Identification and Optimisation of Hopper-Discharge Chute Systems

University of Wollongong
Research Online
University of Wollongong Thesis Collection University of Wollongong Thesis Collections
1977
Identification and optimisation of hopper-

discharge chute systems for bulk granular materials
G. John Montagner
University of Wollongong
Recommended Citation
Montagner, G. John, Identification and optimisation of hopper-discharge chute systems for bulk granular materials, Doctor of
Philosophy thesis, Department of Mechanical Engineering, University of Wollongong, 1977. http://ro.uow.edu.au/theses/1583
Research Online is the open access institutional repository for the University of Wollongong. For further information contact the UOW Library:
research-pubs@uow.edu.au
IDENTIFICATION AND OPTIMISATION OF HOPPER-DISCHARGE CHUTE
SYSTEMS FOR BULK GRANULAR MATERIALS
by
G.J . MONTAGNER, B.E.(N.S.W.),

M..I.E. Aust., A.A.C.S.,
F.A.I.E.A.
Thesis submitted for the Degree
of Doctor of Philosophy
Department of Mechanical Engineering,
The University of Wollongong.
May, 1977
808157
TABLE OF CONTENTS
PAGE
ABSTRACT 6
ACKNOWLEDGEMENTS 8
LIST OF FIGURES 9
LIST OF PHOTOGRAPHS 14
LIST OF TABLES 15
LIST OF PRINCIPAL SYMBOLS 16
CHAPTER 1: INTRODUCTION 18
1.1 Genera I 18
1.2 Outline of the Problem 19
1.3 Review of Previous Research 20
1.3.1 Granular Materials 20
1.3.2 System Identification 21
1.4 Scope of this Research 22
PART 1: UNIFORM FLOW 24
CHAPTER 2: GRANULAR MATERIAL FLOW DYNAMICS 25
2.1 General 25
2.2 Description of the Apparatus 26
2.3 Model 33
2.3.1 Introduction 33
2.3.2 Lumped Parameter 36
2.3.3 Statistical Evidence 40
2.4 Uniform Flow 43
2.4.2 Apparatus 43
2.4.3 Procedure 45
2.4.4 Equation of Motion 47

2
' PAGE
CHAPTER 3: OPTIMISATION 52
3.1 General 52
3.2 Model Formulation 53
3.2.1 Minimum Descent Time 53
3.2.2 Maximum Exit Velocity 55
3.2.3 Additional Constraints 56
3.3 Methods of Solution 59
3.3.2 Polynomial Approximation 59
3.4 FIetcher-PoweI I Algorithm 61
3.4.1 Description 61
3.4.2 Programming Considerations 63
CHAPTER 4: RESULTS AND DISCUSSION - UNIFORM FLOW 64
4.1 Comparison of Model Results with Theoretical 64
Calculations
4.2 Check on the Value of Viscous Drag 66
4.3 Comparison of Model with Chutes ofOther Profiles 70
4.4 Sensitivity Analysis of Model Parameters 72
4.5 Theoretical Scale up Analysis 80
4.6 Design Procedure 94
4.7 Conclusions 95
PART 2: TRANSIENT FLOW 97
CHAPTER 5: TRANSIENT FLOW 98
5.1 System Identification 98
5.2 P.R.B.S./Cross-correI at ion Method 100
5.2.1 Theoretical Considerations 100
5.2.2 Hardware Implementation 105

3
PAGE
5.2.3 Parameter Selection 110
5.3 Programme Development 111
5.3.1 Genera I 111
5.3.2 On-Line NOVA Package 112
CHAPTER 6 : CORRELATION TESTS 117
6.1 Genera I 117
6.2 Apparatus 117
6.3 Procedure 120
6.4 Step Responses ' 121
6.5 High Speed Cine Photography 123
6.6 ’Two-Dimensional’ Transparent Hopper 128
CHAPTER 7: RESULTS AND DISCUSSION - TRANSIENT FLOW 131
7.1 Transient Behaviour of Straight InclinedChutes 131
(Assumptions, Limitations)
7.1.1 Impulse Function for Straight Inclined Chutes 132
7.1.2 Bode Diagram for StraightInclined Chutes 136
7.2 Transient Behaviour of Curved Chutes ofKnown Form 140
7.2.1 Impulse Function Comparison of Straight Chute
with Curved Chutes ofKnown Form 141
7.2.2 Bode Diagrams for Curved Chutes 143
7.2.3 Comparison of the Optimum Chute at Varying
Flow Rates 145
7.2.4 Bode Diagrams for the Optimum Chute with
Varying Flow Rate 147
7.3 Step Responses 147
7.3.1 Step Response for theStraight Chute 149

4
PAGE
7.3.2 Step Responses for Chutes of Known Form 157
7.3.3 Step Response for the Optimum Chute 157
7.4 Predicting Flow Transients 150
7.5 Flow Non-linearities 155
7.5.1 Bin Outlet Flow Resulting from Valve
Perturbations 155
7.5.2 ’Two-Dimensional’ Flopper Results 169
CHAPTER 8 : SUMMARY OF CONCLUSIONS 177
CHAPTER 9: REFERENCES AND BIBLIOGRAPHY 182
CHAPTER 10: APPENDICES 191
10.1 Flowmeter for Granular Materials 191
10.1.2 Flowmeters/Velocity Detectors Investigated 198
10.1.3 Straing Gauge Cantilever Flowmeter 204
10.1.4 Ultrasonic Velocity Transducer 205
10.2 Flowmeter Filter 208
10.3 Camera Timing Lights 212
10.4 Camera Techniques 213
10.5 Computer Programmes 214
10.5.1 System Identification Package 214
10.5.1.1 Algorithms 214
10.5.1.2 User Experience 218
10.5.1.3 Programme Flowcharts and Listing 219
10.5.2 Optimisation Programme 280
10.5.2.1 Algorithm 280
10.5.2.2 User Experience 281
10.5.2.3 Programme Listing 281
10.5.3 Step Responses 312

5
PAGE
10.6 of Viscous Drag

Check on Value ' 312
dv
10.6.1 Viscous Drag Based on a 312
" dt
_ dv 316
10.6.2 V iscous Drag Based on a
" Vds
10.7 Pub Ii cat ions 319

6
ABSTRACT
In an effort to decrease the price-performance ratio in the
agricultural industry for material handling, and recognising that the
hopper-discharge chute forms a part of most such handling systems,
optimum performance of this component becomes a necessity. To achieve
this goal it is necessary to first obtain a detailed knowledge of its
dynamic characteristics.
Operation under uniform flow conditions is analysed and a model
incorporating a generalised drag force is used to formulate the
design criteria to obtain chute profiles suitable for operation in the
’fast’ flow mode and able to achieve a prescribed optimum performance.
The design procedure for minimum transit time is presented in the form
of a computer programme incorporating a sensitivity analysis on all
significant design parameters. The resulting optimum chute's
performance was contrasted with the performance of other chute shapes
more commonly used.
The transient flow characteristics of a model hopper-discharge
chute system under gravity flow are examined. Using the P.R.B.S. and
cross-correI at ion method of identification, the impulse flow responses
for the system under varying conditions of initial flow and chute
geometries are obtained. With the aid of spectral analysis the
magnitude and phase character istics of the identified hopper-chute
model were determined, enabling both the system bandwidth and the
behaviour within that bandwidth to be determined.

7
This information permits the examination of non-linearities in
the flow as well as enabling the determination of the dynamic flow
response characteristics during prescribed controlled operation of
the flow control valve.
This transient analysis was supported by the use of high speed
cine photography to determine the flow characteristics associated with
the operation of the flow control valve. Technigues for conducting
this high speed film analysis are presented.
It is shown that of the chutes tested, the optimum chute has
the most favourable overall performance characteristics both under
uniform and transient flow conditions.

8
ACKNOWLEDGEMENTS
The investigation described herein was carried out in the
Department of Mechanical Engineering of the University of Wollongong.
The invaluable assistance and guidance given by
Professor A.W. Roberts, Dean of Engineering, the University of
Newcastle, and by Associate Professor W.H. Charlton, Department of
Electrical Engineering, University of Wollongong, in all stages of the
work and in the preparation of this report is gratefully acknowledged.
For his many suggestions with the electrical apparatuses
and with the assembler language programme, I wish to thank
Dr. G.W. Trott of the Department of Electrical Engineering.
The Workshop staff’s efforts in constructing the various items
of equipment for the experimental work was appreciated.
Finally, a very special thanks to Rosalie, my wife, for
proofreading, and Maree for her patience in typing this report.
John Montaqner.
9
LIST OF FIGURES
Figure No. Section Title
1 2.2 The testing apparatus.
2 2.2 Hopper orifice arrangement.
3 2.3.1 Pressure distribution on chute walls (Roberts).
4 2.3.1 Pressure distribution on chute walls (Parlour).
5 2.3.2 FIow probIem mode I.
6 2.3.3 Transit time from the step response.
7 2.4.2 Chute profiles for model chute.
8 2.4.4 Acceleration Model - Force balance.
9 4.2 Force balance.
10 4.3 Transport delay from the cross-correI ation response
11 4.4 Variation in limiting angle 0f.
12 4.4 Variation in initial velocity.
13 4.4 Variation in Coulomb coefficient (t ).
14 4.4 Variation in Height to Breadth ratio.
15 4.4 Variation in viscous drag (y).
16 4.5 Scale up of chute geometries.
17 4.5 Chute profiles for (3 x 4) metre chute.
18 4.5 Chute profiles for (5 x 5) metre chute.
19 4.5 Variation in I¡minting angle 0f for (3 x 4) metre
chute.
20 4.5 Variation in initial velocity for (3 x 4) metre
chute.
21 4.5 Variation in coulomb coefficient (y) for (3 x 4)
metre chute.
10
22 4.5 Variation in height to breadth ratio for (3 x 4)
metre chute.
23 4.5 Variation in viscous drag (pi) for (3 x 4) metre
chute.
24 4.5 Variation in limiting angle 0f for (5 x 5) metre
chute.
25 4.5 Variation in initial velocity for (5 x 5) metre
chute.
26 4.5 Variation in coulomb coefficient (x) for (5 x 5)
metre chute.
27 4.5 Variation height to breadth ratio for (5 x 5)
metre chute.
28 4.5 Variation in viscous drag (pi) for (5 x 5) metre
chute.
29 5.2.1 System identification model.
30 5.2.1 Test signal characteristics.
31 5.2.2 Test rig for NOVA identification.
32 5.3.2 Generation of operating runstream.
33 5.3.2 Operating system subroutine hierarchy.
34 6.5 Camera performance details.
35 6.5 Single frame film analyser setup.
36 6.5 Perturbing gate end-section modification.
37 6.5 Grain flow shadow profiles showing effect of
valve perturbations.
38 6.5 Stream thickness transient response characteristi
at bin outlet resulting from valve perturbât ions.

39 7.1.1 Impulse function curves for chute inclinati
6 = 35°.
6 = 40°.
6 = 45°.
42 7.1.1 Straight inclined chute (Time Domain).
43 7.1.2 Straight inclined chute (Frequency Domain).
44 7.2.1 Comparison of impulse response.
45 7.2.2 Frequency domain comparison of chutes.
46 7.2.3 Optimum chute (x) = 0.0.
47 7.2.4 Frequency domain results for optimum chute.
48 7.3.1 Step response straight chute 0 = 35°.
51 7.3.1 Step response straight chute 12 mm flow.
54 7.3.2 Step response chutes of known form.
55 7.3.3 Step response optimum chute.
56 7.4 Comparison of computed and measured step
responses for straight inclined chute.
57 7.4 Predicted flow transients.
58 7.5.1 Grain paths and flow profiles at bin outlet
resulting from valve perturbations.
59 7.5.1 Grain stream velocity profiles at bin out le-

12
60 7.5.2 Slip pianes for M2-D" model hopper.
61 7.5.2 Veloci ty prof îIes 100mm above orifice.
62 7.5.2 Veloci ty prof iI es 50mm above orifice.
63 7.5.2 Veloci ty prof i Ies at chute orifice.
A 1 .1 10.1.1 Dean’s flowmeter for granular materials.
A 1 .2 10.1.1 Granular material flowmeter of Nugent, Iinoya
and Watanabe.
A 1 .3 10.1 .1 Nolte’s flowmeter for granular materials.
A1 .4 10.1.1 Henderson’s flowmeter for granular materials.
A1 .5 10.1.2 Piezoelectric impulsive flowmeter system.
A 1 .6 10.1.2 Linear displacement impulsive flowmeter system.
A1 .7 10.1 .2 Strain gauge impulsive flowmeter system.
A 1 .8 10.1.2 ’Linear’ strobe velocity detector.
A1 .9 10.1 .2 Photodiode height of grain stream sensor.
A 1.10 10.1 .2 Wire paddle wheel photo-encoder.
A1 .11 10.1 .3 Calibration curves for the strain gauge flowmeter
A 1.12 10.2 FIowmeter f iI ter.
A 1 .13 10.3 Camera timing light generator.

13
A5.1 10.5 SI GAN(1) - Generation of Operating Runstream.
A5.2 10.5 SIGAN(2) - Operating System Subroutine Hierarchy
A5.3 10.5 SERV - Device Drivers.
A5.4 10.5 A2D1 - Analogue to Digital Converter Driver.
A5.5 10.5 PLOT - Autoranging/AutoscaIing Plotter.
A5.6 10.5 ATODC - Analogue to Digital Converter.
A5.7 10.5 DATA IN - General Data Entry Routine.
A5.8 10.5 LI ST(1) - Normalise Data.
A5.9 10.5 LI ST(2) - Data Listing and Punching.
A5.10 10.5 C0RREL - General Correlating Routine.
A5.11 10.5 IMPUL - System Impulse Function.
A5.12 10.5 C0NV0L - Convolution.
A5.13 10.5 INVERCON - Inverse Convolution.
A5.14 10.5 SPECT - Power Spectral Density.
A5.15 10.5 XSPECT - Cross Spectral Density.
A5.16 10.5 FRESP - Frequency Response.

14
LIST OF PHOTOGRAPHS
ate No. Sect ion Title
1 2.2 Main testing rig.
2 2.2 Underside view of hopper.
3 2.2 FIowmeter.
4 2.3.3 ’Single' grain transit time.
5 2.4.2 Comparison of chute profiles.
6 3.2.3 Chute stalling (unconstrained).
7 5.2.2 Identification hardware.
8 6.6 ’Two-dimensional’ hopper.
9 6.5 Grain flow profiles showing effect of valve
perturbat ions.
A1 10.1.4 Ultrasonic Velocity Detector Prototype.
A2 10.1.4 Portable Digital Readout Ultrasonic Velocity
Transducer.
15
LIST OF TABLES
TABLE 1 Model results with theoretical parameters.
TABLE 2 Variation in Viscous Drag - y.
TABLE 3 Comparison of chute profiles.
TABLE 4 Range of parameter variations in model.
TABLE 5 Spread in parameter variation with scale up.

