7 Application To Complete Processes


Determination of the total degrees of freedom for a complete process is now trivial.
Connected units sharing a stream lose one degree of freedom from the sum of those for
the individual units. Thus the stream related degrees of freedom is simply equal to the
total number of material and energy streams in the process.

The degrees of freedom for the complete process may be determined by either of two
equivalent procedures.

1. Using the approach above determine the degrees of freedom for each unit. Sum
these and then subtract the number of shared streams to obtain the final count.
2. Count all the streams in the process. Separately count the total number
of extra phases i.e. add up all occurances of phases greater than one in all units.

Method 2 is shown in the example below. All streams represent potential degrees of
freedom and possible adjustable variables, but beside each unit is written the number
lost as a result of the presence of multiple phases in the unit.
Figure 4 - Absorption Process with Solvent Recovery Flash Separator

 Total Streams = 12
 Extra Phases = 3
 Total D.O.F = 9

2.2.8 Inventory
In the preceeding analysis it has been assumed that

1. Material balances will always be somehow balanced.


2. An adjustment which affects only inventory is not as such a control degree of
freedom.

An alternative view of 2 is that inventory control should be regarded as a taking up a


degree of freedom. Under this assumption the number of degrees of freedom will
always equal the number the material and energy streams. Thus for the countercurrent
cascade there would be 4 degrees of freedom rather than 3, the additional one being
the inventory of the liquid phase. This alternative approach does have the advantage of
identifying all the potential control loops, explicit or implicit, which will exist in a
process.

It is quite straightforward to identify when and where inventory regulation should be


provided. Two trivial but useful rules are used in this decision.
 If n streams join together in a process or part of a process over which mass must
be conserved - normally any process - then the flow of only [n-1] these may be
set by control loops other than one regulating inventory within the process or
part process.
 In any unit containing P phases, P control loops - explicit or implicit - must be
provided to maintain P inventories, which may include combinations of phases.
This number allows for a loop to regulate total inventory.

Finally there appears to be three circumstances in which inventory affects strategic


process variables and thus should be counted as a control degree of freedom.

1. In the case of holdup dependent reaction conversion.


2. In situations where inventory determines pressure.
3. The construction of certain types of equipment means that a change in holdup
will affect performance e.g. by changing the surface area in a flooded condenser.

It should be noted that where inventory affects several strategic variables there will still
only be one degree of freedom.

2.3 Control of Separation Processes


2.3.1 Introduction
This module deals with the control of a couple of simple separation processes:

 Adiabatic Flash
 Non-adiabatic Flash

In each of these sections there will be a degrees of freedom analysis, some discussion
on control strategies and then a few examples. Note that distillation columns will be
dealt with in a separate section.
2.3.2 Adiabatic Flash: Degrees of Freedom

Before we can attempt to control this process we have to know how many streams we
are allowed to manipulate i.e. the control degrees of freedom. This can be evaluated by
the equation

d.o.f = unknowns - equations

Unknowns

There are 3 streams in this process. Each stream has unknown composition, temperature
and pressure. Thus for a stream with C components there are

3 [C + 2] = 3C + 6 unknowns

Equations

In an adiabatic flash there are

 C material balances
 C equilibrium relationships
 1 energy balance
 1 equation relating the temperature of the products i.e. they must be the same.
 1 equation relating the pressure of the products i.e. they must also be the same.

Hence altogether there are

2C + 3 constraints

Degrees of Freedom
From the above we can evaluate that we have

C + 3 degrees of freedom

However we know the composition, temperature and pressure of the input stream.
These were considered unknowns for the above calculations but now can be taken into
account. Thus this adds on another C + 1 constraints and so we now have

2 control degrees of freedom or C.D.F

This answer can be compared with that obtained directly using the equation below.

C.D.F = no. of connections + 1 - no. of phases

 Number of connections for an adiabatic flash = 3


 Number of phases = 2
 Hence C.D.F = 3 + 1 - 2 = 2, as expected!

2.3.3 Adiabatic Flash: Control Strategy


It is now possible to devise control schemes for this adiabatic flash vessel. The basic
input-output block has 3 streams. It has been evaluated above that there are 2 control
degrees of freedom which means control valves on two of the three streams. That leaves
one other stream which also has a control valve on it to regulate inventory i.e. to ensure
that the mass balance does in fact balance.

Of the two streams left one would normally have a flow controller and be used to
regulate throughput. This could be any of the three streams. This leaves one other to
regulate a strategic variable.

From a knowledge of the properties of the adiabatic flash, the normal design
specification is the feed and only one further quantity, usually pressure, but temperature
or another flow could also be chosen.

2.3.4 Adiabatic Flash: Examples


For the examples below we will choose to regulate the pressure and one flowrate, not
necessarily the feed.

 Feed Rate and Pressure


 Vapour Rate and Pressure
 Liquid Rate and Pressure
This gives rise to six different alternatives. Each one is shown below with a short
discussion on whether it is feasible or not. Note that the shaded valve indicates a valve
used for inventory regulation. Not shown is the controller and level measurement. Since
all measurements in this example orginate in the flash drum that end of the loop is
omitted to simplify the diagrams.

Feed Rate and Pressure

Figure 1 below shows the alternatives for this set up.

Figure 1 - Feed Rate Controlled

Figure 1 (a) is a very common arrangement and will work well. Both pressure and level
loops have good adjustment-measurement sensitivity. There is some undesirable
interaction because opening the pressure control valve increases the rate of boiloff and
hence affects the level.

Figure 1 (b) will not work. Although as noted above the vapour rate affects level, altering
the liquid rate does not change the pressure of the flash.

Vapour Rate and Pressure

Figure 2 below shows the alternatives for regulation of vapour rate and pressure.
Figure 2 - Vapour Rate Controlled

Figure 2 (a) is another fairly conventional arrangement. Level control is good, although
there is interaction between both flow and pressure loops. That is if the flow of vapour is
increased then the pressure will decrease significantly and so the feed rate will have to
be increased also. Decreasing the vapour flow will have the opposite effect causing the
feed rate to be reduced.

Figure 2 (b) is unworkable for the same reasons given above for figure 1(b).

Liquid Rate and Pressure

Finally figures 3 gives the two alternatives for liquid rate and pressure.
Figure 3 - Liquid Rate Controlled

Both of these schemes will work. However figure 3 (b) should give stronger response of
both pressure and level control loops to their respective adjustments, and hence less
interaction. It would be preferred to figure 3 (a).

Alternative Arrangement

It would be possible to devise other schemes in which the flow control loop acted
indirectly. For example, by adjusting a stream other than the one which is measured as
shown below in figure 4. Such arrangements should be avoided, as should any system in
which the operation of one loop, here the flow control, depends also on the operation
of another, here the inventory loop.
Figure 4 - Another Possibility?

2.3.5 Non-Adiabatic Flash: Degrees of Freedom

As with the Adiabatic Flash we will start by evaluating the control degrees of freedom
from the equation

d.o.f = unknowns - equations

Unknowns

In this example there are 4 streams associated with the flash. There are 3 process
streams each with unknown composition, temperature and pressure, and 1 energy flow.
This gives

3 [C + 2] + 1 = 3C + 7 unknowns

Equations

In a flash with heat input there are

 C material balances
 C equilibrium relationships
 1 energy balance
 1 equation relating the temperature of the products i.e. they must be the same.
 1 equation relating the pressure of the products i.e. they must also be the same.

Hence altogether there are

2C + 3 constraints

Degrees of Freedom

From the above we can evaluate that we have

C + 4 degrees of freedom

Now, as before, we can fix the composition, temperature and pressure of the feed
stream. These C+2 constraints can be included in the above to give

3 control degrees of freedom

Once again this answer can be compared with that obtained directly.

C.D.F = no. of connections + 1 - no. of phases

 Number of connections for an flash with heat input = 4


 Number of phases = 2
 Hence C.D.F = 4 + 1 - 2 = 3, as expected!

2.3.6 Non-Adiabatic Flash: Control Strategy

The non-adiabatic flash separator has an additional adjustment not present in the
adiabatic case, namely a heat input from a heating stream of steam or other utility.
Analysis of the design problem, as shown in the previous section, confirms that there is
a further degree of freedom and therefore a flowrate plus two further variables can be
set. i.e. there are four possible adjustments we can manipulate, the three mentioned and
the material balance maintained with the fourth.

The best procedure to adopt in these examples is the following.

 Put in all flow controllers on the streams which they regulate.


 Check to see if it is now possible to locate the inventory control unambiguously,
if so do it.
 Consider all possible combinations for remaining loops.
o Eliminate those without casual relationships between measurement and
adjustment.
o Evaluate the others.

Note that while it is usually important to locate strategic loops before choosing the
manipulated variable for level, we have found that liquid rate does not affect either
temperature or pressure. All other things being equal, this would be the preferred
choice for the level control loop. As with the adiabatic example, the measurement ends
of the control loops are not shown and the adjustment for the level loop is indicated by
a shaded valve.

2.3.7 Non-Adiabatic Flash: Examples


Below are four examples with the following parameters regulated.

 Pressure, temperature and vapour rate


 Pressure, liquid and vapour rates
 Pressure, liquid to vapour flow ratio and feed rate
 Pressure, temperature and vapour to feed ratio

Pressure, Temperature and Vapour Rate

Figure 5 below shows the best arrangement for this option. In this case interchanging
temperature and pressure loops could result in an inferior arrangement, as the feed rate
would probably not have so direct an effect on temperature, but would still work.
Figure 5 - Pressure, Temperature and Vapour Rate Regulated

Pressure, Liquid and Vapour Rates

Once flow controllers are placed on the liquid and vapour product streams, the third
material stream, i.e. the feed, must be used to maintain the mass balance. This leaves
only the heat input, steam, valve to regulate pressure. This is the only possible scheme
and is shown below in figure 6.
Figure 6 - Pressure, Liquid and Vapour Rates Regulated

Pressure, Liquid to Vapour Flow Ratio and Feedrate

To regulate a ratio requires a flow control loop on one stream, with its setpoint adjusted
by a ratio controller. A flow measurement on the second stream feeds into the ratio
controller which sees both flows and changes the setpoint of the flow controller
accordingly. Thus only one of the ratioed streams has a valve, but both have flowmeters.
In principal the valve may be on either stream.

A suitable arrangement is shown in Figure 7. The flow control valve for the ratio system
has been located on the vapour line in order to leave the liquid line for level adjustment,
as discussed above. Pressure control must then be by the heat input.
Figure 7 - Pressure, Liquid to Vapour Flow Ratio and Feedrate Regulated

Pressure, Temperature and Vapour to Feed Rate Ratio

One alternative is shown in figure 8 below. It would be possible here to put the valve for
the ratio control on either feed or vapour streams. Also it is possible to interchange the
temperature and pressure loops. The best choice would depend on the particular
system: flowrates, component volatilities, etc. and would be determined after detailed
modelling of the process.
Figure 8 - Pressure, Temperature and Vapour to Feed Rate Ratio Regulated

2.3.8 Three Phase Flash


In a three phase flash (or separator) the feed into the vessel separates into two liquid
phases and a vapour phase. An example where this can be used is in the primary
separation of light organics, heavy organics and water. The diagram below shows a
typical three phase separator.
Figure 9 - Three Phase Flash

Now let us consider the degrees of freedom:

C.D.F = no. of streams - no. of interfaces

 Number of streams = 4
 Number of interfaces = 2
 Hence C.D.F = 4 - 2 = 2

Typical control specifications would be feed rate and pressure. Note that two interfaces
must be maintained with two control loops. An example of a control scheme is shown in
the diagram below.
Figure 10 - Three Phase Flash With Control

2.3.9 Countercurrent Cascade Columns


We have already looked briefly at countercurrent cascade columns in Module 2.1:
Controlling Simple Processes. Here is a diagram of one without control.

Figure 11 - Countercurrent Cascade Column

Now let us consider control.


C.D.F = no. of streams - no. of interfaces

 Number of streams = 4
 Number of interfaces = 1 (gas-liquid)
 Hence C.D.F = 4 - 1 = 3

For gas-liquid operations the three strategic variables to control would be

 Column pressure
 One flow rate
 One composition

Below are three examples of controlling a cascade column with comments on how good
or bad the control scheme is.

Example 1 - Good Composition Control

In this first example the control loops are:

 Liquid feed flowrate


 Column pressure - tops flowrate
 Bottoms composition - gas input
 Inventory/level - bottoms flowrate

Figure 11 - Control Scheme 1


This is a good control scheme. All the adjustments have a direct effect on the controlled
variables. In particular look at the composition control of the bottoms product. As you
can see the adjustment and measurement are both at the bottom of the column.

Example 2 - Poor Composition Control

In this second example the control loops are:

 Liquid feed flowrate


 Column pressure - tops flowrate
 Tops composition - gas input
 Inventory/level - bottoms flowrate

Figure 12 - Control Scheme 2

This is not quite as good as the previous example. The pressure, flowrate and level are
as before. However this time the top composition is controlled using the flowrate of the
gas entering at the bottom of the column. Hence this time there will be a time delay
between making the adjustment and the composition changing to reflect this change.
This will be the length of time taken by the gas too travel up the column.

Example 3 - Bad Composition Control

Finally in this third example the control loops are:

 Gas feed flowrate


 Column pressure - tops flowrate
 Bottoms composition - liquid input
 Inventory/level - bottoms flowrate

Figure 13 - Control Scheme 3

In this example the flowrate of the gas is constant and once again the pressure and level
control are as before. However this time the bottoms composition is regulated using the
liquid flowrate entering at the top of the column. So once again there will be a time
delay between making the adjustment and seeing the effect of the adjustment, but this
time it will be the length of time taken for the liquid to travel down the column - a
significantly longer time. Hence this control scheme should be avoided.

2.3.10 Liquid-Liquid Extraction Columns


Finally let us look at the situation where the gas in the previous example is replaced by
another (lighter) liquid to form a liquid-liquid extraction column.
Figure 14 - Liquid-Liquid Extraction Column

In this case note that there is going to be some vapour present at the top of the column.
This gives another phase and so the degrees of freedom analysis gives us:

C.D.F = no. of streams - no. of interfaces

 Number of streams = 5
 Number of interfaces = 2 (vapour-liquid, liquid-liquid)
 Hence C.D.F = 5 - 2 = 3

A liquid-liquid column running full of both liquids would probably use implicit pressure
regulation by leaving an open line to another part of the plant or alternatively may be
vented to atmosphere. This effectively reduces the degrees of freedom to 2. Normally
one flowrate and one composition of controlled along with the two interface control
loops.

The example below shows the flowrate of the lighter liquid being controlled with the
flowrate of the heavier liquid entering at the top of the column being used to regulate
the composition of the top product. The two product streams are used for inventory.
Figure 15 - Liquid-Liquid Extraction Column With Control

2.4 Controller Design and System Modelling

This fourth module is concerned with the modelling and use of both process and
controller. First there is a section on the mathematics behind the controller actions.
Then there is a section on how to create a model from simple building blocks and
then analyse the response. Then finally there is a section on how to relate this
response to the controller settings for optimal control.

