Documentos de Académico
Documentos de Profesional
Documentos de Cultura
Systems
To my sons
Drazen, Dinko, Alan and Ozren
Fran Jovic
Professor of Control Engineering,
University of Zagreb;
Section Manager
Computer Systems and Intelligent Process Control
Engineering,
A TM Zagreb Company,
Yugoslavia
Preface IX
This book reflects the considerable current industrial interest and investment
in process control systems. The use of computer systems in process control
can provide great benefits, and it is estimated that efficiency can be increased
by up to 30%. It is not surprising, therefore, that there have been considerable
efforts by system designers and users to introduce and use such systems.
Process hardware is integrated into a complete production system through
data processing. It is for this purpose that technical specialists (e.g. electrical,
mechanical, electronics, communication and process engineers and program-
mers) are involved in data processing.
The scope of this book is therefore to assist in the selection of computer
hardware and software that match the functional specification of the data
processing component of a particular system. The principal points covered in
this book are set out below.
Part One: Production process hardware for a standard process is outlined
and the information processing hardware is described. Large
mechanical process hardware and process information devices
(e.g. sensors and control elements involved in the process)
create a coherent production unit, or system, which can be the
control unit (i.e. the basic process unit). The hardware processes
are described and the mathematics explained. This enables the
application of control laws in order to linearize the process about
its working point, as well as a stratification of process control
tasks. Linearization allows process control constants to be re-
duced and therefore the information capabilities of the computer
and communication hardware are also presented and analysed.
Part Two: Although the majority of process control tasks can be run using
hardware functions, the natural trend towards software process
control as more cost-effective is presented. The software tasks
in a process control system are dealt with in this part and, in
particular, the specific role of software in data processing. The
design of the information system based on the software instaIred
in the computer is discussed. These systems may be small, large,
x Preface
Fran Jovic
Zagreb
Acknowledgements
Figures 3.2, 3.3 and 3.5, and Tables 3.1 and 3.2 © 1983;
Figure 6.9 © 1980; and Figure 6.17 © 1983.
Part One
Systems, Processes and
the Role of Process
Control Hardware
1
Signals, systems and process control
1.1 INTRODUCTION
3
4 Signals, systems and process control
operator. Since signals carry useful information between these parts, a de-
scription of signals and their information content is also presented in this
chapter. The hardware systems are also described based on the response function
concept and each system part is considered as a system in itself and then the
parts are integrated into the complete system.
Since the basic functions of such a system can be modelled, implemented
or optimized using computers, a brief description of automata is also given.
Processes are classified according to their utilization criteria. Analytical
relations are given for some linear and nonlinear processes in order to support
their design into bigger system parts. The rules for connecting process and
process controllers are presented and analysed according to the change of the
response function and their static and dynamic behaviour.
1.2.1 INTRODUCTION
This part describes a computer system designed for production control and
the analysis and design of such a system. It also explains what is meant by a
production process and a production plant. A production control system is
developed, installed, used and dismantled by man, therefore the roles of the
designer, engineer, operator, etc. are also considered.
Before a production system can be developed and implemented, it must be
decided whether such a system is really necessary. Therefore, a preliminary
(or feasibility) study should be carried out. The essential make-up of any
process is the hardware, which consists of elements such as pipes, reactors,
vessels, wires, valves, motors, etc. The quality, connection and interrelation
of process elements determine the operational performance of the process
since it is these elements which ensure materials and energy storage, processing,
exchange and recycling. Examples include reservoirs, condensers, heat ex-
changers, heaters, pumps, motors transformers, generators, vessels, reactors,
etc. Integrated into large units, or plants, it is these units that enable produc-
tion - with associated energy and materials transfer - to be carried out. Such
units include power plants, liquefied gas storage, machine tools, rolling mills
and pipeline systems. The integration of units in a production plant is comple-
mented principally by hardware elements such as pipes, valves, power cables,
pulleys, conveyors, gears and clutches.
Individual units in a plant are often fitted with sensing and control elements
- typically, temperature indicators, level indicators, pressure indicators, valve
motor controllers, electrohydraulic flow controllers, pump motor controllers,
etc. Measuring, sensing and controlling the process variables (e.g. tempera-
ture, voltage and flow and control of interconnections and control elements)
can be used to control each particular production unit. An integrated group of
A system approach to process control systems 5
preliminary, feasibility
modelling study
Phas:..... _---- ___ ~_ ...... ,
.
/ ... -~-.-.-.-.-.,.,
I analysis and \ , development and
I design of control I i
.
preparation of
process and plant
: system I
j I
hardware
! (
:l
~ realization of
i realization of process
I control system I . and plant hardware
\
',------~--~')
,
r--J-·-·-·j
'. I
implementation
realization
commissioning work
phase
maintenance
r.
1 -iii " ...r block control
'3 ..... l J .
~
- 1 1
; ~
~
"
;.
I
~
1 set-point functional protection data
' lid
~ group processing
,g ........ . controls system
...
C:: . controls and alarms control
ai '
. systems
1 ..1! .
~ 10 1 .1 "..t..
signal conditioners/ a _< . ~
~
"
.~~
~
, tL .1I
l ;; - :.~
iDterf~iDg~~~t
..."... . . . -. · ~
J in terconnection
l J
1
signal 0
transducers ~ drives
l l ~~ oj p~wer plant block
/
/ memory
up to lOIS (bit) ,
\
I
/ operational
\
\
/ memory speed \
/ 100 (bit/s)
\
I \
motives
I \
I
I
I
I
•
\ reflex paths I
\ '4 I
\ muscles ------ ... "- receptors I
/
\ actions ./
"- stimuli /
~ external world /
/
.-
computer control system
" ,,
I \
\
\
I
memory \
I I
108 - 10 10 (bit) \
main memory
I
lOs - 10 7 (bit)
I operation speed \
\
10 6 - 10 7 (bit/s) \
\
I
I
up to 106
,,
\
\
I process I
\ transducers /
'--------_.
\ /
,'--------~~ ./
Figure 1.3 Comparative data on speed and operational memory for human and
computer control systems.
System deliverer Customer
lOO% 50% 50% 100%
I I I
requirement definition
~s I 14% I and analysis (1)
1 22 17% nary design (2)
%18 A l I
,l 19% I main projec
21%
I coding and debu
~ factory testing (4
~s {/\ I
l 5% I hardware installation U
1 13% I 8 12% installation and work (6)
II site installation;
I
e maintenance one
I 12% I ,er handover (7)
Figure 1.4 Typical allocation of the activities carried out during process control
system development: c::::=J, data according to Wolverton (1974); data
0,
according to Koch and Hoffmann (1978); 6., data according to Siemens (1990).
A system approach to process control systems 11
production
process
goods, information and energy. The process is a controlled system and forms
part of a set of production or processing functions executed in and by means
of process hardware. The features of a process are usually measured by process
quantities, or conditions, termed process variables. The control of process
variables is achieved by the control equipment (e.g. electromagnetic valves,
transformer tap positioners). The processes are situated in the production
environment and they are affected by human, time-space, materials/raw
materials, energetic, socioeconomic, informational and environmental aspects
(see Figure 1.5). These aspects are interrelated in a complicated way. Table
1.1 shows the main process descriptors according to the given production
environment aspects.
1. Human aspects mostly determine the mode of process control. The given
number of operators, their work, educational background and qualifica-
tions, and process control responsibilities are dependent on the particular
Table 1.1 Main process descriptors
Human aspects Time-space Materials and Energetic aspects Socioeconomic Informational Environmental
aspects raw materials aspects aspects aspects
aspects
Manually Fast/slow process Fluids Energy-generating New process Process control Environmentally
controlled process process information safe process
Manually/ Concentrated/ Raw pieces Energy- Capacity additive Supervisory Process with
automatically distributed transforming process process removable
controlled process process process information contaminator
Automatically Continuous/ Single parts Process with Capacity Auxiliary process Process with
controlled process discrete process energy supply replacement information unremovable
process contaminator
Automatically Aperiodic/cyclic Sheets and
controlled process process textiles
with learning
Automatically Deterministic/
controlled process stochastic
with learning and process
prediction features
A system approach to process control systems 13
process type and the control system applied. Human aspects seem to be
the most influential factors in proper system functioning and overall pro-
cess productivity.
2. Time-space aspects determine the character of the process type, showing
five main process characteristics: speed (slow, fast), spacing (concentrated,
distributed), continuity (continuous, discrete), periodicity (cyclic, aperiodic)
and determinacy (deterministic, stochastic). Time-space aspects influence
the complexity, automaticity and sensitivity of a particular process.
3. Materials/raw materials aspects show the four types of materials handling
processes: fluid processes, raw pieces, single parts and processes for sheets
and textiles. Materials/raw materials aspects influence the type of process
variables, process sensitivity and the type and extent of process control.
4. Energetic aspects determine the main process categories: energy-generating
process (e.g. a chemical reactor), energy-transfonning process (e.g. a
hydroelectric power plant) and process where the energy is supplied for
materials processing only (e.g. a machine tool). Energetic aspects are
prominent in the calculation of process production efficiency and affect
the type and extent of process control applied.
5. Socioeconomic aspects determine the adaptation to the motives and inter-
ests of people engaged in the production process. Consequently, the main
types of process development that can occur are purchase of new equipment
for a new process, enlargement of existing production capacity or simply
replacement of existing production capacity. All estimates are based on
the analysis of future expenses and on expected benefits.
6. Informational aspects determine the selection of process instrumentation
and data processing equipment.
7. Environmental aspects determine the environment protection counter-
measures.
According to the main descriptors given in Table 1.1 there exist 8100 different
process types. All the processes have four features in common:
process economy;
process capacity;
process speed;
process safety.
Different stages of these process features have led to different stages of process
complexity and these are shown in Table 1.2.
To obtain certain qualitative and even quantitative relations between process
features, and some production environmental aspects, a simplified model is
shown in Figure 1.6. For instance the amount of material on the market de-
creases with the increase of product and investment costs. The extra demand
for goods on the material market promotes production, which in tum increases
the size of the labour force needed to make the goods and thus increases
Table 1.2 Main classes of process complexity
Control source Type of process response Process complexity level Process capacity
operating product
labour market
costs
Figure 1.6 A simplified relation model of the production process and some
production environmental aspects: • >. increasing action;~.
decreasing action.
operational and investment costs. A rise in product market also increases the
production. The model in Figure 1.6 may also be used for quantitative pur-
poses by determining the conversion factors Ai,i+l between specific produc-
tion factors for a particular product, and by determining initial quantities and
time relations ..!lti,i+l between the causes and consequences for specific pro-
duction factors.
The following main categories of process control system are recognized
according to the application field.
1. Energy generation, electric power transmission and distribution systems.
2. Chemical and petrochemical industry.
3. Metallurgical industry.
4. Traffic and transportation systems.
5. Cement, paper and pulp processing industry, and food and fermentation
industry.
6. Environmental systems.
7. Metalworking and electrotechnical industry.
16 Signals, systems and process control
8. Mining.
9. Instrumentation and laboratory equipment production.
The export of engineering products for the above industries covered 70% of
the total world exports for engineering products in 1987 (Bulletin of Statistics
on World Trade in Engineering Products, 1989), making a total of approxi-
mately $168 billion.
The amount of control equipment for the engineering products of these
same industries is calculated to be between 1% and 5% of the total.
1.3 SIGNALS
1.3.1 INTRODUCTION
---------,
output signals data
display
control
board
[[I input
valves
sensors [[I signals
PROCESS
input signals
Analog signals can be either continuous or quantized but digital signals are
always quantized signals. The most important types of signal for the presen-
tation, analysis and execution of control functions are time-discrete signals,
because they represent the actual data scanned, acquired and processed in
process control systems.
Signals represent the result of actions in process control systems and they are
measured and registered at different points in the process or in the process con-
troller. These signals can be put into an analytical form which allows quick
comparison, analysis and the preparation of data processing algorithms in
process control. Quantized signals are analytically described by the z-transform
and continuous signals by the Laplace transform. These analytical tools are
briefly presented in this chapter, in order to support the further system design.
Signals carry information and the information content of signals can also be
given analytically. Data are formed as based on the information content in the
signal processing part of the process control system. Thus information content,
data acquisition, data processing and process control execution are highly
influenced by signals.
Some process control data are communicated to the process system in a coded
form through communication channels in order that their content is not affect-
ed by noise. The basic analytical features of these channels are also de-
scribed. Details of process control data communication are given in Chapter 3.
considered as an analog signal, but if its amount has only quantized values,
then it is regarded as a digital signal. Analog signals fall into two main types.
1. Time-continuous analog signals, such as the pressure gauge signals.
2. Time-discrete analog signals, such as pressure indicator signals.
Digital signals fall into two main types.
1. Binary signals, such as proximity detector signals.
2. Count signals, such as count rate pulses from a liquid flow meter.
These four types of signal are shown in Figure 1.8.
There are some standard signal forms called singular signals which are used
for signal, system and process analyses. These signals can be aperiodic or
periodic, and are given in Figure 1.9 as:
1. The unit step signals, S(t), which has the value one in time instant to and
lasts infinitely long, and before this time instant it equals zero.
2. The unit pulse signal, del(t), commonly named the delta function, which
has the 'infinite' value within a very short time interval to - e, to + e, in
all other time instants it equals zero, and the total signal intensity of the
signal equals one, i.e.
I to+e
to-e
del(t) dt = 1, e ~ 0 (1.1)
3. The square pulse signal which equals one in a given time interval tlo t2
and outside this interval it equals zero.
4. The stationary sinusoidal or cosinusoidal signal, i.e. the signal which
changes its value according to the following equation
u(t) =a sin (COot + CPo) (1.2)
Example
The process liquid tank given in Figure 1.10 should be provided with necessary
signals for its proper functioning. The levels in the tank are:
L 1, level of pump deactivation;
- L2, working level of pump;
- L 3 , level of security output.
Solution:
The following signals should be provided:
Llo digital signal for the low level indication;
L 2, digital signal for level indication and pump on/off control;
L3, digital signal for level indication and pump on/off control;
S 1, digital or analog level indicator and alarm signal for the total shutdown
function.
y - A • time continuous analog
~----~ signal
!
I
I ~ . t
Figure 1.8 Signal parameter values for analog and digital signals.
T
del(t)
J'o+e
,o-e del(t) dt =1
• input
81
, u(t)
a)
b) us(t)
/
/
/
t
us(t)
c)
d)
u(3)z-3
1
_
u(9)z-9
.,t
is its plausibility but it lacks accuracy for long time intervals and it is not
representable to computers.
2. Tables where the signal is presented either as compared to other signals
or in given time intervals; signal tables are usually not easily surveyable
by the process operator, but computed signals can be stored in the com-
puter as tables of raw data.
3. A sum of a series of singular functions, as shown in the three examples
in Figure 1.12; such a description can help in the fast analysis of response
signals in process control.
4. The Laplace transform of the signal time function (see Table 1.3); by
Signals 23
-t = 0, 1, 2, 3, ... , n, n + 1
t.t
S(t) 1 - Z-l
S(t) = S(n)
s
S(t)e at at = an
s-a 1 - az- 1
Z-l
S2 (1 - Z-I)2
t =n
u(t)
u(t)!
I
I
I
I
I
I
I
I
I
I
,,
I
(1.7)
The logarithm basis of equation (1.7) is agreed to be equal to two. Thus the
information unit is defined as a binary amount of the information named
Shannon or bit (binary digit).
Example
Determine the information amount which can be obtained from the occur-
rence of one character out of 16 equally probable character events. Accord-
ing to equation (1.7) where ld = dual logarithm (lOg2)
26 Signals. systems and process control
analog signal
LlA = 1 V
As{t)
~~~----------------~
b)
Tp2 = 0.2 s
I = ld 16 = 4 (bit) (1.8)
The entropy of the signal source is the mean value of the information amount
per character out of n independent characters. It is equal to
(1.9)
The entropy of the analog signal HA (observed over longer time intervals, so
that the measured values are independent) when converted by an eight bit
analog-to-digital converter is equal to, Pi = 1/256,
where I I are explanatory brackets for determination of Ii' The entropy of the
digital signal is one bit. The mean information rate is
R = r . H (bit/s) (1.11)
where r is the mean number of signal characters in a time unit. The analog
signal with the mean number of characters given in Figure 1.13b as 5i1Q
(charfs) has a rate
RA = 5 . 8 = 40 bitls 0.12)
and the digital signal from Figure 1.13a has a rate
RD = 1000. 1 = 1000 bitls (1.13)
For a given signal with a mean information rate r and for a given communication
channel with the capacity C, where r < C, there is always a way of how the
signal can be coded into another message source so that the information
transfer can reduce the amount of errors and be independent of the noise in
the channel. Process signals are unfortunately not coded into other signal
forms (unless processed by a 'smart transmitter'), but the necessary channel
capacity can be calculated according to the relationship
C = B . Id( I + SIN) (1.14)
where B is the width of the frequency band (Hz) of the information channel
and SIN is the signal to noise power ratio.
Example
When B = 20 kHz (an audio channel), S =1W and N = 1 mW, the channel
capacity is
C = 20 . 10 3 Id 1001 = 200 kbitls ( 1.15)
1.4 SYSTEMS
1.4.1 INTRODUCTION
1. Causal, meaning that there was no excitation in the system before t < 0
when regarding the system output at t ~ O.
2. Time instant, meaning that the response is an instant answer to the input
amplified by a given factor a, i.e.
y = au a ~0 (1.16)
3. Dynamic, meaning that the system contains certain memory elements of
analog (e.g. condensers, tanks) or digital type (e.g. on/off valves, switches,
bistables).
//
/
I
den _ k) = {O
1
for n
for n
*"= kk ( 1.17)
Example
The response of a time-invariant discrete system to the unit function is shown
in Figure 1.15. As depicted in Figure 1.I5b, there is a consistent time invariance
of the system response to the unit function. The input function uo(n) is shown
in Figure LISe. The output is calculated as follows:
for n = 0, Yo(O) =0
for n = 1, yo(l) = uo(O)h(1) + uo(1)h(O) + uo(2)h(-I) = 2
for n = 2, yo(2) = uo(1)h(1) + uo(2)h(0) = 7
for n = 3, Yo(3) = uo(1)h(2) + uo(2)h(1) + uD(3)h(O) = 7
for n = 4, yo(4) = uo(1)h(3) + uo(2)h(2) + uD(3)h(1) + uo(4)h(O) = 8
for n = S, yo(S) = uo(4)h(1) + uo(S)h(O) = S
for n = 6, Yo(6) = yo(5)h(1) = 1
where
ho(n, n - k) = wo(n, m) (1.27)
is the weighting function of the time-variant discrete system. (See the
analogous situation for continuous systems in Figure 1.17.)
(a) t d(n)
-...2
hn(n)
1 t I"n n
0 0
+ hn(n)
~. ~: t~~._ _ _ _._
(b) d(n)
n
K K
7 7 8 5
_LI. L1 I
2
2 3 4 5 2 3 4 5 6
(1.32)
and the system response is given as the continuous convolution of the input
and response function, i.e.
(1.33)
(1.34)
where
O>(t, 't) = h(t - 't, 't) (1.35)
Figure 1.17 depicts a way of presenting the calculation of output function y(t)
under time-variant conditions.
F (t)
I
I
I
I t
~~~----~------r---~------T---~T-----"
~
dF(t)
h ( t) =----
dt
Figure 1.16 A response of a linear continuous system to unit function and unit
step excitation.
2. Instant response systems, i.e. systems where the output follows the input
according to the relationship
w(t,'r}
T
Figure 1.17 The calculation of the system output function under time-variant
conditions.
II j X S-l X
---
~ B
Z-l
~
r--- - C :> y
-=== A I
Example
A delay line with the delay of 10 ms when scanned at a rate of 1 ms has the
following properties as a dynamic system
A = C = 0, B = D = 1 and transfer function
(1.44)
Linking subsystems into a system can be done in the simplest way by regarding
the flow chart of signals in a system. The places of subsystems in a system
can be exchanged in linear systems (or linear parts of them). Let us suppose
that the system has the structure given in Figure 1.19a. When the paths of
signals through the system are followed, the input/output relationship can be
obtained
y = H 4H 2H 1u + H 4H 2H 1v + H 5H 4H 2y (1.45)
where
v = H4"ly (1.46)
The same system can be rearranged differently as shown in Figure 1.19b
supposing T = Hi! and 4 = H4"l, or after solving equation (1.45) as
F = (H 1H 2H 4)/[1 - (HIH2H4)(H5/Hl + H3/H4)] (1.47)
as depicted in Figure 1.19c.
5
a) I
I
v
2 4
---- y
,
b)
rl T
1 5
I
I
f1
I
u
~I 2
1 4
~~ .y
- -1
1 "3
- -1
·1
"5 =H- 1
4
"3 =H- 1
1 =H l' 4=H4' 5' 3
c) I--------PROCESS--~-~~---- --,
I ~
I
u ~I
1 2 4
I
y
I I
L _____ ~ ____ ~_~~ __________ J
r---~----~~--~---~~--~----~-I
r
I I
I I
I 1 5 I--- I
I I
I
I I
CONTROLLER I
I I
I
I
I
I
'--- 3 4 - I
I I
L ______ ~ _______ ~ _____ ~ _____ ~ _~~
A control system is made up of a controlled system and its control part. The
control system can either be a closed-loop or an open-loop system.
1. The closed-loop control system, where the control action is made de-
pendent on the measurement of the controlled variable, usually adjusts
the manipulated process variable(s) according to the difference between
the measured value(s) and the reference value(s). The net effect of the
adjustment process is a change of some basic features such as response
time and disturbance sensitivity. The closed-loop control system can be a
feedback or a feedforward system.
2. An open-loop control system does not directly utilize the measurement
of the controlled variable. It consists of the same parts as the closed-loop
control system, but with no direct feedback (or feedforward) connection.
The simple process control system shown in Figure 1.21 is a feedback control
loop that affects the forward control element. The process part includes direct
and indirect controlled systems, while the process control part includes the
feedback elements, the summing junction and the forward control element.
The adjustment of the input variable by the forward control element produces
the controlled variables as an input to the direct controlled system. The direct
controlled variable occurs at the output of a direct controlled system. The
feedback element produces a feedback signal that is subtracted from the reference
variable value at the summing junction, thus producing an error control signal
which is fed into the forward control element to adjust the manipulated variable
to the input variable.
A direct controlled variable is fed to the indirect controlled system and,
together with the disturbance, produces an indirect controlled variable as the
a)
u= 0, y= (j u=(j, y=l u= 0, y=l
u=l, y=2
b)
STEP INPUT OUTPUT INTERNAL STATES
n Ul Yl II ql q2 q3 q4
0 '/1 (j 0 0 0 1
1 1 0 1
2 0 1 1
'"
'/1
'"
----------
''/"1
3 1 0 0 1
'" '/1
4 1 1 '0" ~
5 'I" 1 '0" 0 1 0
----
6 1 2 0 0 1 0
7 0 1 1 0
'" '"
r
c)
I 1 n
• 1 I n
~
'" 0 1 '"
Figure 1.20 A simple set of computer input, output and internal states.
The basic analytical concept of process control systems 39
manipulated
variable
•
direct
controlled
system
input CONTROLLER
variable
t final
control
sensing
element
error
control
signal feedback
element
summing
junction
reference set-point
1--_11--
input element signal
The sets of equations defining the models vary greatly. Some may be as
simple as a single linear differential equation; others will be as complex as
sets of nonlinear partial differential equations. Most will lie somewhere in
between. However, it is worth noting that lumped systems are always modelled
by ordinary differential equations whilst distributed systems have to be modelled
by partial differential equations.
The computing effort required to solve a particular model is not straight-
forwardly related to the number and type of equations, although effort does
generally increase with size. The effort also depends on 'stiffness' which is
best described by saying that to obtain accuracy the computer is obliged to
use very small step lengths, thus requiring greater computational effort.
A good modeller is involved in making compromises between speed of
solution (very important if the model is to be used on-line for control) and
accuracy. For this reason modelling is often described as being both a science
and an art.
REFERENCES
2.1 INTRODUCTION
Postulate 1 The basic hardware elements of the production process and plant
are called the basic process units of the entire process plant. A basic process
unit generally consists of:
large mechanical process hardware that includes all process devices, sensors,
transducers and signal converters;
the controller for the basic process unit that includes all the hardware and
software, where all the procedures, relations and logics, as well as a
connection with the superimposed control system(s), are executed.
41
42 The basic process unit
Postulate 2 A basic process unit is responsible for at least one specific process
function. Any part of the unit is based, designed, manufactured, mounted and
put into action only if it completely fulfils its part in the functioning of the
whole unit.
Example 1
The functions of a crude-oillnatural-gas separator can be split up into the
functions of the controller for the basic process unit, the process controller for
the whole measurement station, the synoptics of the basic process unit and the
special device function as shown in Figure 2.2. The basic process unit is
responsible for three main functions.
1. Processing process signals A, C, D and E and command B, where
flow transmitter data (A) are transmitted directly to process controller
level;
level transmitter data (C and D) are processed for the synoptics of the
basic process unit with the logic functions
Cl = C . (Q21 + 1 Hz) (2.1)
Dl = D . (Q22 + 1 Hz)
where Q21 and Q22 are set/reset sequential functions generated at the
process controller and 1 Hz is a 1 hertz signal generated at the basic
process unit controller for the alarm warning purposes of the local
synoptics;
data from flow pulse transmitter (E) are preprocessed at the level of
special device functions where they are converted into pulse-cumulative
and flow-rate data;
the command (B) is issued according to the logic relation
B=A·L·J1+F (2.2)
where L and J1 are set/reset sequential functions generated at the
basic process unit controller and F is a logic variable generated at the
r-----------------------------------------------------------------,
I I
I I
I I
I superimposed control system I
I I
I I
:L ________________________________________________________________ ...JI
process controller
I
I
connection
IG
process 2~ " 21~
controller 00
"
en 00
-eN
functions ......
~eN
~
'"
eN
'-
~
00
eN
II U .....
eN
.....
'" 00
" 001
'" '"
",,,, 0
00 r-
eN eN
~~
u,~
II
~F 00 h
connection
synoptics
of the
basic
tlrocess
unit
connection
functions II
.......
of the
...... + + + ......
~
basic
+ .......
'"'"01 :.:: r-
+2::: ~~ '"
"
process ...... eN ~
01 ~- ~
~
unit « :3' « 0
Uen'
+ -~ :i~
00
-0
~
,
U
0......