16
LIST OF PRINCIPAL SYMBOLS
The various symbols introduced in the text are listed here in
alphabetical order.
a coefficient
a acceleration
B chute width
f frequency
Fq drag force on grain stream
g acceleration due to gravity
G power spectral density
h impulse response or system weighting function
H average depth of grain stream
H(f) system function
H0 depth of grain stream at entry to chute
HB depth of grain stream to chute width ratio
j /-T
k ratio of lateral to normal pressure along chute wall
keo effective linear pressure gradient down chute wall surface
at zero velocity
m mass
n normal co-ordinate
n number of samples
n(t) noise signal
|N| modulus of normal force
N' normal force per unit mass
N number of clock pulses per sequence

Pn pressure in normal direction
R correlation function
s d isp 1acement
So constant power spectral density
t t ime
t tangential co-ordinate
AT period of clock pulse
V grain stream velocity
Ve grain stream exit velocity
Vo grain stream velocity at chute entry.
x» y Cartesian co-ordinates
X (t ) input signal
y (t) output signal
0 chute slope angle with the vertical
0f limiting si ope ang 1e
T correlation delay time
T coefficient of friction of grain on chute surface
Te equivalent coefficient of friction on stream element
X dummy variable for time
y coefficient of velocity dependent resii stance
P radius of curvature
U) computation variable
CHAPTER 1: INTRODUCTION
1.1 Genera I
The need to design, install and control highly efficient
materials handling systems cannot be denied. Pa low itch (Reference 1)
estimates that in the U.S.A., for example, materials handling
activities cost industry about $125 billion in 1970 representing 25$
of the Gross National Product of that country. Thus, even small
incremental gains in the efficiency of materials handling can yield
significant cost reductions.
The variety of materials currently being handled in bulk is
almost endless. The Australian agricultural industry provides a
range of such activities. The materials handling function only adds
time - position utility to the material; it does not add form-utility
usually.
The viability of the total bulk materials handling system and
its components is strongly influenced by the characteristics of the
material. The ability to delineate the relevant material
characteristics and the ability to describe quantitatively the
mechanics of flow are essential pre-requisites to ensuring this
vi ab i Iity .
In narrowing this investigation to granular agricultural
products, one takes cognizance of the present annual Australian wheat
production requiring the handling of more than twenty million tonnes
per year. For 1973 this amounted to 20,135,000 tonnes with a dollar
value in excess of $348 million (Reference 2).
It is recognised that the techniques developed for these
agricultura I products should quite readily be applicable to other
suitable free flowing cohesive and non-cohesive materials.

19
1 .2 Outline of the Problem
The substance of this investigation forms part of a
comprehensive research programme in progress at the University of
Wollongong to study the bulk handling of granular materials. It is
well recognised that the hopper-discharge chute is a component of
most material handling systems. Roberts (Reference 3) highlighted
the need for determining the dynamic characteristics, with a view to
optimising the design for gravity flow under steady state conditions
and to enable accurate flow control under transient conditions, such
as blending.
In the bulk handling process it is usually necessary to
transport the granular material from an elevated storage bin to
some laterally displaced lower position. The problem of efficiently
utilising the available potential energy, so as to achieve some
specified performance objective, has received increased attention
by Roberts et a I (References 4 to 6 ).
The main objective of the steady flow problem has been to
achieve optimum chute profiles for either minimum transit time or
maximum exit velocity in some specified direction, whilst achieving
stable ’fast flow' .
Use was made of the FIetcher-PoweI I constrained minimisation
algorithm, to optimise the selection of coefficients for the
polynomial approximation method, for evaluating the steady flow
problem.
For the controlled flow case where flow transients are caused
by the operation of the control valve, the complexity of the problem
makes it quite difficult to apply theoretical analysis for the
determination of the dynamic model.

20
The advantages of experimental techniques based on statistical
methods are now well known, being more reliabile than conventional
methods applied to systems that are inherently ’noisy’. This study
makes use of the Pseudo-Random Binary Signal and cross-correI ation
method of analysis to identify the transient characteristies of the
hopper-discharge chute system. High speed photographic techniques
were used to aid the interpretation of phenomena indicated by the
stochastic analysis.
1.3 Review of Previous Research
1.3.1 GranuIar Materi a Is
A number of schemes (Reference 7) have been used in the past to
quantify the steady state ’dynamics’ of the hopper-discharge chute
system with some success. Although, at first sight, an analytical
solution seems straightforward, Roberts (References 6 , 7) showed,
that because of the distributed nature of the parameters involved,
lumped parameter approximations imposed a number of restrictions to
a general solution. A feasibility study (Reference 8 ), indicated that
the Pseudo-Random Binary Signal (P.R.B.S.) and cross-correI at ion
method, might prove to be quite effective in identifying the flow
transients under controlled flow, because of extraneous system noise
insensitivity.
Optimisation studies (References 8 to 13) have concentrated on
either minimising transit time through the chute or maximising
horizontal exit velocity under steady state conditions for gravity
flow, through chutes of varying geometry. Solution methods, based on
variational techniques including, Pontryagin’s minimum principle,
proved awkward since small variations in the initial values of the
Lagrange multipliers produced large differences in the solution. A

21
discrete segment method of solution for the descent curve proved
(Reference 10) efficient only when Coulomb friction was ignored. The
polynomial approximation outlined by Chiarella (Reference 11) was used
to design a chute profile, resulting in improved relative transit
times compared to the more commonly used profiles. However, due to
the low exit angle, instabilities in flow occurred and the method
needed to be modified to incorporate a constraint on the minimum
exit or lift angle in the profile.
Parlour (Reference 14) delineated the parameters necessary to
quantify the mechanics of granular material flow down inclined
straight chutes. The experimentally derived pressure profiles
refined the earlier work of Roberts (References 3, 7) in introducing
the concept of equivalent friction coefficient providing additional
insight into the pressure distribution at the chute boundaries as well
as the total drag force.
1.3.2 System Identification
Recent comparative studies (References 15 to 20) of system
identification schemes have highlighted the difficulty of finding a
general method for system identification, that can yield meaningful
results for a broad class of identification problems. However, given
the usual unavailability of ra priori* information of system
parameters often encountered in actual plant identification, the
P.R.B.S./cross-correlation method has been found to yield the best
overall performance, with regard to reliability, accuracy and
computation speed; suitably modified for on-line identification
(Reference 15).
22
The many advantages of the P.R.B.S./cross-correIation method
of identification must be weighed against the disadvantages of high
equipment cost for on-line analysis, and large data reductions for
off-line work. The former was not a problem in this case as the
necessary equipment was on hand.
In a pilot study (References 8 , 21) the application of the
P.R.B.S./cross-correI at ion identification method to bulk handling
problems was demonstrated. The inherent noise immunity of this
identification method makes it suitable for application to the hopper
discharge chute problem under investigation.
1.4 Scope of this Research
The experimental investigations which form the basis of this
research thesis generally extend previous work in this field. The
steady flow problem is analysed, and formulated on the basis of a
lumped parameter model subjected to drag forces resulting from both
Coulomb and velocity dependent friction. The possibility of flow
instability, mentioned previously resulting from chute profiles
generated by methods based on unconstrained parameter optimisation,
is obviated by using an optimising technique after FIetcher-PoweI I
enabling optimisation under parameter constraints. A programme is
developed which enables an optimal chute geometry to be calculated
given the material characteristics. The inlet conditions and outlet
position are such that a nominated parameter can be either minimised
or maximised. The cases analysed include minimum descent time and
maximum exit velocity in a specified direction.
The main research effort is directed towards the experimental
identification of the hopper-discharge chute dynamic model and an
experimental test of its validity. Pursuing the analysis of Choda

23
and Willis (Reference 22 ), regarding the phenomenon of ffastf and
’slow’ flow regimes, this report concerns itself to the analysis
of the ’fast’ flow phase. In the ’fast’ flow mode the material makes
contact with the chute bottom and side walls but does not make
contact with the top of the chute. In the ’slow’ flow mode the chute
operates completely full, with the material making contact across
the complete perimeter of the chute. ’Fast’ flow is more efficient with
the flow rate out of the hopper being entirely governed by the
characteristies of the hopper orifice. In ’slow’ flow additional
retardation effects are encountered which impair the flow rate and
the chute becomes, in effect, an extension of the hopper.
It is shown that the P.R.B.S./cross-correlation method of model
identification, is suitable to the granular materials handling field,
yielding both a time domain and via Fourier transforms, a frequency
domain description of the dynamic model. Because of the noise
immunity characteristics of this technique, it is possible to carry
out on-line identification by superimposing low energy random signals
onto the plant under test whi 1st the plant is in normal operation.
A side benefit of this method yielded an accurate determination of
in-chute transit time.

Part 1 : UNIFORM FLOW
25
CHAPTER 2 : GRANULAR MATERIAL FLOW DYNAMICS
2.1 Genera 1
The designed transfer of granular solids by gravity flow
through discharge chutes requires the establishment of satisfactory
geometrical characteristics of the chute profile. Roberts
(Reference 3) showed that for efficient operation, the most desirable
mode of flow is the ’fast’ flow mode in which the grain stream
thickness tapers towards the exit end of the chute.
Occasions arise whereby a discharge chute is required to
achieve some prescribed optimised flow performance. A case in point
is that of finding the profile which yields a minimum transit time.
This implies a minimum average stream thickness which is a desirable
feature of discharge chute operation. Another often cited example
is that of maximising the kinetic energy to achieve a maximum
discharge velocity, particularly in a horizontal direction to impart
a maximum throw to the granular material.
Blending and mixing processes incorporating gravity flow bins
and discharge chutes require the transient flow be controlled.
The operation of the control valves cause flow transients, which in
the case of critical blending operations with unavoidable transport
blending lags, give rise to unnecessary out-of-specification mix.
A necessary pre-requisite to solving these problems is the
determination of the dynamic model. The complexity of this problem
was alluded to by Roberts (Reference 8 ) in proposing a pilot study
based on experimental methods, rather than purely theoretical analysis,
to establish the dynamic model. On the basis of the results of that
pilot study, this experimental investigation based on experimental
system identification techniques was undertaken.

26
2.2 Description of the Apparatus
The experimental apparatus is designed to investigate the
gravity flow of the selected granular material through chutes
of different geometries, at various discharge rates and chute
inclinations. The design allows for recirculating operations,
with the grain dischargi ng from an elevated fixed ’funnel flow’
hopper, through an orifice chute system under test to the
receiving hopper, which is the feed hopper for the grain conveyor.
Figure 1 shows the general arrangement of the apparatus.
Construction details of the apparatus, and other ancillary
fabricated equipment are described below.
Main Rig Framework
The main rig (Plate 1) is fabricated from steel angles
welded to form a rigid framework, in which the chutes and other
auxiliary components may be erected as required for each test.
The model ’funnel flow’ hopper, of approximately 0.1m3 capacity,
has the front wall made of perspex to facilitate grain movement
observations. To the base is fitted a steel orifice assembly
with provision for static flow control setting as well as a
dynamic perturbing valve operated by a Philips (PR 9270) solenoid.
The recycling ability is clearly seen (in Figure 1 and Plate 1).
The rig accommodates straight chutes of approximately one and a
quarter metres in length at inclinations ranging from the stalling
angle to approximate Iy 30° from the vertical depending on cross
section. Chutes of curved geometries passing through the same
end points are also accommodated.

27
P.R.B.S,
NOISE MILLET RETURN LINE
GENERATOR
POWE R
AMPL IFI ER
1.1 General arrangement of the testing apparatus
FIGURE 1 The testing apparatus

28
.
k g /m i n
Fo w r a t e
Valve opening mm,
1,2 Hopper discharge characteristics
FIGURE 1 The testing apparatus

GRAIN
RETURN
1 HOPPER
P.R.B.S.
TRANSDUCER
HOPPER
GRAIN
ELEVATOR
CHUTE
FLOW
METER
GRAIN ELEVATOR
RECEIVING HOPPER
POWER
AMPLIFIER
Plate 1 Main testing rig

30
ControI-Perturbinq Valve
The controI-perturbing valve assembly (Plate 2) is bolted
to the bottom of the elevated hopper and effectively controls the
datum discharge rate through a variable sized rectangular
orifice and additionally allows the perturbation of this main flow
with the fluctuating pseudo-random binary signal. A general
assembly of the valve is shown in Figure 2.
The two gates slide in slots under the orifice assembly
butting against each other in the closed position. The slides
are so arranged in close contact with the orifice plate to obviate
jamming by whole grains. The main gate is manually set and
locked with a locking screw in the test position. The perturbing
gate is operated by a Philips electro-dynamic vibration exciter
(type PR 9270) powered from a Philips power amplifier (Type GM 5535).
To the perturbing gate is also fitted a Hewlett Packard
displacement transducer of the 7DCDT series which monitors the
actual movement of this gate. The nominal stroke of the Philips
solenoid of ±4mm could be reduced by the adjustment of a mechanical
stop. The chute supports fitted were fabricated from perspex
enabling visual and photographic observation of flow from the
hopper orifice.
Chutes
Chutes were fabricated from 5mm perspex to give an open
chute of rectangular cross-section with an internal width of
25mm and a wall height of approximately 90mm. The curved chutes
were cut using a bandsaw from full sized drawings glued to the
perspex protective sheet.

F igure 2 H o p p e r -o r i f i c e arrangement
Pl a t e 2 Un d e r s i d e view of hopper
33
IncIi nometer
An inclinometer was used to measure the angle of inclination
of the test straight chute to the horizontal. The device is
located at the base of the chute and a plumb bob enables
inclinations to be read off directly from graduations marked on
the perimeter of the quadrant.
FIowmeter
The design of a number of direct reading electrical output
flowmeters was attempted (see section 10.1). For the present
tests a cantilevered beam fitted with a rectangular flap and
strain gauges was used. Plate 3 shows the flowmeter assembly
with its multiaxis adjustment.
2.3 Model
2.3.1 Introduct ion
Choda and Willis (Reference 22) nominated two flow
regimes, ’fast’ and ’slow’ flow. Roberts (Reference 3) initially
and more recently Parlour (Reference 14) proposed approximate
theories to account for the flow behaviour. The theories, based
on dynamic analysis, gave rise to generalised flow equations,
which were nonlinear in form. An equivalent friction coefficient
was introduced by Roberts (Reference 3) to account for the
drag on the chute bottom and side walls, based on a pressure
distribution as shown in Figure 3.

Pl a t e 3 Fl o w m e t e r
35
Pn
F igure 3 Pr e s s u r e distribution on
CHUTE WALLS ( ROBERTS )
Parlour (Reference 14) improved the approximation by measuring
the bottom and side wall pressure distribution and found that the
distribution was as shown in Figure 4. It can be seen that the
bottom pressure distribution is approximately uniform across the
chute while the side wall pressure increases nonl¡nearly with
depth of gra in .
F igure 4 Pr e s s u r e distribution on
CHUTE WALLS ( PARLOUR )
’Fast’ flow conditions represent the more efficient mode of
operation and will be the basis of modelling. The three researchers
mentioned worked with millet seed flowing through perspex chutes,
and advantage is taken of the availability of their published
parameters to conduct this study.
2.3.2 Lumped Parameters
Although the physical problem is concerned with a distributed
discrete system, Roberts (References 3, 7) has shown that, for the
’fast' flow mode, the bulk flow characteristics may be modelled in
terms of a représentâtive single particle moving subject to
tangentially directed resisting forces.