 Introduction
 Controller Action Mathematics
o Introduction
o Proportional Controller
o Example: Flow Through Pipe (1)
o Proportional Integral Controller
o Example: Flow Through Pipe (2)
 Modelling a Process
o Introduction
o Simple Black Box Models
o Typical Responses
o Theoretical Response
o Analysis of Response
 Tuning a Controller
o Introduction
o Selecting Controller Parameters
o A General Process Model
o Zeigler Nichols Open Loop Response Tuning Method
o Controlling Real Processes
o Gains for Real Processes and Controllers

2.4.1 Introduction

The purpose of this part of the course is to show how settings for controllers can be
obtained from a knowledge of the process to be controlled. This forms part of the
complete control system design procedure. After manipulated and adjusted quantities
have been selected and their pairings, perhaps tentatively, chosen, (as discussed
elsewhere) then values of one or more parameters for each controller must be
determined.

The process with these control loops and controller settings can then be tested, usually
by simulation using a mathematical model of the process, but sometimes with the
`real' process if it is available. The choice of control loops and/or the controller
settings may then be changed if their performance is not satisfactory.

There are three topics to be covered:

 What functional relations or algorithms are to be used in a feedback controller


to relate measured error observed to adjustment made?
 How do we obtain, either experimentally or from first principles, a model of a
process?
 For a process represented by a given model, what controller parameter values
should be used?

These are introduced in separate sections. Finally there is a separate case study section
which includes simulations of processes which can be controlled by feedback control.

2.4.2 Controller Action Mathematics

Introduction

The basic type of controller is the Feedback Controller.

Feedforward controllers also exist but are more complicated to implement. Here we
will describe the use of feedback controllers.
In feedback control the variable required to be controlled is measured. This
measurement is compared with a given setpoint. The controller takes this error and
decides what action should be taken by the manipulated variable to compensate for
and hence remove the error.

Figure 1 - Feedback Control Loop

The advantage of this type of control is that it is simple to implement. Not only does
the feedback control system require no knowledge of the source or nature of the
disturbances, but it also requires minimal detailed information about how the
process itself works. Feedback control action is entirely empirical. So long as an
adjustment is being made in the correct sense then the control system should
remove the effect of an external disturbance.

The disadvantage is that the disturbance has to enter and upset the system before it is
eliminated.

A feedback control loop can have one of two objectives.

 A servo control loop is one which responds to a change in setpoint. The


setpoint may be changed as a function of time (typical of this are batch
processes), and therefore the controlled variable must follow the setpoint.
Figure 2 - Servo Control

 A regulatory control loop is one which responds to a change in some input


value, bringing the system back to steady state. Regulatory control is by far
more common than servo control in the process industries.

Figure 3 - Regulatory Control

Proportional Controller

The first type of controller that we will study is the proportional controller. This
controller sets the manipulated variable in proportion to the difference between the
setpoint and the measured variable. The bigger the difference, the greater the change
in the manipulated variable.
The equation that describes a proportional controller is

where

 ut is the output from the controller, i.e. the adjustment


 is the constant of proportionality, ususally called the controller gain
 ud is the output of the controller at its design conditions, sometimes called
the bias
 ys is the required value of y or the setpoint
 y is the input to the controller, i.e. the measured variable

The advantage of proportional control is that it is relatively easy to implement.


However the disadvantage is that when implementing a proportional only controller
there will be an offset in the output. Thus there is always a difference between the
setpoint and the actual ouput. The reason why this is so can be shown by means of an
example.

Example: Flow Through a Pipe (1)


Below is a diagram of the example.

 F is the flowrate through the pipe


 Fm is the measured flowrate
 Fs is the required, setpoint flowrate
 e is the error between the setpoint and measured value
 Fv is the valve position or controller output
Figure 4 - Diagram of Flowrate Example

Therefore we can see that

Here

So

Let us assume to begin with that Fs = 50 and Fd = 50. If this is true then it can be seen
from the above equation that F = Fs and there is no error. Note that this result is
independent of the value of the gain.
However, let us now consider what happens when the value of the setpoint changes
from 50 to 60 with Fd staying constant at 50. First the relevent equation is shown and
then the table below summerises the results for different gains.

From the above example we can see the problem of using proportional only control,
namely the offset. Note also that there must always be an offset. This is because to
achieve the new steady state the term must have a value and so
there must be an error. There are two ways of eliminating this problem.

 Choose ud to correspond always to the correct output


 Make the gain very large

The first is hard to achieve since it requires very accurate knowledge of the process,
and would require changes whenever the setpoint is moved.

The second leads to problems of rangeability and sensitivity. Suppose the gain is 10,
then measurement noise of 1% of the total range will cause the control valve to move
over 10% of its total travel. This is unacceptable.

Proportional-Integral Controller

To remove the offset integral action is required and so PI control is normally used. It
works by summing the current controller error and the integral of all previous errors.
It may be thought of as a way of automatically calculating the quantity ud. Proper
tuning - described in a subsequent section - of the integral part of a PI controller can
improve its performance.

If the error e is defined as


Then the equation describing a proportional-integral controller is

where

 is the reset time of the controller

Alternatively, we can differentiate this expression to get

Example - Flow Through a Pipe (2)


Again we will use the example of the flow through a pipe to investigate the nature of
Proportional-Integral control.

As before

 F=y=u
 e = Fs - F

So

Since Fs is constant, this becomes

Or
Equations of the form

are very common and have well known properties. Their solution has the form
shown below.

Figure 5 - Response Curve

We note that

 As time goes to infinity, x = a


 The rate of response of the system increases as T (called the time constant of
the equation) becomes smaller

For the flow control system with integral action we see that

 F will eventually become equal to Fs


 This will happen faster if is small and/or is large
2.4.3 Modelling a Process

Introduction

There are two ways of approaching the problem of obtaining a mathematical


representation or model of a chemical process, or indeed anything.

 Create a fundamental or mechanistic model based on knowledge of the


physics and chemistry of the system to be modelled. This can be quite a hard
thing to do, indeed nearly all of a chemical engineering degree course might
be regarded as being about the creation of such models! The advantage of
such a model is that it is basically `right' (provided of course the model
builder's knowledge of physics and chemistry is right and is applied correctly).
Such a model should be robust in that it can be applied again under conditions
of operation different from those for which it was first constructed. If the
process being modelled is modified, then analogous modifications to the
model will enable it to continue to be used.

We will describe briefly some rules for constructing this type of model which
help to ensure that if the modeller's understanding of the problem is correct
then a correct model will be obtained.

 Choose a mathematical form which is convenient (e.g. it is simple or easy to


manipulate) and which represents fairly well the observed behaviour of the
system being modelled. Fit numerical parameters to the mathematical form.

This is a so-called `black box' or `input-output' model, which seeks only to


reproduce the behaviour of the system's output in response to changes in its
setpoint or inputs. The mathematical form chosen may bear no relation to the
form of the equations which truly describe the system. As a result, such models
must be used with the greatest care under conditions in the least bit different
from those at which the original parameters were determined.

The advantage of such 'arbitrary' models is that they can be developed with
little or no knowledge of the system to be represented, and hence complicated
systems can be modelled quickly.
Simple Black-Box Models

Input-output models form the basis of most classical process control theory. They are
usually subdivided according to whether they have one or more than one input and/or
output. We will consider initially only single input, single output (SISO) models,
although some ideas associated with multiple input-output models will be touched on
elsewhere in the course.

The basic SISO model can be thought of as relating an output y to an input u. In


general both of these quanties will change with time, the model must represent how y
responds to changes in its input or inputs.

Typical Responses

Suppose an input u is given a step change at some time, as shown in the figure.
Observations of typical `processes', from aircraft to papermills, suggest that there
are three main types of behaviour which may be seen in an output y.

1. Instantaneous response

The first typical response is called the instantaneous response. In this case y also
responds in a step, but in general of different size to that in u (in any case y will
normally have different dimensions to u) as shown below

The simplest mathematical relationship is of the form:

Classical control theory assumes that behaviour can be represented by linear


equations like the above, and so this is the only type of equation required to
represent this type of behaviour.

In the above equation is called the Gain of the process or model.

2. Lagging response

Here y starts to change the moment that y changes, but the full extent of the
response `lags' behind the disturbance. After a while, y will have responded fully.
The simplest mathematical form which provides this behaviour is an ordinary
differential equation with time t as the independent variable, having the form:

Here is as before the gain, and is called the Time Constant of the equation,
system or model. Because it is described by a single first order o.d.e. this is called a
First Order model, system, lag or response. The interpretation of these parameters is
described below.

3. Delayed response

When u changes, no immediate change in y is observed. However after a time T, y


responds completely to the change in u as in the instantaneous response case.
Mathematically this is represented by a difference equation:

This does not have simple analytical properties, but is easily understood by chemical
engineers as corresponding to a plug flow or pipeline system with residence time T.
It is also referred to as a time delay or pure time delay system.

4. Representing Complex Responses

Complex systems may be reasonably well approximated by combinations of the


above three elements.

Models of such systems can be assembled as networks of the elements as shown


below.

Analytical and numerical techniques are available to work with models constructed in
this way.

Theoretical Response

Classical control theory constructs all its models from sets of linear ordinary
differential equations. (The instantaneous response is the limiting case of the the o.d.e.
where is zero, and the plug flow delay, like the plug flow reactor, is the limit of an
infinite number of first order lags.)

There is no good physical reason why a real process should be well represented by
such a set of equations, except that in the limit of infinitesimally small changes, all
nonlinear equations approximate to linear ones.

However, the theoretical advantage of linear representation is twofold. Firstly, the


whole system may be represented by o.d.e.s, whereas if there were any nonlinear
algebraic equations a mixed set of differential-algebraic equations would be required.
Further, a system of linear differential equations always has an analytical solution, but
more particularly, is amenable to various other types of analysis which cannot be
performed on nonlinear equations. The tuning methods for controllers described later
make use of this type of analysis to obtain generalised equations for suitable controller
settings in terms of parameters of a process model written in terms of the above three
types of behaviour. This is not possible for nonlinear systems.

It should be stressed that if we wish to simulate the behaviour of a process, which


requires only the solution of the relevant equations, and not their analysis, then there
is no particular point in approximating it with this type of simplified approximate
model. A `real' model should be constructed, as discussed later, and solved.

Let us look again at the differential equation which describes first order behaviour.

It is possible to solve this equation analytically to obtain the expression

Here

 yo is the value of y at t = 0
 is the size of the step change in u at t= 0

Note that a graph of this equation gives the response curve shown above under the
section on the lag response.

The first thing to consider is What is the Change in y

This equation can be now be used directly to calculate the new value of the output
variable if the change in u, the gain and time constant are all known. Otherwise it is
necessary to estimate values for the gain and time constant as shown below.
Analysis of Response
It will be shown later in the section on tuning controllers that it is useful to be able to
look at the open loop response of a process and try and estimate the values of the
gain and time constant. Below are notes on how to do this and then you can try it for
yourself in the exercises associated with this part of the module.

1. Estimating the Gain

is known as the gain. It tells us how much the output variable will change per unit
change in the input variable. A large gain implies a large change in y for a given
change in u and hence leads to a quicker response.

To calculate its value we have to consider the system going from one steady state
value to another. Thus we can see what effect a change in u has on the value of y.

After the system has settled down following the step disturbance

So

Or, as shown in the graph below


From this we can see that it is a simple calculation to evaluate the gain of the process
given the change in u and y.

2. Estimating the Time Constant

is the time constant for the process. This is related to the speed of response of the
system. The diagram below shows a graphical method of evaluating its value.

1. The first stage is to draw the initial slope


2. Then the final steady state value is drawn
3. The time at which these two lines intercept is the value of the time constant

Note that this is also the time taken for the output value to travel 63% of the distance
to its new value.
This is shown mathematically below

The following points should be noted about the time constant

 y(t) reaches 63.2% of its final value in one time constant.


 The smaller the time constant the steeper (quicker) the response.
 After 3 to 4 time constants the system is essentially at its new steady state.

3. Changing the Gain and Time Constant

Finally, how does the response change when and are altered but the change
in u stays the same?
The diagram below shows that changing alters the slope of the initial slope and
changing alters the final steady state.

2.4.4 Tuning a Controller

Introduction
In the first section on controllers we looked at two control algorithms for
proportional and proportional integral controllers. In order to implement these
algorithms there are two parameters which have to be fixed, namely

 , the controller gain


 , the integral reset time

In the second section on modelling processes we looked at

 First order input output processes


 Mechanistic models of actual processes
For the first of these we saw that there were three parameters necessary to define the
process. These are

 T, the dead time


 , the process gain
 , the process time constant

The aim of this section is to introduce a method of matching the personality of the
controller to that of the process so as to achieve the optimum controllability. In other
words how do we go from the process parameters to the controller parameters. The
method introduced uses the open loop response of a process and works best with a
delay-followed-by-first-order-lag. There are many other tuningmethods which look at
other aspects of the process in order to tune the controller. A couple of these will be
discussed in a later section.

Selecting Controller Parameters


The best choice of controller parameters depends significantly on the nature of the
process to be controlled. Thinking back to the simple input-output models we can
say that

 Instantaneous Response processes are easy to control. Large gains may be


used, subject to noise constraints. Integral action should be used.
 First Order Response Processes are also easy to control. The tuning method
described below is based on a first order response.
 Time Delay processes are difficult to control. A pure time delay becomes
unstable in principle if a dimensionless gain greater than 1 is used in a
proportional only controller.
 Inverse Response processes exhibit a response to an adjustment like this:
You will see later in the interactive exercises that this is very difficult to
control!

A General Process Model


Let us consider first the simple case of a first-order lag in series with a time delay.
This setup is shown in a diagram below.

If we change u by a known amount and plot the response curve it is possible to


determine the model parameters Ts, Td and the gain from the resulting graph. This
is shown in the diagram below.
Zeiglar Nichols Open Loop Tuning Method
The Zeigler Nichols Open-Loop Tuning Method is a way of relating the process
parameters - delay time, process gain and time constant - to the controller
parameters - controller gain and reset time. It has been developed for use on delay-
followed-by-first-order-lag processes but can also be adapted to real processes.

The method is outlined below.

 Look at the open loop response of the process to a step change in the
manipulated variable.
 Evaluate
o The steady-state gain, (y2 - y1) / (u2 - u1)
o The time delay, Td
o The time constant, Ts

The diagram above shows how to obtain these values.

 Finally substitute these values into the table below to obtain the relevent
controller parameters.

Controller Type Gain Reset Derivative


P Ts / Td - -

PI 0.9 Ts / Td 3.3 Td -

PID 1.2 Ts / Td 2.0 Td 0.5 Td

The Gain evaluated above is the product of the controller gain setting, and the
process steady state gain, G.