............- ......-
eN
c/+
......
« ~. " .....
U
.....
0 "
'-::t::
II
::::: .....
II II
,,2-
...... II
(f ! t !
II
:If ~:'::oo ~oo -,00
connection
Example 2
The functions of a basic process unit for a street traffic control system are
shown in Figure 2.4; the synoptics of the basic process unit and corresponding
process controller functions are also given. The three main functions of the
basic process unit are:
I. Processing process signal A and commands B, C and 0
the data from traffic pulse transmitter A are passed directly to a special
device level where they are converted into traffic flow cumulative and
traffic rate data;
,• synoptics of
the process
controller
~
process
i
,
controller of
the whole plant
from/to other
I j from/to other
basic process basic process
units units
,• synoptics of
the basic
process unit
1 1
II
•
II
synoptics of
the special devices
~ i
logic controller of
the basic process
- special devices
(set-point regulators,
unit counters, integrators)
1 t ~ i
basic process unit:
hardware, process devices, transmitters
Figure 2.3 The structural scheme of main data processing and synoptic system
blocks.
special device level
connection
process
controller
functions
connection
yellow flash
synoptics
of the ··808··.
EJ o~ ..~ BEl 0
basic
process
unit
connection ~
-
II ::: II II II
..... c-s II II U.lii: ..... u. ..... C'-Ic<")
..... c-s .......... - N .....
~
z II II 1I1u. .... II
; ;;
~ ~~ ~~z
~ <J <I<J<J r.:Q:::E uc:lI:::EZ .... ::E <J<J
connection
process
1
<t;
TPT
r.:QUc:l
TLO TLOTLO
data and
commands f13 '" (E21 !:, E21) MFI Gl + (E31 d E31) LFI + (Kl d Kl) M +
Figure 2.4 Basic process unit functions for a street traffic control system: TPT,
traffic pulse transmitters (signal); TLO, traffic light on (command); D, panel
pushbutton; 0, panel signalling lamp; d , panel symbol for the traffic pulse
transmitter.
48 The basic process unit
generation of signalling functions Bl, Cl, Nll, N12 and N13 for the
signalization of traffic lights and working modes of the basic process
unit.
The process controller for the entire system performs the following functions
for the basic process unit of the traffic control system:
receiving process signal A and command states B2, C2 and D2;
issuing signals E21 and E22 from signals L21 and L22 received from
other basic process units.
The split of functions given in the above examples enables consistent process
control of process parts, signalization of states to process operators and
meaningful integration of these basic process units into the entire process.
Postulate 4 The speed of data processing VDB of each basic process unit
controller is faster than the speed of data processing of the superimposed
process controller VDC concerning data processing required for this particular
basic process unit, that is
(2.11)
expressed as the number of logic/arithmetic operations in a time unit.
Postulate 6 The quality of all functions of the process depends solely on the
quality of the functions performed at the basic process unit. The quality of
functions performed at the basic process unit depends on the data processing
ability of the control unit with other parts of the basic process unit being
equal.
Of the many parameters characterized in the basic process unit controller,
the following appear to be the most important:
50 The basic process unit
Common passive
filter
2 ~= 1 ZI = 0
Passive T type 't = RC VII + 2 RCs Z2 = 0
RC filter
V2 = see (*) ZI = 0
Active RC filter
Vl Z2 = 00
Figure 2.5 Some commonly used filters and their transfer functions: A, amplifier
gain factor.
Scanning time differs for various types of data processing systems, but usu-
ally there is a difference in analog and digital input scanning, where digital
signals are scanned mostly upon a change of signal state, whereas analog
signals are scanned at regular time intervals. Thus, scanning time for digital
signals can range from 10 /-lS to 10 ms while scanning time for analog signals
is usually 100 /-ls to 1 s, depending on the number of inputs scanned.
Analog-to-digital conversion time is the time interval needed for the
conversion of the analog value at the scanner input into the digital value at
the analog-to-digital circuit output. Table 2.2 shows the analog-to-digital
conversion time for different types of device, together with their important
parameters.
The data processing time T p is the time needed for the conversion of data
from their raw form (i.e. as they appear in the system after filtering, scanning
and digitization) to the appearance of a digital signal after data processing, de-
pending on each process variable. Data processing algorithms, and their time
responses, are discussed in Chapters 4 and 5. T p can range from 1 ms to 100 s.
The response time T D of a process device after receiving an output com-
mand from the control unit usually consists of:
reception time of the output command. When the output command is given
in a coded form, it also includes the conversion of this code;
Table 2.2 The parameters of analog-to-digital converters
AID converter type on Accuracy bit Conversion speed Integral linearity (%) Differential linearity Conversion error bit
principle of (J1Slbit) (%, LSB)
Table 2.3 Response and settling time for different process devices
conversion of the received command into the powered signal to the proc-
ess device;
response of a process device to the powered signal.
Table 2.3 gives the characteristic response times for different process devices.
The process response time TPR is the time interval that characterizes particular
processes. For most processes, T PR lies within a wide range from 1 ms to
1000 s and usually represents the longest time constant of the whole basic
process unit.
To enable the basic process unit to function, the control unit with its synoptics
and special devices receives the process data, processes it according to given
algorithms, issues commands and set-point values to the process devices and
communicates with the superimposed control level. While the control unit is
performing these tasks, the process parameters change and, due to the serial
nature of logic and arithmetic functions of the control unit, there is always a
time delay between the real and computed process variables of the basic process
unit. To preserve the accuracy of a process variable, the speed of the data
scanning should be given. For a process variable y(t) scanned by the control
system at regular time instants Tj, i = 1, 2, 3, ... , given the accuracy e' of the
observation of a variable y(t) and a maximum speed of change cm = [dy(t)/
dt]max of process variable y(t), the following relationship exists
i=I,2,3, ... (2.15)
where k is a data acquisition factor, usually between 3 and 5, that depends on
the number of time instants taken in variable calculation. For input variable
with cm = 2 Vis, e' = 0.02 V, and k = 3, a data acquisition interval tj+! - tj
equal to or less than 3.33 ms is recommendable.
Digital signals do not operate like analog signals, and are usually of a more
spontaneous nature with their acquisition time being much shorter. A higher
speed of data processing is also required for a digital signal in the primary
54 The basic process unit
protection and process safety of a basic process unit. The count pulse data
differ in their rate, usually between 10 per second and 1 per minute, and are
processed by the control unit of a basic process unit. Pulse signals of a higher
rate, and pulse signals of a special process, are usually preprocessed at special
device level and then fed back to the control unit (see Figure 2.4).
Control functions are performed according to a required process control
algorithm and the precision of their functioning is essential for the overall
response time, stability and process control accuracy of the basic process unit.
A real process control system is described as a linear time-invariant discrete-
time dynamic system with concentrated parameters
X(t k+ 1) = Ax(tk) + Bu(tk)
y(t k) = Cx(tk) + Du(t k) (2.16)
x(to) = Xo
and is theoretically controllable only if its mathematical model (2.16) has the
following five features (Novakovic, 1981).
1. Complete controllability of states and inputs.
2. Complete observability.
3. The same number of control variables and state variables.
4. Minimum sensitivity to parameter and structure change, and to variable
disturbance change.
5. Maximum sensitivity to control variable change.
A complete controllability is usually not fulfilled, i.e. the dimension of the
measurable state vector A(tk) is 1 < n where n is the dimension of the state
vector x(tk). The state vector x(td can be reconstructed or approximated from
the observed variable if a functional F' exists
xlt k) = F'[y(t k'), to < t k' < td, tk ~ to (2.17)
so that xltk) =:: x(tk) where xr(tk) represents a reconstructed state.
implies
x(td = X(tk) tk ~ to (2.22)
for all u(td, tk ~ to
(2.27)
consisting of a series connection of second order elements and an integrator
with
Figure 2.6 The structure of the full order observer .1t =t k+1 - tk = constant.
The dynamics of the error given in equation (2.31) can be modified by the
appropriate choice of K. Usually the observer has to be faster than the plant
so that the results of the states may be supplied quickly to the process. This
is especially important when some external disturbances exist, or the plant's
initial conditions are not known. At other times, the observer may smooth the
measurements contaminated by noise thus being slower than the plant.
The proper K may be solved by using the values of the poles of closed-loop
dynamics described by the equation (Kwakernaak and Silvan, 1972)
(2.32)
The desired poles of equation (2.32) are three eigenvalues ZJ, Z2 and Z3 found
from the equation
det(zI - a - ~C) =0 (2.33)
Basic process unit data processing 57
PROCESS
----------------------------- ~
r--------------------------------------------
state estimator
_________________________________________ J
(2.36)
The block diagram in Figure 2.8 for the case in equation (2.28) with known
K requires 21 multiplications and 19 additions during each sampling interval
in the controller.
Controllers can be designed for optimal control of the process from three
main purposes.
58 The basic process unit
- 6
----------------------------~------- ..... --------- D
!
sample and
hold element
~--------------------------~rk
Figure 2.8 Complete block scheme of the computer,controlled continuous process:
C, equation (2.28); K, equations (2.33) and (2.35).
J ~ 1/2 (% xi Aw Xj + uJ Bw Uj J (2.38)
Each control system when observed as a data processing system follows the
simple and approximate calculation of data processing that can be expressed
as available data processing time for a single processor in its basic process
unit. The available data processing time can be estimated differently for vari-
ous controller functions. When all processing functions have the same priority,
the simple relation for total available time T TOT can be expressed as
N Mi
(2.40)
where
XSAT and XTH are the saturation and threshold levels of the analog-to-digital
converter; usually XSAT = 10 Y, XTH = 5 mY;
Fq is the quantization noise value in dB; usually 60-80 dB;
82 is the variance of the input samples to the analog-to-digital converter,
usually around 10-3 y2; with these data C w is between II and 22 bits.
Digital inputs are scanned in groups in case a change occurs in one of the
input channels. Each group of input channels consists of 4, 8, 16 or 32 inputs.
Digital signals (being a stochastic variable) appear at the rate of one per hour
analog c=:::=v> analog
inputs outputs
data flow path
digital
inputs
input output
scanner control and scanner
and data processing and : digital
count pulse
and process o outputs
inputs data data
algorit~ms
adaptor adaptor
communication communication
I
inputs I outputs
I
I
I
--)
_________________
analog input ....processing
________ ..... ' / I ......
~'-----
-"-...-.. ....._on/off command processing
o
°t l O t 0 / I '-... -"-...-.. processingoTseti>Oint-----
d Igl a mpu processmg /
_____________________________J /1 ................... ..... ................ -..-. .................... '-'-.. , values
_______________ _
I ' ...... ...... ......
count pulse input processing /
___________________
1-'-...-..
I..............
-""" ...
-"-'-...-. ..... """-... ... communication output processing
- -- - - - - - - - - - - - - - - -
o 0 0 0 / -"-"-. "-"-. '-...-.. process safety and protection
_______ ~~~~~~l~~!~O_n_~ny_~t_~~<:':.~s~~ngj -"-"-"-.. -"'-""-' -'-"'-...Jl!'~~~~ ___ _
............... ..... .....
-"'-""-.. -"-'-"-.. process control algorithm processing
................... ---------------------------"-
"-"-.. reserved data processing time
'------------------~--------
hardware-type II I
I
software/ data processing I I hardware-type
I I
data processing : I
I
I I data processing
analog input I
I
1
: data acquisition algorithm
I
I
: analog output
I II.)
module 1 II.) u : module
digital input 1 u process control algorithm JS.... I
I JS I
I .... II.) ~ ....
II.)
module I II.) ....
~ C<l
digital output
I ....., C<l process safety procedure module
count pulse I .S ~
.~
"04-<
~
I
module
I ';:: ...c
local display protocols I "5.-C : output module
I "5
I
operator
1 S I
I S
I I II.)
I
I "0
keyboard I
I
communication protocols I : local display
module iI
I
I t module I
i I
I
I
I I
system bus
supervision
Figure 2.10 The hardware configuration and the calculation of data processing
parameters.
to one per second, depending on the type of process controlled. The probabil-
ity of data acquisition upon request by the ith group of n digital input chan-
nels equals
'A'j = n'A' + (n - 1)'A'2 + ... + 2'A'n-l + 'A'n (2.41)
where 'A' is probability of the occurrence of an event in any of n input chan-
nels, all probabilities are equal. For a small amount of probability, equation
(2.41) can be approximated with the first two terms.
Each event is tested for plausibility as there is a risk of errors due to noise
and cross-interference.
Count pulses are acquired at regular intervals because they must be checked
for the accumulated pulse number. When count pulses are directly connected
to input hardware, a provision is made to enable the collection of all count
pulses. This is usually the case for lower and regular count pulse rates and
mandatory for point of sale sites.
62 The basic process unit
A measuring transducer is the process input device that accepts process infor-
mation in the form of a physical quantity as its input variable and, according
to a definite law, converts it to information in the form of the same or another
physical quantity.
Process hardware for data input 63
i=l i=l
where f(v) is the dependence of output physical quantity f on the quantity v
to be measured and Vb v2, ... , Vi are measurement perturbance factors that
modify the output f(v) in an undesirable way. It is necessary for a linear time-
invariant response to possess a sensor where the second, third and all other
terms on the right-hand side of the equation (2.44) are negligible compared
with the first term, e.g. less than 0.1 %.
process
physical
quantity
signal physical
L quantity
Standard process input devices are categorized according to the types of sig-
nal they issue. These are set out below.
1. Binary device. A device that issues a process signal with a binary quantized
information parameter conventionally denoted as 0 and 1 (binary signal).
2. Digital device. A device that issues a process signal with a quantized
information parameter represented numerically (digital signal).
3. Analog device. A device that issues a process signal with the information
parameter which may assume all the values of the given range (analog
signal).
4. Sampled device. A device that issues a process signal with the informa-
tion parameter which is only intermittently observed and represented
(sampled signal).
The description of standard process input devices and their signals are given
in Table 2.4.
Process input devices are usually installed on site where the process vari-
ables can be detected. Input signals are transmitted from the process into the
control room, or cubicle. For larger distances or where only the sensors are
mounted on site, the amplification of input signals is performed between proc-
ess and control points. Typical solutions for the connection of various input
devices are outlined in Figure 2.12. Input devices usually fulfil the following
functions.
1. Power supply of transducers and signal converters. The supply of power
to transducers is, for the most part, organized separately for each meas-
uring variable through the corresponding devices; transducers are con-
nected by two, three or four wires to the power supply device, thus enabling
individual voltage supply; through power supply a live zero processing of
current signals is possible ('0' = 4 rnA).
2. Testing of process input value (state). The voltage or current of power
supply for transducers is controlled; the limit value of input signal
is controlled against low (or high) preset value; thus the breaking of
Table 2.4 Standard process input devices and their signals
Input device Input variable Information parameter Input signal Common connection characteristics
Binary device Any process variable Quantized information Binary signal Closing contacts
parameter (0, 1) Insulated from any ground circuit or
conductive parts
Volt-free
~ 10 7 operations
Digital device Any process variable Quantized information Digital signal BCD accumulator up to six digits
parameter (digital Insulated from any conductive part
number)
Volt-free
Normally open contacts
Contacts with breaking capacity 10 VA
Minimum breaking current 0.5 AI
48 V DC
Contact resistance:;; 100 mn, ~ 10 7
operations
Analog device Any process variable Any value of a given Analog signal Floating loop (~ 100 mn; ~ 0.5 kV)
range with no common point
Standard DC current 4-20 rnA,
0-20 rnA or voltage 0-10 V
Controller load not exceeding 500 n
Sampled device Any process variable Intermittently observed Sampled signal Closing contacts
information parameter Duration of intermittent signal ~ 50 ms
(analog or digital)
Volt-free (for digital signal)
Floating loop (for analog signal)
process variable instrument adaptor transmission control place
resistance
temperature
ftJh~e~r:m~o~m~e~t~e~r}-~=========:________________
L
-1 A
strain
~ strain ~
. 1L.._g_a_ug_e_-,~----------------<0
flow ~_ flowmeter
electro-
magnetic
_
H3> A ---------------------------------o
_ _ _ _ _ _ ~~ Jdifferentiall
displacement ,----~~--~ - ~transforme~~----------------------------~O
pressure
.~
-1 transmitter
pressure r---- o
direct current
temperature
-~thermocouPle~-------_____i~
1 _ DC/DC L-,
~Iconverterl
~ z=Va.~
c z =Vt, PlJ
Flow
o -lOY
'LgJ-
for other
functions
to limit circuits
Figure 2.14 The plant distribution of process signals circuits: 1LPF I. low pass filter;
~. voltage to current converter.
compensation from the current supply is made for slow varying volt-
age disturbances induced from other conductors or transducers;
the distribution of signals is performed by independent circuits (see
Figure 2.14) which enable the independent and isolated use of one
input signal for more plant functions.
6. Breakdowns report and simulation of process value includes the following:
testing and indication of input signal value and internal power supply
voltage;
lock-in of regulators and limit circuits when disturbance occurs, or
when simulation of process variables takes place; simple transducers
and signal converters (as temperature signal converter, for example)
possess connector pins for the control of input signal, for the calibra-
tion and for the simulation of process input variable; the connection
of test instruments to complex devices, as for example signal con-
verter and distributor, enables the measurement of input signal, simu-
lation of the input signal, simulation of process variable, simulation of
signal for limit circuits, measurement of the state of limit circuit el-
ements and the simulation of binary signals from transducers.
Electric system data of input devices
These include the following features of transducers:
1. Rated voltage of 24 V DC for power supply Us and for control test
voltage Vc.
70 The basic process unit
Final control devices are executive elements of a control system that directly
change the value of the manipulated variable. Final control devices are both
continuous and discrete. They are made out of drives and executive devices.
Examples of drives are valves, actuators, valve positioners, speed variators,
solenoid actuators and servomotor actuators. Drives are connected to the ex-
ecutive devices: shutdown valves, transformer tap positioners, air dampers,
gearing mechanisms and electrohydraulic valves.
The transfer characteristics of continuous-type drives and executive devices
Process hardware for data output 71
feedback.
device
".I:; .,.'
Figure 2.15 The overall control scheme for final control devices.
For the range of applied final control devices, a whole set of supporting
hardware components is required to enable the proper functioning of these
output devices. Special features of output devices are devoted to operator-
process communication functions. These functions are:
Ordinary signals for the hardware of final control devices are given in Table
2.6 for different types of output devices and connections of output devices to
data processing equipment. (Testing and safety of final control devices are
described in Part Four.)
Various commands issued from the control system to final control devices
and other parts of the control system are indicated in Table 2.7. For a basic
process unit, Figure 2.16 shows the standard data on input/output and process
equipment (PI-diagram).
Auxiliary data for process control are obtained during data communication
between a process control station or remote terminal units. These data are
issued either on the side of higher or lower hierarchical level, and either as
process information or process control information (see Table 2.8). Process
information includes:
Designation Feature
Single command Type of command with one output signal, usually volt-free
open contacts
Double command Type of command with two output signals in series, usually
volt-free open contacts
Pulse command Command signal usually of shorter time duration, i.e. 5 ms
to 500 ms
Maintained command Command signal usually of longer time duration, i.e. 500 ms
to 100 s
Persistent command Command signal continually applied to an output device
Starting command Command signal applied to start a plant or a part thereof
Stop command Command signal to stop or shutdown a plant or a part
thereof
Switching command Command signal to switch over the process apparatus from
one to another working state
Set-point (adjusting) Command signal to adjust the state of process apparatus to
command appropriate state
Persistent regulating Command signal of continuous type to keep the process
command regulator in a given operating state
Regulating step command Command signal issued in step-by-step manner
Command signal to enable a selection of specific process
Selective command apparatus
Group command Command signal issued to a group of process apparatus
Broadcast command Command signal issued to all peripheral process control
units
Instruction command Command signal issued to the control system operator
Function command Command signal of a specific system function
411 8------------.. 510
412 8--------------..----;
I
start stop
I
I
I
I
416@ 416©
RTK 410
Figure 2.16 Process and signal paths in a basic process unit: - - , crude-oil
flow; ----, signal flow; 0, transmitter; ©, pump motor; 510, another basic
process unit.
REFERENCES
3.1 INTRODUCTION
78
Stratification of computer tasks 79
3.2.1 EXAMPLE 1
An analysis of control functions of the basic process unit for power supply of
a crude-oil measurement station is given in Figure 3.1 and, as seen from the
synoptics of the basic process unit, the voltage is applied through a voltage
transformer TRI to:
an ACIDC converter and to the field instrumentation as 24 DC power
supply;
- a separate branch controlled with the voltage switch VS3;
a branch fed uncontrolled to other parts of the field consumers.
The consumption of supplied electric power is controlled through a Wb-coun-
ter on a special device level acquiring pulse count signals C. The signals of
supervisory
level
connection
remote
terminal
unit
connection
t t
kwh
t t
special
counter
device
level I DJiJiJ I
"7
connection
process
controller
functions
connection
synoptics
of the
basic
process
unit
connection
functions of N
::r:: ..... .....
the basic ..... + + + +
process unit +
..... ~
~
c<')
~
'<t'
~
>n
~
9 9 9 9
~
9 ....
<: CQ 0 ~
II
,.....
Y
II II
connection
[ ~~
L! L L u
Q~ ~~
....
process data <: ""
CQ 0
vs vs eet vs vs vs
and commands
Figure 3.1 Analysis of control functions for the power supply of a crude-oil
measurement station: VS, voltage switch (signal); EET, electric energy pulse
transmitter (signal),
Stratification of computer tasks 81
voltage switches (A, B, D, E and F) are processed in the basic process unit
and fed to the process controller, where all the necessary set/reset functions
and a connection L28 to other basic process units are executed. For the purposes
of process supervision, all the signals from voltage switches are then fed to
the remote terminal unit.
Figure 3.1 also shows the following main data processing tasks:
integrity, protection and safety of the basic process unit;
- supervision of basic process unit;
- measurement and presentation of process data.
Simple analyses of process control functions are largely dependent on the
structure of the controlled process. Thus, a relatively simple power supply unit
that is independent of other process parts in a crude-oil production field, which
supplies only one measurement station in the field, can be protected and
supervised in a particular manner. Equally, this unit is also included in higher
control levels by sending its data for further data processing to these levels.
The results obtained can be used for the control of other power supply units
in the same crude-oil production field, or even for a group of production
fields. There is no direct automatic action on this basic process unit from
other power supply units, since it is not needed. The provision of direct automatic
action from other basic process units would complicate the structure of
algorithms performed on the process controller at the level of the basic process
unit, as can be seen from Example 2 below.
3.2.2 EXAMPLE 2
The algorithm for control actions of all distribution substations (see Figure
3.2) put in one control centre is highly complex. This is due to the complexity
of the problem, implying that the total power system of generation, transmission
and distribution of electric energy should be represented and computed in the
real-time environment, that is at the time of data acquisition. The proper conlrol
of such a system requires the monitoring and processing of real-time power
system variables, and then the execution of the desired feasible control actions
directly on the system's side remote control equipment. Instead, considering
the system's spatial structure and local control objective characteristics, the
overall network can be reduced to a set of distribution subnetworks as shown
in Figure 3.3. Each subnetwork is provided with independent local control
hardware which consists of a distribution substation feeding one radial load
bus centre. A distribution substation consists of a set of m parallel transformers,
capacitors and reactors.
The power flow is controlled by transformer tap positions, and by switching
capacitors in or out to control the power factor. The aim is to maintain safe
stable operation within the constraints. The load between phases must be kept
82 Stratification of control tasks and data communication
IIOkV bus
transmission
network
,..- ,..
-" .... ",," ,.. .... ,.. "- A ....
llO/35kV
transformers
35kV bus
distribution
t capacitors/
~ reactors
substations
load
} centres
1 to N
in balance. The voltage must be controlled and heat and transmission losses
minimized by adjustment of the power factor.
Each subnetwork is decoupled from the rest of the network. Two assumptions
are made.
1. Constant load demand during data collection, data processing and process
control implementation.
2. Constrained requirements of localized control actions .
•
The formulated control actions are performed as a set of noninteractive local
controls, with little overall effect on the system operations.
The overall control strategy of the radial load bus subnetwork enables the
control to be reduced to its priority functions which are executed independently
and in order of priority. The priorities and their criteria are described as follows
(Arafeh, 1978).
generator
1
generation bus
voltage ratio
transformers
phase angle
transformers
load centre
I capacitor bank
(3.1)
where 1m and Re denote reactive and active parts of the complex power
respectively.
Priority 2 control function Load centre voltage control for specified voltage
schedule V sc ' voltage tolerance VTOL and calculated load centre voltage VL ,
transformers, capacitors and reactors are controlled so that the change in load
centre voltage satisfies
At; = {[Abs(S;)M ~ Abs(Sj)] -I]} ~ MVA TOL for i = 1,2, 3, ... , (3.5)
t
operator
t t t t
process
controller control algorithms are as given in Table 3.1
,
I I"
# L ~ L ~
I I I I I
basic basic protection and control algorithms
process are as given in Arafeh, 1978
unit
,
i ( t t'" t t t'"
~ ~ ~
synoptics
of the
basic
process
unit
process
device
level
v = Vn+ no~ V n; V 0 = V L + Is / Y f
~V=~ +~V
n n
~ n = n - no
Figure 3.4 Analysis of control functions of a voltage ratio transformer: n, new tap
position; no, instantaneous tap pQ.sition; f/' feeder admittance; Y s ' feed to load
admittance; Y, transformer admittance.
86 Stratification of control tasks and data communication
v o +!:::,V0
reference vector Vo
power unit is shown in Figure 3.6. A great many specific control, regulation,
supervision, measurement and protection functions are attributed to different
levels and parts of a thermopower unit, in which numerous functional systems
are installed with thousands of control loops.