37
On the basis of the pressure distribution shown in Figure 3 in
which it is assumed that the vertical pressure is equal to the
hydrastatic pressure, this being consistent with shallow bed operation,
Roberts (Reference 7) derived the following expression for the
equivalent coefficient of friction
Te = T(1 + k jj)
where k = ratio of lateral to vertical pressure
•g = stream depth to width ratio.
Using this concept of equivalent coefficient of friction it was
shown that the resistance to flow was a combination of Coulomb and
speed dependent friction forces.
In the same reference, ana Iysis of the energy loss in the
flowing stream was undertaken and the ratio of work done due to
grain-on-grain sliding and grain-on-chute sliding resulted in the
following proportions.
Ug :Us : Ub = 9.4% : 8.3$ : 82.3$
where Ug = work done dueto grain-on-grain sliding
Us = work done dueto grain-on-chute sidewallsliding
Ub = work done dueto grain-on-chute bottom sliding.
This showed that the major energy loss was against the chute
bottom with the contributions of Ug and Us being quite small. The
lumped model takes into account the friction drag force based on Us
and Ub and by working with the average velocity the grain-on-grain
sliding has a relatively minor contribution and was ignored.
More recently Parlour (Reference 14) having determined the
actual pressure distribution to be as shown in Figure 4 was able to
refine the expression for the equivalent coefficient of friction. He

38
found that on this basis the ratio of work done by the various
components was
Ug : Us : Ub = 10.4% : 11.3% : 78.3$
which further supports the assumption of a lumped parameter model.
Figure 5 shows the co-ordinates with respect to which the
particle dynamics are formulated. The gravitational force acts
in the direction of the positive y axis.
Figure 5 Flow problem model

39
Parlour found that for millet seed in an open channel
perspex chute the following parameters are applicable:-
t = average coefficient of friction
= 0.306
t@ = effective coefficient of friction (includes friction
between grain base and walls of the
chute and the equivalent friction
due to dynamic internal stresses).
xe = t [ 1 + keo | (1 + c v 2)] (2 .1 )
= x[1 + keo g- + keo ^ c v 2]
r
S•inee vO = 7—
H
v Ho
H = V°^° (uniform flow conditions)

bv
T - T ri , i. v0 H0 , keo v0 Hn cv n .. ...
Te - T[i + Keo — gv' + ----- g u--- J (for v > 0)
= T [1 + + c 2v] (for c 2 = 0 same as Roberts') (2.2)
where
_ keo v q Hq
1 B
= keo vp Hp c
C2
Now for millet seed :
keo = 0.346
c = 0.0215 (sec/ft2)
= 0.001997 (sec/m2)
Ci = 0.346 Vo "Tp*
b
c 2 = 6.9096 x ( 10- “) x
b
40
T = 0.306
Te = 0.306 [1 + 0.346 V° pHo + 6.9096 x (10"“ ) x vl

DV D
(2.3)
The limiting angle of theta, 0f, is given by:
0f = 90°-tan ^ T q ) (2 .4 )
>, For the present 0f will be interpreted to be the angle at which
a straight chute would begin to stall or choke as its inclination
with the vertical was increased whilst ’fast’ flow was in progress.
Later it will be shown that this angle constitutes a major constraint
in the design of chutes for trouble-free stable flow.
2.3.3 Statistical Evidence
The validity of the lumped parameter model was investigated,
so as to gain confidence in the optimal chute profiles to be
predicted from it. Visual and photographic observations of the
stream flow within the chute and upon leaving the chute were made.
The test rig was arranged as depicted in Figure 1 so as to
measure the transit time for a single grain particle moving down the
chute. In order to circumvent the difficulty of recording the
impact of a single grain due to transducer sensitivity limitations,
approximately ten grains were released simultaneously. Observation
revealed little interference throughout the flight down the chute
with the grains sliding on the bottom surface and striking the
flowmeter with an impact pattern shown by trace 2 on Plate 4.

41
PERTURBING VALVE MOTION
TRACE 1
FLOWMETER RESPONSE
TRACE 2
Pl a t e 4 "Si n g l e " grain transit time
The required number of grains were placed on the perturbing
valve within the empty hopper and the valve was then stepped open
using a square wave function generator as a driving signal to
the solenoid. The motion of the perturbing valve, as measured by
the linear displacement transducer, was used to trigger a dual trace
storage cathode ray oscilloscope (CRO). The striking of the
flowmeter by the grains provided the second trace, delayed by an
amount equal to their transit time. Plate 4 shows a typical
test enabling the transit time to be measured as the difference
between the two traces.

42
In order to confirm or otherwise, the validity of the lumped
parameter model, tests were carried out as discussed above. The mean
descent time was found to be 0.704 seconds* for the model chute with
geometry corresponding to the constrained optimisation specification
given elsewhere (Section 2.3.1). This compares with 0.700 seconds
calculated by estimating the step response of the chute. It will be
shown (Section 5.2) that the cross-correI at ion determined under these
conditions approximates the impulse response of the chute which, when
integrated, yields the step response. The transit time was taken to be
the delay time to the midpoint of the step rise, as shown in Figure 6.
F igure 6 Tr a n s i t time from the

STEP RESPONSE
As the transit time so calculated for the bulk flow was found to
be within ]% of the ’single g r a i n ’ transit time it was felt that the
lumped parameter model was, at this stage, an adequate formulation
assumpt io n .
* Since it is realised that this study concerns itself with incremental

gains in performance, the necessity arises to quote experimental
results to a rather large number of significant figures in order to
report these incremental differences.
43
2.4 Uni form Flow
2.4.1 Introducti on
A rigorous analysis of the uniform steady state flow of
granular material, down chutes is a necessary pre-requisite to the
implementation of an optimal design. The uniform flow case is of
particular interest, as it is felt that this constitutes the
activity in which chutes are presently most frequently utilised.
The aim here is to present a general solution to the problem of
establishing stable ’fast’ flow in chutes of prescribed geometry.
Given certain known characteristics of the material to be
transported the physical constraints of the chute entry and exit
conditions, we seek a general solution to the chute geometry such
that a choice can be made as to the optimisation of any given
parameter. A single parameter will be chosen for optimisation from
the great many possible variables to illustrate the general solution
method.
2.4.2 Apparatus
All the uniform flow tests were carried out on the model
hopper-chute rig shown in Figure 1. Chutes with profiles as shown in
Figure 7 and Plate 5 were, in turn connected to the testing rig such
that all of them received from and discharged to the same end points.
That is, the straight chute at 40° to the vertical, the parabolic
chute of the form

2
x = cy
Xo
where c =
(Yo)
and both the unconstrained and the constrained (as detailed later)
profile chutes were tested with identical inlet and outlet

44
RX IS (METRES)
F igure 7 Ch u t e p r o f i l e s f o r m o d e l c h u t e
£*Il '’
¿g& - t< »»v *
Jt.î¿¿ *'"T ‘'‘■v’
";.’
;j''i
*J»Í
&5$
OPTIMUM CHUTE
(t = O.O)
PARABOLIC
CHUTE STRAIGHT CHUTE
Pl a t e 5 Co m p a r i s o n of c h u t e p r o f i l e s
46
co-ordinates. As can be seen the hopper, chute and grain elevator
form a closed loop enabling the static capacity of the model hopper
to be effectively increased by recycling the grain under test. The
motor speed controller on the grain elevator enabled the auger
lifting capacity to be varied and, if desired, to be set equal to
the discharge rate from the hopper, effectively maintaining a
constant head above the hopper orifice.
2.4.3 Procedure
In each case millet seed was set cycling round the testing loop
to condition the chute surface, and to filter out husks which appeared
in the body of the material. The husks adhered to the perspex
surfaces and had to be removed to enable proper observation of the
stream flow. The cross-correI ation equipment was used to measure
in-chute transit times. This gave an accurate measure, by way of
the transport delay, of the time taken for the grain stream to pass
from the perturbing gate to the flowmeter. The main ’flow’ gate was
set to an orifice opening of 25mm, corresponding to a flow rate of
4 kg/min. whilst the perturbing gate stroke was tested at both ±2mm
and ±4mm. The capability of the correlation method to cope with this
very poor signal to noise ratio was clearly evident.
As a crosscheck, stream velocity at exit was measured using
both a visual and a photographic technique. Visually, the grain
stream was observed to fall in an arc in front of a card with a grid
and co-ordinate system centred on the chute axis. Using the
equation for free fall
yx‘
Vp = (2.5)
e |_2cos 2 « (y-x tan °0 _
and the observed x, y co-ordinates the exit velocity of the mean
stream flow could be calculated.

47
Photographically the velocity could be determined
knowing the film scaling factor and the framing rate. In all
photographic sequences, a scale is included near the field of
interest, and the camera enabled timing pips, derived from a timing
light generator, to be placed on the film edge.
In all tests visual observation confirmed that the flow was
’fast' flow with stream depth H decreasing uniformly towards the
chute runout.
2.4.4 Equations of Motion
Using the parameters obtained by Parlour (Reference 14)
and taking axes as shown in Figure 8 we proceed to define the
equations of motion of a particle model moving down a chute of
arbitrary radius of curvature p. As can be seen in Figure 5
granular material enters the open channel curved chute of constant
rectangular cross-section with velocity vQ . At an arbitrary cross
section the average velocity of the grain stream is v and the mean
stream thickness is H.
F igure 8 Ac c e l e r a t i o n m o d e l - f o r c e b a l a n c e
An energy balance between two successive cross-sections,
(1) and (2) yieIds
Kinetic Energy (2) = Potential Energy (1+2) - Work done (1+2)
+ Ki neti c Energy (1).
K.E.2 = P . E . i+2 - W . D . i -*2 + K.E. i
i.e., ? m v 2 2 - m g y - W . D . 1^ 2 + i m v i 2 (2.6)
D e f i n e a dr ag fo rc e F q _ |n |+ yv (2.7)
w h e r e N = dmg SinG + dm a^ ( 2. 8)
49
Now
a = /a)'
n
= /(— ) 2
f dt
= )(— )
J at; at;
1 ,dsN ,d0 .
= / - nf dt (For this orientation ds = -A10)
ds
2 d8 Q (2.9)
= -v — COS0
dy
Substituting Equation 2.9 into 2.8
dfl
N = dm g Sin0 + dm (-v2 — cos0) (2. 10)
a dy
Dividing by dm, Equation 2.10 becomes
: _ g Sin0 - v 2
N11 cos0 (2 . 11 )
dy
where
N 1 = normal force/unit mass
Substituting this into Equation 2.7 yields
Fq s xe [g Sin0 - v 2 cos0]+ yv (2.12)
= generalised drag force per unit mass combining
Coulomb and Velocity dependent friction components
Now
W.D. i+2 = /' Fn ds (2.13)

l
Let ds be an element of arc at the cross-section under consideration.
ds = / l + ( ~ ) Z dy
dy '
= J 1 + tan 20 dy
= sec 0 dy (2.14)
50
Substi tuti ng nto Equation 2.13 yields
W .D .i->2 = /2 Fd dy
i cosG
i .e Equation 2.6 becomes
dy + I (2.15)
vz 2 = gy - /v fd Vl
o COS0
re-arranging
FP
v 22 = 2gy - 2 f dy t vi2 (2.16)
cos0
and final Iy
v 2 = *4gy - 2 /Y dy + Vj 2 (2.17)
37 O COS0 7
Equation 2.17 expresses the velocity based on a lumped parameter
model of the grain stream at an arbitrary cross-section for granular
material moving under the influence of gravity subject to a
generalised drag force combining Coulomb and velocity dependent
friction components.
For millet seed, substituting Equation 2.3 into Equation 2.12
yields
FD = 0.306 1 + 0.346 V°D-

^ + 6.9096 x \0~k x y-°-D-— ~
DV D
2 d0
q Si n0 - v — cos0 + yv (2 .1 8 )
dy .
which needs to be substituted into Equation 2.17 to calculate, for
example, the velocity at an arbitrary cross-section (2 ).
From a force balance at an arbitrary cross-section in Figure 8 ,
and using moving co-ordinates defined by angle 0 and radius of
curvature f, the dynamic equilibrium of an element within the stream
flow results in a generalised equation of motion.

Using the symbols as defined above, a force balance yields:—
Te |N| + yv + dm Af = dm g cos0 (2.19)
and as before
N = dmg Sin© + dm An (2.8)

repeated
combining and expressing all variables per unit mass.
xe [g Sin© + An] + yv + A+ - g cosG = 0 (2.20)
Now
d (v)
A+ =
dt
_d_
dt V « ’]
pd20 6f_ d0
( 2 . 21 )
Jdt 2 dt dt
and An = /(di ) 2 ( 2 . 22 )
J dt;
substituting Equation 2.21 and Equation 2.22 into Equation 2.20
yields:—
“ 2 ^ ^ od20 d/ d0 Q n
Te g Sin© + f(^rr) + yv + + dt * dt " 9 cos6 = 0
a dt
(2.23)
re-arrangi ng
d 20 1 ,df . d0
(^ 0 + ^ (Te Si n0 - cos0 ) + = 0
dt 2 f dt dt ) + Te
(2.24)
This is the generalised equation of motion for a lumped parameter
model, expressed in moving co-ordinates and incorporating a
generalised drag force combining Coulomb and velocity dependent
friction forces.
These equations form the basis for the calculation of chute
geometries chosen to optimise some yet to be specified parameter
subject to a number of chosen constraints.

52
CHAPTER 5 : OPTIMISATION
3.1 Genera I
The necessity/desirabiIity of making incremental improvements
in the efficiency of the materials handling task has already been
alluded to in Section 1.1 and once the material parameters have been
ascertained, the additional cost of designing a chute with optimum
geometry is considered to be small compared to the integrated gain
of even a minor efficiency improvement. Capital and operational
costs incurred in changing chutes to suit the product being gravity
feed would seem feasible.
There are numerous situations known to occur in practice
which require the flow through discharge chutes to be optimised. In
discharging material from a hopper into a rail car, one may wish to
maximise material throw, i.e., maximise exit velocity. In some
cases it may be desirable to minimise transit time in order to obtain
the most favourable flow pattern, i.e., ’fast’ flow, which helps to
make the chute self-clearing. The constraint of matching exit
speed and direction of material falling onto a conveyor belt has been
cited for its potential in minimising conveyor energy losses.
In a recent application (Reference 23) use was made of what
was essentially a double-sided tetrahedral chute for the controlled
aerial spreading of crop fertilizer so as to achieve a specified
spreading pattern on the ground, i.e., maximum uniform dispersion.
For practical reasons a designer may choose not to implement
an optimal chute geometry but such knowledge enables him to work
'with certainty'. In effect, the optimal geometry becomes his
’yardstick’ against which to compare his compromised solution.