Gain = * G

Therefore by substituting all the values in for the above and re-arranging we get the
following values for the controller parameters:

Controller Type Controller Gain, Reset Derivative

P (Ts ) / (Td ) - -

PI (0.9 Ts ) / (Td ) 3.3 Td -

PID (1.2 Ts ) / (Td ) 2.0 Td 0.5 Td

Advantages of this method are

 Only a single experimental test is needed.


 It does not require trial and error
 The controller settings are easily calculated.

However there are also Disadvantages

 Experiment is under open loop response and so disturbances may affect the
results.
 Results tend to be oscillatory.
 Does not work well for complex responses - leads to inaccurate tuning model.

Controlling Real Processes


In the real world, unfortunately, the response of a process to a change in one of its
inputs seldom follows the first-order case required for the Z-N tuning.
If we are lucky it may be similar in form but different in detail as shown below.

In this case the tangent should be drawn at the point where the slope of the
response is steepest. Now we have estimates for the parameters and it may be
necessary to change them in order to get the optimum values. To do this it is
possible to use a model of the process and controller to see the effect of altering the
control parameters. There is the chance to do this in the case study section.

However if we are unlucky the response may be like this...


... in which case it is very difficult to control and it may not be possible using a PI
controller.

Gains for Real Processes and Controllers

In a process the measurement y is strictly speaking a dimensioned quantity:


temperature, pressure, flow etc.

The adjustment u is usually a flow, so that the process gain, , will in general have
odd dimensions! This also makes it hard to interpret or compare gain values.

In practice, both measurement and adjustment have a maximum range determined by


the measuring instrument or valve. It is best to work with scaled quantities always
expressed as a fraction or percentage of range, e.g.

As indicated these values are percentages and so are dimensionless values between
0 and 100. Thus it is possible to define a dimensionless gain for the process as

If the value of the gain is large, say 100, then this means that the change in y is 100
times greater than the corresponding change in u. This could lead to y going out of
bounds or else the change in ubeing very restrictive.

Alternatively, if the gain is very small, say 0.01, then for a large change in u there is
hardly any response in y.

What is required is a gain of around 1. This enables both input and output to be used
to their full ranges which in turn improves the controllability.

So if this definition of the gain is used it is clear from a glance if a suitable value has
been obtained or not. In this case simply use the value of the gain from the first table
above along with the dimensionless process gain above to obtain the dimensionless
controller gain.

In practice a controller does not want to deal with meaningless dimensions when
asking for the value of the gain. Therefore a parameter known as the Proportional
Band is used instead.

Firstly remember that you have a value of the dimensionless gain for the controller as
evaluated above.

Now we define the Proportional Band, P, as the reciprocal of the dimensionless


controller gain.

Remember that when specifying a controller setting, always use dimensionless gain
or proportional band.

2.5 Case Study Section

This section of the course reinforces the material taught in the lecture notes. There are
two parts which are outlined below. Firstly there are a couple of simulations which
cover tuning a controller using the Zeigler Nichols open loop method. Then there are
a number of tutorials on all aspects of this part of the course both controlling simple
processes and simple controller design.

 Laboratory Session

There are a number of experiments in the Virtual Control Laboratory which


involves work covered in this section. In particular please work through the
following

o Introduction to Zeigler Nichols Tuning


o Further Examples on Open Loop Tuning
 Tutorial Questions and Answers
o Exercise 1: Controlling Simple Processes
o Exercise 2: Degrees of Freedom
o Exercise 3: Control of Adiabatic Flash
o Exercise 4: Control of Non Adiabatic Flash
o Exercise 5: First-Order Modelling
o Exercise 6: Controller Tuning

2.5.1 Introduction to Zeigler Nichols Tuning Methods

Aims of Experiment
The aim of this experiment is:

 To introduce the Zeigler Nichols open loop and closed loop methods for tuning
controllers.

Theory
There are two theory sections associated with this experiment.

 Open Loop Tuning

2.5.1.2.1 Open Loop Tuning


The Zeigler Nichols Open-Loop Tuning Method is a way of relating the process
parameters - delay time, process gain and time constant - to the controller
parameters - controller gain and reset time. It has been developed for use on delay-
followed-by-first-order-lag processes but can also be adapted to real processes.

The method is outlined below.

 Look at the open loop response of the process to a step change in the
manipulated variable.
 Evaluate
o The steady-state gain, (y2 - y1) / (u2 - u1)
o The time delay, Td
o The time constant, Ts

Look at the diagrams below to see how this differs between ideal and real
processes.

Ideal Process
Real Process

 Finally substitute these values into the table below to obtain the relevent
controller parameters.
Controller Type Gain Reset Derivative

P (Ts ) / (Td ) - -

PI (0.9 Ts ) / (Td ) 3.3 Td -

PID (1.2 Ts ) / (Td ) 2.0 Td 0.5 Td

 Closed Loop Tuning

2.5.1.2.2 Closed Loop Tuning


With some real processes the response to a stepchange or setpoint disturbance
differs depending on the direction or size of the change. In this case it is irrelevent to
look at the the open-loop response to tune the controller. Instead the closed-loop
response ie the behaviour of the system with control, has to be studied.

The Zeigler Nichols Closed-Loop Tuning Method looks at the response of the system
under proportional only control to obtain PI controller5 settings. The method is
outlined below.

 Set up the system with proportional only control and add a disturbance.
 Alter the gain of the process until you obtain the smallest gain which gives
constant amplitude oscillations. This gain is called the Ultimate Gain, .
 Now evaluate the period of these constant oscillations. This is known as the
Ultimate Period, Pu. Please refer to the diagram below.
 Finally substitute these values into the table below to obtain the relevent
controller parameters.

Controller Type Gain Reset Derivative

P /2 - -

PI /2.2 Pu/1.2 -

PID /1.7 Pu/2 Pu/8

The material for these sections has been taken from Module 2.2 - Controller Design
and System Modelling and Module 4 - Additional Material.

Procedure
Please choose which method you require and move on to work through the exercise.

 Open Loop Tuning


2.5.1.3.1 Open Loop Tuning

2.5.1.3.1.1 Introduction
This short simulation has been set up for the user to practise tuning a controller
using the Zeiglar Nichols Open Loop Response Method. It assumes that the user has
covered the relevent material fromModule 2-2 in the HyperCourse.

There are five parts to this exercise:

 Look at the open loop response


 Analysis of the open loop response
 Evaluate controller parameters
 Compare with actual values
 Run process with control

2.5.1.3.1.2 Look at Open Loop Response


The first stage in the procedure is to look at the open loop response for the process
in the graph below:

50
The change in u (the input variable) was : %
2.5.1.3.1.3 Analysis of Open Loop Response
The second stage is to enter the following values:

Change in y? Dead Time? Time Constant?

2.5.1.3.1.4 Evaluate Control Parameters


The next stage is using the above values to fill in the following table:

Controller Type Gain Reset Derivative

P (Ts ) / (Td ) - -

PI (0.9 Ts ) / (Td ) 3.3 Td -

PID (1.2 Ts ) / (Td ) 2.0 Td 0.5 Td

Controller Type Gain Reset Time

Proportional

Proportional Integral

2.5.1.3.1.5 Comparison With Actual Values


Once you have entered your values, please click here to find out how close you are
to the actual answers:

Controller Type Gain Reset Time How Close?

Proportional

Proportional Integral
2.5.1.3.1.6 Run Process With Control

Finally it is possible to try out the above values in the simulation below and hence do
some fine tuning to find the optimum values.

 Closed Loop Tuning

It is assumed that the user has covered the relevant material in the hypercourse.

2.5.1.3.2 Closed Loop Tuning

2.5.1.3.2.1 Introduction
This short simulation has been set up for the user to practise tuning a controller
using the Zeiglar Nichols Closed Loop Response Method.

In this exercise it is simply a matter of using the Java applet below to practise the
closed loop method following the simple guidelines given and then filling out the
form to make sure that the right answers have been obtained.

There are six stages:

 Set up the system with proportional only control


 Enter the Ultimate Gain
 Evaluate the Ultimate Period
 Obtain the controller parameters
 Compare with actual values
 Run process with control

2.5.1.3.2.2 Set Up System With Proportional Only Control


The applet below shows a process which is being controlled with a proportional only
controller. All you have to do is alter the gain of the process until the response
oscillates with a constant amplitude.
2.5.1.3.2.3 Enter Ultimate Gain
Once you have found the Ultimate Gain please enter the value in the box below.

Ultimate Gain
2.5.1.3.2.4 Evaluate Ultimate Period
Now look at the response and evaluate the Ultimate Period. Please enter this value
in the box below.
Ultimate Period
2.5.1.3.2.5 Obtain the Controller Parameters
The next stage is to use the above values to fill in the table.

Controller Type Gain Reset Derivative

P /2 - -

PI /2.2 Pu/1.2 -

PID /1.7 Pu/2 Pu/8

Controller Type Gain Reset Time

Proportional

Proportional Integral

2.5.1.3.2.6 Comparison With Actual Values


Once you have entered your values, please click here to find out how close you are
to the actual answers:

Controller Type Gain Reset Time How Close?

Proportional

Proportional Integral

2.5.1.3.2.7 Run Process With Control


Finally it is possible to try out the above values in the simulation below and hence do
some fine tuning to find the optimum values.

2.5.2 Further Examples on Open Loop Tuning

Aims of the Experiment


The aims of this simulation are:

 To practice tuning a controller using the open loop method


 To see the effects of altering the control parameters

Experimental Setup
For this experiment there is no actual physical apparatus. There are five
different processes which have been modelled for this simulation.

The first three follow the simple delay-followed-by-first-order-lag model. This is


followed by two lags-in-series processes which reflect real processes.

The aim is to go through these one by one and try to decide on the controller
parameters which returns the response to it's setpoint value of zero with the minimal
of disturbance.

Theory
The relevent theory for this experiment can be found in .

It would also be worthwhile to work through the Introduction to Zeigler Nichols


Tuning Methods or at least look at the notes from it if you have not already done so.

Procedure
The first stage of this simulation is to look at some pre-experimental questions.
These will help you work though the exercise and tell you what you should be
looking out for.

Then it is possible to choose a process. As mentioned before there are 5 to choose


from ranging in difficulty from ideal delay-followed-by-first-order-lag to more real
lags-in-series processes. For more difficult responses, for example like an inverse
response, it is necessary to use the Zeigler Nichols Closed Loop Response Method.

Once you have chosen a process you will be shown the open loop response. It is then
an easy matter to analyse this and insert the required gain and reset time and obtain
the response with control. This step can be repeated as many times as necessary until a
thorough understanding of the concepts of control is obtained.

2.5.3 Exercise 1: Control of Simple Processes


Devise control schemes for the following simple processes. It is necessary to use the
hierarchical approach described in the theory secton. Once again the answers will
not be available until the questions have been attempted in class although it is
possible to look at the complete flowsheet before attempting to put control loops on
it.

Acetone Recovery
Acetone vapour in a byproduct gas is absorbed in a nonvolatile solvent. It is then
recovered in a flash separator.

Answer

2.5.3.1.1 Input - Output Structure


The input-output structure is quite straightforward. A flow controller has been
placed on the feed to indicate that this is set upstream. It is clear that the
composition of both product streams will be important, and so quality controllers
have been placed there without indicating what will be adjusted.
It is sensible to show the solvent stream as a makeup. There will always be some
solvent loss to be made good. Figure 1 below shows the control system at this stage.

Figure 1 - Acetone Recovery : Input/Output Structure

2.5.3.1.2 Recycle Structure


At the recycle structure level the air QC can be associated with the rate of solvent
supplied to the absorber. This is the main recycle stream. Acetone quality control will
be associated with the flash, though the details of the loop cannot be defined at this
stage. It is not anticipating too much to have indicated that the solvent makeup will
be determined by some form of inventory control, as the purpose of that loop is to
maintain the inventory of solvent. Figure 2 shows the results at this stage.

Figure 2 - Acetone Recovery : Recycle Structure


2.5.3.1.3 Final Structure
Finally, Figure 3 shows the absorber and flash. Both are standard and
straightforward. A tank is provided to add makeup solvent on level control. Pumps
and compressors have not been shown, their placement being left as a further
exercise. They do not, in this process, affect the general control system structure.

Figure 3 - Acetone Recovery : Final Structure

Chlorine Removal
Chlorine in a nonaqueous product stream is stripped using air, then absorbed in
aqueous NaOH.
Answer

2.5.3.2.1 Input - Output Structure


As in the previous example the input-output level, as in Figure 1, shows a feed flow
control loop and two incomplete composition loops.

Figure 1 - Chlorine Removal : Input/Output Structure

2.5.3.2.2 Final Structure


The process has no recycle as the caustic emerges with the chlorine in the form of
sodium hypochlorite NaClOCl. Note that this has to be treated before being
discharged, although this is not specified as part of the process. The expanded block
diagram, figure 2 below, shows the functional blocks. These are rather
straightforward, so no process flow sheet is shown.

The absorber would run open to atmosphere with the pipework pressure setting the
stripper pressure somewhat above ambient. A blower would obviously be required on
the air supply, so the valvethere would probably be a motor speed adjustment.
Although the absorption of chlorine in NaOH is highly exothermic, the exit stream
from the stripper would be quite dilute, so temperature regulation on the absorber
would be unnecessary.

Figure 2 - Chlorine Removal : Final Structure

Butadiene Purification
Butadiene is extracted from a reactor product using a high boiling solvent which is
then recovered by distillation. The nature of the impurities is not specified. An
assumption made is that they are all removed in the solvent, and the distillation is
required only to recover this. This implies that all the impurities are heavier than
butadiene, which has a low boiling point of about 290K.
Answer

Not Available

Sorry, the solution for this example is not available at present.

2.5.4 Exercise 2: Degrees of Freedom

Question 1

The process shown below involves the partial vaporisation and flashing of a mixture
which forms a vapour phase and two immiscible liquid phases, one organic and one
aqueous. The feed is preheated and partially vapourised in H101 and separates into 3
phases in S102. The vapour product is condensed in H103.
 What is the maximum number of control loops other than level control
loops possible for the process? How does this relate to the actual number of
control loops which you would provide?
 The process is to be provided with a control system to regulate the following
strategic quantities, as well as levels etc:
o Process throughput
o Flash vessel temperature and pressure

Devise a suitable control system. Comment briefly on the justification for, and
mode of operation of, each control loop proposed.

Answer

What is the maximum number of control loops other than level control loops possible
for the process? How does this relate to the actual number of control loops which you
would provide?

Looking at the diagram it can be seen that there are 6 places where control loops are
possible but two of these have to be used for the liquid and interface levels in the 3
phase flash. Therefore it is possible to control 4 different things.

Note that it is not always appropriate to regulate all possible variables since some are
self regulating and it would be a waste of energy.

The process is to be provided with a control system to regulate the following strategic
quantities, as well as levels etc:
 Process throughput
 Flash vessel temperature and pressure

Devise a suitable control system. Comment briefly on the justification for, and mode
of operation of, each control loop proposed.