3.3.1 LEVEL 1
3.3.2 LEVEL 2
3.3.3 LEVEL 3
v, Vo _
~ ~ (t S: + 1 jQ
Y1 Vo
~v
_[ i = \_ ,=1 st + jQ! ~'=I _ _
[i. ~ni~ VSi ~
L V2 m
YI 0 ~ - -
.£... Yi + Y,
1=1
Table 3.2 The control algorithms of a process controller
Table 3.3 Typical distribution of on/off functions and regulation of a 650 MW thermal
power unit
On/off control 650 MW turbo set and On/off control system of steam
generator cooling system turboset
Feedwater pump turbine
Air turbines On/off control system
Main boiler including
oilbumer and coal pulverizer
Gas turbine On/off control system for gas
turbine
Regulation 650 MW turboset Turboset regulator
Feedwater pump turbine Regulator of the feedwater pump
turbine
Air turbines Turbine regulators
Boiler Boiler control
-- --
thermal control of
power thermal
unit power unit
level
--- I
I
boiler
control
I
---I 1---
group of
I I
combustion
function feedwater
process
level control
control
___ .-11 1L ___
L~::-
group of group of
burner 1 burner 2
control control
---~ ~---
burner 1 - 1 burner 1 - 2
control control
---
I I
process I I
burner 1 burner 1
devices and
safety control protection
level devices devices
~~-
U
process control-
'">-
'"
~
supervisory ~
~}]
control district centre
U
data acquisition- ~.~
~
operator's remote terminal
level 3
console unit-direct f4
digital control joI
U
process data
level 2 synoptics or process controller
indicators 4
t- ~ 11 1 t
complex process instrumentation-
level 1 regulator
- process drives control
. . =-
--------
process hardware
A B c D
Table 3.7 Digital output circuit (according to VDIIVDE 3552) applied to Teleperm
ME AS220E on/off control module 6DS 15I0-8AA
A B
* A = relay output
B = electronic output
3.3.4 LEVEL 4
IG
.-16 Mbit
/",/ I RAM
./
,,"
/,.-"2 Mbit
// I RAM
:
" I
" 1
1M /,/'
:
,," 1
/"
"64kByte I
/1 I
/ / : memory :
/' I I
,,," I I
/ I 1
/' I I
..-lkMOS I :
lk ....
/",/ I
I
RAM iI 1
I
,/,/: I
/ 1
.fourbit I
/ / I TTLcouhter
/f ~T~ ?istable I
Clrclpt
:
// 1 I
,.RT~:
// : logic gate 1 1
~--~-=~~-L---1--~----r---~--~------~-------r--~year
3.3.5 LEVEL 5
Since the production of the planar transistor in 1959, the number of elements
in advanced integrated circuits has, according to the prediction of G.E. Moore
(Noyce, 1977), doubled annually (see Figure 3.8). Less intermediate testing is
98 Stratification of control tasks and data communication
Computer systems for process control have been developed based on the con-
cept of functional systems. Such systems are recognized as basic building
blocks when complex process automation takes place. Manufacturers of
computer process control systems share different approaches to these, which
are as follows.
1. Series or parallel data communication concept.
2. Different operator/system communication facilities.
3. More hardware- or more software-oriented implementation of functional
specifications.
The main features for two different manufacturers of computer hardware for
process control are summarized in Table 3.14. A more universal picture of
computer systems for process control can be given as a four-level control
system, although the approaches to the implementation of hardware for these
manufacturers represent a functional system concept. The basic configuration
of each functional system is shown in Figures 3.9 to 3.12, and their main
features are given in Tables 3.15 to 3.17.
Microcomputer-based systems for distributed data processing and
minicomputer systems need fast buses (up to 20 Mbyte/s) for internal data
communication (Warren, 1983). There are more than 140 different bus designs
identified in Europe alone. The most popular are the Multibus IEEE-796, the
VME buses, the VERSA bus and the S-100 (IEEE-696), which are extremely
important in internal system design. Hardware design is influenced by the
need to achieve cost/performance optimization using the standard modules on
Table 3.8 Global characteristics of computer system for process control
Process controller and RAM and ROM memory only Number of inputs: 25-64
instrument multiplexer to ensure maximum reliability
The length of analog input
Interface hardware designed connections < 30 m
in the range of computer
Input scanning rate 1-10 ms
hardware
Temperature span 0-50°C
Active error protection
Humidity span up to 100%
relative humidity condensation
free
Mean time between failure
6000 hours
MTTR 6-24 hours
Degree of protection IP 54
Process computer RAM and ROM memory only Number of inputs = 250
to ensure maximum reliability
Remote terminal unit Memory cycle 0.2 ~s
Parallel job execution
System organization - parallel
Disposition of hardware for
Mean time between failure
fast program changes
5 -10 000 hours
Interface hardware designed
Memory capacity up to 256 k
in the range of computer
words
hardware
Word length - up to 32 bit
Doubled systems or error
protection when in direct MTTR 6-24 hours
digital control
Degree of protection IP 54
Temperature span 0-50°C
Humidity up to 100% relative
humidity, condensation free
District centre RAM, ROM and disc memory Memory cycle 0.1 ~s
Supervisory control Parallel job execution Mean time between failure
5-10 000 hours
Despatch centre Disposition of hardware for
fast program changes Memory capacity 1-4 M
Management control
words
Doubled systems and doubled
discs when in direct digital Word length 32 bit
control
Disc capacity 30-400 M
Interface hardware designed words
in the range of computer
MTTR 6-48 hours
hardware
Temperature range 20-2YC
Humidity range up to 80%
relative humidity
Two types of data communication exist in a process control system: one between
the process control system and simple process devices such as transmitters
and final control elements, and the other between various parts of the process
control hierarchy. Data communication in the process control system is based
on the messages appearing at the transmitter and received at the receiver. A
definite trend in technical systems is towards messages transmitted in characters
coded in a binary form from the binary message sources. Essential for the
proper functioning of process control systems is their error-free operation. But
the communication of data is influenced by different noise and disturbance
effects in communication circuits and media. Therefore the coding of technical
messages is of prime importance if control systems are to function correctly.
Siemens Honeywell
mICro-
computer +
timer
2 X serial
communication
channel
------
~
RW memory + ROmemory+
I/O ports I/O ports
~ ~
I/O circuits I/O circuits
for digital for digital
signals signals
] m
process onloff
operator's
devices and
panel
drives
•
•II
t
process
operator
Figure 3.9 A microcomputer-based process controller.
ebus
analog digital analog digital special serial er
input input input input purpose communica- s
subsystem subsystem subsystem subsystem interlace tion interlace I
s
lb;>
process - drives and transmitters
f Figure 3.10 A microcomputer-based process control system.
to other
• system console floppy disks process
control
systems
system bus
~
2x disk drive f process - drives and transmitters "/ process
control
systems
6"~'~1
•
mass memory bus II •
U ,
or
Cj = k'[1 - s(p)] bit/s (3.11)
Types of connection and communication hardware III
(b )
When the transfer of signal with a power level S is disturbed with a white
noise of power N having the frequency bandwidth B then the channel capacity
is given by the expression (1.14). The power level can be calculated by squaring
corresponding signal and noise envelopes.
(a)
~ Ise~~a:~ce I C C C C c Ise~~~~ce I J
t
~
(b)
communication channel
(b)
(c)
Figure 3.15 (a) A simplex mode of data transfer; (b) a half-duplex mode of data
transfer; (c) a duplex mode of data transfer.
of stop combination can be deliberately chosen (e.g. 4.15 bit intervals). The
main characteristics of the asynchronous communication are simplicity of
hardware, low amount of information flow suitable for small datawords from
5 to 9 bits, and use of transfer speeds of up to 19200 bit/so According to the
types of communication regarding the direction and possibility of simultaneous
reception and transmission, three modes of communications are distinguished
as shown in Figure 3.15.
Types of connection and communication hardware 113
(b)
(c)
point-to-point
multipoint
loop
Figure 3.17 Nine different data communication modes: C, control terminal;
M, master terminal; S, slave terminal.
star connection
multi-point connection
facility into separate subchannels for each low speed terminal (see Figure
3.20b). Time division multiplexers divide a high speed communication into
time-sliced subchannels, as indicated in Figure 3.20c. A dedicated time slot is
provided for each terminal connected to the time-divided multiplexer. Equipment
for a radio link can be divided into radio equipment, multiplex equipment,
antennas and leased-line connection equipment.
Types of connection and communication hardware 117
I
I
I
I
I I
~ line breakdown
I I
I I
I I
I I
data terminals
~
T1O--i-+-,
I
,
data terminals
,.-i'----c T 1
T2 V--~--j 1----l--<lT2
communication
channel I
T n o--~,+--'
I
_I
I
\'--------
(a)
low-speed
comm unication
channels
N - 1
~
,channel
f (Hz) I channell "channel N
f (H ) guard bands
z / (channel 2 '\ channell
I I "
I I
I I
I I
:
I I
I I
I I
Figure 3.20 (a) The use of multiplexers; (b) a frequency division multiplexing;
(c) a time division multiplexing.
The constant c is so adjusted that the quantity s(t) + c always remains positive
so that the simple envelope signal detection suffices to demodulate s(t).
The phase modulation is usually applied in the following form of the line
signal
where g(t) is the desired envelope shaping of the passband pulse, T is interval
between pulses in seconds and <I>j is phase of the ith pulse. In binary phase
Types of connection and communication hardware 119
[G(ro)]2
-~
~ro
,
1t/T
.. ro
v(t) = Af COs[O>,t + d [ 1
six) dx (3.16)
120 Stratification of control tasks and data communication
device bus
I
I I
individual memory
I A/D and interface I microcomputer
- communication facility
- device memory
- sensor linearization
Il
- rerange
II - engineering units
I - diagnostics
Ocmpern<me
sensor + +
1\ I D/A
I
digital
communication conversion
:~Ci~
I
i
communication output
~ell sensor I
I
)
\
4-20 rnA signal and digital signal
to control system
The change of frequency occurs in the line signal with the change of the
binary state (Figure 3.2Id); this is usually called a frequency shift keying
(FSK).
The Bell 202 standard frequency shift technique, which uses high frequency
binary signals superimposed on the instrumentation standard 4-20 rnA
transmitter output, is applied for The Rosemount Model 3051 differential
pressure 'smart transmitter'. The microcomputer controls the operation of the
transmitter and in addition performs the calculations for sensor characteristics
linearization, rearranging, engineering unit conversion, transmitter self-
diagnostics and binary communication. With the addition of a temperature
sensor for higher measurement accuracy it achieves an accuracy of ± 0.1 % of
calibrated span and stability of 0.1 % of upper range for six months. A block
scheme of this transmitter type is given in Figure 3.22.
The digital data transmission can be so arranged that binary O's are trans-
mitted as 0 volts and binary I's are alternatively transmitted as positive and
negative pulses (Figure 3.2Ie). Thus, signals can be easily regenerated as
opposed to analog data transmission where they are to be amplified.
Modem technology is based on its task to form and acquire an analog
signal to transfer the data under the constraints of the communication facility.
Practical suggestions and recommendations 121
These constraints can include signal level, limited bandwidth, noise level and
nonlinear characteristics. The connection of modems to the communication
line and to data terminals is given in Figure 3.23. An example of essential
parts of a frequency modulated modern is given in Figure 3.24. The low pass
filter removes the high frequency components of the rectangular pulse-shaped
data input. A binary' l' causes the mark frequency to be sent and a binary '0'
the space frequency. The band pass filter limits the frequency to the bandwidth
of the transmission facility. The receiver band pass filter removes noise and
interference outside the signal transmission band. The limiter sizes and shapes
the input signal to the frequency demodulator circuit where the original signal
is then recovered in a discrimination process of the demodulator. The low
pass filter removes unwanted high frequency and noise components and then
the output signal of the low pass filter is fed into a slicer performing the
slicing function to again produce rectangular pulses.
The term 'data communication facilities' refers to the transmission lines
using any technology (e.g. cables, radio-connection, satellite-connection,
microwave link) and being capable of carrying data signals.
Communication hardware for distributed process control is organized as
part of a packet switching network (Figure 3.25) either on bit-oriented data
link procedures like HDLC or SDLC protocols (Figure 3.27) or on coaxial
cables like PROCONTROL P12 as shown in Figure 3.26.
It has been recognized from everyday practice (Plecko, 1984) that radio
equipment is lacking in some aspects of process data transmission. For instance:
1. Squelch circuits for automatic correction of amplification at the input of
the low-frequency circuit of radio receivers have too long a time constant
(100 ms), which causes slow response of the receiver; therefore, for a
duplex mode of data transfer with transmitters always in work, squelch
circuits have to be put out of work; for a half-duplex synchronous mode
of data transfer with transmitters intermittent in work, the unfavourable
influence of squelch circuits has to be compensated by idle words added
transmit section receive section
___________________J\ \
r
r
nOise
packet
switching
network
RTU
RTU RTU
parity bit
I I
00 0 1 1 00 II!
I
,,
I
+ 15v ,,
I
,
,
I.
...-----;'
::- 15v
:,
"
100001 110
Figure 3.26 Signal form in a coaxial cable for PROCONTROL P 12 serial bus 18.
124 Stratification of control tasks and data communication
I
control field bits
8 7 6 5 4 3 2 1
flag information
direction transfer
comma~f' response N(R) P/} N(S) 0
of
address (1 frame
transmission
control
876 5 432 1
supervisory
(:omma~fs/responses N(R) P/F SSO 1
information S frame
unnumbered
comman~~s/responses MMM P/F MMI 1
(U frame. f
FCS
flag
Figure 3.27 Basic structure of an HDLC data packet: flag, flag sequence
01111110; address, secondary station address field, 8 bits; control, control field of
8 bits; information, information field; FLS, frame checking sequence; N(S), send
sequence number; N(R) , receive sequence number; Sibits, specification of a super
visory function; Mlbits, specification of commands and responses; PIF bit, check
pointing bit (command = 1 ~ response = 1).
to the message; the duration of idle words has to be larger than the sum
of duration of the transmitter time (10 ms), to be put into 90% of power,
and the squelch time constant (100 ms).
2. When using communication equipment both for data transmission and
speech transmission squelch time constants of all repeators have to be
added to the time constants discussed above; therefore, extra idle words
of data transmission messages have to be added to decrease the data
transmission speed.
3. The use of a public telephone network on occupied lines can sometimes
cause additional inconveniences where induction telephones and improper
telephone equipment damage the modems.
4. The use of a public telephone network on leased lines can induce prob-
lems, since the declared quality of lines may not be reached and the
quality of lines is controlled only by the owner of these lines.
The bases for MAP are the standards of the OSI (Open System Interconnection)
model established by the International Standard Organisation (ISO).
The OSI model possesses seven conceptual layers covering all aspects of
communication. MAP standards on lower levels of OSI are relatively firmly
shaped, while some higher levels lack descriptions (Peter, 1989) (Table 3.18).
The physical level defines electrical connection between local cable networks
including frequency converters and modems in each mode. Thus a
communication on coaxial cable is defined according to the IEEE 802.4 Token
Bus standard. The main characteristic of such a connection is that only one
communication point can transmit a message which possesses a token (firm
defined bit combination - password). The token is passed to the next
communication point after the message has been sent. The communication
speed on such connection is up to 10 Mbit/s.
Data link level specifies protocols in the local network. These can be provided
either by a token bus or by a CSMA/CD (Carrier Sense Multiple Access with
Collision Detection) technique. A communication partner can only start
transmitting if there is no traffic on the bus. Should two partners try to talk at
the same time then each will soon become aware that the signal on the bus is
different from its own message. When this happens both partners break
connection and only re-attempt transmission after a short random time. This
leads to one device getting back to the bus before the other; provided that
some third partner has not gained access meanwhile. At network level the
transmission of data blocks between data sources and higher levels is
standardized.
- Transport level deals with communication aspects of remote terminal units
in the system.
- Session level standardizes the coordination of dialogs and synchronization
of application programs.
- Presentation level is foreseen for data interpretation and it is still not well
defined.
The highest level - application - is in final development. It should enable, for
instance, communication between the mainframe and each programmable
controller at the site (RS511 standard).
Figure 3.28 represents a block scheme of a fully automated MAP production
cell.
Using two out of the seven layers of the ISO reference model (Katz et al. 1989)
direct communication between 'smart' transmitters and valves and the process
computer becomes a reality (Blick1ey, 1990). The communication is standardized
under the name PROFIBUS DIN V19245 part 1 and 2, and is presented
schematically in Figure 3.29.
Table 3.]8 MAP specification and OS] model
GEC Siemens
Router Router
AUTOMAX
Cutter Kit/
Machine
DekitRobot
-
to process to process
device A device B
-
process device C
remote
object
)
definition
\
APPLICATION
LA\ER INTERFACE device B
APPLICATION PROCESS
control supervision
measurement
- - process --
Figure 3.29 Source and remote object definitions in PROF/BUS DIN V 19245.
REFERENCES
Arafeh, S. (1978) IEEE Trans. Autom. Control, AC-23, 2, 333.
BBC Brown Boveri (1982) Publication No. C-1-T lOa, 174ae.
Blickley, G.L. (1990) Control Eng., Jan., 74.
Karlsruhe, 11 Oct. to 17 Nov.
a) IAEA (1982) Course Material on FRAMATOME SPIN System.
b) Mitsubishi PWR Control System.
c) BWR Japan Control System.
References 131
4.1 INTRODUCTION
Powerful central processor units (CPUs) and slow connections to the monitored
process enable process control functions to be executed using software, with
all the advantages and elegance of software problem-solving abilities. Therefore,
software takes a prominent role in process control systems, but, unlike hardware
where interfaces are standardized, only a few operating system software packages
are standardized predominantly as de facto standards like DOS 3.0 or UNIX
System V (Mikovic, 1983). Operating system software is an essential part of
process control software (Figure 4.1). Peripheral device drivers and handlers
are not usually standardized. They have to be tailored to the operating system
since even the most common physical connections like the RS232C interface
are by no means plug-in-and-go features on a universal basis.
Process control software is developed by operating system programmers,
system programmers and user programmers (see Figure 4.1). Some important
software parts have already been standardized and developed by computer
vendors. For example:
software tools;
compilers and interpreters;
- assemblers;
- linker-loaders.
The development and implementation of all the other software parts shown in
Figure 4.1 is discussed mostly in Part Two.
A process control task is implemented by an abstract automaton. For instance,
a logic or mathematical task may be either implemented in hardware or software
as there are no differences in system functioning. However, there are differences
between the software and hardware approach to implementation of tasks for
process control systems. Typically hardware functions are executed in parallel,
while software functions are performed sequentially. The basic software
components are recognized and compared to the equivalent hardware com-
ponents. This comparison can be made in terms of, for example, 'process
control safety', 'power consumption', 'space requirements' 'cost requirements' .
135
136 The relative roles of software and hardware
external
hardware system
software
border
special user
devices programs
~~~~~(-+_ user
i!!! interface
local-area data base
network
display ~§§=t.•_ system programmer
keyboard interface
RS - 232
process
in terface operating system
programmer
interface
4.2 DA T A PROCESSING
Outer functions of data processing are usually defined in broad tenns in the
preliminary design of the data processing system, and in a more detailed form
in the main project itself. Outer functions are implemented by inner functions.
Inner functions of data processing are the detailed functions of data processing
hardware, i.e. firmware and software are the two main ways of implementing
them.
1. Firmware programs or program parts are implemented mainly on read-
only memories (ROM), programmable read-only memories (PROM) and
erasable programmable read-only memories (EPROM).
2. Software programs or program parts are implemented mainly on random
access memories (RAM).
A survey of different data processing functions implemented by different
hardware, finnware and software tools is given in Table 4.1, and a typical
assembly of a data processing function of a digital input signal is shown in
Figure 4.2, which also specifies the individual process functions, controller,
hardware input/output and computer-programmed functions.
4.2.1 HARDWARE
Input time. The time delay between exciting the analog or digital input of a
device or unit and the formed information response ready for data transfer or
data processing at the device output.
Input rate. The maximum number of digital or analog data that enter
uninterrupted into a device in a given time unit.
Output time. The time delay between the input of an analog or digital command
to the device or unit, and the readiness of the unit to carry out the command
function at its output.
Output rate. The maximum number of digital or analog data that pass
uninterrupted through a device output in a given time unit.
Table 4.i implementation of different data processing functions
Hardware: Computer
input/output hardware
units and units - central
subsystems part
Analog input
processing
Analog output
processing
Hardware: Computer
input/output hardware
units and units - central
subsystems part
Digital input
processing
Digital output
processing
Hardware: Computer
input/output hardware
units and units - central
subsystems part
Printing, listing,
key-in
Displaying
Hardware: Computer
input/output hardware
units and units - central
subsystems part
Other functions
!electromagnetic
/' I ~ ;/-.)'·1 digital message : valve
input I
-if 0 :=::=:r --- analyser I
driver I
I
I
message printer
print-out driver
I
, I
I
I I
process: controller digital operating application program operatin~ printer : process
signal I display input system system : command
I
subsystem I
I
,- computer function
•
Figure 4.2 A typical example of digital input signal processing.
Data processing 143
Hardware input! Symbol Input time Input rate Output time Output rate Multiplexer! Sampling
output units and demultiplexer frequency
subsystems frequency
ls r-----------+-----------+-f
1msr-__________-+___________
l~sr-________~------_+.
Ins~~--------~~--------~-L~----------~----~p-e-r-c-o-m~p~o-n-e-n.t
InW ImW lW lkW
Figure 4.3 Microprocessor cycle time and speed of control system components
(American National Standards Committee, 1978).
Software for process control data processing 147
3 relative
computing
power
suppositions b:. 200%
2 - equal cycle time
- equal technologies
- equal instructions repertory b:. 150%
1 0 @] ~100% @] @]100%
50% 050% b:.
25%b:. computer
~--~---+--------~--------+-------~---word
4 8 16 24 32 length (bit/word)
o = bit processing
o = byte processing
A = 16 bit word processing
Figure 4.4 Approximated relative computer power (from (1977) Electronic Design,
21, 11,54).
capacity of data storage and data retrieval media, byte, such as ROM and
RAM, external memory (e.g. floppy disk and disk memory unit);
computing power of the processor, instruction/s, expressed as the number
of instructions actually executed in a given time unit;
speed of program exchange, program/s, that describes the ability of a
computer installation to change the program already in operation;
program efficiency (expressed as instr./byte x instr./error x progr./instr.)
that describes the number of bytes in an average used instruction, multiplied
by the average number of instructions referred to a program error and
multiplied by the average number of instructions in a program;
program span expressed as the number of different programs that are
available in the computer installation.
Tables 4.3 and 4.4 give basic data on computing power for three microcomputer
and minicomputer installations for process control, and it will readily be seen
that, despite the different hardware, all come out about equal when applied to
process control.
The software for data processing in the process control system comprises a
set of programs, procedures, rules and documentation. The extent of software
differs largely for each particular functional unit, which depends on its
hierarchical level, its role in process control and its connection with other
Table 4.3 Computing power data of different process computer systems
Data word size 4,8, 16 12, 16, 18, 24, 32 24, 32, 48, 64
(bit/word)
Instruction word 8, 16,24 16, 32 24,32,48
size (bit/instruction)
Add time register 0, 1-1000 0.OS-2 0.002-0.3
to register (micro- 1 0.1 O.OS
seconds per data
word)
Number of 10-unlimited 70-unlimited 2S0-unlimited
instructions 70 150
Hardware BCD No Yes No Yes Yes
arithmetic
Maximum input/ 30K-4M lOOK-ISM 3M-30M
output rate (words/ 50K-IM 300K-2M 10M
second)
Direct memory Standard optional Standard Standard
access
Numbers printed in italic type are the most common values or ranges
functional units. Figure 4.5 shows the main software functions of most advanced
process control systems.
The executive part of the real-time operating system coordinates the work
of all the software. Requests for activation of a particular program are the
result of a process event or time-programmed function. The operating system
assigns the permission for program execution on the basis of the time of
request occurrence and on the requested program priority. The other parts of
the operating system are software modules termed 'drivers' which enable the
connection of the computer process and a peripheral unit to the central processor
unit. Each particular peripheral unit, or a group of identical units, usually has
a specific driver which is part of the operating system.
L---..
1
executive
:~------------~ operatmg
I
I
I
1
- part system
I
---------1
process intercomputer
software • communication
software
message
switching
software
expert
system
software
extended
real-time
software
-
Figure 4.5 Main software functions of process control systems.
Software versus hardware 151
coaxial Ethernet
~--.--~--- -- -. --
( )
expert
system
Dl II~ designer's
-
station graphic
• • •
II ~
I station
~ c:~ '-..
II ~
r-- / ~L;---·-7
special
fibre optic
Ethernet
I
/ / ._-
process I.- ! -H-
-IT---- remoteI/O
=
..
station '------.------i
=
I
• • • • OLe connection
I
= ~- .. IY
-_ drive system
I i -----.-
Figure 4.7 Distributed control system with communication and expert system
support.
2. parallel processor architecture where all process points are fed to the
common parallel processor and processed in parallel; in such a way critical
process states can be detected very quickly and fed to the other parts of
the system.
Practical solutions today use a third approach, namely intelligent input/output
modules where a type of multiprocessor architecture is applied but without
the lateral redundancy offered by a parallel processor architecture. The problem
with parallel processor architecture is the lack of operating systems (Brajak,
1990) and supporting software tools. Pending a standard practical solution, a
more parallel architecture for process control, supervising and fast expert system
environment is proposed as given in Figure 4.7 with the following components:
operating system UNIX System V, release 4, featuring standard industrial
communication based on ISO open system interconnection, transparent
file system, real-time capability and standard graphics, icon, motifs;
communication software environment that connects different stations ranging
from pes to mainframes enabling
154 The relative roles of software and hardware
(4.2)
j
where Y is the production cost, i is the interest rate, Pj is the jth product cost,
kj is the capital needed for jth product design and w is the unit cost of labour
(Horvat, 1987).
For n identical systems, the total initial action Tn is a function of hardware
H and software S actions, thus
Tn = nH + S = const (4.3)
The minimum marginal costs are obtained from equation (4.2) when
nH =S (4.4)
Synergism of hardware and software is reflected in the amplification of system
functions by using proper hardware/software trade-offs. Fast process actions
like status signal processing are performed by parallel hardware actions. Slow
process actions like a set-point issue are performed by software actions. A
rough calculation shows that for each status process input approximately 50
to 100 bytes of program are needed for filtration compared to the hardware
solution (Huelsman, 1970) given in Figure 4.9 that costs approximately $10.
Thus, synergistic solutions that involve both hardware and software can be
cheaper than only hardware or, preferably, software solutions.
hardware system software system
yes
OR circuit f = f1 + f2 OR structure
A
B
C
D-
f E-
1 F-
G-
H-
A-
B-
C-
f2 Q -
E-
F-
e
H-
1 fJ.f
operational
I
amplifier
O.39fJ.F 47NFI
REFERENCES
American National Standards Committee X3 (1978 ) Computers and information
processing, organization and procedures, 13/SD-2 CBEMA Washington, DC.
Bailey, S.J. (1988) Control Eng., July, 75.