53
3.2 Model Formulation
3.2.1 Minimum Descent Time
Whilst variational methods of optimisation are indirect and
involve the finding of the extremum of a functional it is possible to
reformulate this problem into an equivalent ’pseudo-static1 problem
in which direct methods of optimisation can be applied. In
attempting to set up and solve this problem one encounters difficulties
with classical methods (Reference 12). The physical model is analogous
to a bead on a wire and the mathematical description of the friction
force requires the modulus of the normal reaction force. This
complicates the analysis and the overall problem is numerically
difficult since a two-point boundary value problem arises in the
application of the calculus of variations or Pontryagin’s Minimum
Principle. Solution methods based on the polynomial approximation
outlined by Chiarella (References 10, 11) are found to be
particularly suitable to this class of optimisation problem.
The choice of co-ordinates for the optimal model is an
important consideration; investigations have shown that moving
co-ordinates 0 and y are the most satisfactory. 0 (y) is the chute slope
with the vertical, a function of y and is the dependent or control
variable while y is the independent variable. Referring to the system
model shown in Figure 8 and recalling Equation 2.17:
(2.17)
repeated
The descent time between two successive cross-sections A and B is
given by:
+AB = /B dt
A
jB d£
A v (3.1)
54
Now ds = Sec0 dy (from Equation 2.14)
, rB sec0 ,
+AB = f Z dY (3.2)
A
Since the end points (X, Y) are fixed, the function 0(y ) must
sati sfy
,Y
tan0 dy = X (3.3)
In order to combine Equations 2.17 and 3.2 to arrive at
an expression for the descent time define a quantity
,Y d
w(y) = -/ yv + Te |N| (3.4)
cos
0 L
Note that u)(y) satisfies the differential equation
dio = -[ yv + xe IN11
dy cos0 (3.5)
with w(o) = 0 .
In terms of this variable, Equation 2.17 resulting from the
application of work and energy principles, becomes:-
V2 = X'l 2 + 2 (gy + Co) (3.6)
substituting into Equation 3.2 for the descent time yields
.Y dy (3.7)
^ /vi 2 + 2 (gy + w) cos0
The problem now becomes; find 0(y ) so as to minimise t whilst
at the same time, satisfying the end condition constraint of
Equation 3.3.
55
3.2.2 Maximum Exit Velocity
Sometimes it occurs that a maximum possible exit velocity is
desired for granular material flowing in chutes under gravity. The
loading of railcars with wheat is a case in point. Within this
framework one may aim to optimise either the exit velocity or some
component of the exit velocity such as the horizontal component.
The required profile shape of gravity flow discharge chutes
to achieve maximum exit velocity involves minimising losses.
Formulation along similar lines to that of the particle model
used previously assumes that the initial velocity of the particle
is v, and an energy balance reveals:
Change in kinetic energy = change in potential energy - frictional
losses;
i.e., |m (V22 - V i2) = mgy - frictional losses (3.8)
The frictional loss in going from the origin (0,0) to a
general point s(x, y) is
loss = f5 {t J n | + yv} ds (3.9)

o
where ds is an incremental arc length.
Now ds = Sec0 dy
loss = /s {Te |N‘ | + yv} ^ (3.10)
substitution into 3.8 yields after re-arrangement:
m (v22 - vi2) = 2mgy - 2fs {xe |N| + yv} (3.11)
Recall that a force balance after resolving into normal
components resulted in:
N = dmg Sin6 + dm An
56
N 1 = g Sin0 - V 22 cos0 — ( 2 . 11)

repeated
substituting 2.11 into 3.11 yields:
d0
(V2 2 - v i2) = 2gy - 2^S {ie (g Si n0 - V2 2 C O S0
dy
(3.12)
Now, depending on specified end constraints, that is, entry
and exit angles, entry and exit co-ordinates one could complete
this formulation and solve to obtain the optimum profile to maximise
the exit velocity for the exit angle specified. In the present study
attention is directed towards solving the minimum descent time
problem as defined in Equation 3.7.
3.2.3 Additional Constraints
In order to provide a practical general solution to the problem
of minimising the descent time, a number of design constraints and
certain system constraints need to be considered. Thedesigner may wish
to specify the inlet grain stream velocity and its direction, to
comply with the physical arrangement of the hopper-discharge system.
To ensure flow stability, and make the chute self-clearing, the
angle that the chute bottom makes with the vertical, 0 , needs to be
constrained. This angle normally increases towards the exit or runout
and can exceed the limiting angle, 0f. For ’fast’ flow this angle
determines the stability of the chute system. Should the stream flow
be momentarily blocked, initiated by a si ng le sf icki ng grain, the chute
would stall and choke or at the very least flow would change from
’fast’ flow to ’slow’ flow. Once stalled, the chute would not self
clear, requiring manual grain removal before flow can be re
established. This phenomenon became obvious when the results of
Roberts (Reference 8 ) for the unconstrained minimum descent time were

verified. Plate 6 shows the range of responses resulting from a
momentary grain blockage. Initially the grain stream is flowing
in the ’fast’ flow mode as shown in Frame 1, when a minute
disturbance in the stream causes the flow to experience a sudden
transition to the Tslowf flow mode, shown in Frame 2. The bunching
up of the grain stream here is clearly evident. In order to
emphasise the seriousness of this instability problem, Frames 3 and 4
respectively show statically the initiation of grain flow instability
and the fully choked chute. These were obtained by immediately
shutting off the flow control valve as soon as a flow disturbance
was detected whilst the chute was discharging under a light flow
rate. A momentary delay in shutting off the flow control valve
resulted in the fully choked condition shown in Frame 4. The
inability of the chute to self-clear is evident since the grain
stream is in a stable stationary state.
The designer needs to identify this limiting angle, 0f, for
the material being handled and to ensure that the combination of inlet
and outlet co-ordinates, chute geometry is chosen so as to avoid
exceeding this limit. The programme package developed in Section 3.4
enables the designer to generate appropriate chute geometries
constrained so as to meet this stability criterion. Testing showed
that the unconstrained solution represents an insignificant
reduction in descent time and in a practical situation, of course,
the stability requirement would rule out any consideration of such
a profile. For these reasons the unconstrained chute profile was,
from here on, excluded from any further analysis.

GRAIN
!**■**.>. BUNCHING
ppa*” *,*»
,;^r f '
’" miijr*— •-
tiËÊÊÊm ''*.-- :
TRANSITION TO 'SLOW FLOW MODE
FRAME 1 FRAME 2
FULLY
í*** -.'\;'.'••/ ■/- í, v •’\ W CHOKED
.- • . .r >'• r .
CHUTE
INITIAL FORMATION OF
GRAIN FLOW INSTABILITY |$P§pp
FRAME 3 FRAME 4
Pl a t e 6 C h u t e s t a l l i n g (u n c o n s t r a i n e d )
3.3 Methods of Solution
3.3.1 Introduction
Solution methods to the unconstrained formulation have been
presented in the literature. Charlton (Reference 5) ignored
frictional losses arising from the curvature term and used
variational methods to obtain a solution to the problem
approximation. Chiarella (Reference 10) applied a method,
analogous to the finite element technique for numerical solution
of variational problems, which he called a discrete segment
solution method. The advantages of this method were quickly
superseded by a polynomial approximation method presented by
Charlton (Reference 12) which was effective in handling the general
resisting force. That is, a resisting force comprising Coulomb
drag as well as a velocity dependent drag component. This approach
will be used to develop a solution for the minimum descent time
problem as defined in Equation 3.7 subject to the intrinsic
constraint contained in Equation 3.3.
3.3.2 Polynomial Approximation
Having selected a moving co-ordinate system for the
equations of motion with co-ordinates y and 0 , the problem solution
involves approximations for 0 (y) or tan0 (y) by a polynomial in y.
o m
i.e., tan0(y) = Ao + Aiy + A 2y + .... + Am y
60
d 0_
An expression for , required in Equation 2.18 is obtained
dy
by differentiating Equation 3.13
M
d9 m-i
i.e., sec 20 £ m Am y (3.14)
dy
m=i
re-arranging
M _
d0
cos 2e £ m Am ym 1 (3.15)
dy
m= i
Now substituting Equation 3.13 into the end condition Equation 3.3
and carrying out the integrations yields the approximation:
M ,
y
L A v m+1
Mm,.y____ X = 0 (3.16)
m=o m +1
for which
M-1
M+1
X £ Am Ym+1 (3.17)
Am y M+1 m+i
m=o
where (X, Y) are the exit co-ordinates.
This means that only the coefficients {A0 , Ai, A2, ... A^_^}
are independent. That is, the descent time t is a function of these
M coefficients.
i.e., t = t(A0 , Ai, A2, ... AM-1) (3.18)
The problem of minimising the transit time, now reduces to
one of finding a set of coefficients, {A0 , Ai, A2 , ... ^} which
minimise the expression for t (Equation 3.7) subject to the constraint
Equation 3.3. .
To specify the initial direction one takes note that if y = 0,
Equation 3.13 becomes:
0O = tan 1 A0
61
This implies that A0 must be specified, in which case one
minimises t with respect to the (M-1) coefficients remaining
i.e., t = t (Ai, A 2, A 3, .... A^_.j) (3.19)
The actual number of terms to be considered will depend on
a number of factors including the order of accuracy required, the
computational effort considered appropriate, and the total number of
constraint equations specified in the particular problem
formulation. Some preliminary test runs with M equal to 2, 4, 8
revealed that no significant improvement resulted by considering
M greater than 4. Thus, in all the final design calculations M
was set equal to 4.
3.4 F 1etcher-PoweI I Algorithm
3.4.1 Description
For the constrained minimisation problem considered here,
experience has shown that the F Ietcher-PoweI I algorithm as outlined
in Kuester (Reference 24) provides a fast, computationally efficient
solution method. Basically, it is a hill climbing technique, using
a penalty function to drive the solution back into the feasible
solution space, should it cross the constraint boundary.
The algorithm was programmed by Haarhoff (Reference 24) and
finds the minimum of a multivariable nonlinear function subject to
nonlinear equality constraints.
i.e., Minimise F(x, x 2 ___ X|s|)

(3.20)
subject to G k (x , x 2 ___ X|sj) = 0
where K = 1, 2, 3, . ..., M.
62
The method incorporates the constraints into a modified,
unconstrained objective function which is then optimised by the
unconstrained minimisation technique of Fletcher and Powell.
Derivates of the objective function with respect to the independent
variables are thus required. Inequality constraints can be treated
by use of slack variables and transformations. The algorithm
proceeds as follows:
1) A new unconstrained objective is formulated from the original
function and constraints,
M M
$ = F - E AK G k + B E G K2 (3.21)
K=1 K=1
where and B are constants.
2) A starting point is selected (feasible or nonfeasible) and the
value and derivatives of F and values of G k are determined.
The derivatives can be either analytical or numerical
approximations. The FI matrix in the Fletcher and Powell procedure
is set equal to the identity matrix for the first iteration. The
A k values are determined from the following system of equations,
N M N
3G j 3G k 3Gk 8F . ( 3 . 22 )
E E (A E
3xj 3x{ 3xj 3xj
¡=i j=i i= i
where K = 1, 2, ..., M.
The B value is set equal to some positive number; the authors
suggested a value of 30. Testing showed that the algorithm was
relatively insensitive to variations in B within the range of 30-100.
3) A series of search directions and one dimensional search steps are
then determined per the unconstrained Fletcher and Powell general
procedure for the modified objective function with updating of the

63
À k values for each new iteration. When convergence is achieved,
G k = o and F = 0 and the required function F has been optimised.
3.4.2 Programming Considerations
To use the algorithm to minimise the descent time t, as a
function of the polynomial coefficients, it is necessary to write a
number of subroutines, which define the objective function and the
constraining equations.
Briefly, the programme package consists of a main programme
controlling calculations and able to call on fifteen subroutines.
Initial values for the chute parameters are read in, including
estimates of the polynomial coefficients. The constrained
minimisation is then carried out to yield the minimised descent time
t for certain convergence criteria. Stability of the package when
seeking low convergence criteria of the order of 10 12, was improved
considerably by using double precision variables. Advantage was taken
of a recently acquired plotter and software package to plot the
chute profiles. This proved invaluable in drawing across-the-board
observations of trends not readily discernable from the raw printed
data.
Further details and operating procedures of the programme can
be found in Section 10.5.

64
CHAPTER 4 : RESULTS AND DISCUSSION - UNIFORM FLOW
4.1 Comparison of Model Results with Theoretical Calculation
During the course of the research programme the need for improved
methods for a non-contact velocity detector was considered to be of
great importance. The concept of an ultrasonic transducer was pursued
as it has obvious attractive features. The development of this device
which was carried out in parallel with the main experimental programme
resulted in a prototype (discussed in Section 10.1.4) being completed
and tested towards the end of the research programme. This ultrasonic
velocity detector enabled quick determination of stream velocity at,
say, inlet to and outlet from the chute. Having established that the
exit velocity from the model hopper into the chute entry was
approximately 0.5 metre per second and using the previously derived
design parameters (Sections 2.4.4 and 3.2.1), a chute with appropriate
minimum time geometry was generated. Table 1 summarises the comparison
of the differences between the experimental results and the assumed
theoretical criteria. For millet seed, the limiting angle, 0f, is
approximately 60° and Parlour’s value for the equivalent coefficient
of Coulomb friction, Te , is used. This leaves the coefficient of
viscous drag, y, and the exit velocity, V, as unknowns. By assuming a
range of values for the viscous coefficient, y, it is possible to
generate a number of chute geometries. Each such chute would be
associated with its particular descent time and exit velocity. If it is
assumed that there is no viscous drag, then the percentage difference
in descent time between the theoretical and experimental chute is 30$
and the percentage difference in exit velocity is 44$.
However, it is recognised that although the variation in the
coefficient of viscous drag has little effect on the chute profile at

65
this model scale (see Section 4.5 for a detailed sensitivity analysis)
this variation does affect both the descent time and the exit
velocity. In fact, if it is assumed that y = 2.0 the relative
percentage differences reduce to:
descent time difference = ]]%
exit velocity difference = <2%
Table 1: Comparison of model results with

THEORETICAL PARAMETERS
PARAMETER EXPERIMENTAL THEORETI CAL
Initial Velocity (VINIT) 4 0.5 m/s 0.5 m/s
Coulomb Coefficient (t ) = 0.306 0.306

(based on
Parlour)
Height to Breadth Ratio 1 1

(H.B.)
Limiting Theta (9f) 60° 60°
Coefficient of Viscous
Drag y - 0 1 .0 1.5 2.0
Descent Time .70 .4.8 .54 .59 .62

sec
Exit Velocity V(exit) 2.3 m/s 4.1 3.2 2.7 2.3

m/s
Design Coefficients Ai=-.6108006 7.16043

A 2= 20.8306 -27.32309
A 3=-43.8373 41.2757
A«*= 26.46906 -19.312
Aiy2+A 2y 3+A3y‘t+Aity 5
Equation of Profile x -
2 3 4 5
66
4.2 Check on the Value of Viscous Drag
At this stage it was recognised that an experimental check would
need to be made on the actual amount of viscous drag present in the
model chute. Using a straight chute for this, simplifies slightly
the equations to be integrated and affords a certain amount of cross
check on the curved, optimally generated chute results. Since,
for this chute experimental results were available for inlet velocity,
exit velocity and descent time, as well as the length of the chute,
two related approaches were taken. Firstly, acceleration down the
chute was formulated in terms of dv/dt, and then as a check, it was
formulated in terms of v dv/ds. Figure 9 depicts the acceleration
model resulting from a force balance on an element sliding down the
open chute subject to friction on the chute bottom and waI Is. It is
assumed here that this frictional drag force can be adequately
represented by the two lumped components - Coulomb and viscous drag
forces.
F igure 9 Fo r c e balance
67
A force balance on the element yields
dv
= g cos0 - T0 N - yv (4.1)
dt
= g cos0 - Te g sin0 - yv
= g cos0 - t (1 + + C 2v ) g sin0 - yv (4.2)
dv
Note that ppp - 0 for terminal velocity.
For a given experimental set up Equation 4.2 can be simplified to
f = A - B - f - Dv - uv
where
A = g cos0
B = tg sin0
C = xg sin© ci
D = t g sin0 C2
E = A- B
F = D+ y
which yields
± L = e - £. - F (4.3)
dt v '
Integrating both sides
| 2 dv = /(E ~ ~ ~ Fv )dt
After re-arranging
/2----- ^ ----- = / 2dt (4.4)