Three strategic loops are required, one less than the number determined above. It
would be possible in principle to regulate condenser pressure as well as flash drum
pressure, but this would be pointless, and indeed counterproductive.

Suggested loops and pairings are thus:

 Throughput: FC on feed
 Flash Temperature: TC on H101 steam supply
 Flash Pressure: PC on H103 water

Inventory loops would be required on:

 LC on organics outlet from side of S102


 ILC on water outlet at base of S102

Below is a diagram showing all the control loops in place:


Question 2
The process shown in the diagram below is to be provided with a control system to
regulate the following strategic quantities:

 Process throughput
 Reactor temperature
 Product and byproduct compositions

Additionally, product and byproduct outlet temperatures, column pressure, which is


about 5 bar and relevant levels must be regulated.

Sketch suitable control loops and any other additions or altermatives to the flowsheet
required. Comment briefly on the justification for, and mode of operation of, each
control loop proposed.

Answer

The process shown is to be provided with a control system to regulate the following
strategic quantities:

 Process throughput
 Reactor temperature
 Product and byproduct compositions

Additionally, product and byproduct outlet temperatures, column pressure, which is


about 5 bar and relevant levels must be regulated.
Sketch suitable control loops and any other additions or altermatives to the flowsheet
required. Comment briefly on the justification for, and mode of operation of, each
control loop proposed.

Here is a summary of all the control loops:

 Process Throughput: FC on feed


 Reactor Temperature: TC on preheater
 Product Composition: QC on distillate (TC on top of column)
 Byproduct Composition: QC on reboiler (TC on bottom of column)
 Product Temperature: add heat exchanger to output stream
 Byproduct Temperature: add bypass to heat exchanger
 Column Pressure: PC on water inlet to condenser
 Inventory:
o Level in Reactor? (buffer tank or overflow)
o Level in reflux drum controlled by reflux rate
o Level at bottom of column controlled by byproduct rate

2.5.5 Exercise 3: Control of Adiabatic Flash

Consider all possible strategic control choices for an adiabatic flash separator. Identify
any which seem likely to give rise to problems. Draw complete control systems for all
alternatives.

Hint: This question has already been covered in Module 2.3.


Note that this material has already been covered in the theory section

It is now possible to devise control schemes for this adiabatic flash vessel. The basic
input-output block has 3 streams. It has been evaluated above that there are 2 control
degrees of freedom which means control valves on two of the three streams. That
leaves one other stream which also has a control valve on it to regulate inventory i.e.
to ensure that the mass balance does in fact balance.

Of the two streams left one would normally have a flow controller and be used to
regulate throughput. This could be any of the three streams. This leaves one other to
regulate a strategic variable.

From a knowledge of the properties of the adiabatic flash, the normal design
specification is the feed and only one further quantity, usually pressure, but
temperature or another flow could also be chosen.

For this example we will choose to regulate the pressure and one flowrate, not
necessarily the feed. This gives rise to six different alternatives. Each one is shown
below with a short discussion on whether it is feasible or not. Note that the shaded
valve indicates a valve used for inventory regulation. Not shown is the controller and
level measurement. Since all measurements in this example orginate in the flash drum
that end of the loop is omitted to simplify the diagrams.

Feed Rate and Pressure


Figures 1 and 2 below show the alternatives for this set up.
Figure 1 - Feed Rate Controlled - 1

Figure 2 - Feed Rate Controlled - 2

Figure 1 is a very common arrangement and will work well. Both pressure and level
loops have good adjustment-measurement sensitivity. There is some undesirable
interaction because opening the pressure control valve increases the rate of boiloff and
hence affects the level.
Figure 2 will not work. Although as noted above the vapour rate affects level, altering
the liquid rate does not change the pressure of the flash.

Vapour Rate and Pressure

Figures 3 and 4 below are the alternatives for regulation of vapour rate and pressure.

Figure 3 - Vapour Rate Controlled - 1


Figure 4 - Vapour Rate Controlled - 2

Figure 3 is another fairly conventional arrangement. Level control is good, although


there is interaction between both flow and pressure loops. That is if the flow of
vapour is increased then the pressure will decrease significantly and so the feed rate
will have to be increased also. Decreasing the vapour flow will have the opposite
effect causing the feed rate to be reduced.

Figure 4 is unworkable for the same reasons given above for figure 2.

Liquid Rate and Pressure

Finally figures 5 and 6 give the two alternatives for liquid rate and pressure.
Figure 5 - Liquid Rate Controlled - 1

Figure 6 - Liquid Rate Controlled - 2


Both of these schemes will work. However figure 6 should give stronger response of
both pressure and level control loops to their respective adjustments, and hence less
interaction. It would be preferred to figure 5.

Other Possibilities

It would be possible to devise other schemes in which the flow control loop acted
indirectly. For example, by adjusting a stream other than the one which is measured as
shown below in figure 7. Such arrangements should be avoided, as should any system
in which the operation of one loop, here the flow control, depends also on the
operation of another, here the inventory loop.

Figure 7 - Another Possibility?

2.5.6 Exercise 4: Control of Non Adiabatic Flash

Devise Control strategies for a flash separator with adjustable heat input where it is
required to regulate

 Pressure, feed rate and vapour rate


 Pressure, temperature and liquid rate
 Pressure, temperature and vapour to liquid ratio
 Pressure, vapour to feed ratio and liquid rate

Answers

In this exercise we have an additional degree of freedom hence we can control three
strategic variables as well as the inventory.

Pressure, Feed Rate and Vapour Rate

In this first case the first loops we can place are the two flowrates on the feed and
vapour. This leaves the inventory which has to go on the liquid rate (for mass balance)
and so the pressure is controlled by the heat input.

Pressure, Temperature and Liquid Rate

Once again we can put the flow control on the liquid rate. The temperature has to be
controlled using the heat input. This leaves the vapour rate and feed rate for pressure
and inventory. The inventory can only be controlled by the feed which leaves the
vapour for pressure which is desirable.
Pressure, Temperature and Vapour to Liquid Ratio

This example is really the same as the one above. The only difference is that there is a
ratio controller which measures the flow of both vapour and liquid but alters to the
liquid rate. Therefore all the rest of the loops are as above: pressure on vapour,
temperature on heat input and inventory on feed.

Is it possible to do it any other way? What if the ratio controller had altered the vapour
rate? This would leave the liquid rate for inventory and the feed rate for pressure. This
scheme would also work and the final right answer would depend on the conditions of
the system.

Pressure, Vapour to Feed Ratio and Liquid Rate

In this final example we can firstly put a flow controller on the liquid line. Next we
have to decide which stream the ratio controller is going to alter. Consider first putting
the flow controller on the feed line. This would leave the vapour line for inventory
which would not work. Consider instead putting the flow controller on the vapour line
and then the inventory on the feed. Finally the pressure is controlled using the heat
input.

2.5.7 Exercise 5: First Order Modelling

Question 1
A process with input u and output y is described by the equation:

What is the value of the time, in terms of the time constant , when the process has
reached 90% of it's final steady state.

Initially the process is at steady state with y = 5 units. At time t = 0, the input u is
subject to a disturbance of 5 units.

Sketch the process response using the following values for the gain and time constant
and comment on the results.

 Gain = 2, Time Constant = 10


 Gain = 3, Time Constant = 15
 Gain = 3, Time Constant = 5
 Gain = 1, Time Constant = 5
Answers

The solution to the first order equation is

At 90% of the final steady state value = 0.9

Remembering the definition of the gain

we can rearrange the above equation to get

and so

Hence when the output reaches 90% of the final steady state value the time is equal to
2.303 time constants.

A similar procedure can be carried out for when t = to get the result that at this point
the output has reached 63% of it's final steady state value.

The next part of the question requires a sketch of the respones with the given values of
gain and time constants. It is not necessary to plot lots of different points on each line
although this is possible.

Instead what is wanted is for a few points to be fixed:


 The initial value, y=5 at t=0
 The final steady state value evaluated from the gain
 The point at which t= i.e 63% of it's final value
 The point at which t=2.303 i.e. 90% of it's final value

Please find below a table summarising these points for all 4 processes in the question

Steady State 63% 90%

Process 1 15 (10,11) (23, 14)

Process 2 20 (15, 14.45) (34.5, 18.5)

Process 3 20 (5, 14.45) (11.51, 18.5)

Process 4 10 (5, 8.15) (11.51, 9.5)

Once these points have been fixed it is a simple matter to sketch the rest of the curve.

As can be seen from the graphs

 Increasing the gain increases the final value of y


 Increasing the time constant decreases the time taken for y to reach its final
value.
Question 2
From the following response curves estimate the parameters required

 The Gain

 The Gain

 The Gain
 The Time Delay
 The Gain
 The Time Constant

 The Time Delay


 The Gain
 The Time Constant
 The Time Delay
 The Gain
 The Time Constant

Answers

 The Gain
The gain is a measure of how the output changes in response to a change in the input.
The response above is instantaneous. i.e. the output changes as soon as the input
changes and is the same form of response. Note that the output has a different final
value from the input and so the gain does not equal 1.

The equation below is used to calculate the gain. The values of the parameters are
easily obtained.

 The Gain
The response this time is similar to the one above but this time as the input increases
the output decreases. Thus the process has a negative gain.

 The Gain
 The Time Delay
Again the response has the same form as the input but does not start at the same time.
This time difference between the step change in the input and the response in the
output is known as the time delay.

 The Gain
 The Time Constant

In this example the response is first order. i.e. it is lagging behind the input. To
evaluate the time constant you have to draw the tangent to the curve where the
gradient is greatest - in this case (and all first order cases) at the start of the response.
Where this line has the value of the final value of y is the value of the time constant.
 The Time Delay
 The Gain
 The Time Constant

Once more this is a first order case except it also has a time delay which is easily
calculated.
 The Time Delay
 The Gain
 The Time Constant

This final example is slightly trickier than the rest. This is because it is not simple first
order. The biggest gradient is now no longer at the start of the response. Instead it is
necessary to draw in the gradient and hence obtain approximate values for the
parameters.
2.5.8 Exercise 6: Controller Tuning

Question 1
Below are a range of responses of several process measurements to a step change in
an adjustment of 50% of its maximum range.

If each of the measurements represents an alternative control loop pairing with the
adjustment, rank these in order of desirability with a brief explanantion of your
choice.
Answers

Below is a list of the processes in order of desirability starting with


the least desirable:

Response (e)
This is the least desirable since it shows an inverse response i.e the measurement
variable (controlled variable) increases first before decreasing. This is not good since
the process could be upset and also this scenario is very difficult to control.

Response (b)

Next in order of desirability comes response (b). This is difficult to control for two
reasons. The first is that there is a large time delay befor the adjustment affects the
process. This makes the controller hard to tune and could lead to the process being
upset by disturbances before it is controlled back to the setpoint. The second reason is
the scale of the measurement. The range of the adjustment has been 50% however the
measurement has only decreased by approximately 15%. Thus the response will not
be very sensitive to disturbances.

Response (a)
This response is similar to (b) as regards the scale of the problem. Again the range of
the adjustment has been 50% but this time the range of the response is only
approximately 2.5%. Thus the controller is very insensitive to changes and would not
work satisfactorily. However note that on the plus side there is no delay time which
makes it easier to control.

Response (f)

The advantages of this response are that it is a comparable scale i.e a 50% change in
the adjustment causes a 60% change in the measurement, and it has a definite steady
state at the end. However on the down side it could be approximated to a pure time
delay i.e the response suddenly changes from one steady state to another. Again this
makes it hard to control.
Response (d)

We now have nearly perfect response. There is a good scale on the problem and the
response can be approximated to a first order response making it easy to obtain
controller parameters. However the only disadvantage is that there doesn't seem to be
a definate steady state at the end.

Response (c)

Finally the best response for controlling purposes is response (c). This response has
good scaling, an adjustment of 50% causes a change of 30%, can approximate to a
first order response and there is a definate steady state.
Note that your order may be slightly different but just as long as you picked up the
main points from the exercise. For a process adjustment to be paired with a process
measurement things to look for are:

Good Points

 An adjustment of 50% should give roughly the same change in the


measurement
 There should be a definate steady state
 Response should approximate to a first-order response
 An instantaneous response is better than a large time delay

Bad Points

 Look for scale to check for insensitivities


 Inverse responses are difficult to control
 Large time delays are difficult to control

Question 2
A control room chart record, see the figures below, shows a test on a process in
which a control valve was opened from 10% to 50% of its range and the effect on the
measurement, to be used in a control loop with the valve, recorded. The top and
bottom of the chart represent upper and lower range limits on the measurement.

Recommend settings (proportional band and reset time) for

 Proportional Controller
 Proportional Integral Controller
Answers

First of all evaluate the dimensionless process gain:

 Steady state deviation of measurement = 90% - 20% = 70%


 Change in input = 50% - 10% = 40%
 Dimensionless gain = 70 / 40 = 1.75

Next estimate the time delay and first order time constant using the graphical method
shown below (drawing the steepest slope tangent):
 Delay time = 2
 Time Constant = 2

Therefore the settings are

 Proportional only controller


 PB = 175%
 PI controller
 PB = 194%
 Reset time = 6.6 minutes

Question 3
The temperature of a chemical reactor is to be regulated by adjusting the flow of
heating fluid to a jacket.

The following data is available:

 Heating fluid maximum flowrate = 10kg/s


 Temperature measurement
o Minimum reading = 20oC
o Maximum reading = 220oC

The heating fluid valve is initially 20% open and the measured temperature is 140oC.
The valve is then opened to 40% and the temperature finally settles down at 173 oC.

 If the relationship between heating fluid flow and reactor temperature is


assumed to be linear what is the relationship?
 What is the value of the dimensional process gain and what are its
dimensions?
 What is the dimensionless process gain?
 If the reactor temperature is to be controlled at 180oC using a proportional-
only controller, what percentage manual offset would you recommend?
 Experimental tuning of the process suggests the use of a dimensionless
controller gain, standardised to a unity gain process, of 2.4. To what
proportional band setting does this correspond?
 A theoretical investigation of the process suggests a dimensional controller
gain of 0.2kg/s/oC. What proportional band does this represent?

Answers

If the relationship between heating fluid flow and reactor temperature is assumed to
be linear what is the relationship?

The equation for a straight line is

Temp = aFlow + b

So now we have to evaluate a and b. We know two points on the line

 Flow = 2 kg/s, Temp = 140 oC


 Flow = 4 kg/s, Temp = 173 oC

and so the equation can be evaluated to be:

Temp = 16.5 Flow + 107

What is the value of the dimensional process gain and what are its dimensions?

Dimensional process gain = 16.5 oC(kg/s)


The units are temperature/flow

What is the dimensionless process gain?

To get this multiply the dimensional gain by (kg/s per flow %) and divide by ( oC per
temperature %).