Bodlovic, P. (1977) Koncar Information Globus Zagreb 24, I, 9.
Brajak, P. (1990) Personal communication.
Evans, C. (1983) The Mighty Micro. Publ Gollancz, London.
Horvat, B. (1987) Working theory of costs, Rad, Beograd (in Croatian).
Huelsman, A. (1970) Active Filters, McGraw Hill, New York.
Milovic, M. (1983) Personal communication.
Schindler Electronic Design (1983), March 17, 117.
5
System software
5.1 INTRODUCTION
157
158 System software
programmed
tasks
•
operating
system
/ ~
programmed input/output
tasks units
interrupted. The request has the fonn of a programmed interrupt so that its
priority and the address of the continuation of the interrupted program are
automatically saved at the top of the interrupted program queue at the moment
of the execution of the request instruction. The table of interrupted programs
is usually of fixed length and it is filled bottom-up and emptied top-down,
thus enabling the inclusion of interrupted programs according to last-in-first-
out order, and saving the priority order of interrupted programs because the
last interrupted program has the highest priority in the queue of interrupted
programs.
When programmed tasks use some data input or output unit, they occupy
this unit by requesting it through the operating system that has routines for
input/output requests, for the extraction of the first request out of the queue
and for the omission of the processed request out of the queue. The request
for the input/output unit also contains a unit tag. A separate queue is formed
for each input/output unit by an acceptance routine. This queue is a list of
requests ordered according to the priority programs that request the input/
output unit and satisfies a first-in-first-out rule. Certain data should be given
to the operating system in order to accept input/output requests. These data
are:
tag of the requested input/output unit;
address of the working area for data including the direction of data transfer,
number of data or data volume and data code;
infonnation on execution of program after use of input/output unit that is
usually the relative address of the place where the program has to be
continued.
The program that continues after using the input/output unit is stopped until
the unit executes its task, thus disabling the change of the working area before
data are read in (transmitted) or written (accepted).
Figure 5.2 illustrates the routines for data input/output. The acceptance routine
of the input/output subsystem accepts all these requests and should this unit
be free the program proceeds to the program for input/output communication
and sets the integer variable of input/output occupation. Any input/output
communication program uses the routine for extraction of the first request out
of the queue at its beginning. Omitting a processed request from the queue is
done by the routine shown in Figure 5.2. The routines for input/output requests,
the extraction of the first request from the queue and the omission of the
processed request from the queue must not be interrupted until completed.
This feature is executed by these routines using the highest priority level.
An operating system contains the components that handle job management
functions (or executive functions), system management functions and data
management functions. Tables 5.1-5.3 outline the particular functions of an
operating system.
I/O unit request from programmed task
acceptance routine of
I/O request
(I/O RR)
no
I/O unit
I/O unit communication program
immediate free?r----------- - ------- - ------- --- - - -..,
return to I
I
programmed task I
I
I
I
deletion of
processed
request out of
the ueue
yes no
call of program
program to be continued?
continued
I
I
I
I
1 execution of
I continued
I
I program
I (BPR)
I
i
1______________________________________ J
I
Operating systems for process control can be divided into time critical
operating systems and time uncritical operating systems.
The time critical operating systems are met in some process control systems
on the lowest hierarchy level.
Figure 5.3 illustrates how most systems specify typical time response ranges
(shown by a horizontal bar) and shows that the mean response times peak at
approximately 150 J.ls (actual data in Table 5.4).
162 System software
percentage of system
%
150J,ls
I ... .....
50
, -~
/ I ,
, I "
40 , I ,
, I ,
, " II \\
, I \
30
I : \
, ~I \
I I \
20 / ! \
, I \
/ I \
10 //
/ " .....
.,.'
"
5 10 20 50 100 200 300 1000
Company System name Supported Size (kbytes)* Speed (Ils) Supported Comments
processors languagest
* The first number gives the size of the kernel, the second the minimal size of the operating system
t B = Basic C = C language F = Fortran P = Pascal
Structure and functions of real-time operating systems 165
exchange; --+,
Figure 5.4 Global structure of real-time operating system.
program exchange.
~, data
user's driver
program N M
monitor interruptI
waiting
list
system :
clock to tl
3
~
inte rrupt
t2
4
I
:
software
priority
:::::: ~ t31 levels
5
I
- tI~~}~
:~t;:::::;
6
priority levels
J
3
4
to t1
t
sy stem cloc k interrupt
I
I
store 110 I
I/O queue I
.
I
I
I
I
I
r-------::;;o..r"I/O
queue
I/O processing empty yes
via common
resource~
routines. The sequence of saving and subsequently restoring all the contents
of processor registers is initiated after each change of processing control.
Data flow activities exist that are not processor controlled, e.g. data transfer
between memory and disk or floppy disk units. These transfers are executed
at the highest priority by using hardware (cycle stealing).
The program activities at different program priority levels for a particular
real-time system are given in Figure 5.8, where the scheduler is triggered by
a timer interrupt to perform either the interval control routine at the priority
level 3 when there happens to exist a real-time data acquisition, or to proceed
to level 4 priority when there happens to exist off-line data processing. Having
timer interrupt
, / ,
I ,
sc heduler , program progra program
I
I
,I
~~. yes (real-time : yes (real-time data
,i I
I I data acquisition) : acquisition) ~---
Figure 5.8 Some program activities at different priority levels (Mikovic, 1980).
170 System software
A B c D
scheduler and
system programs
II
II
interrupts and
DMA transfer
I data
acquisition I data
processing
/
~.x_
~
I
I
I
I
I
( \
I
I
I
I
I
I
I
I
I
I
I
basic interval
....
I
I
'....
I
+"
timer interrupts
--==-==::.--~-~--------------- +
01 I I c I A B
basic interval
D " A I I I I[ ~t
B c
basic interval
D
If---- -~----iI
end
prepare
block
~~._______
. un_b_IO_Ck____~C
scheme of program tasks varies for different real-time process control systems.
The stochastic part of program execution tasks lies in the process interrupts
caused by status and count-pulse inputs. Status data are usually processed at
priority levels higher than measurement data and count pulse data because of
the need to associate time tags of higher accuracy to status data to present a
chronological event record, or to intervene faster in the process after receiving
emergency status data.
Usually a task can assume four distinct states:
processed state when it is being processed in the central processing unit;
suspended state when it is put into inactive state;
ready state when it waits for the assignment of the central processing
unit;
waiting state when it is blocked waiting for other tasks to perform (i.e.
make free a certain I/O unit).
The exchange of states is performed through operating system routines (Figure
5.10), and an example showing the execution of three tasks is given in Figure
5.11.
A multiprocessor configuration involves the communication of many single
processors through a global memory, i.e. a memory accessible to all the
processors. The communication is based on send and receive mechanisms to
or from this global common pool. Figure 5.12 presents the difference between
priority
TASK
CD ® @Q 0@ G)®®
~ ~
p p
W
a S
p p
b R
c R
Figure 5.11 Task priority control: 1, starting task a at initialization time; 2, task a
starting task b; 3, task a starting task c; 4, call of task a waiting for ending of
tasks band c; 5, call of task b waiting for ending of task c; 6, ending of task a
call; 7, ending task a; 8, ending task c; 9, ending task b. Task state: S, suspended;
P, processed; R, ready; W, waiting.
local
end (A) mailbox
J
r ceive (A)
local and global data traffic. The synchronization of tasks can be realized
through a polling mechanism where a cyclic task is started in a goal system.
This cyclic task searches for messages in the global mailbox and takes
appropriate actions upon them. The interrupt method makes use of the interrupt
(OUT-command) to start the task in the goal system whenever a corresponding
message has been found in the global mailbox.
An operating system makes an overhead in data processing systems that
can be calculated as 20% of the time (Mikovic, 1980) for a single processor
microcomputer installation. The time response overhead is approximately
1 ms in data acceptance, compared with 100 f.ls in a minicomputer operating
system (see Figure 5.7).
When written in C critical components of the operating system can add
approximately 20% more time overhead and make program writing a lot more
comfortable (Rus, 1990).
Multiprocessor microcomputer systems show approximately the same time
overhead characteristics (Borevkovic and Cvitas, 1983) since the most critical
time response functions are done at the highest priority level by a single
microcomputer in a system.
Operating system data are used to build up and enable a flexible and modular
exchange of system functioning when applied to different process control
functions. For example, a microcomputer control system has the following
sequence of programs and data:
1. System parameters
2. Operating system
3. User's routines
4. RAM memory programs and stack
A more detailed sequence is given in Figure 5.13.
System parameters are set by the statement EQU. The symbols that have to
be defined are:
XSTL - stack length, 80 bytes plus maximum stack length for each
priority level - an approximate total between 150 and 200
bytes;
XPNl, XPN2, XPN3, XPN4, XPN5, XPN6 - maximum number of
programs for each priority level;
XRTC - address of system clock;
XRTW address of bus control watch-dog;
XBAS basic time interval (ms);
174 System software
2K
user's programs for control
algorithm realization and
calculation (macros and algorithms)
3K
( initialization routine )
4K
function generation tables
and constants
KrROM
J
1-8
Figure 5.13 PROM and RAM usage for a microcomputer control system
(from M. Mikovoc, Internal Report, No. 1651, R. Koncar).
XRWM - RAM space detennined for system stack, system tables and
parameters; the calculation of space is given according to the
relation
6
XRWM = XSTL + 108 + 3 L XPN (5.1)
i=l
System parameters may be set up in the following form
XPNI EQU 2; 2 programs at priority 1 level
XPN2 EQU 2; 2 programs at priority 2 level
System software 175
System software for process control systems includes real-time system software
for on-line and off-line application. The on-line system software covers operator-
system communication, debugging routines, utility routines (mathematical
176 System software
system user's
programs programs
I
read write
analysis read write analysis
and and
read-in print-out
,I
l
I
r
+--5
level 2
driver
6,
communication
control
+-7 ,-7
driver I ... driver N
--~j~- ~A-~-----~-,--
-8 -8 - 8 -8
---------
1 -8
-
Table 5.6 Definitions, variables and tables for a minicomputer operating system
110 unit
interrupts
r- --- ~-----------------1
feedback
r-+-
I scheduler
1
I/O unit to user's
1
1 drivers f------- request program
1
f
I I
1
1
1
user s read/write I eXIt
I I.
program -----r request 1
to user's
program
request I processor I
f
1 1
1 I
I I
I request I
I ~ entry I
,
1
Ii
processor 1 scheduled
I program
I
I starts
user's
program I executive
control
1
I
U.
finished I program
1 restores
1
interrupted
1
I I program
1
1 • I
operatmg system
•________________________ J
application
programs
Figure 5.16 Application programs and program mechanisms in the PROZA 11 DIR
(Debeljuh and Vranit, 1982).
by personal computers and provide fine and effective solution for many non-
risky smaller process control applications.
REFERENCES
ASEA LME Automation Production Information 0011, PK1 962, Nr 88610023. ALRX-
11 Real-time Executive Programmers Reference Manual.
Borevkovic, B. and Cvitas, Lj. (1983) Koncar Strucne Informacije 30, 1-2,47.
Cihlar, z. (1980) Proceedings of the 9th Yugoslav Symposium for Measurement and
Measurements Equipment, Prishtine, Part II, pp. 539-51.
Debeljuh, V. and Vranic, P. (1982) Koncar Strucne Informacije 29, 1-2, 4 .
182 System software
report definition
report key-in
report files
- acquisition
- timing
e--1-\
(
screens
1
report blocked - screen reports r-
report list r- - printer protocols
( keyboard
}t- +
( mimics key-in l J mimics
~
operator
list of mimics ) ~
display of mimics
command
dispatcher
functions
~ database primary ( copies )
data processing
timer update
I/O functions
- local process I/O
~.
- RTU data
communication
'1
J
- r.roc~ss control
timer update
J unctIOns
- timer processing
( printer
control
- control of system
JI
responsibility
modifications
of responsibility
1/ 0 hardware
6.1 INTRODUCTION
Each process control system demands its own set of application programs
and possesses its own database. Simple process control systems require a modest
volume of application programs - 2 kbytes (Mikovic, Internal Report No. 1651,
Rade Koncar), whereas complex systems require 250 kbytes (Debeljuh and
Vranic, 1982) or more. Many complex systems are split up into smaller func-
tional systems since the cost of process control systems increases exponentially
with program complexity (Laduzinsky, 1982). Application programs use program
modules already developed, or macro-instructions, higher languages (e.g.
FORTRAN, C, Pascal and Ada) of functional symbols (ASEA Information
YLDA 78-08; STEP, Siemens). Some can be produced using different tech-
niques, tested, linked together and tested again, but extended real-time programs
are usually written in higher languages.
Databases for real-time applications are normally small and fast refreshed.
Extended real-time bases and knowledge bases are complex and require a lot
of data processing to be properly updated and maintained. Distributed databases
are found in distributed process control which require special methods and
techniques of updating - these include independent recovery of site failures,
resilient protocols and a complete point-to-point communication system
(Martinovic, 1983).
183
184 Application programs and databases
real-time systems. This is because they answer questions on where input and
output data are fed and what logic or other data processing functions should
be performed in what time restrictions. Table 6.1 defines an input and output
data signal.
More complex applications, e.g. positioning systems, can be solved by
combining specific application programming modules in a given operating
system that also possesses driver routines for the connection with hardware.
Figure 6.1 shows a block scheme showing software control of the analog
output.
A more systematic approach to programming is necessary for a larger control
system where previous techniques might prove to be cumbersome. A special
set of symbols for use in program function diagrams has been developed by
ASEA, Sweden.
A language tool, TLM from Siemens, using a successful combination of
logic and arithmetic (analog) variables, enables structuring of control functions
and parametrizing of process values, checking of control functions, use of
graphic features and protocolling of process values.
Application program tasks 185
system initialization
,-
I
I
: {routine
: ---- active at
,--_ _-'-_ _ _-, level 3
calculation of
control function PAG
and ramp generator
synchronization for process
values output analog
variable [ control to
enlarged ,------1.-------------,-- operating
detail of MEH system
__ analog unit~_-I set-point for
output I analog output
control I unit
IL-----r___--'
I
____ {cont~ol to
I
I
I
I
I mom tor
'-
Figure 6.1 Block scheme of positioning system with a software detail of the analog
output control (from M. Mikovic, Internal Report, No. 1651, R. Koncar).
primary data
communication with
peripheral
units syst~~~_o_p_e_ra
~Place
.....tor's
communication
with
basic
data
interactive mechanisms
"-
"-
"-
'\
- data read-in by means of
\
\.
\ standard protocol on display \
\ - direct read-out of data
on display
- acceptance of characters
.. \
\
from keyboard
\
\
,
data read-in/out
mechanisms \
/
,/
,/
Figure 6.3 Interactive communication with operator's control place (Debeljuh and
Vranic, 1982).
188 Application programs and databases
Example
An industrial data transmission unit can be defined by the following charac-
teristics.
1. Computer: an M68000-based multiprocessor unit, up to 640K memory
and separate interrupt unit.
2. Hardware interfaces: communication unit, local console pushbuttons and
indicators, local printer for process data print-outs, bus extension, process
interface modules for up to 256 digital inputs, 128 analog input and 64
binary output, real-time clock.
Structure and timing requirement of application programs 189
Real-time tasks
Instantaneous input of process data - analog + + + +
Instantaneous input of operator's data
digital + + + +
analog + +
Data processing
logic + + + + +
arithmetic + + + + +
digital filtering + + +
complex + + + +
Data presentation
real-time data presentation + + + + +
post-mortem analysis + + +
chronological event recording + + +
Data output and operator's intervention into
process
direct control of important process points + + + +
control of on/off process points + + + +
control of set-point values + +
remote control of process points + + +
Communication facilities
data and commands communication + + + +
message switching + +
Extended real-time tasks
Process state calculation and estimation + +
Process state simulation and modelling + +
Process identification and optimization +
Expert system tasks
Automatic start-up + + +
System parametrizing + + + +
Functional diagnosties + + + + +
192 Application programs and databases
/
./
./
./
./
./
./
./
./
activation ~
activation
l~/I ./
./
./
./~
t
device activation device application
driver driver programN
application
program
,
activation application
program 1
process activation~ time-delay
I/O unit module
activation
operator's activation activation
console
L/~:J
Figure 6.4 Activation of application programs.
change of
pointer of
digital input
table (DIT)
activated
by operator
print-out
program delete the
(P5) known message
out of digital
input table
procedures are shown in Figure 6.6. The programming chains for a specific
class of messages are given in Table 6.3 and a complete schedule of data
processing of digital signals is given in Figure 6.7. The four subsets of data
processing chains of the application program for the acceptance of messages
and the direction of data processing (see Figure 6.6) are:
API = (AI' A2, A3) (6.1)
AP2 = A4 (6.2)
AP3 = As (6.3)
AP4 = (A6 , A7 , Ag) (6.4)
A stochastic schedule of the execution of program subsets API, AP2, AP3
and AP4 exists, since the change of digital input signals is of a stochastic
nature. The program chains for already printed out messages to get out of the
queue are A 10 and All. The put-in and get-out of table (named DIT) is
completely synchronized if at the end of the data processing interval the number
of executions of subsets N(APl) and N(AP4) equals the number of chains
N(A IO ) and N(A ll ), that is, when
N(API + AP4) = N(A IO + All) (6.5)
The beginning of the execution of a program chain Ai denoted Al and the end
of the execution of a program chain Ai denoted Al are connected in the fol-
lowing manner
(6.6)
The given characteristics of the described operating system enable the inter-
ruption of a lower priority program by a higher priority program; however, it
cannot be activated until the interrupted sequence is completed. The overlap
of program chains inside the same program is therefore not possible and the
determinacy of the program P2 is preserved. Naturally the program chains are
serially connected so that after the last task of a program chain Ai the first task
of the program chain Aj is continued.
There are two ways of communicating between application programs and
operating systems: direct and indirect communication. The indirect commu-
nication of programs is performed via the database.
Class of data processing Subclass of data processing Chain of'tasks Number of tasks chain
Process interaction and print Message received in empty T] < T2 < T3 < Ts < T7 < TIO < T]4 < T]S < T]6 < T]7 A]
out digital input table
Message read-in into last free T] < T2 < T3 < Ts < T7 < TIO < TI3 A2
location of digital input table
There are still free locations T] < T2 < T j < Ts < T7 < T]O < T]4 < T]7 Aj
in digital input table
Message transmitted to Number of locations in digital T] < T2 < T3 < Ts < Tg < T]2 < T]3 A4
program for process input table unchanged
intersection through digital
outputs
Message transmitted to Number of locations in digital T] < T2 < T3 < Ts < Tg < T]] < T]3 As
program for production input table unchanged
registration
Message to be printed out Message read-in into empty T] < T2 < T3 < T4 < T7 < TIO < T]4 < T]S < T]6 < T]7 A6
digital input table
There are still free locations TI < T2 < T3 < T4 < T7 < TIO < TI4 < TI7 A7
in digital input table
Message read-in into the last TI < T2 < T3 < T4 < T7 < TIO < T!3 A8
free location of digital input
table
Message of nonconnected Number of free locations of TI < T2 < T3 < T4 < T6 < T9 A9
digital input digital input table remains
unchanged
Deletion of already printed After deletion of message the TI8 < TI9 < TI3 AIO
out message from digital digital input table remains
input table empty
After deletion of message TI8 < TI4 < TIS < TI6 < TI7 All
from digital input table the
print out of left over
messages is activated
program
acceptance of status priority
signals from process levels
data and proceedings of PI 5
,
processing level I status states to
application programs
~
data
production
processing level 4 P4 3
registration
~
data operator
processing level 5 print-out on teletype P5 3
~/~ P3
r- ::~~--.J P4
! """-..1 L
! 1 P5 r --I
I'- ___________________ _ _ _ _ _ _ _ _ J I
postpone
_~ send) (\
flag = I flag set
program PI
internal variable flag
message - I
flag = t,3 flag reset
EMTI 1 •
indirect transmission direct flag r eset
of message to P2 ifDIT< N
t
message
receiver ( DIT<N )
internal variable DIT
program P2
Hi is the worst case amount of time when the ith task chain is blocked
or slowed down by other tasks by using or needing to use the same
devices or resources.
Structure and timing requirement of application programs 201
task 1 task 2 t
resource Rl --"::"'1>
task 3 task 4
resource R2
task 5 task 6
lOms
.'4non-resource
6ms
Of • 4 ~resource3ms
, lms......
The longest time is when the ith task chain is blocked or slowed down, and is
Bi = L TPR j [(TTC;I TTC j ) + 1] + BDi (6.9)
j=l
because the jth task chain can be executed (TTC;/TTCj ) + 1 times during a
given execution of the ith task chain (resource blockage); the blockage BDi
due to the use of devices by other task chains has to be calculated for each
particular case. Figure 6.9 shows an example of a guaranteed response time
for three task chains.
In a given example TPR i = 1 ms, TDi = 13 ms and TPNR i = 6 ms for i =
1, 2, 3. Bi = 2·2 ms + 6 ms = 10 ms for i = 1 and 2 because the blockage
caused by devices in other task chains is only 2 . 3 ms = 6 ms; the task chain
1 is, for example, slowed down once in its device D 1 action by the D 1 action
of the task chain 2 and once by the Dl action of the task chain 3, because the
total time requirement of task chains according to equation (6.9) equals 2·2
ms = 4 ms. B3 = 2·2 ms + 6 ms = 10 ms also for the task chain TC3. This
is due to the last task of the task chain TC3 being blocked by the first tasks
202 Application programs and databases
of both task chains 1 and 2. With the help of the relationship between differ-
ent tasks.
TTCj = X· (relative TTC j) (6.10)
where x can be chosen to select any set of times that maintains the relationship
between different task chain response times, equation (6.7) becomes
ratej = TPNR j . X· (relative TTCj) - TPR j - TDj - Bj (6.11)
By putting the calculated data into equation (6.11) and by letting relative
TTCj equal 1
ratej = 6/(x . 1 - 1 - 13 - 10) ms, for i = 1, 2, 3 (6.12)
When x = 42 the relative rates equal 1/3 and their sum equals 1. The guar-
anteed response time is 42 ms. If the task chains are actually restarted that
often, then the total processor usage is (21/42) . 100 = 50% rather than (7/
20) . 100 = 35% when each task is run separately.
Actual relations in the system time response include the operating system
overhead consisting of:
request handling time of the task chain that consists of the start task chain,
start task, request device, handle device interrupt and semaphore syn-
chronization time;
tennination handling time of the task chain;
timer interrupt handling time.
Each of these time-consuming operations has to be added to the required time
consumption given in equation (6.7).
When multiple or parallel (simultaneous) use of devices takes place then a
single semaphore is not enough. The problem when multiple or parallel
(simultanwous) use of devices takes place then a single semaphore is not
enough. The problem can be solved by using an array of semaphores
(Tanenbaum, 1987) which defines the state of all the devices in simultaneous
use.
Basic process software enables the connection of the operator, process and
superimposed or subordinated process control system by global system functions
such as those illustrated in Figure 6.10. The use of global system functions
depends on the actual process control application and the hierarchical level of
the process control system. The basic process software consists of:
database;
- programs for autonomous functions (process data acquisition, process data
(a) process data
process communication to
data higher
display hierarchical
level
process commands to
data process and control
acquisition of commands
execution
(b)~
I--------------r--l:~~:der
•
operator
•
I
process
Figure 6.10 (a) Main parts of basic process software. (b) The structure of a basic
process software.
204 Application programs and databases
r assignment word
I
type of
. • process data
.
I
. , process data
. , I apparatus point marking
markmg 000 - 999
(r; -
level 3 /S - single status data
~
00 - 99
level 2 marking double status data
{
marking 00 - 99
A - analog data
0-9
!
C - counter data
level 2
level 3
Name of status information Graphic representation Database parameters to be maintained for process
supervision and control
D
state of digital status that must be
maintained after control time delay,
single command command code, control function (open,
close, start, stop, switch on, switch
off), code of location and apparatus,
codes of peripheral units for command
display and recording, display
description
I
double command
'I..I"
10--
I,
2. Connecting tables between programs for visual data display or data print-
out and a process database.
3. Other minor groups of connecting parameters and definition tables as
well as system tables, among them specifically:
lower level hierarchy computer systems;
process apparatus;
process data points for the historical periodic record;
video display and printer lists;
system configuration and dynamic system data;
definition tables of group data;
mimic displays;
trend displays;
Database and basic process software 209
operating
system
application application
program 1 program 2
database
table
Data from a process database can be shared among many application pro-
grams. The support of database share can be organized by the program sup-
port inside the operating system and software mechanisms built in at the
beginning of those application programs that share a common database table
with other application programs. An example showing data sharing between
two application programs is given in Figure 6.12. A typical global structure of
a database, as shared by application programs in a teleprocessing environment,
is given in Figure 6.13.
These include:
1. Integration and control of data content: implies the coordination and
controlled integration across the number of application program areas to
210 Application programs and databases
SER FUNCfIO S
.,
•
\
\
II ,,
\
,
II
\ ,
\
~'''''''' \\
" .
'\
PROCESS I/ O
obtain efficient use and consistency of data structures. Thus, data dupli-
cation and data relationships are finally controlled.
2. Complex data structuring: implies some forms of complex file structur-
ing. For example, hardware pointers, inverted indexes or logic iden-
tifiers that provide a variety of entry points and access paths through this
complex structure to enable the efficient retrieval of data for a variety of
applications.
3. Retrieval and maintenance of data: provides the user with ready programs
of adding to or modifying the structure, retrieving and amending existing
data and storing row data in the database. The data manipulation facilities
should provide for the organization of parts of the database and for the
analyses and summarization of data contents and usage.
4. Data independence: implies the complete separation of the storage and
organization of data from the logic and processing of the various appli-
cation programs using these data.
Database and basic process software 211
Process data: the group of process data is formed from the data acquired
from remote terminal units or local process subsystems.
Derived data: are obtained from raw analog and binary data (count and count-
rate data) by applying arithmetic and logic transformations to two or more
process data.
User's data: are put into the system by the user. Read-in of these data usually
includes a part of software that realizes this data processing algorithm.
The access and use of a database subsystem in the program system PROZA
11 D/R is given in Figure 6.14. The organization of the database follows the
hardware organization of the process control system because the data of each
212 Application programs and databases
data classes
status data
process ---1---- --- ---
analog data
data ---1--- --
counter data
plausibility of,
user's measurement data
data
auxiliary
data
derived
data
operator
class 1
process input are uniquely determined with their physical input into the sys-
tem. The data key for each process signal contains the data class (e.g. status,
analog or binary variable) and a description of three level hardware con-
figurations through which signals are fed to the system (see Figure 6.15).