1 ( E - — - Fv ) 1
68
In order to maintain the flow of this analysis, the balance of
the solution of this equation continues in Section 10.6.1 and
ultimately results in
E J_
f 2Fv i v 2 - E(vi + y2 ) - 2Fa(v2 - V i ) + 2c ] ^ 3 ["Fvi2 - Evi + c l ^
L 2F v i v 2 - E(vi + v2) + 2Fa(v2 - Vi) + 2c J L F v 22 - Ev2 + cj
= e (f2 - ti) (4.7)
Now given vi, v 2 and t we seek a value of y, the viscous drag
coefficient, such that Equation 4.7 is satisfied. Use was made of a
Rosenbrock (Constrained Rosenbrock Hill Algorithm - Reference 24)
hill climbing subroutine to find the value of y to minimise Equation
4.7 expressed as L.H.S. - R.H.S. = 0. Some difficulties were
encountered as the extremum point is not very pronounced and may in
fact be a saddle point.
It was decided to operate on the function squared since this
sharpens the extremum point and stabilises the numerical solution
with vi = 0.5 m/s
v2 = 2.7 m/s
t = 0.735 sec.
The minimisation revealed a value of y of 1.85. It is well
recognised that experimental equipment for measuring these parameters,
including the ultrasonic velocity detector, have a resolution limit.
This result also needs to be considered in the context that all the
parameters could not be measured at the same time. A sensitivity
analysis based on a variation of ±3% in just the three variables above
was considered mandatory. This revealed the spread in the value of y
to be from 1.78 to 1.88 using combinations of these parameter variations
to yield the ’best case’ and ’worst case’ value of y. This result
would seem to support the assumption that for the test environment the
69
coefficient of viscous drag, y, is of the order of 2.
As mentioned earlier a cross-check can be made by considering the
force balance to yield an acceleration term as a function of velocity
and displacement rather than velocity and time.
Using the same right hand side of Equation 4.3 we rewrite the
acceleration expression as
dv c c _
ds v
Ev - c - Fv2
• 2 dv r •
i .e., v T - = Ev - c - Fv (4.8)
ds
Re-arrang ing
v2dv
= ds
Ev - c - Fv2
v2dv
= ds
C 2\
F(— v - -p - vz )
v2dv
= ds
Ev.
-F(v2 + ^ -
vzdv (4.9)
= -Fds
Again in order to maintain the flow of the analysis, the develop
ment of this equation continues in Section 10.6.2 and ultimately results
in _E_
2F
(V 2-V 1 ) ifl£i + c - Ev 2 2
I f v1 2 + c - E 1
2a \2¥
r 2Fy1v2 ~ E (v i+ V 2 ) - 2 F a (V 1 - V 2 ) + 2c 1
[2FV1V2 ” E(v i + V 2 )) +
+ 22Fa(V
F a ( V 11--V
V22 )) +
+ 2c
2c J
= e ~ F ( s 2- S i ) (4.15)
Given vi V 2 and (S2-S 1) we seek a value of y the viscous drag
coefficient such that Equation 4.15 is satisfied. Use was again made of
the Rosenbrock Hill climbing algorithm to minimise Equation 4.15
expressed as (L.H.S. - R.H.S.)2 = 0 to avoid any similar problems
with insensitive extremum conditions.

70
The minimisation with
vi = 0.5 m/s
v 2 = 2.7 m/s
s 2 - Si = 1.145 metres
As mentioned earlier a sensitivity analysis based on a variation of
±3$ in just these three parameters revealed the spread in the value
of y to be from 2.22 to 2.33 again using combinations of these
parameter variations to yield the 'best case' and 'worst case’ value
of y. These results are summarised in Table 2. Clearly under the
testing environment the experimental results support the assumption
of a coefficient of viscous drag equal to approximately 2 (y = 2.0).
Table 2: Variation in viscous drag - m
Acceleration based Acceleration based

dv dv
on dt on v-r-
ds
y 1.83* 2.28**
max. y 1 .88 2.33
min. y 1.78 2.22
* converged for initial y between 1.4 to 5.
** converged for initial y between 0.5 to 4.
4.3 Comparison of Model with Chutes of Other Profiles
In order to compare the performance of the model chute with
chutes of known profile^ transit times were determined as described
previously from cross-correI ation transport delay times. Advantage
was taken of the availability of a hardware correlator as well as the
package programme (Section 10.5) to determine the cross-correI at ions

71
simultaneously using both procedures. Figure 10 illustrates a typica
experimentally derived impulse function for the hopper-chute system.

* /
The transport delay, dip and impulse peak are clearly evident. The
transport delay was taken to be the time elapsed to the centroid of
the impulse response as represented by the cross-correI at ion.
CENTROID OF IMPULSE
F i g u r e 10 Tr a n s p o r t delay from the

CROSS-CORRELATION RESPONSE
Table 3 summarises the comparison of descent time based on the
above interprefation for chutes of known form. The descent time for
’single gra i n s ’ is included to contrast it with the design flowrate
of 4 kg/min. For the ’single grain’ case the improvement in descent
times for the constrained chute is approximately \% compared to the
parabola and 2 1 compared to the straight chute. For the designed
flowrate of 4 kg/min the improvement in descent time calculated by
taking the average of the three measurement methods is 2 % compared
to the parabola and just under 5% compared to the straight chute.

72
It can be seen from the above that even at this small scale level
significant improvements in descent times are possible for the optimally
designed chutes when compared to the more commonly used chute forms.
Table 3: Comparison of chute profiles
DESCENT TIME BASED ON EXIT

4kgm/mi n VELOCITY m/s
Single Step Rxy Rxy Single 4kgm/
Gra in Response (package) (H.P.) Average Gra în mi n .
Straight Chute .719 .763 .721 .72 .735 3.1 2.7
Parabo1ic
Chute .712 .707 .721 .72 .716 2.7 2.4
Constra ined
Chute .704 .700 .707 .70 .702 2.7 2.3
4.4 Sensitivity Analysis of Model Parameters
Results could be relied upon with confidence if a sensitivity
analysis on the design parameters revealed that parameter variations
correspond to a prescribed change in chute performance. In order to
assess this performance variation each of the design parameters was
varied over a range that was considered practicable for the model.
Table 4 indicates the range of variation in parameters considered.
For millet seed, the limiting angle, 0f, is known to be approximately
60° and hence a variation in this parameter of ±5° was considered
adequate. Subsequently a variation of ±1° was used to assess any
influence on chute geometry resulting from the inability of measuring
this angle precisely. For this analysis and that of Section 4.5 the
±1° variation resulted in an insignificant change in chute geometry
and hence performance,and the results from the 60° case only are
reported here.
73
The assumed range for initial velocity (VQ ), height to breadth
ratio, (HB), and coefficient of viscous drag, (y), were considered
adequate to cover all likely combinations. The range in Coulomb
coefficient (t ) considered included Roberts' value of t = 0.464 based
on the approximate model of Figure 3 and Parlour's value of x = 0.306
based on the more precise model of Figure 4. In addition, for
comparative purposes, a value midway between the difference of the two
and Parlour's value (t = 0.35) was used.
For each analysis one variable was held constant whilst each
of the remaining variables was; in turn, varied through its
respective range. The results are presented in graphical form in
Figures 11 to 15. Since the chutes of parabolic form are known to
have favourable performance characteristics, parabolic profiles as
defined in Section 2.4.2 are included in each plot for comparative
purposes.
Table 4: Range of parameter variations in model
Limiting Angle (0f) 55° 60° 65°
Initial Velocity (VQ ) (m/s) 0.5 1.0 1 .5
Coulomb Coefficient (t ) .464 . .306 .35
Height to Breadth Ratio (H/B) 0.5 1 .0 1 -5
Coefficient of Viscous Drag (y) 0 0.5 1 .5
Some general observations can be made regarding the resulting
model chute geometries.
1. There is little change in chute geometry with changes in viscous
drag (y) between 0.5 and 1.5. However, this variation in y does
74
0. Q -I---------- 1---------- 1-----------i_______1_______i_______i_______i_______ i_______i______ i---------- i-----------i—

ii
- 0 . 1
- 0 2 -
-0- 3
AXIS (METRES)
-0. 4
-0. 5
- 0 6 .
- 0 . 7
- 0. 8
-0. 9
D-Ü 0-1 0-2 0.3 0.4 0.5 0.6 0.7
X AXIS (METRES)
Figure 11 Variation in limiting angle 0f

75
AXIS (METRES)
Figure 12 Variation in initial velocity

AXIS (METRES)
76
77
AXIS (METRES)
Figure 1*1 Variation in height to breadth ratio

78
Q . Q -4____ i____ 1____ i____ I____ i____ I____ i____ I____ i____ I____ i____ I____ i____ L
RXIS (METRES)
0.0 0- 1 0.2 0.3 0.4 0.5 0.6 0.7

X AXIS (METRES)
Figure 15 Variation in viscous drag (u )

79
have a significant effect on the exit velocity and on the descent
time. The maximum spread in exit velocity was 34$ while the
maximum spread in the descent time was 21$.
2. There are slight changes only in chute geometry with change in
limiting angle (0f) between 55° and 65°, together with only minor
variations in exit velocity and descent time. The spread in exit
velocity was 1$ and the spread in descent time was only 0.3%.
3. Large variations in chute geometries result from changes in inlet
velocity (Vo) becoming increasingly steeper or more concave with
an increase in inlet velocity. This effect is non-linear and
results in a spread in exit velocity of 1.8$ and a spread in
descent time of 16$.
4. Large variations in chute geometries result from changes in
Coulomb coefficient (t ) but with only a slight spread in exit
velocity and descent time. The spread in exit velocity was 6.5$
while the spread in descent time was 2.2$.
5. Large variations in chute geometries result from changes in
height to breadth ratio (HB) but are associated with only a slight
spread in exit velocity and descent time. The actual spread in
exit velocity was 2.8$ and the spread in descent time only 0.9$.
Confidence in these observations would obviously be increased
if they were seen to hold for larger scale chutes. While it was not
possible at this stage to carry out an experimental study of large scale
chutes, a theoretical analysis of such a scale up was considered
essentia I.
80
4.5 Theoretical Scale Up Analysis
Scale up analysis was considered by way of comparing the model
geometry with chutes whose outlets were (3 x 4) metres and (5 x 5)
metres away from their inlet, as shown in Figure 16. These
orientations were chosen firstly because it was felt they would be
realistic sizes for production chutes and secondly to contrast equal
leg with the unequal leg type chute geometry.
— .... r ........ i1 T m '

x
i
• — \—
1 X = 0,75 m
Y = 0.9 M
SCALE : 20 MM TO 1.0 M
X = 3.0 M
, Y = 4.0 M
X = 5.0 m
! Y = 5.0 m
f Y
F i g u r e 16 Sc a l e u p o f c h u t e g e o m e t r i e s
For each of these chute end conditions similar profiles were
generated as shown in Figures 17 and 18 and included a straight chute,
and a parabolic chute, and profiles for both the unconstrained and the
constrained optimised chute. Each of these chute profiles was in turn
subjected to a sensitivity analysis similar to the model chute using
+he same range of parameter variations as described in Table 4.

81
(METRES)
T RX1S
Figure 17 Chute profiles for (3x4) metre chute

(METRES)
RX IS 82
Figure 18 Chute profiles for (5x5) metre chute

83
Figures 19 to 28 detail, in graphical form, the variations in a
chute’s geometry that result from this sensitivity analysis. Again
the parabolic chute form is included in each plot for comparative
purposes.
Some general observations across the range of chutes considered
can be made with Table 5 summarising the variations of descent time
and exit velocity.
Table 5: Spread in parameter variation with

scale UP
PARAMETER DESCENT TIME EXIT VELOCITY
BEING (.75x.9)m (3x4)m (5x5)m (.75x.9)m (3x4)m (5x5)m
VARIED Chute Chute Chute Chute Chute Chute
fi % % % % %
Viscous Drag
Coefficient 17.6 33.4 46.1 34.2 60 15 A
Limiting
Angle .3 .18 .54 1 1 .6 .3
Inlet
VeIoci ty 16 8.8 7.4 1 .8 .28 .6
Coeff icient
of Coulomb
Friction 2.3 .72 1.6 6.5 .64 3.5
Height to
Breadth Ratio .92 .08 .24 2.8 .49 .69
1. With increasing scale up the coefficient of viscous drag (y)
increasingly affects the geometry of the chute particularly
with y > 1.0. The descent time variation increases from a spread
of 17.61 to a spread of 46.1 % for the (5 x 5) metre chute. At
the same time the spread of exit velocity increases from 34.2%
to 75.4%.
84
(METRES)
AXIS
F i g u r e 19 Va r i a t i o n in limiting angle 0f
FOR (3x4) METRE CHUTE
(METRES)
R XI 5
-4. 0
0. ü 1 .ü 2. D 3. ü
X RX I S (METRES)
Figure 20 Va r i a t i o n in i n i t i a l v e l o c i t y
86
(METRES)
AXIS
Figure 21 Variation in Coulomb coefficient (t )

87
AXIS (METRES)
Figure 22 Variation in height to breadth ratio

RX 1S (METRES)

89
AXIS (METRES)
Figure 24 Variation in limiting angle 0f

90
(METRES)
AXIS
Figure 25 Variation in initial velocity

91
(METRES)
AXIS
Figure 26 Variation in Coulomb coefficient (t )

(METRES)
AXIS
F igure 27 Variation in height to breadth ratio

93
(METRES)
AXIS

94
2. For changes in the limiting angle (0f) between 55° and 65° there
is little change in the geometry across the size range considered.
The descent time spread remains insignificant with a maximum of
0.5^, while the exit velocity spread has a maximum across the
range of 1.6$.
3. Changes in inlet velocity (VQ ) from 0.5 to 1.5 m/s across the
size range of the chutes have a decreasing effect on the geometry.
The descent time spread drops from 16$ to 7.4$ and the exit
velocity spread drops from 1.8$ to a minimum of 0.28$ for the
( 3 x 4 ) metre chute.
4. Changes in the coefficient of Coulomb friction (x) from 0.464 to
0.306 across the size range of the chutes have a decreasing effect
on the optimally constrained geometry. The descent time spread
drops from 2.3$ to a minimum of 0.72$ for the (3 x 4) metre chute
and the exit velocity spread drops from 6.5$ to a minimum of
0.64$ for the (3 x 4) metre chute.
5. Changes in the height to breadth ratio (HB) from 0.5 to 1.5
across the size range of the chutes have a decreasing effect on
the optimally constrained geometry with only a slight spread for
the (5 x 5) metre chute. The descent time spread is insignificant
across the whole range being always less than 1$. The exit
velocity spread drops from 2.8$ to a minimum of 0.49$ for the
( 3 x 4 ) metre chute.
4.6 Design Procedure
Consideration was given at this stage to the possibility of
formulating a set of design guides. This might take the form of
nomographs cross directing the designer through the design process
for chutes operating under uniform flow conditions in the TfastT flow
95
mode. Because of the large number of variables Involved, any such
scheme would need to represent a compromise in order to avoid making
the use of the nomographs unwieldly due to the sheer number of
reference points. In contrast the exact solution using the design
programme package is relatively straightforward. The required input
parameters are read into and depending on the computer used, the
execution time and cost are relatively small. In the case reported
here, the execution time of a complete sensitivity analysis, as
detailed in Section 4.4, was of the order of 5 minutes. The
programme package itself is programmed in the FORTRAN language and
could be readily implemented on any modest computer with a FORTRAN
comp i Ier.
For these reasons, perhaps the most satisfactory approach for
solving a particular chute problem is to use the programme package
with the complete sensitivity analysis. This provides the designer
with the complete range of possible solutions enabling him to choose
the solution that best fits his physical and environmental
constra ints.
4.7 Cone Ius ions
From the preceding analysis of the operation of the hopper-'
discharge chute under uniform flow conditions, the following general
observations can be made.
1. For hopper-discharge chutes operating under uniform flow
conditions the work of Roberts and Parlour has been extended,
to incorporate the limiting angle, 0f, in their design, resulting
in chute geometries suitable for stable ’fast’ flow operation.