 Valve scaling is 0.1 (kg/s per flow %)


 Temperature range is 200 oC
 Temperature scaling is 2 (oC per temperature %)

Dimensionless gain = 16.5 * 0.1 / 2 = 0.825

Easier way...

The dimensionless gain really has units of temperature range % / flow range %.

 Change in temperature is 33 degrees which is 16.5% of range


 Change in flow is 20%
 Dimensionless gain = 16.5/20 = 0.825

If the reactor temperature is to be controlled at 180oC using a proportional-only


controller, what percentage manual offset would you recommend?

For the answer to this question determine what flow will give a temperature of 180 oC
and convert this to a valve position.

 180 = 16.5 Flow + 107


 Flow = 4.42 kg/s

This represents 44.2% of the valve range and this must be the controller output with
zero error to achieve the required temperature.

Experimental tuning of the process suggests the use of a dimensionless controller


gain, standardised to a unity gain process, of 2.4. To what proportional band setting
does this correspond?
The standard gain refers to a process with unity gain. Divide this by the process gain
to get the actual required dimensionless gain:

 2.4/0.825 = 2.91

This process has a gain of 0.825, which is less than one, so the required controller
gain will need to be greater than the standard gain, so you need to divide by the
process gain to increase it.

Proportional band is the reciprocal of dimensionless gain, expressed as a percentage:

PB = 1/2.91 * 100 = 34%

A theoretical investigation of the process suggests a dimensional controller gain of


0.2kg/s/oC. What proportional band does this represent?

This uses the same approach as the more involved procedure for the third part of this
question above.

Dimensionless gain = dimensional gain {(kg/s)/oC } * {oC/temp%} / {(kg/s)/flow%

= 0.2 * 2 / 0.1

=4

So PB = 100/4 = 25%

3 Control Systems For Complex Processes


The two modules in this part of the course deal with control systems for complex
processes. First of all we look at the control of distillation columns. Then we develop
the concept of the hierarchical method for devising control schemes for complex
processes. The case study section contains examples to practice the concepts
learned and contains interactive exercises.

 Module 3.1: Control of Distillation Columns


 Module 3.2: Extension of Hierarchical Decomposition
 Case Study Section

3.1 Module 3.1: Control of Distillation Columns

 Introduction
 Degrees of Freedom Analysis
 Controlling Pressure in Distillation
 Controlling Tops Composition in Distillation
 Distillation Column Control Example

3.1.1 Introduction
The aim of this module is to introduce the control of distillation columns. We will
start by analysing the degrees of freedom to establish how many and which control
parameters it is possible to control and/or manipulate. Then we move on to discuss
different ways to control the two most important parameters: composition at the
top of the column and the pressure of the column. Finally there are a number of
examples showing different control structures.

3.1.2 Degrees of Freedom Analysis


We will use the method developed by Professor Ponton in his Paper

Degrees of Freedom Analysis in Process Control, Chemical Engineering Science,


1994, Volume 49, No. 13, pp 2089 - 2095

to determine the number of control degrees of freedom in a distillation column. There


are two equivalent procedures based on the equation -

C.D.F. = Total No. of Streams - No. of Phases Present + 1

All we have to do is count all the streams in the process. Separately count the total
number of extra phases i.e. add up all occurrences of phases greater than one in all
units. The number of control degrees of freedom is the difference between these two
numbers.

Figure 1 below shows this method.


Figure 1 - Degrees of Freedom Analysis of Distillation Column

 Total Streams = 8
 Extra Phases = -3
 Degrees of Freedom = 5

So the number of degrees of freedom is 5. However, a typical control strategy for such
a process would use only 4 of these - feedrate, column pressure, top and bottom
composition. This is because the column and condenser are normally maintained at
the same pressure.

However, a valve could be placed in the line between. This would actually be
undesirable as reducing the condenser pressure will decrease the temperature driving
force available from the cooling medium.

3.1.3 Controlling Pressure in Distillation


In a distillation column it is usually necessary to regulate the pressure in some way.
Below there are five different methods described for doing this.
 Vent to Atmosphere
 Cooling Water
 Flooded Condenser - 1
 Flooded Condenser - 2
 Partial Condenser

One thing to note is that in none of them is a valve simply placed on the vapour line.
This would lead to the use of a large expensive control valve. Instead the pressure is
controlled indirectly involving the use of the condenser and/or reflux drum.

Vent to Atmosphere

Figure 2 below shows the easiest way to control the pressure in a column operating at
atmospheric pressure.

Figure 2 - Vent to Atmosphere

In this case the cooling water flow stays constant and the reflux drum is vented to
atmosphere. Thus the reflux drum and hence the top of the column are at atmospheric
pressure. The advantage of this scheme is that it requires one less control valve. The
disadvantage is that the tops have to be subcooled so that a minimal amount of vapour
is lost through the vent. Hence more energy is required from the reboiler when the
reflux is added to the top of the column.
Cooling Water

Figure 3 shows the most common method for controlling the pressure - adjustment of
the cooling water flow.

Figure 3 - Cooling Water

In this case if the cooling water flow is increased then more vapour is condensed and
the vapour pressure is reduced (and vice versa).

Flooded Condenser - 1

Figure 4 shows the classic flooded condenser approach.


Figure 4 - Flooded Condenser - 1

Again in this setup, as with the first example, there is no valve on the cooling water.
Instead the valve is in the liquid line between the condenser and reflux drum.

If this valve is closed then the condensed vapour i.e. liquid will build up and flood the
condenser. This has the effect of reducing the heat exchange area, thus reducing the
amount of vapour being condensed and hence increasing the pressure.

The valve can then be opened, the liquid level will fall, increasing the heat exchange
area and hence decreasing the pressure.

Flooded Condenser - 2

Figure 5 shows an alternative arrangement for a flooded condenser.


Figure 5 - Flooded Condenser 2

The first thing to notice about this setup is that the reflux drum and condenser are at
the same level. The second important point is that the vapour line, on which there is
the control valve, is very small in comparison with the overhead line. If the valve is
opened there is a small escape of gas into the reflux drum. This pushes the liquid
level down in the drum and up in the condenser, flooding it and reducing the heat
exchange area as in the last example.

Therefore to increase the pressure the valve is opened and to decrease the pressure the
valve is closed.

Partial Condenser

The final example is the control of a partial condenser.


Figure 6 - Partial Condenser

The above scheme is used if the overhead product is required as a vapour.

3.1.4 Controlling Tops Composition in Distillation


As well as pressure, the other parameter most likely to be controlled is the
composition of the tops product. The reason is that the final product will most
probably come from the top of the column and it is important to know its
composition. Again, as with pressure, there are many different ways of controlling
the tops composition. Three methods are described below.

 Reflux Rate
 Reflux Ratio
 Distillate Rate

Reflux Rate
In this first example the reflux rate is adjusted to control the composition of the tops
product.
Figure 7 - Reflux Rate

As the amount of reflux is changed so the temperature profile in the column changes
and hence the composition.

Reflux Ratio
The second example uses the reflux ratio as the control parameter.
Figure 8 - Reflux Ratio

When designing a distillation column it is usually the reflux ratio that is determined.
This can be kept constant throughout operation by using two flow indicators and a
ratio controller.

Distillate Rate
The third example is for high purity tops. It uses the distillate flowrate to control the
distillate composition.
Figure 9 - Distillate Rate

It can be shown that for a high purity column i.e. one with a large reflux, that the
composition of the distillate is sensitive to the distillate flow but insensitive to the
reflux rate. Therefore for a high purity column the control scheme outlined above is
used. It should be noted that tight control on the level in the reflux drum is required
using the reflux rate.

3.1.5 Distillation Column Control Examples


The following examples describe alternative control strategies of fairly standard
form.

 Pressure, Overheads Rate and Composition


 Pressure, Bottoms Rate and Composition
 Pressure, Bottoms Rate and Overhead Composition, With Partial Condenser
 Pressure, Overhead Rate and Bottoms Composition
 Pressure, Bottoms Rate, Overhead Rate and Composition

In all cases actual composition controllers are shown. These could of course be
replaced by inferential measurement from temperature, with or without cascade of
a slower analyser. Unless otherwise stated, it has been assumed that the feed rate to
the system is not available as a manipulated variable.

Pressure, Overheads Rate and Composition

This is a fairly standard configuration for a single product column, i.e. when the
bottoms streams is a byproduct, recycle or goes to further processing.

Although the overheads composition is regulated by adjusting the steam rate at the
base of the column, the response of the column to heat input changes is quite rapid,
and so this strategy is acceptable.

Pressure control on condenser cooling water is shown; of course any other pressure
control scheme would be acceptable.

Figure 10 - Overheads Rate and Composition


Pressure, Bottoms Rate and Composition
This is the analogous situation to the previous case, in the rather less usual
circumstances where a main product is withdrawn from the bottom of the column.

This does not work well, since either the bottom level, as here, or composition, has to
be regulated by adjusting the reflux rate. In either case the loop involves a long delay
due to the hydraulic lags on each tray.

It is probably marginally better to regulate composition by steam rate since this is a


more important quantity than level, although the two loops could be interchanged with
the steam adjusting the level, which is quite a good scheme, and the reflux
manipulating the bottoms composition, which is very poor. Fotunately this is an
unusual requirement, as main products normally come from the top of columns for
other reasons.

A standard flooded condenser pressure control system is shown.

Figure 11 - Bottoms Rate and Composition


Pressure, Bottoms Rate and Overhead Composition, With Partial
Condenser
This is not a particularly common strategy, but the arrangements for a column with
partial condenser are typical. The pressure in such a system is almost always
manipulated by a valve on the vapour product line. There is no reflux drum, and
reflux rate is often set implicitly by adjusting the cooling load on the condenser.

Figure 12 - Bottoms Rate and Overhead Composition, With Partial Condenser

Pressure, Overhead Rate and Bottoms Composition


This scheme should work satidfactorily as all adjustments are made at the same end
of the column as the related measurements. The pressure control scheme is the so-
called hot gas bypass. Note that the layout of condenser and reflux drum shown is
critical to the operation of this method, which is actually a variation on the flooded
condenser approach. The bypass is a very small pipe which bleeds vapour into the
reflux drum where it does not immediately condense. The pressure in the
system rises as the bypass valve is opened.
Figure 13 - Overhead Rate and Bottoms Composition

Pressure, Bottoms Rate, Overhead Rate and Composition


Since three regulated quantities are specified, the feed to the unit must be available
as an adjustment. Apart from this, the arrangements are similar to those of the first
example. Level control on the column base is not very satisfactory due to the lags
between the feed and the bottom of the column, but any other arrangement would
be worse.
Figure 14 - Bottoms Rate, Overhead Rate and Composition

3.2 Module 3.2: Extension of Hierarchical Decomposition


3.2.1 Introduction
In this module we aim to introduce an alternative method for the design of control
systems for complex processes. In the traditional approach the process engineer is
presented with a complete flowsheet and is required to specify the structure of all
the loops.

There are two disadvantages to this method.

 The complexity of the task.


o It is very difficult to choose the control structures for a complete
flowsheet from amongst the large number of alternatives.
o It is made even more difficult in that there is no defined procedure for
tackling this problem.
 It is impossible for any feedback from the control study to be taken into
account.
o Decisions about equipment will already have been taken.
o There will be considerable reluctance to change the process flowsheet
to accomodate possible control problems.

For these reasons a Hierarchical approach to the design of control systems will be
used. In this method controls are placed, as far as possible, on the flowsheet in its
simplest form. As the flowsheet is elaborated more loops are added.

This approach has significant advantages over existing practice.

 The task of placing all control loops on a complete and detailed flowsheet is
reduced to more manageable proportions.
 Attention is concentrated on strategically important control loops rather than
those of secondary importance such as level in holding tanks.
 Decisions on process and control system structures are taken together,
allowing the latter to influence the former in a manner not possible with the
conventional approach.

Below is a reference to a paper which has been published on this material.

Ponton, JW and Laing, DM, 1993, A Hierarchical Approach to the Design of Process
Control Systems, Chemical Engineering Research and Design, Vol 71, Part A, 181-
188.

3.2.2 Outline of Procedure


This hierarchical approach is already well established in the conceptual design of
chemical flowsheets. In the Douglas scheme the designer starts with a single block
showing only feeds and products and proceeds by sequential elaboration through
the following steps:

 Input-Output structure
 Recycle structure
 Separation sequencing
 Energy integration

The method used for designing control systems employs a similar set of steps:
 Feed and product rate control
 Recycle rates and composition
 Product and intermediate stream composition
 Temperature and energy balance control
 Inventory regulation

As with the design outline above, these controls are placed on the flowsheet at
different stages in the structure.

 Input-Output Structure
 Recycle Structure
 Functional Subsystems Structure

Advantages of this hierarchical approach are:

1. At any one time only a subset of the overall problem is considered. This
simplifies the problem somewhat and avoids the intellectual overload inherent
in the conventional approach.
2. The more important strategic control loops are emphasised, i.e. those
affecting the economic performance of the plant such as product rate, overall
conversion and product quality.
3. Less important decisions are left to the end, i.e. those associated with
operability such as regulation of inventory.
4. By proceeding in parallel with both process and control system design,
alternatives for both may be considered at each step.

We will now go through these steps in more depth and discuss what should be
considered at each level.

3.2.3 Input-Output Structure


At this stage we can only look at control of the feed and product. However this
affects the basic shape of the process and so control decisions at this stage
determine the whole strategy. Things to think about at this point are:

 Identify likely disturbances


o magnitude
o timescale
 Ranges and Scaling
 Sensitivities i.e Steady-state gains

3.2.4 Recycle Structure


The next stage in the design of the control system is to elaborate it further to include
the recycle structure. In some cases this may be done in more than one step. Things
to consider at this stage are:

 Sensitivities and ranges


 Interactions i.e RGA etc
 Functional controllability
 Multiple forward paths and possible inverse response
 Approximate dynamics
o Time delays
o Approximate time constants
o Simple models

3.2.5 Functional Subsystems Structure


The next stage is to include sections such as feedsystems, reactors or separation
schemes. Important points which can be added to the lists above are:

 Simple dynamic unit models


 Standard control schemes investigated for subsystems

This last point covers systems such as feed/mixing systems or separation schemes. It
includes equipment such as flash units or distillation columns. These are discussed in
a separate module on control of such units.

3.2.6 Discussion
At each of the above three stages it is important to understand what each stream is
there for and to ensure that as much control as possible has been included. Thus the
following steps are used:

 Consider each stream on the flowsheet at the current level.


 Why is it there?
 What will decide its flowrate?
 Can a complete, partial or tentative control system be added to manipulate its
flow and meet the above objective?
 Draw in as much of the control system as possible.

This is an extremely straightforward procedure. It is easy to see what all streams do


when the flowsheet is in a simplified form. On a complete PFS or ELD it will be
much harder to identify important streams and their interrelationships. Similarly, the
purpose of streams is most clearly established in the designer's mind at the point when
he has just identified them in the process design.