The data are partially fed into the fast memory of the central computer and
partially on to system disks. Data descriptors and parameters that demand fast
access and frequent use are stored in the fast memory. These data are ordered
in the same way as keys in the key tree, i.e. according to hardware structure.
Data descriptors and parameters that are rarely used are stored in disk media.
The access to these data is also done through the key tree by hash ad-
dressing, which does not impose serious limits, especially when a hardware
specification is not well known. According to Figure 6.15, the key is formed
by three memory symbols Klo K2 and K 3 , each for every key level. The
pseudostochastic transformation of this address is done by applying the following
function
Database and basic process software 213
level 1
1 1 1 1 data class (process data)
- centralized computer
- communication unit
at centralized computer
Figure 6.15 The key of the addressing tree in the system memory.
F(K K
" 2,
K)
3
= [R(K" K22w, K 3 )· A] + 1 (6.14)
Where the bracket [N] determines the greatest whole number of N equal to
or less than N.
A collision situation appears when two or more keys are transformed into the
same goal address. Two methods are applied against this process:
214 Application programs and databases
1. Connection of the whole primary bucket to each goal address. B keys can
be stored into each primary bucket. The value B is deliberately chosen.
The number of address keys AK, address space A and bucket length BL
are connected by
A=[:J (6.15)
where [N] denotes the smallest whole number greater than or equal to N.
2. When a primary bucket is full an overflow bucket of the length BL/2, is
reserved, where all overflow keys are stored. The address of the overflow
bucket is stored in the primary bucket. When the first overflow bucket
is full, a second overflow bucket is formed with its address stored
in the first overflow bucket. Thus, a list is formed for each goal
address that begins in the primary bucket with overflow buckets as its
elements.
The application of hash-coded addressing for 21 600 keys with the space that
is 24% greater than the minimal required memory space has shown the fol-
lowing:
10 18 448 cases it was possible to do a read-write operation with one
disk access;
in 2686 cases two disk accesses were demanded;
in 440 cases three disk accesses were demanded;
in 26 cases four disk accesses were demanded.
~ intelligent
~ r
in telligen t
operator's
I/O process I/O process
interface
unit unit
,
•
III
II J
-------- -------
processl/
Figure 6.16 The concept of a distributive intelligent I/O control system.
Table 6.6 Failures and repair design in distributed control systems
Figure 6.17 Two phase commit protocol with timeout transitions and undeliverable
message transitions: ---, deliverable message states; - . - . - ., timeout
transmissions; - - - -, last message undeliverable.
218 Application programs and databases
centre 1 centre 2
+ centralized computer
TL TL
, I
I
I
I
+
,+
front distributor
: (doubled if necessary)
I
I
remote terminal
units
process
--------- - ,I
safety reasons (Lichtberger, 1990) while a knowledge base needs more ex-
planation - see Part Three of this book.
A distributed configuration with a concentration point is shown in Figure
6.18. A front-distributed concentrator, DS-8FD, is located between a network
of remote terminal units and a centralized computer. DS-8FD can also be
realized as a double configuration. The basic task of a distributed concentrator
is data acquisition and administration from remote terminal units and their
distribution to centralized computers. A summary of the basic functions of the
distributed concentrator is as follows:
data acquisition for centralized computers;
message switching to centralized computers;
supervision of data transmission through appropriate protocols;
transmission of on/off commands and set-point commands from the cen-
tralized computer to remote terminal units;
generation of time synchronization upon request from centres and its de-
livery to remote terminal units;
communication speed conversion.
The basic programming structure of a DS-8FD system is illustrated in Figure
6.19.
The database system is a queue-oriented system, each queue containing a
given number of queues of different priority for each remote terminal unit.
The contents of these queues are mapped into an independent queue-system
TL to centralized computer TL
communication
program C2
sorting
routine
request scanning
processor processor
TLn
to RTU's
Figure 6.19 Basic program structure of the DS-8FD computer: TL, multiple
transmission lines; RTU, remote terminal unit.
220 Application programs and databases
for each centralized computer. Thus, a complete state of all remote tenninal
units is mapped into a queue-system of centralized computers. The DS-8FD
computer contains two independent queue-systems each for one centralized
computer. The organization and appearance of queues in a DS-8FD computer
is shown in Figure 6.20. This organization enables selective data acquisition
in centralized computers according to data priority.
FIFO 3 - detail
FIFO 1
FIFO counter
FIFO 2 RTU
buffer 1
buffer 2
iiiil J RTUn
FIFO 1
FIFO 2 RTU 1
centralized
computer 2
buffer m
The variables that characterize the state of accumulated mass, energy or infor-
mation are termed 'state variables' and denoted by x; the variables that char-
acterize the exchange of mass, energy and information with the environment
by means of control system hardware and software are termed 'interactive
variables' and denoted by y; the variables that characterize the quantity of
supplied mass, energy and information are termed 'input variables' and de-
noted by u. The models of real processes contain:
equilibrium equations such as equations (6.16), (6.17) or (6.18);
process state equations (see Chapter 2);
inequality relations that include limitations and specificities of the given
process (see Chapter 1);
probabilistic data (see Part Three).
Table 6.7 Models used in process control systems
The quantity characteristics of a given space and process are termed 'process
parameters', which influence the formulation of a model shape.
The mathematical model of a real process can thus be expressed as func-
tional L
L{ x(t), yet), p(Vi> t)} =0 (6.19)
where x(t) are state variable values, yet) are interactive variable values and
P(vi' t) are process parameters. Terms Vi (i = 1, 2, 3) are space coordinates
and t is a time variable. A static model of the real process is characterized by
dm/dt = dE/dt = dl/dt = 0 and equation (6.19) is simplified as a functional M
of the form
M{x, y, P(Vi)} =0 (6.20)
The process parameters p determine principally the relation between x and y.
The process parameters p are almost always deterministic quantities contain-
ing a small stochastic component like, for example, pipeline friction coeffi-
cient. Thus, a separate presentation of some examples of stochastic and deter-
ministic models (as applied to the process control systems in Table 6.7) is
taken as realistic only for cases where a deterministic model can be separated
from the stochastic model of a process.
The other part of the extended real-time software supports operator-system
interactions of a set of programs and routines that enable the presentation of
the process state and process variable trends in a way that is given in Table
6.7 as, e.g., consumption presentation, mimic and trend diagrams, failure model
presentation and system state and system variable trend reports.
The extent and amount of programming needed for the application of ex-
tended real-time software are estimated as being about ten times the efforts of
the complete process model software presently available. The main reason is
that additional adaptation work is needed. For instance:
fonnation, formulation and additional programming efforts for establish-
ment of the appropriated database of extended real-time software;
algorithmization and formatting of system tables for approximate operator-
system communication for the fulfilment of extended real-time functions;
adaptation of the existing computer system to the requirements of the
model software. The usual requirement for computing equipment con-
sidering different process models is given in Table 6.8.
A static model based on algebraic relations between the process variables may
suffice for the description of a stationary state of the process and the optim-
ization of plant operation, when plant response times are negligible versus the
minimum period of significant external perturbations. The model can be suitably
applied for the optimization of electroenergetic and water supply systems, and
partly so for natural gas transportation systems, for thermoelectric power plants
224 Application programs and databases
t = t max
I-:r--
ti +! t
ti
U
-r
I
F{x.1 tt~-
i) I
:i
jlF (xi+!' t i+1)
-·--------+F-'R-l-.G~\=-------F--="..:..R.:...'-'G..:...,--=,: . . - - - - - - -
-
- ' - -
it L-----'
- x
------"----------1 x=l
F = (F'+F")/2+[v.(F'-F")+(C~-v~)(R'-R")-(C.-v.)G"~t-(C.-v.)G'~t]
1 1 1 1 1 1 1
2C.1
where Ci - sound speed in a fluid
vi - fluid velocity
Figure 6.21 The series of calculation steps for gas transportation model: R fluid
density; G, friction coefficient; F, fluid mass flow.
REFERENCES
process data
system
operator
process
,
•
acquisition topology
determination
real-time
process state software
estimation extended
real-time
software
extended real-time
database
operator
communication
programs ,• process
operator
Borevkovic, B. and Cvitas, Lj., (1983) Koncar Strucne Informacije, 30, 1-2, 47.
Debeljuh, V. and Vranic, P. (1982) Koncar Strucne Informacije, 29 1-2,4.
Heninger, K. (1980) IEEE Trans. Software Eng., SE-6, 1, 2.
Jovic, F. (1989) Proc. MIPRO '89 Conference, Opatija, PU Section, 6-1.
Koncar, R (1980) Documentation of the Process Control System DS-80.
Laduzinsky, A. (1982) Control Engineering, Nov., 64.
Leinbaugh, D. (1980) IEEE Trans. Software Eng., SE-6, 1, 85.
Lichtberger, Z, (1990) Personal communication.
Lundberg, M., Goldkuhl, G. and Nilsson, A. (1979) Inf. Systems, 4, 1.
Martinovic, A. (1983) CEP, Feb., 72.
Skeen, D. and Stonebraker, M. (1983) IEEE Trans. Software Eng., SE-9, 3, 219.
Tanenbaum A.S. Operating Systems, Prentice-Hall, Englewood Cliffs, NJ.
Tunon, J. (1982) Control Engineering, Oct., 110.
Wolverton, RW. (1974) IEEE Trans. Comput., C-23, 615.
Zelenko, B. (1976) Informatica Proceedings, Bled, 5, 111.
Part Three
The Man-Machine Interface
7
Reduction and visualization of
data and procedures
7.1 INTRODUCTION
231
Table 7.1 Aspects and characteristics of human use of a process control system
7 .2 OPERATOR-SYSTEM COMMUNICATION
The problem here is one of data reduction and it is useful to consider the
following systems on the basis of their size, since this affects the amount of
information to be presented to the operator.
1. Small control systems, where the number of status and counter state data
is in the order of 200 or less, and the number of analog data is 100 or
less.
2. Medium control systems, where the number of status and counter state
data is between 100 and 2000 and the number of analog data between 50
and 1000.
3. Large control systems, where the number of status and counter state
data exceeds 2000 and the number of analog data exceeds 1000.
Large control systems demand a special communication between operator and
control system, and are often used, for efficient process control on the operator's
side, as a set of medium control systems tenned 'functional systems'. Small
control systems usually do not present any special problem to the operator-
system communication, i.e. the number of data presented or the number of
commands issued. The basic characteristic hardware functions of medium control
systems concerning the reduction of data number, visualization of data and
the issuing of commands are:
exchange of infonnation between operator and process by the central
computer and peripheral equipment;
logging of process data and historical process data performed by the central
computer and peripheral equipment;
234 Reduction and visualization of data and procedures
system
( operator
central
process (~,)
me~ory
umts
Loperator
system
cons0!s--
unit
A I controller
system bus
I 1
"
fast
I I
semigraphic
memory colour
unit console
bus/bus •
• ( productio
converter
II planning a
supervisio
I
data
I
data
I
process
input output input!
units units output units
f•
•
I ~
process process
command signalization
,
console console I
• to /from process
devices/sensors
process
operator
that has just been acquired but has not yet become part of the long-term
memory. The work by Peterson and Peterson (1959) indicates how simple
words and three words are recalled after short time intervals (Figure 7.2).
This ability of humans emphasizes the need for simplification and reduction
of data presented to the operator. The reduction of process data requires more
236 Reduction and visualization of data and procedures
correct recall
(percentage)
100
'\\
\ one word
\ --____~~73~o/.:o_____71% 71%
'q68% 0------'0
\,
50 \\
~%
""""o.! 5% three words
___ 20% 18%
-<>-----c:>-_ 16%
----0
3 6 9 12 15 18
retention time (seconds)
Figure 7.2 Correct recall of one or three words after short time intervals.
percentage difference to
physiological average value of
working efficiency
%
60+----------r--------~----------+_------~
40+-~~--~~--------~----------+_---------1
40+---------~--------_+----------~~~~~1
60T----------r---------+--------~~~~~~
\ ,
, ....
80----------~--------~--------~------~
6am 12 6pm o 6am
time of day
The control levels or priority levels of a power station are illustrated in Figure
7.5. For instance, the control of the power station is handled at the supervisory
level enabling the coordination of power plants (e.g. monitoring the overall
station efficiency, maintenance, staffing) and the control of separate power
plants involves the control of the entire plant as a system such as plant start-
up, plant load change policy. The control of plant subsystems enables the
efficient and quick operation on separate plant's subsystems (e.g. burners,
cooling system, heat exchangers) and system component controls enable the
separate operation of each plant component (e.g. pump, motor, valve).
Organization of hierarchical levels of data presentation usually varies between
two and four levels of data presentation.
correct repeating of data associative memory
direct after reception often refreshed
Table 7.4 Communication functions and channels for the three control levels
power
station
power station control
functions
other functions
power
power plant 1 power plant 2 plant
control
functions
system
plant plant plant ••• control
subsystem 1 subsystem 2 subsystem 3 functions
component
component 1 component 2 component 3 control
functions
First level This keeps the operator informed that the system is working without
perturbation and what has to be searched for with the highest priority in the
case of perturbation.
Second level This exhibits the structure of the process, or plant, that is
controlled, indicating the principal energy and material flow and processing
states and including an alarm summary with the last received alarm.
Third level This enables the operator to see those process details that actually
exist and those process and alarm states that have actually happened.
Fourth level Additional data are shown for the supervision and control of
specific process components, e.g. analog values, limit values and zone values
of a specific process variable.
Table 7.5 details a typical four level process data presentation for normal
and alarm process conditions.
An important view of operator-system communication is established by
report issuance that serves the process state documentation, and the initiation
of the necessary process state calculations. The report issuance can be divided
Visualization and data logging 241
into process event documentation and collective process data issuance (Table
7.6).
Thus, visualization and data logging can be broken down into two areas:
Information level Type of display Display hardware Information level Type of display Display hardware
Analog values:
process primary 1 Alarm indication CRT display
values such as 2 Not displayed when passing of Operator's console
voltage, current, 3 limit value
flow, level, etc.
process globai 4 Completely CRT display 3 Text indicating CRT display
values such as net displayed Measuring the cause of
net frequency, Permanently instrument perturbation on
valve position, etc. displayed Analog recorder CRT alarm field
Possible selection Numeric indicator
XY plotter
Counter state, 4 Possible selection Measuring Usually not
Increments instruments displayed
Counter Display
244 Reduction and visualization of data and procedures
2. Measuring values
Communication of Issuing of measuring and Periodically or on demand
measuring values limit values
Supervision of Issuing of selected Periodically or on demand
measuring values measuring values
Statistics Recording of mean and On demand and often
maximum values graphically on CRT screen
or on analog recorders
3. Counter state
State of process All counter states Periodically or on demand
variable counters
The numbers and percentages are based on the US and European market. VRC, vertical
retransmission control; LRC, line retransmission control; CRC, character retransmission
control; HDX, half duplex transmission mode; FDX, full duplex transmission mode;
SYN, synchronous transmission timing; ASYN, asynchronous transmission timing
Table 7..8 The main features of semigraphic and graphic display units
LS
VC
FS
~ TT
---
-- --
- -- --- -
FT
Figure 7.6 A simple hypothetical basic process unit: VC, valve command; FT, flow
transmitter; IT, temperature transmitter; FS, flow switch; LS, level switch.
Valve 2 2
command
Flow (high) (nonnal to high)
transmitter 2 3 2
(low) (nonnal to low)
Temperature (high) (nonnal to high)
transmitter 2 3 2
(low) (nonnal to low)
Flow switch 2 2
Level switch 2 2
Figure 7.8 The usual format of the alphanumeric video display data.
states can be supported by the time tag, so that the obtained print-out can be
used for official supervision and arbitrations. Fast, complex and mutually in-
terconnected processes (e.g. electrical energy generation, transmission and
distribution systems) are then provided with such functional recording systems
usually termed chronological event recorders, that are primarily used for the
time detection and the time-of-event discrimination. Such chronological event
recorders are usually installed for a particular process control. For the large
interconnected systems exact time distributions have to be provided to satisfy
the demand on precise time-of-event control in a process.
The data on the process state and on the change of process state appears on
the screen of the operator's console usually iri order of appearance to enable
faster decisions by the operator and production supervisor. An example of the
process data print-out is given in Figure 7.7. The visualization of process data
on alphanumeric video displays greatly resembles the print-out formats (Figure
7.8), but it can be modified to present data in different forms like those given
in Figure 7.9.
The most efficient data visualizations are obtained by using semigraphic
and graphic video terminals where process mimic diagrams can be shown and
NETWORK INPUTS 03 - JULY 81
CONS 1 PROD 1 PROD 2 CONS 2
-------0 -------o--------o---------o----------q------------o
I P = 43.0 P = 0.1: I
l F = 9820 F =0 I : PROD 5
I I
PROD 4 I I
I
r---9 PF == 43.0
53928
, I
lI P = 42.0
I I
PROD 3 I r-----o-- F = 2110 I
---0--------- 0 I :
P = 28.2 I I I
F = 67514 I <? ----------------- <? CONS"3---------- ~
I
I
PROD 6:
P = 43.0 O----~
F = 5450 I
I
II
, I
I I
I
I • I 1
b- ---0----------------- 0 ----Q----Q---0---.L.---.Q PROD 10
CONS 4 PROD 7 PROD 9 P = 16.0
P = 42.0 P = 41.0 F = 2269
F = 11412 F = 774
Figure 7.9 The pseudo mimic formatting of data on alphanumeric video display:
PROD 1, natural-gas production point 1; CONS 1, natural-gas consumption point
1; P, pressure (bars); F, flow (cubic metres per second).
Visualization and data logging 251
dioptres
t
accommodation
28 year distance
8T-----------------~C------------------------- 12.5cm
6
16cm
41 year
4
25cm
2
50cm
1 2 3 4 5 time
(s)
Figure 7.10 The accommodation step response with time and age of operator for
yellow-green colour (Krueger and Muller-Limmroth, 1981).
provided with up-to-date process data. Such an interface enables the presentation
of the process to the operators in a way that they can:
see the actual structure of the process and the interactions between specific
basic process units mainly by means of the static part of the mimic
diagram;
estimate the quantitative relations between process variables mainly by
means of a variable part of the mimic diagram that presents actual and
historic process model data.
The ability of a mimic diagram to enable the fast magnification of process
details by calling mimic diagrams of specific process parts of complex processes
increases the speed of decision making by the operator in the case of vast
numbers of process data. The efficiency of visual presentation can be increased
by approximately 40% in the information content (Smrkic, z. (1962) personal
communication) by a colour video display unit. The human eye has different
reaction times, depending on colour. The cause is possible due to chromatic
aberration where corrections of about +2 dioptres is needed for red, and -2 for
blue (Figure 7.10). Thus, yellow-green video display screens are preferred for
processes where fast operator's reaction is expected.
252 Reduction and visualization of data and procedures
Features Colour
Colour
between the operator and the computer via an interface. The quantitative measure
of merit of the dialog can be calculated when the content of information sent
by each of them is compared to the content of information received by any of
them. There are crucial points in this dialog design:
1. Creation of environment for the user to use the computer properly.
2. Creation of safe and comfortable dialogs. This implies, for example,
standard keyboard style of terminals, the same special characters all over
the plant control system, dangerous commands to be typed character by
character.
3. Involvement of the user into system and dialog design process.
The goal of a dialog is the recognition of process states. The recognition of
process states is based on previous knowledge of the process state
interdependencies, as opposed to the recall process by an operator. The result
is a higher percentage of retention of the operator's knowledge of process
states (Figure 7.11). The success of dialog therefore depends on a menu-
driven dialog, user involvement in the system and dialog design process and
on available computer graphics. The design of an operator system dialog based
on Pascal, such as CCITT language, named MML (man-machine language)
(Hombach, 1982), facilitates the operation and maintenance of computerized
process control systems.
A fully successful operator-system interaction can be designed for coarse
processes that change slowly with plenty of informational feedback. There are
environment critical processes that degenerate fast and with much interdependent
complexity where operator's control is impossible, as shown in Figure 7.12.
To enable operator control, an adequate design of basic process units, computer
hardware and software must be made (ASEA-ATOM, 1982) and expert system
engineering included (ABB, 1989). Using knowledge base engineering methods
a further reduction in data volumes is achieved by pointing from process facts
retention(%)
100
90 .--~
80
70
60
50
40
30
20
10
20 min 1 h 2h 24 h 48 h
y expert
system
,
support
all manual ; ~
control I ~ control concept
limit 1// change
high
I
I
I
I
I
I
I
I
I
I I
increasing control
difficulty
low
~--------------L------------------------- __~x
slow-stable-simple fast unstable
complex
REFERENCES
ABB (1989) GRADIENT - Graphic Dialogue Environment, Publ. No. D CRH 133
89E.
ASEA-ATOM (1982) Special Features of the Control Equipment for the ASEA-ATOM
BWR Appendix to IAEA Guide Book on I & C. ASEA-ATOM, Feb.
Dallimonti, R. (1976) Instrumentation Technology, May, 39.
Hornbach, B. (1982) IEEE Trans. Commun. COM-30, No.6, 1329.
Krueger, H. and Muller-Limmroth, W. (1981) Arbeiten mit dem Bildschirm-aber richtig,
Bayersche Staatsministerium fur Arbeit und Sozialordnung, Munchen.
Livingston, W. (1982) IEEE Trans. Power Appar. Syst., PAS-lOl, No.8, 2543.
Peterson, L. and Peterson, M. (1959) Journal of Experimental Psychology, 75,
193-8.
Woodson, W.E. and Conover, D.W. (1966) Human Engineering Guide for Equipment
Designers, University of California Press, Los Angeles.
8
Process management and control
8.1 INTRODUCTION
Once the problem of hardware for process control had been solved the next
move, in the 1970s, was to provide monitoring and simple control systems.
Inevitably, things have advanced since then and one definite trend is towards
the integrated automation of specific processes. Such automation is absolutely
necessary for fast and complex processes (e.g. electric power generation,
transmission and distribution systems or natural gas distribution systems) and
consists principally of two possible system behaviour steps:
1. Recognition of situation for automatic small-size system adaptations
and the small-size control action and automatic reporting on the action
to the operator.
2. Automatic on/off control action on recognition of specific process state,
such as alarm state or danger, and reporting to the operator on actual on/
off control actions performed.
The second step is performed on the basis of process control algorithms simpler
than the first action that requires fast, accurate and updated action, model
calculation and model verification algorithms because it usually represents the
fast and complex action on process behaviour. A relatively simpler case of the
first step can be executed for slow processes where the recognition of process
action is modelled and reported to the operator who then decides and/or un-
dertakes the appropriate action.
Four cases of process control actions considering the role of the operator
are given schematically in Figure 8.1.
257
C,
process
operator control process
• system
control
action
(a)
simple
protection
(b)
•
-----------
• control
action
(c)
II process
model
•
dialog t-~~~--c-o-n-tr-o~l--:;~------~------~~
, II
I
I action
I (d)
I
expert
system
,,
\
process
__ A
model .JI
_______
'-
Figure 8.1 Four cases of process control action: (a) control action by operator's
decision; (b) automatic, by control equipment; (c) control action by operator's
decision based on process modelling; (d) automatic, by control equipment with
feedback of information to the operator or automatic through expert system or
through the dialog between operator and expert system.
Process states 259
total number of
process states
controlling
solvable
process states process states
working nonproductive
states states
normal-to-alarm
50% 1-4~--------~E:---------:3~+range
25%
10%~--+-~~~~=--=~==~~~~---------L.1
t alarm to out
of range low
tout of
range low
dead zone of data processing,
for example 2% of rated
signal amount
Figure 8.3 An example of the data processing function for analog signals.
Process states 261
VCI
VC3
V2
- - - - - - - - - - -'---+----, PS
Figure 8.4 Safe and dangerous process states and corresponding actions to be
taken: FSI, inputfiow switch ('1' ~ FLOW); FS2, outputfiow switch ('1'~
FLOW); PS, pressure switch ('1' ~ HIGH PRESSURE); VI, input valve; V2 output
valve; VCI, input valve command ('i' = OPEN, '0' = CLOSE); VC2, output valve
command ('i' = OPEN, '0' = CLOSE).
Actions Process state Status signals Control and operator actions Supposed process state
number FSl FS2 PS
The basic reason for this complexity is that given N status process signals
of a basic process unit that independently describe independent process variables
and process states, then a total of ST process states is
ST = 2N (8.1)
All possible ST states should be analysed and adequate decisions for each of
them prepared. These decisions, executed by the operator, range from very
simple automatic actions (e.g. automatic action on a button after receiving
a signal on the video terminal) to sophisticated operator-system dialogs that
involve process modelling and emulation.
Even more process states than given with equation (8.1) appear when analog
and counter state signals are taken into account.
Distinguishing k groups of signals, each group with an equal number of
states, the total amount of states ST is
k
ST = I1 Nfi (8.2)
i=l
where Ni is the number of signals in the ith group each with Ii states.
Thus, a recognition of process states should be preferred since a very large
amount of process states is expected.
The interaction of an operator with the process can be caused by three main
reasons, as viewed from time requirements (Table 8.3).
1. Fast interaction (e.g. start/stop or on/off) operations on final control ele-
ments, named control operation; the time for these operations ranges from
a few seconds to a few hours.
2. Slow interaction named planning operation (e.g. load and demand pre-
diction, process simulation for optimization purposes, economic produc-
tion, distribution and process security); the time r<:mge for these operations
extends from minutes to a few days.
3. Very slow interactions named following up (e.g. operations caused by daily,
weekly or monthly change in production, demand, load, power and ma-
terial exchange); these operations range from a day to a few months.
The main tasks of various computer control systems at different hierarchical
levels of control are given in Table 8.4. The control levels below substation
or remote station are usually equipped with synoptic panels with dedicated
operator-system communication and used both for synoptics of the basic process
unit and the process controller.