96
2. The lumped parameter model has been found to be adequate for
analysing the uniform flow of millet through chutes subject to
generalised drag forces.
3. For the model, the generalised drag force included a viscous
drag component with a coefficient of approximately 2 (y t 2.0).
4. Even at the small scale tested the optimum chute geometry
exhibited favourable performance characteristics (minimum
transit time) whilst maintaining the desired 'fast’ flow
pattern compared to chutes commonly used. These incremental
improvements in performance represent potential operating cost
savings.
3. A design procedure in the form of a package programme has been
presented which enables specific solutions for chute design to
be obtained incorporating facilities for a sensitivity analysis
on all the significant design parameters.
Overall the investigation has shown the potential of discharge
chutes as flow controlling devices in the gravity flow of bulk
granular solids. For steady flow operation it is possible to design
chute profiles to achieve prescribed optimum performance by the use
of direct mathematical methods.

Part 2 : TRANSIENT FLOW
98
CHAPTER 5 : TRANSIENT FLOW
5.1 System Identification
The necessity of determining the transient performance of chutes
was alluded to in Section 2.1. The delineation of a mathematical
model of the chute’s dynamic behaviour would enable the prediction of
chute performance under forced operation of the flow control gate,
and highlight those parameters which affect the frequency response
of the chute.
Knowledge of the transient phase during blending operations
would enable one to determine what proportion of the mix is out of
specification and hence subject to waste or reprocessing. In a
multi-input blending operation knowledge of the longest transient time
constant would determine the appropriate earliest time to take a mix
sample for compliance to specification. If the blending operation
involves controI-vaIve-chute subsystems with widely differing time
constants consideration could be given to the timing order of
operation of these subsystems when changing product mix so as to
minimise the waste. Pressure for better quality control and increased
resource utilisation would seem to indicate a need for these
incremental gains in efficiency.
Experimental identification of the transient performance of the
hopper-discharge chute system was undertaken. Use was made of the
P .R.B.S./Cross-correI ation method for reasons given below.
The field of identification and process-parameter estimation
has developed rapidly during the past decade. In a survey paper;
Astrom (Reference 25) reviewed the state-of-the-art/science of system
identification schemes, stating that what was needed were
comparative tests performed on similar data. Isermann (Reference 26)

99
compared the performance, computation time and overall reliability
of six recursive identification and parameter estimation methods,
using three simulated processes. Sardis (Reference 27) compared the
computational and convergence properties of six popular on-line
parameter identification algorithms, by compiling and evaluating
the results of tests on two fourth-order discrete, dynamic systems.
In each case the results are similar and according to
Isermann (Reference 26, page 99)
MFor general linear processes, COR shows most
advantages ...; very good performance, shortest
computation time, 100 per cent overall reliability
with no problems with poor convergence or instability..."
COR here is defined as correlation analysis with least squares
parameter estimation. Whilst this paper presents a comprehensive
delineation of the advantages of the correlation method, Sastri
(Reference 28) draws attention to some of its disadvantages, and how
they may be overcome. In the practical application of the cross
correlation technique a number of errors can arise due to the
existence of imperfect input transducer characteristics, non-ideal
test input characteristics and wide band and output drift. His
findings are of importance in the selection of parameters as
described in Section 5.2.3 and are summarised as follows:
1. Make sure that the ratio of pseudo-random binary coded signal
bandwidth to that of the system bandwidth is of the order of
14:1 and is bias corrected.
2. Set the digit interval (AT) to one quarter of the smallest time
constant of interest.
3. Compute the cross-correI ation for at least two periods of the

100
test signal, and use the reference phase. Additional comparative
inferences were drawn from References 29 to 33.
Roberts (References 8, 21) showed, in a pilot study, that by
making use of the P.R.B.S./cross-correIation technique’s many
advantages, it appeared to be suitable for application to the bulk
handling field. In particular, it appeared suitable for use in
determining the transient flow behaviour of granular material
discharging from a hopper through a discharge chute. Support for its
application here was gained by its successful application to widely
diverse fields including impulse response testing of reactor systems,
identification of flotation plant dynamcis, and for process control
in the steel industry (References 34 to 39).
5.2 P,R.B.S./Cross-CorreIati on Method
5.2.1 Theoretical Consideration
While most aspects of the theory of random signal analysis and
system identification are well documented (References 40 to 42), these
are not generally in a form directly applicable to an experimental
programme. The salient aspects relating to the present experimental
investigation are outlined below as a self-contained coherent
presentat io n .
Figure 29 shows, schematically, the system under test, with
input signals x(t) and input noise n(t) and with an output signal or
response of y(t). The hopper-discharge chute is, in our case, the
system or equipment whose dynamic characteristics we seek. For the
present, it is assumed that the system is linear and deterministic.
The input signal x(t) is the test signal (P.R.B.S.) which is injected
into the system at some predetermined point (flow perturbing gate).
The signal n(t) represents an extraneous noise signal which could

INPUT NOISE
X (t) --
SYSTEM
INPUT
F igure 29 System identification model

102
arise in one of a number of ways. This includes via the normal
operating signal and any random noise signal originating from some
unknown source that appears as an input to the system.
Since it is assumed that the system is linear, it follows the
superposition principle and the response is unique for a given input.
The system response, given by the convolution integral is
„00
y(t) = f x (t - A) h(A) dA (5.1)
o
Where h(A) in 5.1 is the system’s impulse response or ’weighting
function’. The time variable X is often referred to as the age
variable. In the case of random signals, the statistical properties
are described by correlation functions. For the system identification
problem, two correlation functions are of importance, the auto
correlation function Rx x (t ) of the input signal and the cross
correlation function Rxy(t ) of the input and output signals x(t) and
y (t).
For the input signal x(t) the autocorreI at ion function is
expressed by
Rxx(x) = E {x (t ) x (t + x)} (5.2)
where E denotes the ’expected value’.
AIternateIy
Rx x (t) = Lim jJ x( + ) x(+ + T) d+ (5.3)

T-K» 2T -T
In a similar manner, the cross-correI ation function is expressed by
Rxy (T ) = E {x (t ) y (t )} (5.4)
103
Combining Equations 5.1, 5.2, 5.4 and simplifying this becomes
00
Rxy(x) = f Rxx^T — X ) h (X ) dA (5.5)
It is common in engineering applications to deal with discrete
signals of finite length. For such cases Equation 5.5 can be
expressed in digital form as

J
Rxy(j) = t R x x U “ i + 1) h ( ï) AA (5.6)
i= i
for j = 1, 2, 3 ... k where
k is norma I Iy <
10
n being the number of data values.
Equation 5.6 can be expressed in matrix form as follows
Rxy ~ [Rxx] h AA (5.7)
where
Rxy = cross-correIation column vector
[Rxx] = lower triangular autocorreI ation matrix vector
h = system weighting function column vector
AA = time interval between correlation estimates.
For any arbitrary random signal input x(t) one can identify the
system’s weighting function h(t) by solving the matrix equation
(see NOVA programme).
Rxv (5.8)
h = [Rxx] AX
If the input signal approximates ’white noise’*, often taken to
mean that the input signal bandwidth is very much greater than the
* Strictly, ’white noise’ is practically unattainable since it

implies constant power at all frequencies.
104
bandwidth of the system, the identification problem is greatly
simplified. The autocorrelation of a ’white noise’ signal is an
impulse function. Namely
Rxx = SQ 6 (t ) (5.9)
where S0 represents the constant value of the power spectral density
for that frequency. Substitution into Equation 5.5 and simplifying
leads to
RXy(T) = SQ h (t ) (5.10)
re-arrangi ng
h(x) = ^ (T) (5.11)
Thus the system's 'weighting function' can be obtained directly
from the cross-correI at ion function.
Figure 29 illustrates the addition of extraneous noise signals
n(t) and the respective 'weighting function' of the signal and the
noise. Here the cross-correlation is given by
,0 0 CO
Rxy(T) = £ Rxx(x - A) h (A) dA + £ Rxn (x - A) hn (A) dA
(5.12)
where Rxn(x “ is +he cross-correI at ion function relating the
signals x(t) and the noise input n(t). Now provided these two signals
are statistically independent, then their cross-correlation function
reduces to the product of their mean values. Namely,
Rxn<T - A) = x n (5.13)
Making this assumption and considering the input to approximate
'white noise’ Equation 5.12 simplies to
Rxy(T ) - So h (x ) + c (5.14)
105
where c is a constant depending on the mean values x and n.
Now if the mean value of either or both signals is zero then
this equation reduces to the simple form of Equation 5.10, since c
would then be zero.
Thus by ensuring that the input signal has a zero mean the
system’s impulse function can be determined in the presence of input
noise, that is uncorrelated with the test signal. The unsatisfactory
requirement of long averaging times for ’white noise’, makes it
generally unsuitable for practical identification studies. An
alternate signal, having the desirable characteristic of an auto
correlation function approximately an impulse, or delta function, is
the pseudo-random binary signal (P.R.B.S.). The Iiterature relating
to test signal selection is voluminous but the following were found
useful (References 43 to 55).
Figure 30 illustrates a P.R.B.S. which is readily generated
using a feedback shift register having simple plus-minus amplitude.
The mark-space switch occurs in a random manner at various multiples
of a clock period AT. In this case NAT the sequence time is 15AT
and the signal pattern repeats itself after this time interval.
The autocorrelation function and corresponding spectral density
of this P.R.B.S. are also shown in Figure 30. Leary (References 56,
57) and Brown (References 58 to 60) provide a good introduction into
the identification techniques based on the above formulation.
5.2.2 Hardware Implementation
Two hardware implementations were used in the experimental
investigation. Plate 7 shows the Hewlett Packard equipment package
comprising a 3722A noise generator, a 3721A correlator and 3720A

106
X(t)
P •R , B .S • ( N = 1 5 )
+ a
- a
Aï
1 SEQUENCE = SEQUENCE
R x x (t) N. ¿T REPEATS
Autocorrelation
P o w e r (log)
k
power spectrum
(sin x / x ) 2
3 dB
Shaped spectrum
1 >
1
1
1
1 log f
0.45 f f 2f 3f
c C c c
F igure 30 Te s t s i g n a l c h a r a c t e r i s t i c s
i CORRELATOR
SPECTRUM DISPLAY
(fissa P ■R •B IS I
fSpI NOISE
.u,J GENERATOR
• ' -.— V '
o fa fo ’o l [ -'—
^ J>
STRAIN GAUGE
BRIDGE AMPLIFIER
m w
X -Y RECORDER
Pl a t e 7 Id e n t i f i c a t i o n hardware
108
spectrum display. This enabled correlation tests to be carried out
quickly with an immediate result visible on the C.R.O. screens. This
feature is important avoiding time wasted on data reduction for an
experimental test run that was faulty, particularly when carrying out
tests to determine parameters to be used.
Use was made of a NOVA 1200 minicomputer to implement the
alternate identification scheme as shown in Figure 31. Here both
system input and output signals are sampled on-line via an analogue
to digital converter. The assembler language package, written for
the mini, enabled all the relevant correlation functions to be
calculated. Additionally, the power spectral density functions,
the system’s impulse function, the system’s frequency response and
direct convolution analysis could be calculated and plotted. The
latter is particularly useful for testing the accuracy of the
identified model, enabling the comparison of the actual system output
with that obtained by convoluting the system’s input with its impulse
function; in other words, the predicted output.
In order to impose the P.R.B.S. onto the perturbing gate a
power amplifier, Philips GM5535, and on electromechanical solenoid,
Philips PR9270 were used. The possibility of input contamination by
this power train was obviated by measuring the actual displacement
of the perturbing gate, with a linear displacement transducer,
H.P. 7DTDC, and using this as the P.R.B.S. input to the system in
preference to that obtained directly from the noise generator. This
pseudo-random binary displacement signal was checked to ensure that
its autocorreIation function had suitable delta function characteristics.

109
P iR •B iS ì
NOISE MILLET RETURN LINE
GENERATOR
E
::rC - r l v HOPPER
*• V
* v'MILLET--',
, ^ f • ,
POWE R >'^ .‘ ‘ / --
AMPL IFIER
^ s'
. - i - j 1 7 < , v *• - FLOW CONTROL VALVE
zn
PERTURBING / X
TRANSDUCER
STRAIN GAUGE
FLOW METER
TO NOVA
A/D CONVERTER
OUTPUT
SIGNAL
F igure 31 Test rig for NOVA identification

5.2.3 Parameter Selection
To obtain a reasonable estimate of the system ’weighting
function’, it is necessary to examine the effects of the various
parameters involved. In order that Equation 5.11 be applicable it
is necessary for the clock pulse (AT) of the P.R.B.S. generator to
be selected such that (1/AT) is approximately ten times the bandwidth
of the system. Since the system bandwidth is initially unknown,
it is necessary to establish, by systematic trials, the minimum value
of AT for a consistent RXy(T) to be obtained. Here the quick-look
facility of the H.P. correlator was extremely useful. There is a
confIicting practical Iimitation imposed by the frequency response
characteristics of the electromechanical perturbing gate system.
The sequence length, or P.R.B.S. period, NAT where N is the
number of clock pulses per period, has to be chosen in relation to
the 'settling time'* of the system. Exploratory investigations indicate
that a sequence length of about five times the longest time constant
of interest in the system is necessary. While longer sequence
lengths may be used, correspondingIy longer computation or
averaging times are necessary which may introduce errors, particularly
in cases where the system parameters vary with time.
The delay time increment (AX) between correlation estimates,
should be sufficiently small to permit minor variations in the
* The settling time is often defined as the time for a signal to

settle to within either ±2 or ±5$ of its steady state value
including any dead time.
’weighting function’ to be detected, but should not be greater than the
Nyguist period* in order to prevent errors due to aliasing (Reference
40). Experience in the hopper chute identification study indicated
that more reliable results are achieved when (AX) is set equal to
(AT).
The remaining parameter in the analysis is the selection of the
correlation experiment time (nAA), where n is the number of samples
taken. A sampling time of at least one sequence must be employed
(i.e., nAA = NAT) but in practice longer experiment times of
approximately two or three sequences are generally more satisfactory,
particularly where the signal to noise ratio is very poor. This may
occur where the perturbing amplitude has to be kept low, compared to
the normal operating signal in order to prevent out-of-specification
product material being produced whilst the plant is operating
normally.
5.3 Programme Development
5.3.1 Introduction
The necessity of having available a computer programme to perform
the large statistical data reductions is obvious. During the period
of this study the computing facilities available at the University of
Wollongong varied enormously. This development is, in part,
responsible for the programme development.
* The Nyguist or folding frequency is defined as (^y) anc* 's

maximum frequency that can be detected from data sampled at time
spacing AT (seconds).
112
The I.B.M. 1620 available initially severely limited the scope
of any statistical identification programme attempted. This was
soon complemented by a small mini, i.e., NOVA 1200 with 8K-16 bit
words of core. At about this time, the hardware correlator became
faulty and had to be returned to the manufacturer. These last two
facts provided a strong incentive to write an identification package
that duplicated the capabilities of the hardware correlator providing
a cross-reference and extend them to include system computations such
as convolution. Because of the smaI I core size and with a view to
the eventual use in a closed-loop control system, assembler language
was used. Just prior to completing this project a large Uni vac 1106
became available, which offered the ability of performing very large
data reductions because of its, by comparison, enormous core
storage (132K-36 bit words).
5.3.2 On-Line NOVA Package
An assembler language programme written for the NOVA line of
mini computers has been developed which enables the following
functions to be computed:
Rxx autocorreIation of the input signal.
Ryy autocorreI at ion of the output signal.
Rxy cross-correIation function.
Gx power spectral density of the input signal.
Gy power spectral density of the output signal.
Gxy cross-spectral density function.
H impulse function or ’weighting function'.
G (jco) Bode frequency response, magnitude and phase,
X predicted input by inverse convolution of the
model with the actual system output.