The above procedure is best shown as a worked example. There are four of these in a
separate section including one which is interactive and so requires the user to think!!

The final stages in the design of a control system are to consider energy integration
and inventory control.

For energy integration the general principal is to derive a control system for an
integrated energy recovery network from the unintegrated form by replacing valves on
utility streams with bypasses. The case of energy integrated distillation can be rather
more interesting. For example, in unintegrated distillation systems reboiler heat load
is often used as a manipulated variable for bottom product composition regulation.
Condenser heat load is used to regulate column pressure. However, what happens
when the condenser of the first column forms the reboiler of the second?

We regard the regulation of inventory, i.e. ensuring that all mass balances do in fact
balance, as the lowest level of control, to be achieved after all strategic control
systems have been specified. It is not usually difficult to place level control loops
once the number of alternatives has been reduced.

Consider a process or subsection with n inputs and outputs which contribute to a


single total material balance. (n-1) of these may be set by strategic control loops for
composition, temperature etc. One remaining stream must be used to ensure material
balance via level control in a liquid system or pressure control in a gas system or
implicitly by some inherent control mechanism such as the weir on a distillation tray.

3.3 Module 3: Case Study Section

The aim of this case study section is to reinforce material taught in Module 3: Control
Systems For Complex Processes. We start with three worked examples on developing
control systems for complex processes using the hierarchical method. These are
presented in order of difficulty. This is followed by three exercises. The first is
an interactive exercise on building up a control system. Choices are made along the
way and helpful hints and tips are given to help the user. The second is an exercise on
the control of distillation columns. The third is an exercise on heat exchanger
networks.

 Worked Examples of Control Schemes


o Example 1: A -> B
o Example 2: HDA Process
o Example 3: Styrene from Ethyl Benzene
 Tutorial Questions and Answers
o Exercise 1: Chloromethanes Process
o Exercise 2: Control of Distillation Columns
o Exercise 3: Control of Heat Exchanger Networks

3.3.1 Example 1 - A -> B

Process Description
In this first example we will consider a process turning a single feed A into a single
product B. The flowsheet consists of

 A reactor
 A flash separation unit

To begin with, however, we are concerned with the first level of complexity - the
input/output block.

Input-Output Structure
In this simple example there is one input stream and one output stream. The critical
decision to take at this stage is whether we regulate the feed or product rate.

Figure 1 shows the structure for feed rate control.

Figure 1 - Feed Rate Controlled

Figure 2 shows a similar method for controlling the product rate i.e. both the
measurement and adjustment are of the same stream.
Figure 2 - Product Rate Controlled

Figure 3 shows an alternative way to control the product rate. In this case the product
rate is measured as before. However, in order to maintain this value the feed rate is
adjusted.

Figure 3 - Alternative Product Rate Control

Clearly each strategy will lead to quite different control systems. It is reasonable to
suppose that there will be circumstances in which the choice of control strategy at this
point will affect the process design. For example, regulation of product rate, as shown
in figure 3 would favour a process design that minimised hydraulic lags, e.g. with a
reactor which ran full of liquid, and the minimisation of holding tanks. All these
alternatives should be explored individually using the criterion discussed in the
introduction. i.e. thinking about what the likely disturbances might be and how
sensitive the system is to these changes.

It is not implied that controllers will be of any particular type. For example, regulation
of product rate by adjustment of the feed is unlikely to be achievable by a simple PID
feedback scheme. It is assumed that when individual controllers are designed an
appropriate feedforward, model-based or miso/mimo scheme will be adopted.

It may be useful at this stage to determine whether inventory regulation is now


defined. In this process, with one input and one output, where either has been chosen
as an adjusted variable, then overall material balance requirements fix the other. This
may be indicated by putting in the part control system shown in figure 4. Doing this
will prevent any erroneous attempt to use that stream as an adjusted variable in
another loop. Thus future choices are simplified by taking all decisions which can be
made at earlier stages.
Figure 4 - Inventory Control Added

3.3.2 Example 2 - HDA Process

Process Description
The second example we are going to study is the Hydrodealkylation of Toluene. The
main reaction is

However there is also a side reaction which produces unwanted diphenyl.

Below is a flowsheet of the complete process. There then follows a description of the
process giving more information on the role of each part of the equipment.
Figure 1 - HDA Process Without Control

The homogeneous reactions take place in the range 600oC - below this temperature
the reaction rate is too slow - to 700oC - above this temperature a significant amount
of hydrocracking takes place - and at a pressure of about 35 bar. An excess of
hydrogen - a 5/1 ratio - is needed to prevent coking.

The toluene and hydrogen raw-material streams are heated and combined with
recycled toluene and hydrogen before they are fed to the reactor. The product stream
leaving the reactor contains hydrogen, methane, benzene, toluene and the unwanted
diphenyl. We attempt to separate most of the hydrogen and methane from the
aromatics by flashing away the light gases.

We would like to recycle the hydrogen leaving in the flash vapour, but the methane,
which is produced in the reaction and enters as an impurity in the hydrogen stream,
will accumulate in the gas-recycle loop. Hence a purge stream is required to remove
both the feed and the product methane from the process.

Not all the hydrogen and methane can be separated from the aromatics in the flash
drum and therefore we remove most of the remaining amount in a distillation column
- the stabilizer - to prevent them from contaminating our benzene product. The
benzene is then recovered in a second distillation column and finally the recycle
toluene is separated from the unwanted diphenyl.

To attempt to place control loops on this flowsheet as it stands would be rather


complex so once again we will use the hierarchical approach.

3.3.3 Example 4 - Styrene From Ethyl Benzene

Process Description
Styrene monomer is produced from ethylbenzene - EB - by the reversible reaction
shown below. The reaction is carried out in a vapour phase catalytic tubular reactor.
The feed to the reactor is vapourised and then superheated in a furnace to about
900K. The reactor operates at about 2 bara.

Conversion is about 30%.

Two side reactions of EB also take place, one to benzene and ethylene, and the other,
with the hydrogen formed in the first reaction, to toluene and methane.

Superheated steam is fed to the reactor along with the vapourised EB at a ratio of
1.7:1 of fresh EB feed. This serves a number of purposes: as a heat carrier, as a diluent
reducing product partial pressure and shifting the reaction equilibrium, but most
significantly to increase and prolong catalyst activity.

The boiling points of the organic liquids are

 Benzene 353K
 Toluene 384K
 Ethyl Bezene 409K
 Styrene 418K

The separation section first removes gases and separates water from organics in a 3-
phase flash after cooling to a suitable temperature. Styrene is then separated from the
other species by distillation. Ethylbenzene is then separated for recycle and the mixed
byproducts exported. Note that it is not possible here to avoid taking important
products from the bottom of columns.

3.3.4 Exercise 1: ChloroMethanes Process

Introduction
The third example in this section is a process for the manufacture of
dichloromethane [DC] and trichloromethane [TC] by successive chlorination of
monochloromethane [MC]. Tetrachloromethane, also known as carbontetrachloride
[CTC], is produced in significant quantities as a byproduct.

As mentioned previously, this process will be used as a practical example on the use
of the Hierarchical Approach to Control System Design. Thus at various stages
options will be given as to the next step forward. If the wrong option is taken further
advice will be given until the point in question is understood. If you have not looked
at the previous three worked examples then it is recommended that you do so. Also it
is your own best interest to carry out the following procedures properly and not just
skip from page to page!

Process Description
The reactions which describe this process are

The ratios of the three organic products is determined by the ratio of chlorine to fresh
monochloromethane feed. This is set as a ratio for a given production programme and
is not adjusted by online feedback control.

The HCl is exported as gas to another plant. A certain amount of monochloromethane


is acceptable in this stream but is economically undesirable. Dichloro and higher
products must be avoided.

The adsorption column C101 uses CTC already present in the process as a solvent to
remove the heavier organics from the reactor offgas.
The reactor, absorption column and recycle condenser operate at 5 bara. The rest of
the process is at ambient pressure.

As before we will look at the whole process without control before going back to
basics and starting from the input/output stage.

Figure 1 - ChloroMethanes Process Without Control

Input-Output Structure
The figure below shows the input-output structure for this process. At this stage it is
difficult to give a number of options to choose between but if you are in any doubt
as to where this stucture came from then please activate the link below for further
information.

Further Explanation
Figure 2 - Input/Output Structure Without Control

Now it is necessary to add control to this structure. As with the first two examples we
have to decide whether to control the feed or product rates.

In the process description it is mentioned that the ratios of the three organic products
is determined by the ratio of chlorine to fresh monochloromethane feed. Thus it would
seem sensible to set this ratio using a controller.

The next issue is concerned with the four product streams. What control is required
for them? A short discussion is included on this subject but before looking at the
answers please read through the following points.

 Is the temperature of these streams important?


 Is the flowrate of these streams important?
 What about the composisiton?
 Finally, think about inventory?

Product Control Discussion


The most inportant thing to note is that all these streams are product streams and so
the most important parameter is the composition. We have already been told to avoid
any DC and TC in the HCl stream. Thus it would seem sensible to put
composition indicators on each stream.

At this stage, however, we have not been given enough information about the
interactions between the streams shown so we do not know where the control valves
should go.

There is no point in putting flow controllers on any of the streams. It would have been
a different matter if we had been required to produce a certain amount of products.
But in this case that was not mentioned as one of the strategic aims. In fact, by
including complete control loops at this stage, we will severely limit the number of
choices at subsequent stages.

At this early stage in the design, inventory is not an issue. It is best left until all other
decisions have been taken. An important point which should be restated here is that
only one stream can be used for inventory purposes. Not enough information has been
given so far to enable any decision on inventory to be made, thus no inventory should
be added at this point.
3.3.5 Exercise 2: Control of Distillation Column

Develop control schemes for distillation systems where the following quantities are to
be regulated, in addition to column pressure.

 Feed, high purity distillate composition, bottoms temperature


 Distillate rate, distillate composition, bottoms composition
 Distillate rate, bottoms rate, bottoms composition
 Bottoms rate, distillate composition, bottoms composition

Go to Answers

Feed, high purity distillate composition, bottoms temperature

This control scheme is straightforward. First put a flow controller on the feed. Since it
is high purity overheads the composition is regulated by adjusting the distillate rate.
This is achieved with temperature in the top half of the column but some distance
down as an inner loop cascaded with an analyser. Bottoms temperature is controlled
with the steam rate. Finally the inventory loops - reflux drum level controlled by the
reflux rate and the reboiler level on the bottoms rate.
Distillate rate, distillate composition, bottoms composition

Distillate rate is set by flow control so reflux must be used for drum level. Overheads
composition can be regulated by boilup rate which is quite satisfactory because of the
low vapour inventory in the column. But how to do bottoms composition? This could
be done on bottoms flow, but reboiler level would then have to be by feed adjustment.
Unfortunately this has a long series of hydraulic delays from tray to tray. It might be
better to do reboiler level with bottoms and bottoms composition by feed, although
this is an unusual scheme!

Distillate rate, bottoms rate, bottoms composition

Both product rates by flow control. The level in the reflux drum can be controlled by
the reflux rate. The bottoms composition is controlled by the reboiler steam rate. This
leaves the level in the reboiler either floating or unsatisfactorily controlled by the feed
rate.
Bottoms rate, distillate composition, bottoms composition

Distillate composition controlled conventionally either by distillate or reflux rate


depending on purity, the other adjustment used for reflux drum level. Bottoms
composition by reboiler steam rate leaving the same difficulty with reboiler lever as
before.
3.3.6 Exercise 3: Control of Heat Exchanger Networks

The objective of this exercise is to devise suitable control schemes for the refinery
heat exchange systems shown below. It is required to regulate all product stream
temperatures and that of the desalter drum, which is an adiabatic reactor.

Question 1

Figure 1 - HENS 1 without Control


Answer

As mentioned previously, the objective here is to control the temperature of the


output streams. For a flowsheet without energy integration the control loops are
easily added, since in general they are independent and do not interact. The control
valve is placed on the utility stream so as to leave the process stream flowrate
unaffected.

In this example the temperatures of streams h1, h3 and h7 can be controlled in this
way. These all have utility streams to be used as the final stage in the control
procedure.

For the other process streams there is one heat exchanger used each time which has
another process stream as the cold medium. It is not possible to change the flowrate of
these streams so some bypass arrangement must be provided to manipulate the heat
load.

Shown in the diagram below is a simple answer to the problem. Bypass systems could
well be more sophisticated than the single valve arrangement shown. They could
involve flow measurement and ratio adjustment to minimise flow disturbances,
although such a system would incur a capital cost and reliability penalty.

The next decision, therefore, is which stream do we bypass? The choice, obviously, is
between the cold stream or hot stream. The answer, shown below, is the hot stream as
this will have a more direct effect. Also this does not interact with the cold stream
which in turn affects a number of other exchangers.

The next piece of equipment to consider is the desalter. Remember that this is a
reactor so there is no way of bypassing the whole thing. The only alternative is to
bypass the previous heat exchanger as shown below. The disadvantage of this is that it
disrupts the flow of cold fluid through the exchanger and hence affects the
temperature of stream h1. This can be compensated for by the cold utility as described
earlier.

The final thing to be discussed here is the output from the furnace. The usual way to
control this is via the flowrate of fuel as shown below.
Figure 3 - HENS 1 with Control

Question 2

Figure 2 - HENS 2 without Control

Answer

Example 2 is rather more complicated although the same principles apply. As can be
seen the heat exchangers have increased in number and are more fully integrated
with rather a lot of stream splitting.
The first thing to note is that once again streams h1, h3 and h7 can be controlled using
the utility streams. For the others we bypass the final heat exchanger in the series each
time. This would seem logical as it is the output temperature from this exchanger that
we are interested in.

Note that once again it is the hot stream that we are bypassing. Also note that the
furnace is being controlled in the same way as before.

The difficulties arise when we consider the desalter. In this example the input stream
has been split into 4, so instead of bypassing 1 heat exchanger we are bypassing 4.
This has an effect on streams h5, h6, h1, h3 and indirectly on h7.

h5 and h6 are affected the most as the heat exchangers in question are the ones which
directly control their temperature. It may be possible to keep the flowrate through
these exchangers constant by using flow controllers on the relevent section of stream
c1. That would leave streams h1, h3 and h7 which are controlled by the more flexible
method of using utility streams.

This example highlights some of the difficulty involved in controlling integrated


plants. You may save money on equipment and energy costs but you also have to
ensure that the resulting plant is feasible from the point of veiw of controlling the
process.

Figure 2 - HENS 2 with Control


Part 4: More Advanced Concepts
 Module 4: Further Controller Design
 Module 5: Development of Mechanistic Models
 Module 6: Multivariable Systems
 Module 7: Frequency Response Techniques
 Case Study Section

4 Module 4: Further Controller Design


This module introduces alternative ways of tuning a PI controller. The first couple are
based on the ZN open loop method. Then the closed loop method is discussed along
with a few variations.