The operator-system dialog is performed primarily to satisfy the function
Operator/process interaction - actions, commands and timing 265
National Production
grid Supervision of: Consumer prediction Reporting and accounting
consumers production schedules statistics
Inter- production balance planning following up of efficiency
connected reserves planning of reserves fault analysis
utilities process network coordination of
Operation and overhauls
control of plants
Utility Production
Supervision of: Consumer prediction Reporting unit accounting
consumers production schedules statistics
Regional production balance planning following up of efficiency
grid reserves planning of reserves fault analysis
process network coordination of
Operation and overhauls
control of utility
Group Production
of plants Supervision of: Short-term planning Reports on production
consumers according to accounting data
District production directives statistics
grid process state
plant components
Operation and
control of
substation
Plant Control of Work planning Sequential event
variables, recording
sequential start/
stop functions,
automatic system
restoration,
protective
functions,
supervision of
process variables,
auto/manual for
local equipment
Common functions
Selection of the computer
Keybord and CRT selection
Basic mimic diagram
Dialog and system selection
Acoustic alarm quitting
Copying
Interrupt
Process Operation
planning
Selection of mimic diagram
Deletion of time controlled from a selected system
reports Alarm quitting
Memorandum issuance
.l Process shutdown
RCS . PCS I
RCS" .. PCS
CRT Control Manual/computer switchover Control
Functions blockage Control of remote stations On-off control
Process data limits read-in Printer and print-out selection Set-points control
Control of stations status System time Status control
Control of CRT Data read-in Data read-in
expert
system
T4
process
1. Start command that requires the transition of the process from OFF to
ON state.
2. Stop command that requires the transition of the process from ON to OFF
state.
3. Activate command that requires the transition of the process from inac-
tive to active state.
4. Deactivate command that requires the transition of the process from
active to inactive state.
5. Unlock command that requires the transition of the process from un-
available to available state.
270 Process management and control
6. Lock command that requires the transition of the process from available
state to unavailable state.
7. Notice command that requires the process control system to recognize a
state transition and to take any necessary action.
The basic process states transitions and process commands are given in Figure
8.7.
Particular process states (e.g. an unavailable state) can be reached by issu-
ing commands and by changes in process equipment and software behaviour
due to errors and failures that were not previously predicated. In such cases,
after the system has been repaired, particular process states are put into the
Table 8.6 Some standards for operator-system dialog (optical signalization)
type of colour specific colours 6 + black and white only for characters purple, blue, blue green,
from 20' of visual green, red, yellow
angle; for smaller
angles only red, green,
blue and purple
saturation saturation not for coding
light light 2 different types alternative or redundant green-light green
Colour coding
contrast colour contrast 2 different types colour on coloured positive (colour on neutral
background not background)
recommended negative (neutral on colour
background)
neutral 2 different types positive (white or grey on
contrast black)
negative (black on white
or grey)
Operator/process interaction - actions, commands and timing 273
The working body (system) is the process body observed separately from all
the other mutually interrelated bodies (systems) and where they are all treated
as the system environment. The process is a macroscopic reflex of outer,
variable and microscopic changes in the working body. The process states can
Process protection and automatic actions 279
active command
process
process
active inactive
state
state
process process
on off
state state
(a)
lock command
process process
available unavailable
state state
unlock command
Figure 8.7 The basic process states transitions and process commands (a), (b) -
optional transitions.
Table 8.11 Specific features of process control equipment for command and
process signalizations
o 2930 42434647 63
o A I B Ic I D
1 E
47 GI H I I I K
I L 1M
01 2 13 42 43 6061 63
Figure 8.8 An example of the division of the CRT display for operator-system
communication: A, mimic identification number; B, type of dialog; C, reserved for
operator's reminder system; D, time of day; E, last unaccepted alarm; F,
individual mimic display; G, type of computer and display unit; H, key-in of
alphanumeric characters; I, procedure error; K, text of the selected process value;
L, computer message; M, function designation.
change in the process variable caused by the change in another part of the
process;
change in the process variable during repair, damage or unit shutdown.
The protective limit signal generated at the output of the signal limit sensor is
fed to the logic operation circuit where it is combined with other protective
limit signals to generate (1) appropriate signalization to the operator and (2)
signal of protective action condition. The signal of the protective action condition
is evaluated by redundant logic circuits (e.g. 'two out of three' circuits, or
similar) and an output protective signal is generated on this basis to the control
system. The control system generates the appropriate output protection signal
with the required power and time requirements to the final control element.
The protection systems are designed for various processes. The criteria for
their design include the following minimum requirements.
1. The existence of process conditions that require protection.
2. The existence of process variable to be monitored to provide protective
action.
3. Exact limit values and levels for each process variable that require pro-
tective action.
4. The margin between each protective limit and the level considered to
designate unsafe operation.
5. Protection performance requirements such as protection system response
time, system accuracy and ranges of process variable conditions (normal,
abnormal, accidental) both in magnitude or/and rate (Figure 8.11).
Process protection and automatic actions 283
exciting
level
I
I
I
I
I
I
I
I
limit signal I
-------------------------~
I
signalization
•
logic
level
II
II I
I
~~
control
level
I
final control element
---___ ___---__-_-__-_-..::~:, -_-__-_.,.-__- _-__ ____ __________ t __
I
accidental
range ac cidentallevel
I........
unsafe
operation
range safe operation level
- -- - --- ----protective----- - --- - - -- ~
r
limit range 1?!otective limit level
abnormal operation range
,
PU
•
Figure 8.12 An example of a static protection system for a small generator
unit: ~. three phase measurement transformer; ::t>-.OR circuit; CD. voltage
measurement sensor; -cD. current measurement sensor.
286 Process management and control
p
Pr -+----~------------_T------------~-----------------
REFERENCES
DIN Coding of Information (1979) CRT Display Working Place, DIN 66234.
Wilhelm, R. (1979) IEEE Trans. Autom. Control, AC-24, 1, 27.
9
The role of the operator in process
control systems
9.1 INTRODUCTION
There are two trends in the role of an operator in process control systems:
to use automated process control systems as fully as possible and to de-
crease the number of operators and the skill required of them;
to collect a broad number of fairly skilled, trained and schooled operators
from a 'field' operator to a general system operator and process planner,
and to support their work by the appropriate process control systems.
These trends, to possess a fully automated process control system and fairly
skilled process operators, are interrelated since only on the basis of efficient
automation can there be a full use of the operator's skill. This interrelation
materializes when microcomputers are used in process control in such a sense
that there appears a fully automated process which requires less labour.
Production processes are becoming increasingly specialized but there is a trend
to group them together. The variability of production conditions and technologies
with their fast development and competition for energy, raw materials and
human resources has become so powerful that it limits further expansion of
already-installed processes and forces new processes to appear and take over.
The knowledge of a particular type of production seems to depend on the
experience and skill of workers in the process, whose gift, speed of reaction
and creativity help, by using the same amount of raw material, energy or
efforts in the production, to increase the quality of products. Such workers
can be supported by computerized process control systems, and their effec-
tiveness is usually increased by collecting a set of relevant process data and
relations. An operator's creativity can be supported by adequate process control
devices that increase insight into process states and behaviour, and enable
fulfilment of process commands, as well as increased productivity made poss-
ible by fast calculations and personal experiment. The most promising tool for
operator support today is expert system software since it uses artificial
inferencing similar to human reasoning.
The operator's role in a process control system may be presented as four
groups (Table 9.1), and are as follows.
287
Table 9.1 The operator's role in the process control system - functional subsystems
Process + + +
manipulation
Process + + +
supervision
Process + + + +
execution
Process + + + + +
quality control
+ applicable
not applicable
Artificial inferencing tools 289
The operator's role in process control is often very formalized for the follow-
ing reasons:
preservation of the unified manipulation with process control and produc-
tiOl~system;
simplification of the design of process control and production system;
acceptance of specific human reactions and physical and mental features
and limitations;
simplification of the training of operators and their prequalification, if
necessary.
Table 9.2 details the formalization of the operator's role. Three main char-
acteristics seem to exist in operators observed as users of process control
systems:
1. Operators prefer analog to digital data and process presentations.
2. Operators prefer traditional process control procedures, geometry and
colours to new solutions on either one of them.
3. Operators are selective in their choice of tools and means of control,
choosing only the most efficient.
The first characteristic, regarding the design of control panels and displays,
enables an overview of a process state by bar graphs and process mimics.
Traditional solutions require a specific design of process controller panels,
regulator front panels and emergency and alarm control panels.
The formalization of the operator's work can be done effectively by using
formal operator-system dialog such as the CCITT man-machine language.
In such a way, all four reasons for formalization of the operator's work in
process control systems can be solved.
i=!
and for E being any event holds there (see Figure 9.1), then
292 The role of the operator in process control systems
(9.4)
The supposition that events E!l HI ( i = i, ... , n) are mutually exclusive means
that
n
(9.6)
i=1
The probability of event E can be calculated if we know the probability of HI
(usually named hypotheses) and conditional probabilities of events E in relation
to hypotheses HI (i = 1, ... , n). By using the relations (9.2) and (9.6) we can
obtain
(9.7)
which is the Bayes formula applicable to a range of practical tasks where the
probability of the conditional hypotheses Bi is sought, while the event E has
happened which usually appears with one of the hypotheses Bj (j = 1, ... , n).
Example
The case of a masoot warmer of a thermoelectric power plant is given in
Figure 9.2 where four different events are taken into account and put into
probabilistic relation with four hypotheses. This is a case for a binary expert
diagnostic system for which the corresponding data in the knowledge base are
given in Table 9.3 as well.
masoot
8 - filter
wanner ~
@- pump
water
180°C
masoot 7 bar
Figure 9.2 The functional scheme of a masoot warmer in a power plant.
p(H4:E) P(H4: E)
294 The role of the operator in process control systems
Table 9.4 The calculation of rule value (sideways chaining) for the case of the boiler
feed preheater from Figure 9.2
RV = I.!P(Hi:E) -
P(Hi:notE)1 (9.8)
i=i
Calculating rule values according to formula (9.8) from the given example in
Table 9.3 data can be obtained as given in Table 9.4. Thus the indication of
the most probable hypothesis is given. By scanning the facts for activity or by
asking about their state it can be found that FACTI = '0' so that new a priori
Bayes probabilities can be calculated, as given in Table 9.4, new rule values
obtained, etc.
Artificial inferencing tools 295
P(H:E)
P(H:R)act
P(H)
P(H:E) Relative
O-------------~r-------------~r_--------~O_~
-5 does not actual response
+5
respond process
response
change
in P(H) ' \ .
f
final P(H)
current
P(H)
M2
o
Figure 9.4 The five probabilities used for a given hypothesis.
The rule values for all items of evidence should be recalculated to allow for
the change in probabilities that have taken place given the last process response.
Then the minimum and maximum values which each hypothesis may yet
attain should be calculated (see Figure 9.4). Essentially there are five quantities
for each hypothesis; each of these quantities is a probability, namely:
P(H) is the current estimated probability of that hypothesis being true;
P max is the current maximum probability that this particular hypothesis
could attain if all the remaining facts went in its favour.
P min is the current minimum probability which a particular hypothesis
could attain if all the remaining facts worked against it.
Ml is the upper threshold criterion for accepting a particular
hypothesis calculated as a proportion of P max before any fact has
been acquired at all.
M2 is the lower threshold criterion for rejecting a particular
hypothesis calculated as a proportion either of P max or P min before
any fact has been acquired at all.
Artificial inferencing tools 297
N---
prior hypotheses
probabilities for each Hi
extract p(H.)
N-- HI
p(H) start
I
evidence-fact H"
P max ' Pmin
,...
-
1 Fj k
--- rt rule
p, V-- calculate RV values
and
~Vi.
pn
~cture
qt··· qt
check-ups
no
most likely
outcome found
W
report
I
The most likely outcome has been found if there is some hypothesis for which
P min is greater than Pmax for any hypothesis. The likely conclusions are those
hypotheses for which P min is greater than MI.
Uncertain conclusions are those hypotheses for which P min is less than Ml
and Pmax is greater than M2. These items are potential for resolving the
uncertainty.
No inference is possible (or the hypotheses are false) for those hypotheses
where Pmax is lower than M2.
The flow chart of the proposed procedure is given in Figure 9.5.
298 The role of the operator in process control systems
The ability of a control system to collect and process a large quantity of data
often obliges process control engineers, managers and even system engineers
and designers to implement some sort of programmed features that can be
used for the control of an operator's work. The usual requirements that can be
expected from the system to control the operator's work, without affecting
freedom, dignity or privacy, seem to be as follows.
engineering
1 Gbyte
tool (CAD)
expert
100 Mbyte systems
supervisory
10 Mbyle control and
data acqui ilion
systems
1 MbYle
process
controllers
andPLCs
100 kbyle
process
10 kbyte devices
1 kbyte
G\
C \
II) ~ \
~ lS ,
E'" ,
II)
"
envelope of
operator-machine
ystem re pon ibility
severity
of
damage
t
- - ----
~
benefits
•
probability of occurrence
~:
even ts evaluated
as unlikely
to occur
Figure 9.7 The plant operations system.
REFERENCES
Naylor, C. (1989) ABB Publication No. 0 CRH 1333 89 E, GRADIENT Graphical
Dialogue Environment.
Livingstone, W.L. (1982) IEEE Trans. Power Apparatus and Systems, PAS-101, 8,2543.
Part Four
System Design
10
The feasibility study
10.1 INTRODUCTION
A feasibility study is carried out on some small, many medium and almost all
large process control systems. The process owner, user or designer may use
this study to formulate, state or clear up the main points of a future computerized
process. The results may be described as follows:
the ways and means of process computerization to obtain certain tangible
benefits from computerization;
the organization and coordination of the process that has to be changed to
allow computerization with given benefits;
application of other technical means to raise productivity and lor the quality
of products;
changes in the organization of the firm where the process is performed to
increase its productivity and/or the quality of products;
changes in the level of operator's and manager's knowledge of process
computerization in order to secure foreseen benefits;
data and data sources used for the calculation of the proposed computer-
ization of the same or similar processes;
the most effective computer configuration and plans for its implementa-
tion.
A feasibility study begins by issuance of the design task that states (1) the
process to be computerized, (2) possible tasks of process computerization, (3)
time and space margins for process computerization, and (4) service or the
person responsible for the coordination of work on the feasibility study.
The design task is usually prepared by the process user, owner or designer
but the preparatory work may be carried out independently by a preliminary
study.
The core of the feasibility study is the cost/benefit analysis and all efforts
have to be pointed towards ensuring its correct and unbiased calculation. A
proposal document usually contains the following data.
305
306 The feasibility study
1. Introductory note that describes the organization and type of work on the
feasibility study and the assessment of main study goals.
2. Proposed plan of work, including overall phasing and content of each
main phase of work.
3. Areas of investigation describing
contacts with customer experts;
listing of all potential benefits studied;
listing of all data sources, data transmissions, data control facilities
and procedures;
main content of the computer database, its access, updating and cor-
recting;
future process development;
hardware and software potentially available for the system imple-
mentation;
system implementation schedule;
process control and supervision and data communication;
process optimization; modelling and expert knowledge;
customer organization;
process control power supplies and emergency power supplies;
process control safety and reliability and maintenance;
process control security;
software requirements;
process control timing requirements;
training of personnel;
other data in the process control implementation.
4. Content of the feasibility study report specifying
functional specification of the required control system;
proposed computer control system and communication system for data
and speech transmission;
comparison of the proposed system with other possible solutions;
estimated costs for a phased development of the control system;
tangible and intangible benefits to be gained from the proposed control
system;
implementation schedule;
hardware and software required for the system implementation with
their possible sources;
performance of the proposed system according to functional specifica-
tion, particularly regarding its availability, reliability, security and safety.
The efficiency of a feasibility study technique depends on the type of process
studied, its computerization state, and on staff resources of the process owner,
user and/or designer. Usually there are five stages in a feasibility study (see
Table 10.1) and a simple flow chart of necessary tasks is given in Figure 10.1.
Table JO.1 Five stages of a feasibility study
Data collection Initial design Preparation of System design revision System finalization
requirements and and specification
questionnaires
control system
and costs benefits
identification identification of
of requests existing and
and data acquisition future process problems
•
preparation of
functional
•
identification of
potential solutions
.-
specification and benefits
preparation of
preliminary
•
estimation of
main benefit values
design
l
preparation of
costs
•
quantification
of benefit amounts
1
calculation of
~ r--
•
research of
industrial support
costs
for real benefits
costs/benefits
analysis
•
estimation of
system feasibility,
preparation
of report
The estimation of data volumes and flows is the first step in a system design.
The most practical way to do this is to split up the control system into
stratification levels and connect it to a specific functional system that will
Data volumes and flows 309
level 1
level 2
process
plant
execute specific process control tasks at each level, as well as starting with
the estimation of data volumes processed at each functional unit. The tasks at
each stratification level may demand different functional units; it seems that
the process control functions for a specific process require the same or a
similar type of functional unit. Figure 10.2 shows an example of a two-level
process control stratification with two types of functional system, where the
continuous process controllers may be the same. When a particular section of
the process is assigned to the specific functional unit then, based on this fact,
there is a known number of process data and commands connected to this
functional unit.
Data and commands are determined according to the type of process used
and its performance. The usual distribution of process data and commands to
various functional system is outlined in Table 10.2. The definition of each
functional unit is also important, since the interrelations of the various process
units may require additional data transmission from other units. Dislocation of
these units requires a standardized bus structure in a distributed process control
system, or a standardized communication network, for the communication of
process data. Thus, a micro location of each unit has to be defined to prepare
for the required communication facilities.
Input and output process data volumes are defined for each functional
unit, usually as the decimal number representation of each signal or command.
Thus, a necessary consensus on data accuracy can be easily obtained with
process designers. Additional effort and mutual understanding among control
system designers and process designers must be made if a consensus on data
flow estimates is to be obtained. Some processes require a low rate of data
acquisition, and allow slow command executions. For example, one scanning
310 The feasibility study
Measurement
system x
Supervisory
system x x x
Chronological
event recorder x x
Remote
terminal unit x x x x x
Process
controller and x x
basic control
unit
Process
regulator x x x x
Process
governor x x x x x
Special
instrumentation x x x
system
Positioning
system x x x x
Protection
system x x x x
of process data takes 10 minutes or more, and 30 seconds or more for on/off
command execution for a natural-gas network. Data flow estimates are important
for functional units with a large number of data inputs and outputs and with
processes that generate and require control in small time intervals, such as the
control system of a DC-motor positioning unit in a hot rolling mill.
An example of estimates of data volumes and flows for a functional unit of
a crude-oil gathering station is given in Table 10.3, whereby two data flows
can be distinguished: the greater between a functional unit and controlled
process and the lesser between a functional unit and a superimposed control
unit.
Stratification of process control and basic system design 311
Table 10.3 Estimation of data volumes and flows of a crude-oil gathering station
start
freezing
process
recovery
sequence
1200-2400BD
~-------'-------------
1200-2400BD
Investment costs include all the relevant control system costs shown in Table
10.5 for a four-level computer process control network consisting of:
318 The feasibility study
Controllers 8 1 2 12
Remote terminal
units 9 6 2 4 21
Modems 0.5 0.5
District centre
computers 5 6 0.5 2.5 14
Dispatch centre
computers 2 4 0.5 3.5 10
Communication
system 31 6 5 42
Totals 55.5 17 10 17.5 100
Hardware costs
These include costs for every major hardware element of the whole control
system, except those for process instrumentation.
Software costs
These include all costs needed for the design, implementation and documen-
tation of software, excluding the costs of the on-site implementation, installation
and all costs that appear one year after the initial system installation.
Implementation costs
These include the effort and materials needed for the connection of each
computer unit to the system.
Benefits 319
Project leading
These costs include all the effort and materials needed for the planning, phasing
and supervision of all activities on the computer control system development
and implementation.
Installation costs
These include the cost of the effort and materials needed one year after the
initial system installation.
10.5 BENEFITS
Intangible benefits
Intangible benefits are expressed as (1) better working conditions and
organization, (2) greater insight into process condition, (3) knowledge of
320 The feasibility study
Total costs %
planning, process expansion and (4) similar benefit elements that can sometimes
be estimated as a half of the overall tangible benefit gained by computer
control.
Tangible benefits
Tangible benefits are estimated in money terms by means of an approximate
calculation of the expected goal of the computer control system application.
The following four methods of benefit estimation can be used:
1. An analytical method that includes the knowledge of analytical expres-
sions of the process to be computerized and analytical formulation of
goals obtained by computerization; by means of such analytical expres-
sions an exact estimation of benefits can be obtained.
2. Experience of the process experts that can indicate some of the best aspects
of possible benefits obtained by the installation and proper use of the
control system in a given process.
3. Literature searches for descriptions of actual control systems that indicate
the minimum expected benefit for the process under consideration.
4. The opinion of independent consultants involved in a specific production
process and plant similar to the process under consideration.
There are many tangible benefits, which may be categorized as:
process production increase;
decrease of operating expenses;
decrease of investment expenses.
A particular benefit may influence more benefit categories, especially intangible
benefits (e.g. better control and work methods and good working conditions
based on efficient decisions at all process operating levels) can be expressed
in money terms after the installation of the control system and can assume a
significant part of the total benefits. Nevertheless, the point of benefit analysis
is to calculate tangible benefits only.
Benefits 321
Table 10.7 Survey of main tangible benefits and the reason for their appearance
A survey of the main tangible benefits and the reason for their appearance
are given in Table 10.7. Since there are large variations in the level of applied
computerized control and in the type of process involved, the main benefits
given in Table 10.7 have to be specified in more detail for each process or
basic process unit and evaluated in money terms by any of the four methods
of benefit estimation.
An example of the quantification of benefits by better knowledge of system
components states is illustrated in Figure 10.5 for computer control of crude-
oil production. Table 10.8 gives the statistical data on the main causes of
failure for three methods of crude-oil production, as well as the mean time
between failures and the time estimated for repair of the failure. A supposed
percentage benefit in the production increase gained by better knowledge of
system component states is given with the following expression
APR 100
=- - [ PRa,b,e,
8760
L ARdAi + PReAR4/A4 + PR b L ARdAi
3
i=! i=5
7
(10.1)]
eruption
PRa
failure on
submersible
hydraulic
pumps
Table 10.8 Estimated values or upper and lower bounds of MTBF and MITR before
and after control system installation for crude-oil fields
The comparison of costs and benefits gives the main results of the analysis:
the feasibility of the planned computer installation;
the extent of the computer hardware and software to be installed and its
installation phasing in order to fulfil the expected benefits.
A partial computerization of a process that can exhibit higher benefits is
sometimes preferable over the total planned computerization. Yet, only a realistic
cost/benefit analysis is the basis for the implementation of any computer
control system.
REFERENCE
Kessler, R. (1989) Automatisierungstechnische Praxis, 31, 10, 461.
11
Computer control system design
11.1 INTRODUCTION
324
Communication design 325
Content of process Process design Process crossings design Process object documents Process design and
design documentation documents documentation object documents
Standardized process part Standardized process part Typical object documents Typical object documents
documents documents
Technical conditions for Technical conditions for Technical conditions for Technical conditions for
system erection system erection system erection system erection
Safety protection
documents:
sources of danger
danger estimation
safety design
countermeasures
Basic auxiliary Process isometry map Crossing isometric map Object isometry drawing Design task
documentation
Geographic map for Geographic map for Geographic map for Geographic map for
process approach process approach object approach system overview
Tracing schema(s) Tracing schema(s): Tracing schema(s): City planning conditions
with surrounding Crossing surrounding object surrounding needed
orientation longitudinal map map
List of contents
route profile Longitudinal route profile longitudinal profile
basic design data and Characteristic sections characteristic sections System technical
erection conditions needed needed description
Mechanical drawings
Process crossing schemas Design data on process Design data on objects List of applied technical
with other public utilities in crossing points such as names, numbers, norms and standards
revision numbers
Geological and
geomechanical reports on
soil type and carrying
capacity
Basic equipment Amount and estimate of Amount and estimate of Amount and estimate of List of pressurized tanks
documents equipment equipment equipment
Amount and estimate of Amount and estimate of Amount and estimate of List of equipment with
works as per unit of works as per unit of works as per unit of detailed specification of
work work work danger sources
Mounting and testing Mounting instructions for Mounting instructions for Mounting instructions for Authorization documents
instructions given technical given technical given technical
Report of work for local
conditions at system conditions at system conditions at system
authority
erection site erection site erection site
Nomination of workers
Testing instructions for Testing instructions for Testing instructions for
given system technical given system technical given system technical Document on correctness
conditions conditions conditions of work execution
Building work Building work Assembly log-book
instructions for given instructions for given
system technical system technical
conditions conditions
Technical conditions Typical drawings of all needed details (buildings, electrical, mechanical) Technical conditions for:
Instructions for erection and control of all phases and categories of work protection at work
Annual testing instructions protection of life and
health of people
environment protection
explosion proof
protection
protection of roads,
electrical and
electronic cables
328 Computer control system design
master
station
process
process
Figure 11.1 An example of a connection of two data transmission units to a
master station through a radio link: IT~Rxl, radio transmitter/receiver; [Qj, data
transmission unit; ~,multiplexing- emultiplexing equipment; Dc antenna
equipment; iI, j;, h, h, the frequencies of the radio link.
P = Po - Ko + Go - F + Gp - Kp - C - Z - B - Ar + Arg + V + Wo (11.1)
where
Po is the transmitter power at the receiver output or the peak power at the
output for pulse modulated systems;
Ko is the cable attenuation of the transmitting antenna;
Go is the transmitter antenna amplification factor because of its direction
and compared to the isotopic radiator;
F is attenuation due to propagation;
G p is the amplification factor of the receiver antenna compared to the
undirected antenna;
Kp is the thermal noise of the receiver antenna;
C is the thermal noise constant and equal to 174 dB m;
Z is the noise number of the receiver;
B is the bandwidth influence of the modulating signal;
Ar is the noise gain factor of the receiver;
Arg is the signal gain factor and automatic gain control of the receiver;
V is the gain factor determined by the modulation system;
W 0 is the psophometric constant of the approximate value of 2.5 dB that
is not applicable to data transmission.
An example of the power balance of the radio link for data transmission is
shown in Figure 11.2, which shows an approximate signal-to-noise ratio of 87
dB for the fading of 20 dB.
The design of radio links varies for different transmission frequencies. The
most used frequency bands for data transmissions are (1) super high frequency
band (SHF) from 3 to 30 GHz, (2) ultra high frequency band (UHF) from 300
to 3000 MHz and (3) very high frequency band (VHF) from 30 to 300 MHz.
A similar procedure may be applied to these frequency bands for radio link
calculation. The basis for calculation is given in equation (11.1). The basic
unknown variable is F, the attenuation factor of the radio link or the magnitude
of the radio wave at the receiver location. The basic procedure for the calculation
of the radio link attenuation factor is given in Table 11.3. An example of the
radio link calculation is illustrated in Figure 11.3 and further detailed in Table
11.4.