113
Y - predicted output by convolution of the model
with the actual system input. This enables
a specific control policy to be implemented.
Additionally, alI the necessary supporting and peripheral driver
software had to be developed. This package, of course, with its
versatility, has a utility beyond the present problem, evidenced
by its presentation at a computer conference (Reference 61). The
programme is arranged so that an operating system (Figure 32) can
call upon as many subroutines as there are specified tasks. Once
started, the operating system asks a series of questions in
hierarchial sequence to define all the subroutines that will be
called to perform the tasks.
To enable subroutine modifications to be made with a minimum of
re-assembling, the programme has been relocatably written.
Assembly language made possible the minimisation of the programme
size, so that the maximum amount of memory was available for data
storage, and was convenient for writing input-output operations
requiring a mixture of interrupt and non-interrupt routines.
Data input subroutines have been written for the analogue to
digital converter, the teletype reader, the high speed paper tape
reader and the cartrifile magnetic tape unit. The A to D subroutine
accepts analogue signals from any two, out of 16, consecutive
channels, representing the input to the system and the response from
it. At the initialisation stage the operator specifies the number
of samples to be taken and the sampling time interval. The former is
only limited by the core size, while the latter is limited by the
maximum frequency of the real time clock.

114
F igure 32 Generation of operating runstream

115
Since it was felt that trends could be more readily discerned
by reference to a plot rather than to a listing, a plotting routine
was written so that any of the results could be plotted on a
standard +eletypewriter unit. The programme scales the data
amplitude to limit the ordinate to six and a half inches across the
page. The abscissa representing either time of frequency
increments, can be any length, as the paper feed is in continuous
roll form.
Data collection and manipulation is handled by two subroutines.
Firstly, the data is loaded into arrays, counted, and the number of
correlations set at one tenth of this number. Then the unbiased
signal mean is calculated and the data normalised to zero mean to
simplify subsequent formulae and calculations, as detailed in
Section 5.2.1.
The statistical functions for signal analysis and system
computation are related according to the flow chart shown in
Figure 33 and have been programmed from the digitised versions of the
formulae derived in Section 5.2.1. Details can be found in Section
10.5.1.
In implementing an experimental programme based on the preceding
linear theory one is conscious of the need to adequately consider this
linear constraint. During the testing programme it was realised that
the chute system had non-linear effects due to the motion of the flow
control valve. In formally identifying the model, as detailed in the
next section, the experimental programme was designed to check this
out.
F igure
33
Op e r a t i n g
system
subroutine
hierarchy
ir
T
jk
System Computation Signal Analysis Data Manipulation Data Collection

116
117
CHAPTER 6 : CORRELATION TESTS
6.1 Genera I
An experimental programme was designed to use the P.R.B.S.
cross-correI ation technique to identify the system model, in this
case, the dynamics of the hopper-discharge chute system. Using
random perturbations of the hopper flow valve and the P.R.B.S.
cross-correI ation analysis the impulse flow responses for the system
under varying conditions of initial flow and chute shapes were
obtained. Information obtained indicated that certain flow non
linearities were associated with the motion of the flow control valve.
Techniques were developed to enable the examination of these flow non
linearities as well as enabling the determination of the dynamic flow
response characteristics during prescribed controlled operation of the
flow control valve.
An important adjunct was the development of the system
identification package which enabled prediction of system
performance under prescribed input conditions. Once the experimental
run had been completed and the data collected, the package also
enabled the inclusion of any necessary data manipulations, such as
digital filtering, the removal of any bias; and made the data itself
available in a form suitable for later evaluation, stored on magnetic
or paper tape.
6.2 Apparatus
Plate 7 shows the identification hardware which, together with
the apparatus and other ancillary equipment shown in Figure 1 and
Plate 1, were necessary to carry out the cross-correI ation tests.
This arrangement is similar to that used to determine the transit
times for the uniform flow case. The flow control gate was used to
set the required datum flow of material, while the perturbing gate
was actuated by the P.R.B.S. generator via the power amplifier and
the electromechanical solenoid. The perturbing gate input signal
(x (t )) was measured by a displacement transducer, while the flow
meter located near the end of the chute detected momentum changes of
the material in transit (y(t)). The cross-correI ation was obtained
for these two signals, RXy(i), using the above identification
hardware and also using the NOVA package as a cross-check.
Both the cross-correI at ion signal RXy(T) from the correlator and
by means of the complementary Spectrum Display unit, the power spectra
density GXy(f) were recorded graphically using an x-y recorder.
Because of the relatively good noise immunity of this identification
technique, the grain elevator could be operated in the closed circuit
mode whilst carrying out the identification without noticeably
affecting the results. This was of significance since the
experimentation times in some tests needed to be much longer than the
capacity of the hopper would allow. Also the maintenance of a
relatively constant head above the orifice was desirable to eliminate
any extraneous effects due to a low grain head.
The use of the NOVA identification package with the NOVA
situated in another building required special considerations. The
programme package’s data col lection needed to be co-ordinated with
the initiation of the P.R.B.S. testing. The ability of the programme
package to be remotely started facilitated this situation. This
ability of being able to programme the mini computer at what amounts
to the hardware level, assembler language, was seen as one of the
advantages of using the low level language.

119
The Noise Generator (H.P. 3722A) provided the requisite
pseudo-random binary signal. Each of the parameters sequence length
(N), clock period (AT) and amplitude (A) could be varied over a wide
range. This low energy signal was amplified in the Philips power
amplifier (GM 5535) before passing to the Philips solenoid (PR 9270).
This was capable of exerting a maximum force of 3.5 newtons per amp
(r.m.s.) on the perturbing valve.
The momentum signal y(t) measured by the strain gauge flowmeter
required a carrier amplifier for operation. For this a Tektronix
(3C66) dynamic bridge amplifier with a carrier frequency of 25kHz
was used.
The correlator (H.P. 3721A) had a bandwidth of D.C. to 250kHz
and enabled quick determination of correlation functions with 100
delay points being displayed at any one time and with a total delay
capability of 1000 delay points.
The complementary Spectrum Display Unit (H.P. 3720A) was
particularly useful enabling the frequency domain conversion of the
correlator results to be displayed in any combination of linear or
log. versus linear or log. scales. The frequency interval displayed
was directly tied to the selected correlation delay time and the
total range of frequency displayed was always two decades. Both the
correlator and the spectrum display unit were fitted with direct
signal outputs to drive an x-y recorder having an automatic pen-lift
faci Iity.
120
6.3 Procedure
Using the test rig shown in Figure 1, identification
investigations were performed for chutes of rectangular cross
section, of various straight and curved shapes. A wide range of
datum flows and straight chute s Iope angIes were incorporated in the
test programme.
Preliminary testing was necessary to ascertain appropriate
equipment parameters and operating procedures. In each case the
main flow control gate was preset to a fixed datum flow rate and the
perturbing gate was operated in a pseudo-random binary mode via the
power train described above.
The quick-look facility of the identification hardware enabled
repeated testing of a given chute configuration until one was
satisfied that the result was stable and repeatable. At this point
the discrete correlation function in dot form and the spectral function
in combinations of log. and linear scales were recorded. For many of
the tests, the final experimental run using the correlator package was
accompanied by a simultaneous analysis using the NOVA identification
package. This duplication made the results not only useful for cross
checking but also made them complementary; the NOVA Identification
package having system computation capabilities beyond those of the
correlator package.
Since it is recognised that the experimental identification
programme involved stepped movement of the flow control valve, some
insight into the flow characteristics associated with this valve motion
necessitated supplementary investigations. These took the form of step
responses and high speed cine photography.

6.4 Step Responses
Step response tests were performed on the hopper-discharge
chute system and the results compared to those obtained from the
cross-correI ation tests. In each case the perturbing gate was
stepped, open and shut, while the main flow control gate was open
allowing the rated throughput through the chute.
The apparatus as shown in Figure 1 was used to perform the
step responses. The perturbing gate system was activated by a low
frequency square wave generator, while the flow response was recorded
using a high frequency ultraviolet (U.V.) recorder. Both cases of
no main flow and with a main flow of 4 kg/min were tested. In order
to maximise the trace on the U.V. recorder in registering the step
responses, the ga Ivenometer dri ve amplifier was offset to ignore the
D.C. level of the main flow for that test. To improve the keeping
properties of the resulting U.V. record, it was sprayed with a fixing
solution.
The relative magnitude of the parameter signal compared to the
’grain noise’ signal necessitated the pre-filtering of the signal
from the strain gauge bridge amplifier to the U.V. recorder. This
filtering was achieved using a variable frequency Butterworth filter
(design details in Section 10.2) which provided maximally flat
filtering with a 40 dB/decade attenuation beyond the selected cut-off
frequency.
The results of this step testing will be contrasted in Section 7.4
with the predicted performance of hopper chute system based on its
ident if ied mode I.
In support of these results for the step testing, high speed

N)
K)
F i g u r e 34 Ca m e r a performance details
123
cine photography was used to observe the characteristics of the grain
flow through the bin orifice during the motion of the perturbing gate.
6.5 High Speed Cine Photography
A 16 mm Hycam high speed motion picture camera, model K2004E-230,
was used. The camera incorporates electronic speed control and has
a high speed rotating prism optical head to provide framing rates of
up to 11000 frames/second. Performance details are shown in Figure 34.
The stop motion projector (Seimens model 2000) was arranged as
shown in Figure 35. By adjusting the relative position of the projector
and mirror tilt angle, a correctly oriented magnified image could be
projected onto the glass top table. Since a reference grid was
included in all filming seguences, a specific magnification ratio
could readily be obtained, 1X, 2X, 4X being commonly used. As the
film was projected one frame at a time onto the glass top table a
sheet of tracing paper was used to plot successive positions of
selected grains for each time increment. The framing rate was
precisely known since the film edge had timing pips as detailed below
and thus the velocity of specific grains could be calculated.
F i g u r e 35 S ingle frame film analyser setup

124
Philips quartz halogen and agraphoto filament lamps were used
for illumination with approximately 4 kw being normally required.
To facilitate timing measurements the camera was fitted with
neon discharge tubes that produce timing marks on either side of the
film edge. A timing light generator to provide one lamp with 50Hz
pulses and the other with 1000Hz pulses was designed and built
(details in Section 10.3). Exposure settings were estimated using a
Weston Mark IV light meter of the reflective reading type. The more
appropriate spot type light meter was not available necessitating some
trial and error to obtain reasonable exposures using the Weston meter.
Using this information a series of tests were designed to
observe the grain stream at the orifice exit, at the chute exit and,
as discussed below, in a ’two-dimensional’ hopper. The film was
analysed using a number of devices including the stop-motion projector
mentioned above, as well as a 3M micro-film reader-printer (model 500)
which enabled magnified photocopies to be obtained suitable for later
analysis. Film spools of 30 and 120 metres length were used
depending on the framing rate required and the observation time sought.
For example, to photograph three complete cycles of step openings using
a framing rate of 1000 frames/second required approximately 110 metres
of film. Details of exposure selection and camera technique can be
found in Section 10.4.
The signal from the H.P. displacement transducer was used to
trigger one of the timing lights (described above) in the high speed
camera and yielded a stripe on the film edge corresponding to the
opening and shutting of the perturbing gate. This provided a timing
reference for the step responses.

125
The high speed cine photographic analysis of the flow through the
bin orifice during the perturbing gate motion confirmed the
observations of Parlour (Reference 14). The grain stream flow issuing
from a hopper is asymmetrically influenced by the sudden opening and
closing of the flow control gate. In order to lessen the effect of
the perturbing gate end section, it was sharpened as shown in
Figure 36 and its stroke was halved. The effect persisted and the
high speed film revealed that the grain profile flowing out of the
hopper orifice was still asymmetrical and linked.
PERTURBING GATE
AFTER SHARPENING
F igure 36 Pe r t u r b i n g gate e n d -s e c t i o n
MODIFICATION
Figure 37 and Plate 9 record the shadow profiles comprising the
effects of gate opening and closing.

Pl a t e 9 Gr a i n f l o w profiles showing
EFFECT OF VALVE PERTURBATIONS
127
FLOW CONTROL PERTURBING

VALVE VALVE
OPENING CLOSING
PERTURBATION PERTURBATION
F igure 37 Gr a i n f l o w s h a d o w p r o f i l e s s h o w i n g
EFFECT OF VALVE PERTURBATIONS
In an attempt to emphasise the difference between these two
perturbations a plot of stream thickness versus time has been
extracted from the grain profiles shown in Figure 37. The overshoot
accompanying a closing perturbation is clearly evident in Figure 38.

128
co
co
LU
LU
CC
I—
CO
Figure 38 St r e a m t h i c k n e s s t r a n s i e n t r e s p o n s e c h a r a c t e r i s t i c s
AT BIN OUTLET RESULTING FROM VALVE PERTURBATIONS
Further analysis will be deferred for the present being
continued in Section 7.5 after the correlation identification results
have been interpreted.
In order to pursue this asymmetrical behaviour and gain further
insight into its causes it was decided to construct a ’two-dimensional'
transparent hopper.
6.6 'Two-Dimensional' Transparent Hopper
In order to observe the flow pattern above and around the bin
orifice, a small scale, 70 mm front wall to back wall transparent
hopper, as shown in Plate 8, was constructed. The step opening of the
perturbing gate was activated by a 240 volt washing machine solenoid
and return spring system. A mixture of approximately 50$ ink
blackened and plain millet seed was used to aid in the photographic
analysis, providing a better contrast.
The testing programme included similar stepped opening and
shutting of the perturbing gate on the '2D' hopper as was used for the
129
Pl a t e 8 "Tw o -d i m e n s i o n a l " h o p p e r
main hopper-discharge chute system. Both cases of no main flow,
i.e., only the perturbation, and with main flow present were tested.
Using a reference grid behind the falling grain enabled the grain
paths to be plotted before, during and after a gate opening and
closing.
130
The analysis techniques used here were similar to those used to
analyse the grain flow from the main hopper orifice. in addition,
grain movement adjacent to the front perspex wall could be observed
and photographed. The shearing planes for various perturbing valve
positions were observed and velocity profiles along specified
reference axes were computed.
Depth-of-fieId is a function of stop and lens diameter. The
selection of the required framing rate using a particular film
imposes an upper limit to the depth-of-fieId for a given intensity
of illumination. The availability of approximately 4 kw of suitable
lighting necessitated a narrow depth of field. This often resulted
in the particular grain that was being plotted on the glass top table
disappearing from view into the body of the hopper, requiring the
sequence to be restarted with an adjacent grain; one of the most
frustrating aspects of this experimental investigation.
It is recognised that this narrow hopper distorts to some extent
the flow pattern within the funnel flow bin but this was considered
an acceptable compromise in order to gain additional insight into the
hopper’s influence on the chute’s dynamic performance. Having
recognised the importance of these related investigations to the
cross-correI ation identification scheme, their further analysis will
however be deferred until after the main experimental analysis; the
cross-correI ation test.