 Introduction
 Open Loop Methods
o Controller Parameter Correlations
o Example of Controller Parameter Correlations
o Systems With No Steady State
 Closed Loop Methods
o Closed Loop Zeigler Nichols
o Damped Oscillation Method
o Example of Damped Oscillation Method
o Relay Based Tuning Controller
o Example of Relay Based Controller Tuning
 Real Controllers
o Compensated Derivative Action
o Anti-Windup Controller
o A Practical Controller Algorithm
 References

4.1 Introduction

Essentially all tuning methods are based on a model process for


which optimal controller settings have been determined. Tuning procedures are thus
a matter of matching a real or simulated process to the tuning model.
Largely, the more sophisticated the model, the better the controller performance will
be.

The simplest model, already discussed, is the Zeigler-Nichols delay plus lag model.
This can be used in either open or closed loop form. The original open loop ZN
parameters have already been described and will not be further discussed here. In
practice the ZN open loop settings are unduly conservative. Higher gains (smaller
proportional band) should be used with most processes.

Closed loop tuning, Zeigler-Nichols style, with full amplitude continuous oscillation,
is not a satisfactory approach for most actual processes! However, it is perfectly
satisfactory for use on a simulated process. Other methods with reduced amplidude
oscillation or with the process under partial control are more suitable for use on-line
and form the basis of a number of self tuning controllers.

Some processes cannot be tuned directly by open loop methods because they do not
have a stable steady state. Such systems are said to be open loop unstable. A simple
example of this is level control where in the absence of some sort of feedback, a
vessel will either run dry or overfill. A rather different approach is required for such
loops.

The notes below describe tuning methods for single loops. For tuning multiloop
systems each loop should be tuned separately, with all other loops disconnected, i.e.
without control. In general, tunings obtained this way will have to be modified to
allow for loop interactions, see multiloop control systems.

4.2 Open Loop Methods

Described below are a number of open loop tuning methods. First a couple of
correlations are shown based on the ZN open loop response. Then a method for tuning
systems with no steady state is described.

4.2.1 Controller Parameter Correlations

A number of alternative models to the delay plus lag have been proposed for both
open and closed loop tuning, e.g. delay plus two lags, three lags, etc. Howver, the
Zeigler-Nichols model is in fact perfectly satisfactory, but better controller parameters
may be derived for it.
For a model system with delay time Td and first order lag T1, the fractional dead
time Tf is defined as:

Tf = Td/(Td + T1)

This is used as parameter to determine settings of dimensionless overall loop gain,


the product of controller and process gains:

= controller gain * process gain

and dimensionless integral action time:

= controller reset time / (Td + T1)

Settings for PI controllers have been proposed by Lopez ( see ref 2) and Cianconne
and Marlin (see ref 1). Estimates of their values, taken from published graphs, are
given in the table below.

These settings appear to be much more satisfactory than the simple Z-N settings
which are also shown in the table below.

Ciancone Lopez ZN (open)

Tf

0 1.1 0.23 5.8 0.4 - -

0.1 1.1 0.23 5.8 0.5 8.1 0.33

0.2 1.8 0.23 3.1 0.6 3.6 0.66

0.3 1.1 0.72 2.1 0.7 2.1 1.0

0.4 1.0 0.72 1.7 0.8 1.35 1.32

0.5 0.8 0.70 0.91 0.9 0.9 1.65

0.6 0.59 0.67 - - 0.67 1.98


0.7 0.42 0.60 - - 0.43 2.31

0.8 0.32 0.53 - - 0.25 2.64

These settings were obtained by minimising the integral of the absolute value or
square of the error following a disturbance.

4.2.2 Example of Controller Parameter Correlations

Analysis of the open loop response of a first order process gives the following
parameters:

 Delay time = 5
 Time Constant = 20
 Gain = 3.5

Hence evaluate the fractional dead time and from this the controller parameters using
the three correlations in the table above.

 Fractional Dead Time = 5 / (20 + 5) = 0.2


 Ciacone Parameters:
o = 1.8
o = 0.23
o Controller gain = 1.8 / 3.5 = 0.514
o Controller reset = 0.23 * (20 + 5) = 5.75
 Lopez Parameters:
o = 3.1
o = 0.6
o Controller gain = 3.1 / 3.5 = 0.886
o Controller reset = 0.6 * (20 + 5) = 15
 ZN Open Loop:
o = 3.6
o = 0.66
o Controller gain = 3.6 / 3.5 = 1.029
o Controller reset = 0.66 * (20 + 5) = 16.5
4.2.3 Systems With No Steady State

These are best tuned using a closed loop procedure. If a step change is used in open
loop, then the equivalent time delay is easily identified. However, at the end of the
time delay period the process output will rise (or fall) continuously until some
physical limit is reached.

The average rate of change during this time, normalised by dividing by the size of the
input step, may be taken as the reciprocal of an equivalent first order timeconstant and
used to estimate parameters using Z-N or either of the above methods.

A difficulty arises is determining the steady state gain of this kind of process, since it
has no steady state. In fact the time constant as calculated above already includes a
sensitivity which is somewhat equivalent to process gain.

4.3 Closed Loop Methods

Now we will move on to show a couple of closed loop methods. First the basic
Zeigler Nichols method and then a simple closed loop tuning procedure with damped
oscillation.

4.3.1 Closed-Loop Zeigler Nichols

The classic closed-loop Zeigler Nichols tuning procedure is to advance the gain of
proportional only controller until the process is oscillating continuously at a constant
amplitude.
The gain required to achieve this (the ultimate gain, ) and the period of oscillation
(the critical period, Pu) provide parameters from which controller settings are derived
as shown below.

Controller Type Gain Reset Derivative

P /2 - -

PI /2.2 Pu/1.2 -

PID /1.7 Pu/2 Pu/8

Note that it is not necessary here to estimate the steady state process gain as the
controller gain, or proportional band, settings are exressed in terms of that actually set
on the controller to make the plant oscillate.

Allowing a real process to oscillate between anything but rather tight limits is clearly
not likely to be acceptable except for control loops which are noncritical. Two
alternatives are therefore used.

 Allow a damped oscillation which dies away as the process approaches a


setpoint under control with nonoptimal, but known, controller settings.
 Enforce a limit on the amplitude of oscillation.

The first of these is similar in principle to fitting an open loop model to a delay plus
lag, except that it is mathematically a bit more complicated. A simplified approach is
discussed below.

The second approach is applied to self tuning controllers.

4.3.2 Damped Oscillation Method

Obtain a controller setting which keeps the process within acceptable limits. Use only
proportional action. Note the gain setting on the controller, .

Disturb the process with a step change. This is most conveniently done by changing
the controller setpoint. The behaviour of the process will be similar to the figure.
Damped response of process under non-optimal proportional control

The process is oscillating rather less than it would if the controller gain were . It is
also oscillating rather more slowly, since increasing the gain of a controller speeds up
the response of the process to which it is attached.

The period of oscillation, Pd, e.g. the time between successive peaks or troughs, will
thus be slightly greater than Pu, but not much. Hence we can say that typically:

Pu is approximately equal to 0.95 Pd

We can estimate the increase in gain required to cause the system to oscillate
continuously if we know the amount which the oscillation decays between half cycles.
This would be the ratio between the amplitude of e.g. the first peak and that of the
trough which follows it. This is not so easy to estimate, as the zero point of the
sinusoidal respose may not be obvious. Instead take the ratios of peak-to-trough and
following trough-to-peak distances. The gain would need to be increased in proportion
to this ratio to cause steady oscillation.

4.3.3 Example of Damped Oscillation Method

In the response shown above the proportional controller had a gain of 1.5.

There are 4 complete sinusoids in about 18 time units, so

Pd is approximately equal to 18/4 = 4.5

Pu = 0.95 * 4.5 = 4.28

The first peak to trough distance is about (340-80) = 260 units

The following trough to peak is (80-260) = - 180 units

The ratio of their magnitudes is thus 180/260 = 0.69

The gain should be increased by this factor, i.e:

= /0.69 = 1.5/0.69 = 2.17

Hence the ZN PI controller settings are:

Gain = /2.2 = 2.17/2.2 = 0.99

Reset = Pu / 1.2 = 4.28/1.2 = 3.57

4.3.4 Relay Based Tuning Controller

Another way to make a process oscillate is to use a control system which is


intrinsically oscillatory. An on-off control system has this property, which is why it is
seldom used in chemical process control.

However, restricting the size of change which the controller can produce will restrict
the amplitude of the oscillations. These can thus be small enough not to upset the
process, or its operators!
This controller has a parameter deadzone, which defines the minimum change in the
measured variable required to produce a change in output. The other parameter
defines the size of the output of the controller. This is set to be + output/2 when the
controller is on and - output/2 when it is off.

The algorithm for the on-off controller is as follows.


error = setpoint - measurement
if (error>deadzone) adjustment = output / 2
if (error<deadzone) adjustment = - output /2

The size of output can be quite small, although it must be enough to swing the
measured variable across the range of the deadzone. This latter must be larger than
any noise in the process.

This controller will oscillate at a period equal to Pu. (Actually if the output is too close
to the value required to swing the measurement across the whole deadzone, the period
will be slightly longer. It should be enough to cover 4 or 5 times the deadzone.)

If the observed amplitude of the oscillations (maximum to minimum) is a the ultimate


gain of the process can be shown to be:

= (4 * output) / ( * a)

4.3.5 Example of Relay Based Controller Tuning


We can use the above algorithm in an actual experiment. In the example below the
first order process parameters are:

 Delay time = 5
 Time constant = 20
 Gain = 3.5

Other information required by the program is a value for the controller deadzone =
0.01.

Now let us run the experiment with the output of the controller restricted to 4. This
gives the following response.
From this response we can determine the value of a, the observed amplitude of the
oscillations and the period of oscillation.

 a = 3.1
 Pu = 19

and hence evaluate the ultimate gain and the tuning parameters.

 = (4 * 4)/( * 3.1) = 1.64


 Controller gain = 0.747
 Controller reset = 15.83

Now let's try the same process again this time restricting the controller output to 10.
From this response we can determine the value of a, the observed amplitude of the
oscillations and the period of oscillation.

 a = 7.75
 Pu = 19

and hence evaluate the ultimate gain and the tuning parameters.

 = (4 * 10)/( * 7.75) = 1.64


 Controller gain = 0.747
 Controller reset = 15.83

As you can see the two examples give the same values for the tuning parameters. How
well do these values actually control the process at a setpoint of 0.5?
This approach can form the basis of a self tuning controller which is switched into on-
off or relay mode to determine Pu and and hence tuning parameters which then get
set into the controller.

4.4 Real Controllers

Real controllers differ from the ideal ones used in most exercises in several ways:

 Their inputs and outputs are dimensionless (0-100%, 4-20mA, +/- 5V ...)
although measurements are normally displayed in the control room in
appropriate units.
 They work with proportional band rather than gain.
 Derivative action is not implemented as a true derivative.
 Limits are put on the amount of integral action applied by the controller.

The first two points have been discussed earlier.


4.4.1 Compensated Derivative Action

True derivative action is neither possible nor desirable. If a true derivative controller
saw a step change in a measurement it would have to produce an infinite output. Thus
a small but sharp noisespike, very common on practical electrical measurements,
could produce a very large, sudden and spurious, change in a process adjustment. this
is obviously highly undesirable, and so the derivative term on any controller must be
modified to prevent this happening.

Derivative action is sometimes omitted altogether. In previous tuning examples we


have use only PI control. In practice derivative action should be used only when very
precise control is required onmeasurements which are known to be reliable. If in
doubt, use PI only.

All real controllers use what is called compensated rate action. This introduces
another parameter which is used to damp the derivative action to minimise the effect
of noise spikes.

Most controllers nowadays are implemented digitally. To obtain derivatives a


computer based controller would require numerical differentiation. This is an
unsatisfactory procedure, since it also introduces noise. The rate compensated
controller is described entirely by o.d.es, and so is implemented using only
numerical integration.

Here is the theory of the compensated controller.

We require:

Without performing numerical differentiation.

Let:

Solve for z by numerical integration.

Then for small :


is called the rate compensation parameter, and can either have a fixed value or be
made, e.g. 0.1 * deadtime of process.

This corresponds to a form of filtering of the derivative term which prevents it goint
to infinity in the presence of a step disturbance.

4.4.2 Anti-Windup Controller

Consider what will happen if for any reason an error persists in a controller
measurement. A simple integral action integrator would just keep on building up to a
larger and larger value. If later the measurement error is removed, the integral is still
there and will take about as long as the original error persisted before it unwinds back
to a sensible value.

To prevent this happening. the size of the integral has to be limited. Either an arbitrary
maximum may be set, or else integration may be turned off in certain circumstances.
These would typicaly be when eithet the measurement or controller output reach the
limit of their ranges.

The phenomenon is called windup, and the solution anti-windup.

4.4.3 A Practical Controller Algorithm

To illustrate these points, we provide a Fortran 90 subroutine used to simulate a PID


controller. This was used in off line simulations, but could equally well be
implemented in a process control computer to run a process in real time.
subroutine control_pid (in,out,setpoint,pband,Ti,Td,reset,rate)
! Proportional+integral+derivative control
! pband is proportional band, Ti, Td integral, deriv action times
! reset holds integral ; rate holds compensation LPF
! *** must be distinct variables
! *** must not be changed by calling program!!
! out must be given a value at time zero to initialize integration
! all variables scaled 0-100
! setpoint must be in this scale too!
! If pband is negative, the controller is reverse acting
use times
doubleprecision :: in,out,setpoint,pband,Ti,Td,rate,reset
intent (in) :: in,setpoint,pband,Ti,Td
intent (inout) :: out,reset,rate
doubleprecision :: xin, xout, error, gain, action, alpha, edot,TTi
!
TTi=Ti
if (Ti<=0d0) TTi=1d20
gain=abs(100/pband)
action=1 ; if (pband<0d0) action=-1
alpha=0.1*Td ! rate compensation parameter
call scale100 (in,xin) ; call scale100(out,xout)
error=(setpoint-xin)*action
if (time<=0d0) then ! set initial conditions
reset=xout/gain-error
rate=0
end if
if ((xout<100d0) .and. (xout>0d0)) reset=reset+deltat*error/TTi
! .... antiwindup check
! Derivative stuff...
edot=0 ! default
if (alpha>0d0) then
call solve_ode1 (rate,error,1d0/alpha)
edot=(error-rate/alpha)/alpha
end if
! Finally...
xout = gain*(error + reset + Td*edot)
call scale100 (xout,out)
return
end

4.5 References

1. R Ciancone and T Marlin, Tune controllers to meet plant objectives, Control, 5,


50-57, 1992
2. A Lopez, P Murrill and C Smith, Tuning PI and PID digital controllers, Inst and
Control Systems 42, 89-95, 1969
3. KJ Astrom, Adaptive feedback control, Proc IEEE, 75, 2, 185, 1987
4. KJ Astrom and T Hagglung, Automatic tuning of PID controllers, Inst Soc
America, Research Triangle Park, NC, USA, 1988.