80 dB m
60
Go
40
r~
p
20 0
F
0
Arg
-20
~~~r-----r----
Gp :
I
-40 L.-_~ I
f L.._ ~
I
I
I signal/noise signal/noise
-60 I I ratio without
,L.. __ JI ratio with
fading fading
r;~,---.1.-1----L1_
-80
-100 V I
-12 .J Ar ~----------------------
Wo
~
-160~ B
-180P
Figure 11.2 An example of signal-to-noise determination for a specific radio link.
~ ~ I-- t::;?
~ I--
~
v--- V /"
f-"
-- I
~
~ V V
~ ~ 1j'V
---------- -------------V\----
V- ---- ---- - ~ ---- 1"""-
V ~ V'v
----- t-'"
....-- ~
--- ---- f' ./\
~~ ~
-- --- / \1
II
~ \
1\ V
/
~/
I "
Ml 0= 17.223km M2
Figure 11.3 Profile of radio connection MI-M2 from Table 11.4.
Table 11.4 Short-form calculation of radio connection between locations MI and M2
Quality of connection
Calculation of connection
Analog input data Periodic data acquisition With internal cycle < 1 min
Signal filtering Corresponding filtering
algorithm
Supervision of the signal Dead band of 0.5 + 1%
dead band
Reporting on demand To superimposed level
Status input data Change of state control With internal cycle < 30 s
Status signal processing After shorter time delay
Reporting on demand To superimposed level
Count rate data Change of state control For maximum input rates
Data processing Acceptance of all input signals
Reporting on demand To superimposed level
On/off commands Acceptance of command According to own algorithm or
from the superimposed level
Control of on/off output 0.2 to 10 seconds after
command acceptance
Control of on/off command Usually optional function
execution
Set-point value Acceptance of set-point According to own algorithm or
value from the superimposed control
level
Control of set-point value According to given address
Programming of time A time-programmed function
function normally in small steps
Table 11.8 Messages from a master station and data transmission unit
trans01ission
direction
analog data 8 analog values DTU master
station
..
lon/off command
-
a set point
2 3
Figure 11.4 Data transmission by messages, the example of message block types:
m start code; IADRI, address word; 51617 ,5,6,7 - 8 bit data word.
The respond message from the data transmission unit contains the fixed address
of the data transmission unit that serves as the identification label.
The communication of on/off commands and set-point values is performed
by interrupting the scanning cycles. An example of different data exchanges
is given in Figure 11.4.
Data transmission between data transmission units and a master station has
to be protected against noise and crosstalk on communication lines and links.
These are some recommended data on security measures. For example, code
security with single horizontal and vertical parity, and with single parity and
BCH code, bit length checking by supervision of the bit phase position and
the pulse and pause length, and finally start code restriction of the receiver.
Data transmission units 339
Description of Functions and tasks not completely Do not begin with design
functions and described until all present and future
tasks New functions added later functions and tasks are
known
Description of Description not based on any specific Specify hardware system
hardware hardware system
Description of Lack of precise catalogue data on Contact device and
process devices process devices and final controlling element producers and
and final elements obtain exact data
controlling
elements
Description of Vague definition of data processing Insist on unambiguous
data processing functions description of data
functions processing functions
Description of Standard data given only Insist on proof of each
mechanical given data
construction,
power supply
and working
conditions
Specification of Signal representatives, mean distances, Check each signal,
each input main process variables, standard work distance, process variable,
signal, distance, and equipment specified and specify actual work
process and equipment needed
variable, work
and equipment
needed
The design study of control and dispatching centres usually contains the fol-
lowing subjects.
1. Hardware description.
2. Software description.
3. Description of function perfonnance.
4. Hardware and software specification.
Hardware available on computer market and additional data on process re-
quirements are used for the improvement of hardware description obtained in
the basic system design phase.
Software description has resulted from the functional description of the
system given in the feasibility study by its extension and specification and by
adding the parts connected with the software already available on the market.
A control and dispatching centre represents the highest level of the produc-
tion control system where all strategic decisions concerning process control
are executed. Strategic decisions are taken by the production management and
represent the official production policy. The connection between strategic de-
cisions and their implementation in everyday production practice is maintained
by process dispatchers through their control and dispatching console usually
supplied with the following:
1. Process video display monitor(s).
2. Printer(s) and tenninal(s).
3. Hard copy unit.
4. Message exchange facility.
5. Functional and alphanumeric keyboards.
6. Telephone connection facilities.
The state of the process is shown on the print-outs and video display moni-
tors. Data on the process state are gathered in the centre by the communica-
tion equipment or the common system bus from remote tenninal units, or
distributed data processing units and controllers. The refreshment of the pro-
cess database is perfonned in time intervals between 5 seconds and 1 minute
for analog and status data and between 1 minute and 10 minutes for counterstate
data. The acquired process data are presented on monitors in the fonn of
mimic diagrams or data sets. Important process states presented on monitors
can be documented by copies issued by a hard copy unit. Process data in the
database are analysed and composite alarm states issued as warnings, alanns
or critical alanns for each process event that requires such data processing.
The established alarm is displayed to the dispatcher on monitors or a print-out
of a line printer or printer. Historic data fields are automatically created for
certain important process data during the last 12, 24 or 48 hours, as required.
342 Computer control system design
VI COLD
AIR
HOT
STEAM
(8 bar) It V2 FROM/TO
COWPER
OXYGEN
(12 bar)
TO BLAST HOT
FURNACE f - - - - - - - - . . : >....... AIR
Figure 11.5 Cold air, steam and oxygen control of a blast furnace - a block
scheme for panel presentation.
These historical data fields can be shown on monitors and logged at the op-
erator's demand. The dispatcher can issue set-point settings and on/off com-
mands to process final control elements on the basis of the given process
strategy, process data, historical logs and his/her own control and process
knowledge and experience. The issuance of set-point settings and on/off
commands is characterized by time and process variable values and both are
controlled by the dispatcher on process monitors. The automatic process soft-
ware mechanisms ensure the process integrity.
Design of control software is based on process algorithm definition (see
Figure 11.5 for a part of blast furnace control), detailed control scheme (see
Figure 11.6 for the same control part as given in Figure 11.5) and on imple-
mentation. An example of a PID control algorithm software module is given
in Figure 11.7.
The dispatching centre should also allow the following additional func-
tions:
1. Message switching between operators at various points in the system.
2. Centralized execution of larger engineering calculations.
3. Issuance of preliminary pro forma invoices on quantities of consumed or
sold product to customers.
4. Statistical elaboration of dispatchers' work.
5. Modifications and developments of software for remote terminal units,
district centres and control or dispatching centres.
Designing control and dispatching centres 343
OXYGEN
0,
close:
r---~---,
valve 2 ~---6---------""1
i
lb J
L _______ ( _
L _______ Y!l_____ J
ss
valve 2
__ -. alarm +
u
manual
j L~
.Q._--V
Figure 11.6 Cold air, steam and oxygen control of a blast furnace - details of
functioning.
The features and functions of the centre are based on real-time operating
systems that usually provide multiprogramming, priority scheduling,
multitasking, disk-based operation, checkpointing, power failure restart and
contingency exits (Table 11.10). These facilities improve the performance of
the system, increasing its peak load capacity while maintaining its ability to
meet absolute real-time deadlines.
The servicing of input/ output data processes is a very difficult aspect of
application programming. The operating system can support the efficient solution
of input/output programming efforts and requirements by supplying a com-
prehensive input/output device independence. Device independence is sup-
ported by the system software as outlined in Table 11.11. The goal of device
independence is to provide a package of services that allows the construction
of a task that can substitute devices routinely without necessitating a single
change in the task's code.
344 Computer control system design
CONNECTION PlAN
17EA YHO
19EA YHU
~
15EA YN
18EA YH
+
rrnfd r:
n
2EA TN/TA
+
XD e-- Norm. x
n
~
I ~
n
1 EA KP
3EA TV/TA Y 3M
~
4EA V
j-B-G]- -=--
j
L n
13EA XDIF f----
HI' - Norm. X
DY
28EB
ACHD
YNF
• t
16EA YR
Multiprogramming Queue building for system Concurrent usage Lack of dead time intervals as The multiprogramming of
resource demands of system resources compared to single tasks is accomplished by
programming systems dividing available memory
into a number of named
fixed partitions
Priority scheduling Scheduling of task according Faster execution of Tasks can be scheduled by Interrupts are created by
to their priority by means of task on higher system clock intervals - such process events thus
interrupts priority level systems lack fast response and enabling faster system
possesses a lot of overlay response
Multitasking Multiprogramming of two or Parallelism of Better condition of system data Multiprocessor systems
more tasks that need to execution of processing need intertask parallelism
communicate among actions such as in order to exploit the
themselves and synchronize data refreshment availability of more than
their activities and searching one CPU
Disk-based operation The extension of the Enables common Building of overlay tasks, Disk-based configuration
executive main memory and file system, check- extension of main memory needs a development of
data interchange medium by pointing and rapid abilities while still maintaining special software measures
means of external disk-based initiation of tasks system response time for conditions of disk
operation requirements memory failures such as
file copying and
maintaining
Checkpoint Preemption and rolling out The loading of The system possesses much Optional per task-based
of lower priority tasks to processor with as higher workload still operation
disk and their later roll-in much work as it maintaining real-time
and restoration at the can possibly absorb commitments
previous preempted point
Power failure restart The ability of system to The reduction of Service disruptions due to The operating system has
smooth out intermittent shutdown periods power failure are often to support the register
short-term power due to automatic lengthy, reducing system restoration by notification
fluctuations by tripping power failure equipment effectiveness and of power failure to some
volatile registers contents restart demanding auxiliary power users' programs by means
and their restoring after supply of contingency exit
power restoration mechanism
Contingency exits Subroutines automatically Improvement of System lacking contingency Contingency exit
entered after occurrence of structural design of exists are poorly responding to conditions are synchronous
an anticipated condition or a program and I/O terminations and to illegal such as an attempt to
anticipated asynchronous response efficiency instructions execute illegal instruction
condition of a task or asynchronous such as
an I/O termination
Table 11.11 Device independence means usually supported by system software
I/O language I/O service request for Besides the device I/O language communicates
logical interface with independence it makes with the file system and
devices (like OPEN, CALL, higher language record I/O package
GET, PUT, CLOSE) implementations on
existing operating system
possible
Common file A collection of system File system manages file The substitution of similar
system services needed for I/O storage, i.e. finding, device-types can be regarded
transaction between opening, closing of files, as less flexible than common
programs and named holding file name directory, file system
protected collection of and redirecting a named
records file
Record I/O Managing of buffering, Using the I/O language the The entire process is
package blocking and device control user task communicates its transparent to the user level
functions I/O requirements to record task
I/O package
Record I/O builds the I/O
requests to a device driver
to which the data are to be
sent
Device drivers Device drivers accepts Drivers accept requests New devices can be built Error recovery is included
requests and performs the from the file system, record and included into operating when needed
physical functions implied I/O and user tasks. system
in the requests issued to it
Logical unit Logical unit numbers and The task connects the Higher level languages and Systems based on over 16
numbers physical unit directory are logical unit number to a comprehensive I/O services kwords usually include
Physical unit tables that enable the inter- device driver and a simplify user task both higher languages and
directory group change of devices during physical unit number programming I/O service capabilities
the runtime
12
Cost-effective system selection
12.1 INTRODUCTION
As outlined in Chapter 1 the main effort, aside from that of the system designer
(61 %), comes from the system user (39%).
Three factors contribute to reducing the time required to manufacture and
assemble digital control system.
1. Application of system analysis and synthesis.
2. Application of a continuous spectrum of digital and computer hardware
solutions ranging from relay logic, as the simplest solution, to distributed
parallel processor architecture (Brajak, 1990) as the most advanced.
3. Application of feasible software tools and procedures, including system
design and sophisticated system simulation using workstations and personal
computers.
The application of system analysis and synthesis is discussed in Chapters 10
and 11, and the application of feasible software tools and procedures on
p. 358. The procurement and testing of hardware is outlined on p. 352.
Overall, the tendencies in production and marketing of hardware and software
are the separation of investment cost from software costs after system com-
missioning, and from annual expenditure figures; an increase in the number
of original manufacturer equipment solutions on the market; and an effort to
standardize system components.
Investment costs for hardware and software are given in equation (4.1)
in Chapter 4, where the total sum of hardware and software expenditure or
efforts is constant. Efforts can also be described in terms of manpower or
engineer months, or similar units. The initial expenditure Tn for n identical
systems consists of hardware and software expenditures, that is
nH + S = Tn (12.1)
where H is investment costs of one system and S the total investment costs of
software. The optimum marginal costs of hardware ~H and software ~S
expressed as
350
Introduction 351
n~H = ~S (12.2)
are fulfilled in practice only for simple repeatable on/off process controllers,
process control systems and similar control systems of a rather low complexity
and high repeatability of functions. More interesting results are obtained when
discussing the optimum of marginal costs for unrepeatable control systems,
that is
~H = ~S (12.3)
Equation (12.3) explains the natural tendency of the marginal cost of un-
repeatable process control systems to be the least expensive when hardware
and software marginal costs tend to be approximately equal.
Since there are other expenses (e.g. the cost of software one year after
system operation, SI' and annual expenditure costs, AI), equation (4.1) in Chapter
4 can be extended to
H + S + SI + Al =T (12.4)
The type of expenditure can be minimized bearing in mind that usually
Al = aH + b (12.5)
where 0.05 < a < 0.3 is a servicing costs factor of hardware and b is a constant
number that for known systems is usually between 0.1 T and 0.3 T. The
software costs one year after system operation, S I' are
(12.6)
where 0.05 < c < 0.2. Thus the minimum marginal costs of equation (12.4)
can be set for
This result shows that by using equations (12.5) and (12.6) the minimum cost
of one control system can be obtained when marginal costs for hardware and
software are equal and when a total variable part for hardware servicing costs
equals the variable part of the software cost one year after installation. Usu-
ally this one year for software is not enough for proper system functioning.
On the contrary, some software systems possess the ability of open-endedness.
These open-ended control systems (e.g. add-on functions, sensors and ac-
tuators) are different for costing since they require a complete separability of
physical and functional system realization. This separability can be obtained
by applying software system solutions such as flexible functional units,
multilevel system organization and system modification flexibility as referred
by the OSI concept. The minimum marginal cost appears when
where e is the depreciation period, a' is the servicing cost factor of hardware,
b' is the cost factor of operators referred to hardware and software, f is the
cost factor of hardware extension per year and d' is the cost factor of software
extension per year.
Digital control systems designed for general application are mostly manu-
factured as modular units consisting of modules, racks, panels, cubicles, or a
group of cubicles (Figure 12.1). The distribution of the equivalent number of
transistor functions and basic application design varieties are illustrated in
Figure 12.2. The degree of system integration D given is dependent on the
approximately estimated number of equivalent transistor functions T as the
greatest integer of the following expression
D = [log T - log Yo + 0.9] (12.9)
where Yo is the mean number of equivalent transistor functions of the applied
integrated circuits. Thus, for the system with 108 equivalent transistor func-
tions and the mean number of 104 equivalent transistor functions of the ap-
plied integrated circuits, an approximate fourth degree of system integration
can be expected. Thus, the matching of different process control functions to
the appropriate degree of system integration is primarily defined by the required
number of equivalent transistor functions. The complexity of integrated circuits
plays an important role, since the same number of equivalent transistor functions
can be obtained with a different degree of system integration using different
integrated circuits in the basic system modules, i.e. by the application of
small, medium, large and very large scale integrated circuits.
Nevertheless, all process control systems have to be provided with a power
supply unit, an input/output interface facility and a visual display presenting
the process state and controller state, units composed of elements with a different
degree of complexity of integrated circuits and other electronic, magnetic,
optoelectronic and optical components.
The systems are usually based on the general principle of internal hardware
compatibility which dictates that data exchange must be performed through
standardized bus terminations, lines and connections. The standards applied
for bus conceived systems are not strictly obeyed and accepted, yet the ma-
jority of systems recognize separated data, address and control wires or cables
for parallel conceived buses. Process control modules are uniformly designed
and usually purpose-made (e.g. status or analog input modules). There is also
a tendency to design universal modules with some microprocessor-controlled
functions. A universal module includes status and analog input and output
functions and a communication facility with a central data processing unit
through a system bus connection. The mechanical design of control systems
integrated circuit a basic unit with
a different number of
transistor equivalent functions
rack up to 30 modules
--
number of transistor functions
10 2 10 4 10 6 10 8 10 10
application systems
10 12
~ ____ ~.,02
: : 10 6
integrated circuit
I :
~----- : I
basic module
t::::_::~_O': controller on a board
computer on a board
t:::::~_____-~.Ir10
controller
rack
computer
i
r----------- 10
5 9
controller with I/O interface
mounting panel
: ~ computer with I/O interface
r----------- controller with I/O interface
I 106 10 10 and peripheral units
cubicle 1 ____________ ~ computer with I/O interface
: ~ and peripheral units
r------------ 7 lOll and external memory
L_______________ }O~ I computer system with I/O
group of cubicles
: ~ interface peripheral units
1
1--- - -- -------- ~ and external memories
is mostly based on standardized frames for modules, on racks for the plug-in
of the group of modules and on cubicles for the rack mounting. Process input/
output connections from/to modules are made on the front side of the mod-
ules, which are firmly tied to the basic frame in the rack. The connections to
the process are elastically designed to prevent bursting and wearing of wires
and cables under vibrations and shocks sometimes present at the process site.
The systems designed for the special conditions of electrical, mechanical and
chemical interference demand double and sometimes multiple shielding of
electronic components because of noise and crosstalk effects caused by elec-
tromagnetic incompatibility, or by aggregates of dust particles in the atmos-
phere. The module components are regularly tested for their correct working
functions before being mounted into modules or systems. Incorrect compo-
nents are correlated with the supplier's name and their serial production numbers
to avoid possible systematic errors in system functions. Specific control sys-
tems demand specific components with higher temperature and/or humidity
and shock/vibration margins to be built in and applied. There are still some
components of process control systems that have shown specific advantages
concerning their availability, power consumption features and resistance to
Buying and testing hardware 355
Preference
environmental damage. Table 12.1 outlines some of the commonly used, pre-
ferable and avoidable components for mounting and use in process control
systems. A similar survey of some input/output devices for process control
systems is given in Table 12.2. Some mechanical input components are not
reliable enough for use in process control systems and, wherever possible, are
interchanged and substituted by more reliable electronic, optoelectronic, optical
and magnetic components.
The hardware part of a process control system is optimized in both its
reliability and price by decreasing the number of integrated circuits and other
electronic parts, enabling simple hardware production and maintenance. The
application of large and very large scale integrated circuits decreases the overall
component number, and changes the type of efforts needed for the control
system design from hardware to software. The development and implementa-
tion of the classical software function in hardware elements allows an easier
programming of process control functions and a faster program execution.
The basic building blocks of control systems are modules. The complexity
and lay-out of modules vary from manufacturer to manufacturer. Some attempts
356 Cost-effective system selection
Function Preference
Table 12.3 The production of double-layer (DL) and surface mounted devices (SMD)
printed circuit modules
DL SMD
change of hardware functions and in smaller repair time intervals. The dis-
advantages of bus-oriented and modular systems and devices are their price
margins, where purpose-made hardwired and cabled systems are more con-
veniently priced for smaller devices and systems containing less electronic
and mechanical parts and demanding less labour for their production.
Overall tendencies in the production of hardware for process control systems
and devices seem to be:
increasing complexity of system modules;
increasing use of microprocessors and microcomputers;
replacement of software functions by firmware or hardware built-in
functions;
increase in the number of specialized systems and devices and their available
functions;
increasing standardization of modules, devices and communications.
Designing and testing software for process control systems is based on the
following system specification documentation.
1. Functional specification, which should give a precise and concise de-
scription of the system's outer functions.
2. Software requirement specification, which should give a precise and
concise description of the requirements of the software function of the
system.
3. Data specification, which should cover the description of each signal and
command that will be presented at any time to the system.
4. Software/hardware interface specification, which should cover the
description of each hardware part such as register, input/output unit
or similar point where data are identified by software functions of the
system.
Data processing specification describes the status, transformation and margins
of each data processing function in the system. This specification is based on
the graphic representation of software objects and relations, such as that given
in Figure 12.3 for the HOOD method (Heitz, 1989).
The global content and interrelation of the different types of documentation
are given in Table 12.4, where it can be seen that there is a considerable
volume of documentation needed to satisfy the correct design and implemen-
tation of software. A special region of activity represents the development of
software tools and systems for the design and testing of software (Figure
12.4). The activities for the software design and testing in Figure 12.4 are
given in Table 12.5. The overlapping of activities (symbolically denoted in
Figure 12.4 by combinations of letters) are as follows:
HOOD Buhr
I~ entry
1\ N=e J ~
Name
/
active object task
Name Name
Op ... 1
Op ... 2
I
passive object package
data data
~~~ EiJ~ EiJ
'use' relation with 'use' relation with
data flow data flow
except 1
I •
exception exception
L
generic unit generic unit
(not represented) alarm (not represented)
alarm
'\ Name
entry
interrupt interrupt
Figure 12.3 Comparison of graphic representations of HOOD method and Buhr
symbols for software tasks and modules.
Table 12.4 Designing and testing software for process control systems
Definitions Process definitions Software definitions Data definitions used Data processing Interface definitions
used used definitions used used
System verbal Introductory Global software Data processing Global data Global interface
description functional description specification processing description
specification specification
Sensors Function, number and Limits, ranges, Connection tables; Variable status; Registers; bit and byte
types alarms, print-out, addresses; names variable locations, tables of
specification, timing transformation address equivalences
Final control Function, number and Limits, ranges Connection tables; Variable status; Registers; bit and byte
elements types indications, timing addresses; names variable locations, tables of
transformation address equivalances
Computer peripherals Function, number and I/O actions, timings, Connection tables; Data status and Registers; bit and byte
types indicators, print-outs files formats; data transformation locations; tables of
sources and address equivalences
destinations
Computer input/output Function, number and I/O actions, timings, Connection tables; Data status and Registers; bit and byte
units types constraints data files, formats, transformation locations; tables of
data sources and address equivalences
destinations
Control activities
control state Overall definition of Precise definition of Definition of status, Control data I/O address for control
system control states system control analog and counter margins, inputs and outputs
condition states inputs data for control evaluations,
states comparisons and
tables
control response Response scheme Types of control Definition of algorithms Control response I/O address for system
response for control response data processing interface
time constraints Time margins of Time delays and Timer and delay data Timer data Timer interface
control response functions specification processing addresses
operator control Operator control Operator-system Operator data Operator data I/O addresses for
modifications specification control relations processing operator interfaces
Data acquisition Function, number and Content of Location of data, data Processing I/O modules addresses
status types signalization; alarm parameter specification algorithm
margins
analog Function, number and Content of Location of data Processing I/O modules addresses
types signalization algorithm
counter Function, number and Production content Location of data, data Processing I/O modules addresses
types parameter specification algorithms
System reports and
visualization
reports Report functions Time requirements, Report data Report processing
content, scope _ specification specification
VDU content VDU functions Time requirements, VDU data VDU data 'processing
content, scopes specification specification
Mimic contents Mimic functions Time requirements, Mimic data Mimic data processing -
contents, scope specification specification
Table 12.4 (continued)
System construction
reliability Overall system usage Global reliability Detailed reliability
specification estimate Security interface
security Overall Global Security data Security data specification
operator-system usage requirements specification processing
specification
human factors Overall specification of Verbal description of Operator's data and Operator-system Operator-system
operator-system operator activities operator-system communication data interface
interface communication processing specification
System development
constraints
time System development Approximate System development
planning development time timing
resources Overall financial System price Split-up of system
resources for system costs
installation
methodology Overall description of Definition of Detailed data on system System test data
system erection acceptance test testing processing spec.
System maintenance System maintenance Maintenance timing Maintenance crew and Maintenance software
support and requirements tools specification and procedures spec.
System evolution Expected system Anticipated system Hardware and software
evolution changes modules needed for
system expansion
system
documentation
elaboration
according to
Table 12.4
of system C
definition software
of
application
software
software technical
specification prepared
main design of
software
EF
software F
production
FG
production of
documentation
and system
delivery G
Figure 12.4 Division of activities on process control software design and testing.
Table 12.5 Software design and testing
Micro- Mini-
computer computer
process
events
emulation
data and time
prom
programming
iI~~:' ~~~;:
t
ware design is only applied to new software modules or module parts. The
total software production cycle is given in Figure 12.5.
The essence of process control systems is their interfacing towards pro-
cesses, which implies that a simpler programming can be obtained by the
orientation of system software and operating system towards I/O process data
processing. In such cases the concept of the automatic generation of the process
database can shorten the time for software development when the specifica-
tion of data is known and the decrease in software costs is possible. However,
user's real-time programs, extended real-time programs and parts of operator-
system communication programs can still present the largest part of the modified
or newly produced programs for each new application. Approximately one-
third to one hour of an engineer's time is still needed for each instruction of
368 Cost-effective system selection
This requires the specification of test procedures for the minimum number of
modules or units that form a routine. The integration of modules or units is
tested by simulating other software components already tested during the
program unit testing.
The test of the program must include all details of the test plan concerning the
order in which new modules are to be interfaced and included in the proven
system components. Requirements for testing the software subsystem as it
grows with the addition of new modules must also include detailed test stimuli,
along with the desired results of these stimuli on the part of the program
being tested. The test of the software system includes all the necessary data
on program stimuli and responses with off-line and on-line simulating process
data and man-machine communications data.
This must include all the necessary data on software system behaviour at the
time of system acceptance test when installed.
The order of software design and testing is given in Figure 12.6, in which
the work is as follows:
Designing and testing software 371
1
all technical specification
of software
development
test ,b~
/ development
test b~
program b3
level 1
integration
.. te-st-an-d--
pre operation
b3
software b 1
system level
!!.l!~~~!.i9!.l
test and
/ preoperation
b4
6
a12
system operation
and maintenance
Figure 12.6 The order of software design and testing (Boehm, 1976).
ar2 - the change of code and test of program unit level due to change
of function of program unit level;
a~l - the change of program unit level code due to changes in routine
level code;
a~2 - the change in program unit level function and test due to changes
in routine level code etc.;
b} - the work on program unit level coding and debugging;
b~ - the work on program unit level test development;
so that the total time needed for software design and test efforts on program
unit level T SEU is equal to
(12.10)
where
Ii is the inverse of the number of engineers or programmers engaged
in the program unit level coding and debugging and
l~ is the inverse of the number of engineers or programmers engaged
in the program unit development test.