5 Module 5: Development of Mechanistic Models


 Introduction
 Procedure
o Example 1: Flow into a Tank
o Step 1: Draw a Diagram
o Step 2: Material and Energy Balance Equations
o Step 3: Rate Equations
o Step 4: Equations of State
o Step 5: Count the Equations and Unknowns
o Summary
 Example 2: Fed Batch Reactor
 Example 3: Sparged Steam Water Heater

5.1 Introduction

A complete discussion of how to develop chemical engineering models is really an


complete course in chemical engineering, with practical experience besides! The
objective of this section is thus only to show that there is a good, as opposed to a bad,
way of constructing models. If this approach, which is systematic, is followed, then,
assuming that the model builder's chemistry, physics and chemical engineering
understanding of the process being modelled is correct, then there is a better chance
than otherwise that the final model will be correct.

The basis of this approach to model building is that the equations which constitute a
model are not arbitrary mathematical entities, but have a consistent physical basis.
There are certain types of equation which describe different aspects of a model. A
knowledge of this helps to ensure that all equations are written down.

Another point is that the set of model equations should be written without regard to
how they might be solved. A confusing feature of some of the classical textbooks in
chemical engineering is that the author, who knows the answer he wants to reach,
slips steps in the solution process in amongst the construction of the model. The way
to avoid confusion is to write all the equations down without attempting to perform
any algebraic manipulation or simplification.

This section might therefore be subtitled: How to avoid algebra. The final models
may be less elegant than those in some text books, but the computer will have no
difficulty in solving them.

5.2 Procedure

The procedure which should be used to construct these models will now be
introduced using a simple example of the flow into a tank.
5.2.1 Example 1: Flow into a Tank

Step 1: Draw a Diagram

This is the first thing to do in attempting to model any system. Write down,
conveniently on the diagram, but also separately with definitions, units etc., all the
quantities which you can identify, assigning each an appropriate symbol. These helps
to ensure that you identify and understand all the variables in the problem. Also
identify any quantities which you expect to know or be able to choose, such as
dimensions, physical properties or known feeds.

The figure below shows a tank whose outflow, through a pipe in the base of the tank,
depends on the level in the tank. Two feeds are shown, but these are assumed to be of
the same material.

Here we might identify the following initially.

Known quantities:

 F1, F2 - known mass flow rates (e.g. kg/s) into the tank
 A - tank cross section area, m2
 - fluid density, kg/m3

Unknowns:
 M - mass holdup (e.g. kg) in the tank at any instant in time
 L - mass flowrate out of the tank
 h - level (e.g. m) in the tank

Other variables will be found to be necessary as we proceed with building the model.
These can just be added to the list.

Step 2: Material and Energy Balance Equations

Balance equations are the most fundamental type of equation, and have the
advantage that we know exactly how many there must be in any problem. About
every element which we model, there will be one material balance for each chemical
species of interest present, and, if temperature effects are being modelled, one energy
balance.

In this case we have one species and no temperature effects. We are modelling a
single element, namely a tank, and so there is one balance equation.

Note the following.

 This is an unsteady state or dynamic (i.e. time varying) system, so the balance
equations are ordinary differential equations.
 These must have the form:

 In this equation quantity must be an extensive quantity, e.g. kg, moles,


Joules etc. Do not try to write so-called material balance equations involving
temperatures or concentrations.
 Flowrates must similarly be kg/s, kmol/hr, Watts, etc.
Step 3: Rate Equations

The flowrates on the r.h.s. of the material balances are typically determined by rate
equations which relate them to some driving force or potential variables. Here for
example:

This says that the outflow rate depends on the pressure P at the base of the tank, the
downstream pressure Pa and a discharge coefficient.

There must be a rate equation for every unknown material or energy flowrate which
appears in a balance equation.

In the above equation there are some new variables to add to the list. Cd and Pa are
typically known quantities. P is an additional unknown. It is an intensive quantity in
the thermodynamic sense and physically represents a driving force which can cause an
extensive quantity, e.g. mass, to flow. Other driving force variables are temperature
(heat) and concentration (flow or reaction of a chemical species).

The general form of a rate equation is usually:

Step 4: Equations of State

In this context we give this name to the class of equations which relate intensive
thermodynamic quantities to each other and to extensive quantities. A classic equation
of state (e.o.s.) is the ideal gas equation, which does not however apply here. It relates
intensive temperature and pressure and extensive molar amount.

We look here for an e.o.s. which relates the intensive potential variable P introduced
above to extensive an holdup variable, or more generally, a relationship amongst
several such variables.

Note first that:

And also that:


These equations together provide a relationship between the intensive driving force
variable P and the extensive mass holdup M.

Equations of state are sometimes not entirely obvious to identify, and they sometimes,
as here, introduce further variables. Their general form is:

If h and had not been identified as variables of interest we might have here gone
straight to a single e.o.s., namely:

Step 5: Count the Equations and Unknowns

If the model has been correctly formulated then there will be the same number of
equations and unknowns.

There are 4 equations above. The unknowns in summary are M, h, L, P. Hence the
model appears to be correct.

Summary

Notice that there are three categories of equation:

 Balance
 Rate
 EOS

There are also three kinds of variable:

 Extensive holdups
 Extensive rates
 Intensive potentials
The balance equations involve holdups and rates, the rate equations rates and
potentials, and the e.o.s. potentials and holdups. This establishes a circular
relationship as shown in the figure.

5.2.2 Example 2: Fed Batch Reactor


The second order reaction:

is carried out isothermally in a tank which is initially filled with B, and to which A is
gradually added as the reaction proceeds. When the reaction is sufficiently
complete, the product is run off. The feed rate fA kmol/hr of A is a known function
of time. The emptying process is not to be modelled.

Balance Equations
There is one for each component. These will be written in molar quantities.

Here the m are the mass in the tank in kmol and R is the rate of reaction in kmol/hr.

Rate equation
At this stage we have to define the rate R:

R = k CA CB V

Here V is reactor volume and k a specific rate constant in suitable units.


Since the feed rate of A is not an unknown there is only one rate equation.

There are thus, so far, 4 equations in 7 unknowns: mA, mB, mC, R, CA, CB and V

Equations of State
At least 3 further equations are required. It is clear that the concentrations and
molar holdups are related, so that:

It is less obvious that an equation can be written for V in terms of the molar densities
of the species, which are assumed to be known:

This completes the model.

5.2.3 Example 3: Sparged Steam Water Heater


A supply of hot water at temperature T is required at a rate W (kg/s) and this is to be
provided by sparging steam at temperature Ts at a rate S (kg/s) through a mass M
kg of water in a tank. The tank is fed with cold water at temperature Tc.

As before the first thing to do is to draw a diagram of the process as shown below.
Known Quantities : M, Tc, Ts, W, S

Unknown Quantities : C, H, T, qc, qs, qw

The objective of the exercise is to obtain equations so that all the above quantities can
be calculated.

Balance Equations
The material balance can be written formally as:

(It is, however, intended that the amount M of water in the tank shall be constant,
i.e.:

So in this case the material balance reduces to an algebraic equation from which the
required cold water feed rate C kg/s can be determined.

C=W-S
This does not affect the model building process, although it will affect any solution
method.)

There is a single energy balance differential equation:

Here H, in Joules, is the energy holdup in the vessel and the q are energy flowrates
(e.g. in W) associated with the various water and steam streams.

Rate Equations
These are somewhat different fron their classical form, but the following can be
thought of as defining each of the three q variables above.

Cp is taken as an average heat capacity, as a known constant latent heat and the
implied enthalpy reference temperature is zero on the temperature scale used.

Equations of state
Since it is not always immediately obvious what e.o.s. are required, it is useful at this
point to count up the equations and unknowns so far identified as a guide to the
number of equations of state required. Note that this will indicate
the minimum number of equations required, since when these are written it may be
found that they introduce further unknowns.

In the 5 equations above the 6 unknowns are : C, H, qc, qs, qw, T

One further equation is needed.

The temperature in the tank, T, is related to the total enthalphy by the equation of
state:

H = M Cp T

We thus have a complete model with 6 equations and unknowns.


6 Module 6: Multivariable Systems
This module introduces a method for pairing up measurements and adjustments for
multivariable systems.

 Introduction
 Analysis of Interaction
 Definition of the Relative Gain Array
 Calculation of the Gain Matrix
 Calculation of the Relative Gain Array
 Interpreting the Results
 Alternative Method for 2X2 Systems
 Strategies for Reducing Loop Interaction
 Tuning Multiloop Controllers

7 Module 7: Frequency Response Techniques


This module covers a more mathematical frequency response technique of tuning
controllers although in a more practical than mathematical way. There are three
sections in this module. The first introduces the concept of frequency response and
defines the key terms. Next the Bode Diagram is described in more detail. Finally
there is a section on process controllability. The Case Study Section gives the chance
to plot bode diagrams and relate them to the actual control of the process.

 Module 7.1: Frequency Response


 Module 7.2: The Bode Diagram
 Module 7.3: Process Controllability - under development

8 Part 4: Case Study Section


The aim of this case study section is to reinforce material taught in Part 4: More
Advanced Concepts.

 Exercises
o Exercise 1: Mechanistic Modelling
o Exercise 2: Relative Gain Arrays
 Laboratory Session

There are a number of experiments in the Virtual Control Laboratory which are
linked to this part of the course.

o Introduction to Zeigler Nichols Tuning


o Closed Loop Tuning
o Mixing Process
o Non-Linear Process
o Comparing Controller Tuning Methods
o The Blending Process
o Frequency Response Simulation

9 Appendices
9.1 Hot Water Tank Experiment

9.1.1 Aims
The aims of this experiment are

 To show how a control system can be set up


 To introduce the use of on-off control

This example is based on an experiment in the control laboratory. It is proposed to


control the temperature of the water in a tank using on/off feedback control. This is the
simplest and most commonly experienced type of control. It is possible to determine
the relationship between the precision obtained and the amount of oscillation.

9.1.2 Theory
There are two theory sections associated with this experiment:

 Setting up a Control System, discussed in Module 1.2


 Operation of the Control Loop
There now follows a description of the action taken by the controller. The algorithm
which controls the on-off control loop which operates the heater has the following
steps.

The temperature of the water is measured.

It is then compared with the desired temperature.

If the actual temperature is too cold then the heater is turned on.

If the actual temperature is too hot then the heater is turned off.

If this procedure is used then it would lead to the situation where the heater was
switched on/off as soon as the water was either too hot or too cold. The result would
be reasonably tight control around the setpoint but the heater would be switched on
and off rapidly. This is unsatisfactory as the switch is sure to wear out.

One way to get round this disadvantage is to build hystersis into the control
algorithm i.e. a deviation from the setpoint known as the dead zone.

Thus the operation of the control system now has the following steps.

The temperature of the water in the tank is measured.

It is then compared with the desired temperature.

If the actual temperature is below the sepoint by a specified amount then the heater
is turned on.

If the actual temperature is above the setpoint by a specified amount then the heater
is turned off.

9.1.3 Pre-Experimental Questions

Below are a number of pre-experimental questions. Please think about the answers
while you are running the experiment. The answers will be given after the simulation has
been run.

 What is the effect of altering the flowrate through the system?


 What is the effect of altering the deadtime of the controller?
 What are the disadvantages in using on/off control?
 Can you think of any situations where on/off control could be used and others
where it definately can't?

9.1.4 Experimental Apparatus

Below is a diagram and photograph of the experimental apparatus.

Note that it consists of

 A tank of water. Cold water enters at the


bottom of the tank and the hot water
overflows out the top thus keeping a constant
volume. The flowrate can be altered as part of
the experiment. The tank is insulated to reduce
heat losses to the environment.
 A temperature measurement device.
 An electric heater.
 A controller. This adjusts the power to the
heater either turning the supply on or off.

Hot water demand rate Required deviation from the setpoint


[kg/s] [%]
1.0 5.0

9.1.5 Procedure
Before the experiment is run, it is necessary to give some information about the system:

 The temperature of the cold water entering the tank is 20oC.


 The setpoint temperature of the hot water leaving the tank is 40oC.
 There is a fixed heat input of 200W.
 The flowrates in and out are equal hence the mass in the tank stays constant at
10kg.

To carry out the experiment simply choose appropriate values for the hot water demand
rate and the dead zone for the controller and click on Run Experiment. This is done by
clicking on the button using the mouse and dragging the cursor to the desired value.

The computer then takes these values, runs a program and produces the results. These
come in the form of a graph showing
 The measured temperature
 The setpoint temperature
 An indication of when the heater is on

It is possible to run the experiment any number of times. Please try different hot water
demand rates e.g. flows of 0.5 and 1.5 kg/s and different precision of control e.g.
dead zones of 1.0 and 20 %.

9.1.6 Answers
Click here for the answers

 What is the effect of altering the flowrate through the system?

As the flowrate through the system changes, the rate at which the water changes
temperature is affected.

When the flowrate is small (0.5kg/s) the water cools down at a slower rate. This means
that the heater does not have to be switched on as much.

However when the flowrate is large (1.5kg/s) the water cools down quicker due to the
increased flow of cold water entering the tank.

Note that with the higher flowrate more power is required to heat the tank (the heater is
on for longer) and so this leads to more oscillations in the heat input.

 What is the effect of altering the deadtime of the controller?

The deadtime of the controller determines the accuracy of the temperature of the water.
For example when the deadtime is 5% the temperature will oscillate between 38 C and
44 C. Likewise for 20% the temperature will oscillate between 32 C and 44 C.

Notice that with on/off control it is impossible to have a deadtime of 0%. There has to
be some offset between the setpoint and measured variable.

Finally with 1% the temperature is controlled between 39.5% and 40.5% although there
is some overshoot due to the amount of power added.

 What are the disadvantages in using on/off control?

The disadvantage in using on/off control is the oscillation around the setpoint. Although
on/off control is very simple to implement there will always be this oscillation.
Therefore there has to be a trade off between the accuracy required by the controller
(how important is it to be exactly at the setpoint) and the amount of wear and tear on
the valve.

 Can you think of any situations where on/off control could be used and others
where it definately can't?

The main situation where on/off control is used is in a domestic heating system. The
boiler is switched on when the water is too cold and off when the water is too hot. This
works well since it is not important for the temperature to be exact.

One area where on off control will not work is with controlling flowrate. It is not a good
idea to have a flowrate oscillating between fully on and fully off!!

9.1.7 Summary
In summary therefore, this is the simplest and most commonly experienced type of
control, although not the most commonly used in chemical processes. Advantage of this
type is that it is inexpensive and extremely simple. The disadvantage lies mainly in the
oscillatory nature of the control with oscillations becoming more rapid as precision is
increased. Therefore this type of control is suitable only for those applications where it
can be used alone and close control is not essential. It is also applicable only to systems
which have a significant capacity as will be seen in a later example.