The time needed for total software efforts T SE is equal to (Figure 12.6)
4
TSE = Ai . Lo + L (Aj+l + Bj ) L j - 1 + A(; Ls (12.11)
j=l
simple group
leadership region ~
internal boundary group
membership region space
external boundary
~ external environment
compound group
minor boundaries
complex group
minor boundaries
complicated
group common group boundary
common group region
Simple group
A simple group consists of group space, including a leadership and membership
region, an internal and external boundary, and an external environment.
Compound group
A compound group consists of the same basic group elements as for the
simple group above, and an additional internal and external boundary, which
means more priority sub-levels.
Human factors in system assembly 375
Complex group
A complex group consists of the same basic group elements as for the simple
group above, and additional minor boundaries splitting the leadership and
membership region into smaller subspaces, which means more subgroups.
Complicated group
A complicated group consists of the same group elements as the complex and
compound groups, in addition to at least one group region common to some
other group.
,
predictivity predictivity
•
,
mass
...
•
~
open
groups
t i
t
crowd
,
•
<1
ttttttttt •
parties
1 ~ enclaves
(~i~!~ it 1 ";;;1
•
~-,_/
t (: i) t
Y
1 1
't~~
,
I
I
I
completely organized
groups
Figure 12.8 The quantification factors for open groups and enclaves.
Human factors in system assembly 377
has a special position different from those held by other members, and knows
precisely the responsibilities and privileges of all the team members.
The distribution predictivity may also be expressed in percentages. In a
software team with a project manager and three equally responsible and
privileged programmers, the distribution predictivity Dp is
Dp = Ns/NM = 2/4 = 0.5 or 50% (12.12)
where Ns is the number of slots (project manager, programmers) and NM the
number of team members (four team members). The higher the distribution
predictivity, the better the group efficiency. Group efficiency is the potentiality
of working economically. Comparing the number of slots with the number of
members, a given group can be termed as 'overmanned', 'fully manned' and
'undermanned' .
Any working team is always under certain group dynamics, in which there are
three main forces:
1. Group cohesion as an organizing force, including internal organizing
proclivities.
2. External pressure and agitation as disorganizing forces.
3. Internal disorganizing proclivities.
The strength of these forces is difficult to measure and evaluate. The amount
of external pressure and agitation can be taken as a constant in an organized
society and may influence the group in a way that causes absences of group
members in a group work. The percentage of such absences measured over
long time intervals seems to be about 3% (Berne, 1975). Internal disorganizing
proclivities mostly manifest themselves as absences of group members in a
group work. The amount of absences due to internal, disorganizing proclivi-
ties measured over a long period of time seems to be about 9% (Berne, 1975).
Thus, the regular group with a proper group cohesion seems to possess the
ability to attract its members for 88% of the productive group work, as measured
by the time parameter.
A working team is organized for particular working activities and its mem-
bers pass through a series of processes from the start of the group's existence,
the most important of which is understanding the particular group states and
elements. The internal process in a group results from the tension between
two sets of mental images:
1. The image of a group member about what the group represents (e.g. a
working team for the production of software for the microcomputer on-
line positioning system).
2. The image of a group as one of the members sees it (e.g. working alone
378 Cost-effective system selection
on the positioning problem with very loose connections with other team
members).
Group organizers must try to achieve the reality of a group existence as close
as possible to the image of what it should be. Because of differences in the
images of group organizers and a group member, tension may exist. The basic
function of group organizers is to regulate group work by issuing the working
canons that can be the same for several working teams. The basic characteristic
of teams working on the design of computer control systems is a rather short
period of existence compared to other types of working teams. Thus, a process
control working team will not produce heroes, as other social groups tend to
do, since the group does not last long enough for this to develop. A process
control working team possesses a number of rules and processes that last a
much shorter time than with other working groups but the following are
particularly important:
1. The role of a programmer's personality is much more emphasized.
2. Listening to the other team members is more emphasized.
3. A programmer's training and schooling are much more important and last
longer (Weinberg, 1971).
4. Negligence is the programmer's worst enemy (Gramatke, 1989).
The existence of a group rests on group authority and its culture. Group au-
thority is defined by a leadership and group canon. There are three types of
leaders corresponding to the three aspects of the group structure.
1. Responsible leader who fills the role in the organizational structure and
officially corresponds to parental elements of group member psychology.
2. Effective leader who makes decisions and corresponds to adult elements
of group member psychology.
3. Psychological leader who fits the group's image of a leader and actually
corresponds to emotional elements of group member psychology.
Programming teams seem to have a very flexible leadership since in specialized
programming knowledge there is always a possibility of anyone taking over
the effective leadership.
Group culture is part of the group canon and an important factor in group
cohesion. Group culture can be divided into the following:
1. Technical culture used principally in productive work and including all
sorts of useful artifacts, techniques and intellectual operations.
2. Group etiquette that deals with standards of behaviour and ways of pre-
senting an acceptable persona and of reinforcing and guiding other team
members.
Human factors in system assembly 379
Figure 12.9 The symbolic scheme of common regions of emotional, rational and
traditional parts of group culture: P, parental, traditional, etiquette part of
individual and group behaviour; A, adult, technical culture part of individual and
group behaviour; C, child, emotional, group character part of group behaviour.
according to Weinberg (1971) and Couger and Zawacki (1979), they are listed
with almost equal importance as material compensation, work challenge,
working and social conditions, and competence of control and execution.
Teams members are in everyday, mutual interactions (named transactions)
that can be usefully and adequately classified according to complementarity,
direction, pureness and intensity. Complementary transactions can last for
an unlimited time; crossed transactions can stop this process and force the
formation of transactions at other psychological levels.
The structuring of a team member's time includes the following stages
(Berne, 1967).
1. Withdrawal, including fantasies and imaginations.
2. Rituals, ceremonials and ceremonies.
3. Activity, work procedures.
4. Pastimes, group jokes and talks.
References 381
5. Games.
6. Intimacy.
7. Script.
The corresponding classification of behaviour, and three possible ego-controlled
states, limit the number of possible individual behaviours. An individual may
seek social contacts and time-structuring in a team and will primarily struc-
ture the time to obtain the maximum pleasure from work in the group. After
participating in the group, an individual will correct his/her work in the group
according to natural flexibility and adaptability. The share of each team member
is thus programmed by mental group image, social habits, idiosyncratic mani-
pulating patterns, specific long-term goals or, more simply, by the group culture
and individual character structure. An individual team member usually takes
the initiative in group work when recognition is given to the image of lead-
ership. This leads to a series of efforts to match the group image and individual
script where each phase of adaptation is well defined.
There are some generally accepted rules and specific situations concerning
programming teams.
1. The essence of what makes a good program is the same as that which
makes good friendship, understanding and support.
2. When writing programs the programmers are actually tested for their
understanding of the programming language.
3. Good programmers are trained, not born.
4. Teaching of programming should be based on teaching understanding.
5. Computer languages tend to force programmers to think and behave
differently when executing a specific task.
An actual example of software team structuring is presented in Figure 12.10,
and staffing of the project is given in Figure 12.11. The most crucial moments
in project staffing are time instances t3 and t6, since then there are urgent
changes in team structure.
REFERENCES
hardware
system integration documentation production
,
system programming system integration putting into work system
~l ~ and testing work
database programming minor corrections
~
system
technician
definition
team leader II ~
II I
system
designer
I I
I
I
i
software
designers II
II :-------t
hardware design
and engineering
staff
system
technician L
I
II
Figure 12.11 The staffing of the project illustrated in Figure 12.10.
384 Cost-effective system selection
13.1 INTRODUCTION
385
all from One deliverer
tot.1 deliverer
standard
application
programs
Figure 13.1 Relative weighting factors for decision criteria of process control
system deliverer.
mechanica l, h ydraulics
38%
costs (man-months)
software 175%
./
(
hardware
,--''------..
115%
100%
planned
~ ~ ~
17%
-----
53%
5 25
actual
Figure 13.3 The usual uncertainty of planned and realized efforts and time for the
design of process control systems (Koch and Hoffman, 1978).
The mounting and installation procedure usually copes with two main problems:
1. Provision of the adequate installation team, installation equipment and
tools to secure adequate installation work and operations.
2. Design of adequate installation rules and additional equipment and
accessories to secure adequate behaviour of the installed equipment in
dangerous, polluted, noisy and electromagnetically incompatible process
control environments.
To solve these two problems, the following general procedures may be adopted.
388 The integrated approach
maintenance
'd~ "./
" "
""
, I
oe'it'>b ,," ¢O /
./ ....'/)0 I
".... ~'" I
~~J$' /
.~~ / ~
'" / 0
,
I
Figure 13.4 The relative costs of software errors at different stages of system
design and functioning (IEEE, 1981).
type of hardware
switches I----~~~~mnnnn'll"m~~~
meters
instrument
cables
signal Zm2
transmitter _
:
'....1 - - - - transmission line - - - - - i• • 1_
,
signal receiver
F (f}Jdb
unshielded cable
o /"II.
-40 ./"" V- /~
/'
/
shielded cable
-80 I
/' measurement
./ sensitivity limit
-120
."
"
-160
;;
""
10
I ~fiIter
I '
I
I
i.
1 ..... ------1-----
I
• I accumulator
eqUIpotential I
ground
lead :
I cable
zero lead
protective lead
connection at side
of leading body
functional grounding of process
ground devices
1'\,
10°
3 ~ ,\,
F (w) 10- 1 r'\ 1\ \r\ \ lA,
Y ~ ~
\"'r" \ \f\h,
,,
I'
,,
,
, ,
10- 1
-I"-
analog filter delay
T/2rrT I T/T
3 ~
F (w) 10-1
'\ \ )",
l
3 ~ /'r'I '
\ I
10 2
w
wT
Figure 13.10 The filtration effects of input and output signals: F( m), filter
amplitude characteristics; m, circular frequency; T, sampling time; T 1 , time
constant of analog and digital filters; .1T, AID integrator time constant; T, pulse
width of the instantaneous AID converters.
Testing of process control systems installed at the site usually begins with a
step-by-step procedure of commissioning specific parts of the system, starting
from the basic process unit level. The need for testing lies in the system's
faults and behaviour not consistent with the system requirement specification.
The main causes of system faults, and remedial procedures, are given in Table
13.1. The testing procedure and commissioning of the whole control system
require about 11 % of the total efforts in the system realization (Koch and
Hoffman, 1978). For this reason, and in view of further problems in system
maintenance, the study of system faults and the reliability of the system is
considered here in more detail.
Testing and reliability 397
Reliability r(t) is the ability of a functional unit to perform its outer function
under stated conditions for a certain period of time. The reliability is expressed
as a probability, that is
The failure rate z(t) is the ratio of total number of failures in a given time
interval to the number of correct functioning units, that is
z(t) = [F(t + L\t) - F(t)]/R(t) L\t (13.5)
The instant failure rate z(t)~t--*l = A(t) is
A(t) = f(t)/R(t) (13.6)
The mean life time mL(O, 00) equals the mean time between failures for
irrepairable units, that is
Remedies of faults
A B
F Interaction faults
Table 13.1 (continued)
Remedies of faults
c D E F
Measurement and
calculation of
external interference
to components,
modules system and
communications;
shielding, screening
and grounding of
I/O circuits,
sensors, control
procesing system
and peripherals
Change of shielding Component and
and grounding; modules redesign;
better choice of lay-out redesign
system lay-out
where Lltk is the particular unit down-time from the beginning of the output
to the time when the unit was returned to service, K is the total number of
outages for N units that are repairable, thus K ~ N;
• repair time, tR
K
tR = L LltRk (13.9)
k=l
where LltRk is the time spent on the actual maintenance of the kth unit; excluded
are waiting time for spare parts and for getting the maintenance personnel as
well as for system recovery time;
• per cent availability, Ap
Ap = l00(t T - to)/tT (13.10)
where tT is total time the N units were on test, that is
tT = N· tt (13.11)
where tt is the time of the system test;
• mean time between failures, MTBF
MTBF = (t T - to)/K (13.12)
and
• mean time to repair, MTTR
MTTR = to/K (13.13)
The scheme of time domains represented for system reliability under test
conditions is given in Figure 13.11.
System failures can be and usually are random events. There is inevitably
the tendency to estimate, calculate and predict their rate in order to organize
the production process and personnel needed for its correct working. For a
great number of systems and system functional units with a rather small rate
of mutually independent failures, a very good approximation for the unit failure
rate can be given by a nonstationary Poisson process where the nonstationary
failure rate A.'(t) is given as (Jovic, 1972)
A.'(t) = An + Bne - ent (13.14)
Reliability R(t) can thus be represented using equations (13.4) and (13.6) as
A.'(t) = R'(t)/R(t) (13.15)
or as
total time
~---------~--------~
(' "
operable
down time time
repair awaiting
time repair operating idle time and
time time system switched!
off time '
waltmg
time for
~. ':i0:~on
waiting,
time for
spare getting:
parts maintenance
staff ;
system maintenance I system available
Practical data on the reliability of process control systems and their parts
are given in Table 13,2 for MTBF and the mean time to failures, MTTF, for
fast repairable systems (Moore et al. 1978; Data General Corporation, 1976;
Musa, 1980),
For
A'(t) = const = An (13.18)
MTBF is
MTBF = l/An (13.19)
The reliability of function of serially connected units Rs is equal to
Rs = R\ . R 2 · R 3 · .. •· Rn (13.20)
The reliability of function R p of at least one unit out of n parallel connected
units is
(13.21)
402 The integrated approach
h = hardware
s = software
The failure rate calculation of a hardware module of the process control system
is given in Table 13.3.
The reliability of software should be expected to increase its MTBF with
time so eventually no single failure should be expected after some time. A
practical case shows that:
sometimes some program or data sensitive faults cannot be found by any
practical process of testing (SouCek, 1989);
some software faults, although well known, are not removed from the
installation and process operators seem to get used to them;
some new software faults are created by the process of system changes
due to the unfulfilment of system requirements or the implementations of
changes in system outer specifications.
Software failures in a real-time process control system application are
automatically detected and the system restarted in a much shorter time than
after hardware faults. This is shown as the shorter mean time to repair, MTTR
(Table 13.2). Software testing procedures must follow the usual software fault
sources (Goodenough and McGowan, 1980):
construction errors, i.e. faults of software components as implemented, to
satisfy expected specifications;
Table 13.3 Hardware module failure rate
There are many ways of commissioning a process control system. For example,
using a step-by-step procedure, or making the entire system run immediately
System commissioning 405
at lower or simulated production. This task is one of the last major and time-
consuming operations in control system design, resulting in the user obtaining
an insight into the functioning of the whole system. There are many rational
and irrational expectations and worries concerning control system functioning
and features (Koch and Hoffman, 1978). They are set out below.
In addition, the users who make their own application programs expect to
have easy learning and fast programming facilities, easy change and add-on
programs, and better transparency, portability and issuing of documentation.
The commissioning procedure may demand fulfilment of some of the user's
specified desires that may not yet have been included in the system design.
The advantage of the software-based control system lies in the possibility of
fulfilling these desires, even if the system has already been designed and
installed. The contractual side of these last modifications is a matter of mutual
consent between the system designer and the system user, taking into
consideration the facts previously put into control and system functional
specifications.
The commissioning and the final testing procedure are usually performed
together. Some parts of the process control system can be commissioned after
the final test, while others are still in the course of testing procedures.
Commissioning usually requires the operation of .the whole production process
with a decreased rated production and a cumulative increase of production in
agreement with the process operator and the control system supervisor. Some
additional measuring and data logging faCilities can be used for the purpose
of process safety and control. These facilities can be pen recorders and simple
visual displays, or can be made by a separate computer measuring and recording
system (see Figure 13.12). In such a case, the separation of analog and status
signals from process devices must be provided.
The preparatory work for system commissioning should be based on
drawings showing the manner in which the equipment is to be affixed, grounded,
accessed and connected to the electrical power supply, air-conditioned and
supplied with all other technical means. These drawings and required works
are done separately from system installation drawings and work descriptions,
and should be described and prepared by the control system contractor. The
406 The integrated approach
~--,---,...--~
recording system
production
process
system user is usually obliged to prepare all other officially required permissions
for system work and system connection to any public utility network or system.
The expenses incurred during system commissioning usually fall on the
customer.
If as a result of such tests the control system functioning and feature are
found to be defective, or not in accordance with the contract or contract
amendment or any written material issued and mutually agreed upon, the
system contractor should with all speed and at his/her own expense make
good the defects or ensure that the works comply with the contract. If required
by the user, the necessary tests should be repeated thereafter at the contractor's
expense. The installation provision of any instruments, apparatus, raw material,
408 The integrated approach
water, fuel, lubricants of all kinds reasonably required for the final adjustment
and taking-over tests should be provided by the user.
As soon as work on the control system is completed in accordance with the
contract, and has passed all the taking-over tests made on completion of the
system erection, the user takes over the system and the guarantee period
starts to run. The supplier issues a certificate, named a taking-over certificate,
and certifies the data on which the work has been completed and the system
has passed the test.
13.7 MAINTENANCE
Seminar topic System operator System Maintenance System dispatcher Seminar duration
engineer and engineer (days)
programmer
I
total/ /
I
1/ I
/
/
/ 1
I I
( 1
operable : down:
r::=L, 1 1
operating idle switched scheduled inoperable external loss time
off :maintenance
I
I
I
1
1
i.hsupplementary
routme
maintenance maintenance
1 ,
\
, ",
r-....L.-----.---.--.," "'" f---,..----,I
"\,
system system rerun miscellaneous awaiting repair system
productive test ' , repair recovery
1
1
1
I
1
1
1
I
I
I
I
I
I
~
maintenance
..1
1
.'.
I
available
1 unavailable I
• •
1
h/y =hour/year
Maintenance 413
probability (%)
100
90
80
70
60
50
40
30
20
10
minutes
10 1 3 10 2 3 103
The external loss time is the down-time caused by faults outside the
functional unit.
The rerun time is the part of the operating time that is used for reruns
necessitated by faults or mistakes in operating.
A mistake is a human action that produces an unintended result, which
occurs by the operator's intervention in an inappropriate way, or the operator's
nonintervention at an expected time. Statistical data for electric power
distribution (Datzkevic and Meltzer, 1978) show the distribution function Fit)
of time intervals of dispatcher interventions for frequency reglilation on busy
days as
1- e-o·76tl.88 for 0 < t :s;; 1 h
Fd(t) ={ (13.22)
1- e-o·76t °.82 for 1 h < t < 00
The probability density function of electrical network disturbances as a function
of time of an actual 110 kV network is given in Figure 13.14, compared to
414 The integrated approach
data given in equation (13.22). The proper behaviour of the process operator,
or the fully automated process control system, can produce the calculated data
indicated in Table 13.7. In other words, the responsibility of the system
functioning, and thus of the system maintenance, lies with system operator for
most computer control systems. Therefore, the training and education level of
system operators and maintenance staff is a most important factor in the system
operation.
The training of the maintenance personnel at the user's site is justified
when the process control equipment is widely used. A better and more reliable
service will normally result by employing user's personnel for this purpose.
New skills and upgrading of skills are required in the field portion of the
system. Maintenance, calibration and servicing of basic process control units,
communication units and process instrumentation is the most required action.
Involvement in trouble-shooting and in preventive maintenance techniques
requires a more efficient use of time and work scheduling and an efficient
References 415
t
MTBF data
MTTR data
required availability
--
data on rough calculation of the the working
system number and composition strategy of
f-+
,
failure of maintenance the production
critical state and repair crews process
mutual relations
concerning
responsibility
,
and competency
maintenance protocols
and maintenance books
,
for evidence of system
state and maintenance
-- --
designer
dat a designer's data
on spare parts working tools maintenance
and maintenance i--.. and materials instruction
equipment
,
the choice and
schooling of maintenance
staff and their working
tasks and responsibilities
REFERENCES
Baisch, R. and Hellwig, F. (1979) Ind. Prod. Eng., 3, 95.
Cmkovic, I. (1989) Automatika, 30, 3-4, 67.
(a)
:C
Type of Mounting scheme
equipment and border Comment
I
Protected t}m - measured
~
temperature m temperature
sensor
P ,
M :
,
Unprotected ~' t}m - measured
temperature m I temperature
sensor I
r:
P I
I
,
________~M~=~I~~C~___________
Pressure M ' C Pm - measured
transmitter pressure
P
Pm I IV instrument
,
valve
,
Flow Qp - measured
transmitter
Q~ flow
---~------
M~ C
r'
(b)
Type of Mounting scheme
equipment and border Comment
Pm pressure
Pm : measurement
IV" t}rn - temperature
Analytic ~m 'S"'~T~ measurement
instrument' ~ IV - instrument
P : valve
M ' c SO - sampling
device
cv - control valve
PM M---: C ___ M and actuator
Pneumatic control
valve ~
.
' ~
)-
I I
PM - pneumatic motor
EP - electro-pneumatic
_ _ _ _ _ _ _ _ _ _ _ _ .-!-:_E_P_ _CV
_ _--,--:-__c-:-o_n_t_r-:-o_l-,d_e;-vl_·c_e_
Electrically
C :x
EM - electric drive
A control, protection
and signalling
operated control CV -
valve
- M--rf
~ J. devices
E~;-- CV - control valve and
,E actuator
E--·;":·_-C ET - electric terminal
Connection part
~
terminals ET CT I CT - control terminal
part
Figure 13.16 Maintenance, the division between control and process equipment.
For the maintenance responsibility part in (a) and (b): P, process; M, mechanics;
C, control; E, electric
(a)
Type of equipment Mounting scheme
and connection and border Comment
Functionally defined E CB CB - control
border on board
control (OD~
!®®i
C ECS - electric
panel I® ®!
control and
ECS Ll:;:-=:0::;_.::..)_ _ _ _ _----1 signalling
.. part
,---------- - - - - - - -
J:J
-------------,---
Total maintenance
CB - control board
responsibility
1
JCT CT - connection
terminals for
- CB
control board
E C
Electric DC e - fuse
power supply E -I- C
R, 0, T - DC power
e I supply terminals
R~'-
T
o---t---~-----
(b)
Type of equipment Mounting scheme
and connection and border Comment
Electrical AC
power supply E .j, C
(monophase)
-:<2> ~:t:: I
j
e - fuse
Electrical e - fuse
three-phase EN - belongs to
power supply control
maintenance
for smaller
installations
FC I Po device
j c Po, ~o, c - nominal values
j of auxiliary
1 ~o medium
1
Data General Corporation (1976) Reliability Report on the Data General Corporation
Nova Line of Digital Computers.
Datzkevic, Yu. G. and Meltzer, M.M. (1978) Reports to the Academy of Science of
Uzbekh SSR, 15.
Eitz, A.W. and Heining, U. (1989) Automatisierungstechnische Praxis, 31, 416.
Gai1braith, J.K. (1977) The Age of Uncertainty Houghton Mifflin, Boston.
Goodenough, J.B. and McGowan, C.L. (1980) Proc. IEEE, 68, No.9, 1093.
IEEE (1981) Trans. Software Eng., SE-7.
Jovic, F. (1972) Registration of Nonstationary Poisson Process Data, PhD thesis,
Faculty of E1ectrotechnics, University of Zagreb.
Koch, G.R. and Hoffman, R.H. (1978) Ang. Inf, 6, 248.
Moore, C. et al. (1978) IEEE Trans. Power Appar. Syst. PAS·97, 4, 1115.
Musa, J.D. (1980) Proc. IEEE, 68, No.9, 1131.
Puend1er, K. (1977) ElektrizitiitswirtschaJt, 76, Part 17,579.
Soucek, B. (1989), Personal communication.
VDI/VDE 3551 (1976) Recommendation for Security to Disturbances of Signal
Transmission by the Installation of Process Computers.
Author index
419
420 Author index
Naylor, C. 294
Tanenbaum, A.S. 180
Novakovic, B. 54
Tunon, J. 185
Noyce, R. 97, 98
Control centres 337, 338, 341-5 Data input process hardware 62-70
Control economics 180 see also input devices
Control equipment 11 Data logging 241, 405
Control operations 264, 265 protocols 241, 244
Control system designers 309, 383 Data maintenance 210
Control tasks 78 Data management functions 159, 163
Control terminals 113 Data output process hardware 70-3,
Controlled systems 11 74
Controllers 28, 55, 57 actuators 71-2
Conversion routines 176, 178-9 commands 73, 74-5
Cost/benefit analysis 305, 317-19 electronic control devices 72
benefits 319-23 final control devices 70-3
investment costs 317-19 Data presentation 238, 240, 242-3,
operating costs 319 244
Count pulses 61, 171 Data processing 136-47, 157
Crude-oil gathering station 310, 311 algorithms 17, 51, 59, 136
Crude-oil measurement station 79-81 basic process unit 49-62
Crude-oillnatural-gas separator 42, 44 computing power 143, 147
CSMNCD technique 125 flow chart 192, 193, 194, 195
Cycle stealing 168 functions 138-41, 335
Cycle time 143, 146 hardware 137
safety of equipment 70
schedule 198
Data, definitions 6 software 147, 149-51
Data acquisition 53, 170, 202, 218, speed 53-4
309 variables 59
speed 329, 339 Data reduction 233, 235-6
Data communication 101-21 Data retrieval 210
asynchronous 112 Data transmission 325, 328-30
channel capacity 101, 105, 110-11 code security 338
circuits 10 1 units 330-40
codes 101 Data volumes 308-11
duplex model 113 Data-oriented work 5
facilities 121 Databases 162, 202-14
half-duplex mode 113, 337 access to 209
hardware 111-21 basic process software, and 202-14
modes 112-13 data plausibility 214
open-ended 124-5, 126-8 distributed, see distributed databases
security 113 facilities 209-11, 211
simplex mode 113 failure types 215, 217
smart process devices 125, 130 knowledge database of binary expert
synchronous 111 systems 292, 293
see also communication networks: organization 211-13
communications Datawords 111
Data concentration points 218 Debugging routines 177, 178
Data descriptors 212 Demultiplexer frequency 143
Data flows 308-11 Derived data 211
Data independence 210 Device independence 343, 348-9
424 Subject index