131
CHAPTER 7 : RESULTS AND DISCUSSION - TRANSIENT FLOW
7 •"I Transient Behaviour of Straight Inclined Chutes
In order to confirm the suitability of the cross-correI ation
method of process system identification as applied to the hopper-
chute system, a series of tests were performed on the model
equipment as outlined in Section 6.3. At this stage it is assumed that
the flow behaviour through the chute is quasi- 1inear. That is, for
small variations about the datum flow, the variation in flow is
Iinear Iy proportionaI to the position of the control valve.
It is assumed that any noise present is statistically
uncorrelated with the test signal so that the simpler Equation 5.14
for the cross-correIation applies.
Identification of the impulse function was performed for a range
of chute inclination angles for the straight chute each with eight
initial flow settings. This linear modelling in the time domain was
supplemented by a frequency domain analysis.
Assuming the process to be ergodic* the cross spectral density
Gxy(f) is obtained from the Fourier transform of the cross-correI at ion
function RXy(t ).
that is GXy(f) = /“ Rxy (T)e’J*(27Tf+)dt
Under the conditions embraced by Equation 5.14 in which
h(x) Rxy» +he system function is proportional to the cross spectral
density function, H(f) « GXy(f).
*|f the time average of a random signal is equal to the ensemble

average then the process is called ergodic. (Reference 42, p.87)
132
7.1.1 Impulse Function for Straight Inclined Chutes
Figures 39 to 41 show the impulse response or weighting function
h (t ) curves obtained by cross-correI ation analysis for the straight
inclined chute under datum flows ranging from 12mm to 57mm valve
opening corresponding to 1.6 kg/min to 10 kg/min flow rate respectively
for inclinations of 35°, 40° and 45°. Some general observations can be
made.
1. As the inclination of the chute with the vertical (0) increases
the descent time increases as indicated by the transport lags.
2. For all flow rates and inclinations tested the flow experiences
a ’dip’ or negative going pulse prior to the positive going pulse
normally anticipated for a linear system. The phenomenon is
apparently associated with the re-distribution of the grain paths
during the transient flow through the orifice of the bin and is
analysed in Section 7.5.
3. As the main flow is increased relative to a constant P.R.B.S.
step the response becomes more oscillatory indicating the
presence of non-linear characteristics of the transient flow.
4. For the 35° inclination with its relatively high flow velocity
the response shows more oscillatory behaviour for all flow rates.
5. As can be seen there is a variation in the form of the response
for a given inclination at various flow rates indicating non-linear
characteristics of the transient flow. However, as is also
apparent the pattern is substantially constant for small
variations in the preset flow. This last fact supports the quasi-
I inear assumption about a datum flow.

133
rp . -
in
8.9 kg/min.
l,Ll ,.*****.•••• ••• * * «« i •*•**» *4«l* •• m
•l#..**#**««*,
in
'•‘• / . i *• * •*, .«.* • t - • ••**»'*' *‘ *,#**#l**^l «Nil •**«•••*
6.4 kg/min.
Va l v e o p e n i n g m m
CO
A m p l i t u d e v /d i v
• **«• •••«*• « CO
CM
................................................................. ^ •• ••h ,
*• • ....... .
4.0 kg/min.
-«-•«• 4##. » -
m
h- CM
...... .. *•...+
or
^ •••• #
1.6 kg/min,
•'♦,**•* *»•*»
nj
• U«#*•<»,< /
r 1 I
0.0 0.25 0.5 1.0 1.25 1. 5
T ime ( t ) $e c ,
F i g ure 39 Im p u l s e f u n c t i o n c u r v e s for c h u t e i n c l i n a t i o n G - 35'

134
W + *««r1 ....... ..
r
in
8.9 kg/min.
I• ...... ....... ... in

# *
• »
».•
" *il*..*.*-'*”***».*4*“*s*. ... .. * ♦ « • « * ....'

. LD
# «
. 6.4 kg/min.
/d i v
mm
.... ••• .... 00
•»« ....
-iI fV)
opening
v
Am p l i t u d e
Va l v e
*'**%#*^##«%\**#»*#<#*4
CM
H1■«*<‘n ^ *nr‘,*t>»<'t‘><**» »**'*••' ro
f
• ** •• • « • • • * *
4.0 k g /min.
LD
V S ...* * ...... 04
CT>
* *./* ***.... .....«•»••».» ,...**•*** **..«.**.
1.6 kg/min.
•— ............#
r T
------- 1
0.0 0-25 0.5 0.75 1.0 1.25 1.5
T ime ( t ) $e c ,
F i g u r e AO Im p u l s e function curves for chute inclination 8 = A0°

135
..... r
4 •« in
8.9 k g / m i n .
* . . . ...»
in
** Bfc . . ^ • •• •• •• • •• /•' 4 in
6.4 kg/min
/d i v
mm
........ •4 •• . _ •. 4- CO
oo
opening
v
Am p l i t u d e
Va l v e
.*....... Osi
w. ••• 4*« H CO
4.0 kg/min
in
CN
J**»» *,•••*%•%•• 4»«f

as
•• * H*» •
1.6 k g / m i n .
%* ....s*•
. CM
r T
0.0 0.25 0 .5 0.75 1.0 1.25 1.5
T i m e (t ) S e c ,
F i g u r e 41 Im p u l s e function curves for chute inclination 0 = i|5°

136
6. All h( t ) response curves show the same constant transport delay
time for a given inclination which is associated with the time of
travel through the chutes. The approximate transit times for the
three inclinations 35°, 40° and 45° are 0.70, 0.77 and 0.89 seconds
respectively. In this respect the flow perturbation and cross
correlation analysis is a useful method for determining the average
velocity of travel through the chute system. The utilisation of
this fact for an open channel bulk granular material flow meter is
ind icated.
7. Since the flow through the bin orifice under opening and closing
perturbations of the valve has asymmetric characteristics, the
initial flow reduction depicted in the h(x) curves may be
accentuated by the closing perturbations of the valve when its
motion is controlled by the P.R.B.S. generator.
In order to facilitate comparisons between inclinations Figure 42
shows the h(x) records of the straight chute for a flow rate of
4.0 kg/min for inclinations of 35°, 40° and 45°. Clearly the pattern
of the impulse response is substantially the same. The varying
transport delay with increasing inclination is also clearly delineated.
7.1.2 Bode Diagrams for Straight Inclined Chutes
Additional information concerning the transient flow
characteristics may be gained from cross spectral density analysis,
under the conditions previously described (Section 7.1), where the
system function H(f) is a function of the cross spectral density
GXy (f). (H(f) oc Gxy (f) .)
In order to assess the effect of chute inclinations on the
frequency response of the hopper-discharge chute system use was made of

j___________ I________________________I___________ i___________ I
••• 0 = 45°
kg/min . flowrate
A m p l i t u d e v /d i v
• • * • ,
0 =40°
»•V
.............. I,
» • ..............Z’““
••
4.0
0 =35°
*,••••••••••• .•••j
r ---------- 1----------
1 ------------------------------------------------ j
0,0 0.25 0.5 0.75 1.0, 1.25 1.5

T ime ( t) Sec .
LaJ
FIGURE 42 Straight inclined chute (time domain)

138
the hardware correlator to obtain the Bode magnitude and phase versus
frequency plots for the straight inclined chute at inclinations to the
vertical of 35°, 40° and 45°.
Figure 43 illustrates the resulting Bode magnitude and Bode
phase plots.
1. The frequency response based on the -3dB point for each inclination
is essentially the same at 11 Hz . Using the low frequency level
as the zero asymptote the spread in the -3dB points was from
10.5Hz to 11.2Hz for inclinations of 45° to 35° respectively.
This means that step flow adjustments with a period of less than
91 msec (i.e., y p Hz) would be significantly attenuated.
2. As indicated the numerical Fourier transformation implemented
both in the hardware correlator and the identification package
tends to introduce end effect errors at the higher frequency end
of the curves. This results as a consequence of Shannon’s
Sampling Theorem (Reference 41, p .188) limiting the frequency
discernable for a given sampling time interval.
3. The attenuation corresponding to each inclination varies from
4.6dB to 7.5dB for inclinations of 45° and 40° respectively. These
attenuation interpretations are relative to the D.C. level of the
datum flow rate and are influenced by equipment parameter selection
particularly the strain gauge flow meter offset.
4. The phase versus frequency plot indicates that it is a direct
consequence of the delay lag which contributes a term in the
response of the form e^"*". Consequently this type of plot does not
contribute significantly to the understanding of the model dynamics
and will be dropped in the remaining Bode diagrams.

MAGNITUDE dB
4.0 k g /m i n .
FLOWRATE
-18Q
-360
LU
CO
<
X
Q_
-540
-720
-900
•5 5 50
LOG FREQUENCY Hz
FIGURE 43 St r a i g h t inclined chute (f r e q u e n c y domain)

140
5. Ignoring the end effects mentioned in 2. above, the high
frequency asymptote suggests a minimum phase system of about
fourth order.
The results above confirm that the P.R.B.S. cross-correI ation
method of process system identification can effectively be used to
determine the hopper-chute system model. Since it is knownthat the
valve motion produces asymmetric flow patterns as a result of its
disturbing influence on the granular particles falling through the bin
opening, the identified system weighting function needs to be used
with care. The system response to a prescribed movement of the control
valve can be predicted using this system model and convolution,
provided that the control valve movements are not so large as to be
influenced by the non-linear characteristics of the chute system. In
addition the transport lags inherent in the weighting function curves
provide a useful measure of the transit time and hence average velocity
of travel through the chute.
With the aid of spectral analysis it is possible to determine the
magnitude and phase characteristics of the system function H(f),
enabling both the system bandwidth and the behaviour within that
bandwidth to be analysed.
7.2 Transient Behaviour of Curved Chutes of Known Form
In order to compare the transient behaviour of the straight chute
with chutes of known curved profiles, cross-correI ation tests were
performed on the parabolic chute and the optimally generated chute
mentioned earlier. As was established in the uniform flow case the
paraDolic chute has favourable performance characteristics justifying
its inclusion in this comparison. Each of the three chutes were tested
with the same initial and end points.

7.2.1 Impulse Comparison of Straight Chute with Chutes of Known Form
Figure 44 shows a comparison of the impulse functions for chutes
of known curved form each operating at a datum flow rate of 4.0 kg/min.
Some general observations can be made.
1. The transport lags as depicted indicate that the optimal
chute has a marginally faster descent time, as expected
from the results on the uniform flow case.
2. The parabolic chute response is the most oscillatory;
not a particularly favourable characteristic with regard
to stab iIi ty.
3. All chutes exhibit the characteristic ’dip’ in flow
as mentioned for the straight chute case.
4. The optimum chute impulse response function shows the
most favourable rise-time performance with a smooth step
transition from the ’dip’ to the peak value; a
favourable characteristic for transient operation. In
blending this would minimise out of specification mix. It
appears that in generating a chute profile to minimise
the descent time under uniform flow conditions one obtains
a chute that performs well under transient conditions. This
is probably due to the fact that the design is based on
energy considerations constrained by boundary forces such
that lumpiness or grain stream bunching is avoided to
sustain stable ’fast’ flow as previously defined.
Clearly the optimum chute profile generated for minimum transit
time has favourable transient characteristics when compared to the
more commonly used straight and parabolic chute profiles.

STRAIGHT CHUTE (40°)
**v
4.0 kg /m i n . flo w r a t e
***t»«• * •*
•— « . ,1
,* / **•«•** ** , ,*• *' 1
/d i v
v
PARBOLIC CHUTE
Am p l i t u d e
•**•» .. „ i*•
ai••
.• § ,*
.... * •*
«• • ♦«•**•••**«*
OPTIMUM CHUTE
( t = 0.0)
• ><••• ***.,.,....•••.. V.../**
••• •
r ---------- 1---------- T 1
0, 0 0.25 0,5 0.75 1,0 1.25 1.5
T ime ( T) Sec .
FIGURE 44 Co m p a r i s o n of impulse response

NO
143
7.2.2 Bode Diagrams for Curved Chutes
Again recourse is made +o the cross spectral density analysis
to gain additional insight into the curved chute transient performance.
Figure 45 depicts a frequency domain comparison of the curved chute
performance characteristics. Some observations can be delineated for
a flow rate of 4.0 kg/min.
1. The parabolic chute has the highest bandwidth having an
upper frequency cut-off of approximately 14Hz. This is
consistent with its oscillatory impulse response function.
2. The straight chute has the least attenuation at low
frequencies up to approximately 4.0Hz being -3dB down at
4.5Hz and at 8.8Hz. The slight difference in characteristics
for the straight chute between Figures 43 to 45 are a
consequence of improvements in experimental technique
between the first results taken in 1973 and the later
ones taken in 1976.
3. The optimum chute has the highest relative attenuation
for low frequencies and the lowest frequency bandwidth
of 6.8Hz if the straight chute’s ’dip’ at 5Hz is
ignored. Th is is seen as a consequence of the
mobilisation of the available kinetic energy to achieve
the constrained geometry for minimum transit time.
Clearly the parabolic chute exhibits favourable
transient characteristics with the optimum chute not
significantly inferior. It should be remembered here
that the optimum chute had significantly better transit
time characteristics under uniform flow conditions.

144
STRAIGHT CHUTE (40°)

MAGNITUDE dB
F i g u r e 45 Frequency domain comparison of chutes

145
7.2.3 Comparison of the Optimum Chute at Varying Flow Rates
Recognising that the chute flow characteristics are non-linear
it was felt desirable to obtain impulse function responses for a range
of flow rates. Some observations from Figure 46 follow:-
1. For all flow rates tested the characteristic ’dip’ is
again present.
2. For low flow rates with consequent relatively high perturbing
flow rate compared to main flow the impulse function is of
characteristic delta form indicating a sharp rise-time with
little or no oscillatory components.
3. The descent time increases marginally with increasing flow
rate indicating the non-linear operation mentioned.
4. At the designed flow rate the impulse function
characteristics are still well behaved in terms of
rise-time and oscillatory characteristics.
5. With increase in flow rate beyond the design point
the impulse function exhibits increasing oscillatory
behaviour. This characteristic has a direct
consequence on the stability of the chute flow when
coupled with the usual requirement of variable flow
rate capability through the chute. It is suggested
that for stable flow the designed flow rate would need
to be the maximum required flow rate for the optimally
generated chute profile.

2
Am p l i t u d e v /d i v
F igure
46
Optimum chute
(
t
)
= 0.0
Va l v e opening mm
9V l
147
"7-2.4 Bode Diagrams for the Optimum Chute with Varying Flow Rate
Figure 47 illustrates the fluctuation in chute
bandwidth with increasing main flow. Some observations
1. The design flow rate corresponds to the highest attenuation
and the lowest freguency bandwidth. For flow rates of
1.6, 4.0, 6.4 and 8.4 kg/min the respective bandwidths are
12.6, 7.0, 11.1 and 9.2Hz.
2. For small flow rates, e.g., at 1.6 kg/min the bandwidth
is a maximum at a value of 12.6Hz.
3. At approximately ±50% of the designed flow rate the
bandwidth is similar and at least 50% higher than at
the design point.
4. The frequency response to the breakpoint is flattest
for the designed flow rate; a desirable characteristic.
The results above contrasting chutes of curved form with the
straight inclined chute show the straight chute to be inferior in
a number of respects. The optimum chute profile has the fastest transit
time and the more favourable impulse function characteristic when
compared to the parabolic and straight chute forms. The parabolic
chute has the highest bandwidth and the most oscillatory response
and is considered preferable to a straight chute under these conditions.
7.3 Step Responses
It is well recognised that additional insight into the flow
transient behaviour not obvious from the h(x) curves, can be gained by
integrating the impulse responses to obtain the equivalent step
response. Rise-time, overshoot, and oscillatory responses become
readily evident. From a practical point of view, process changes in

148
MAGNITUDE dB
F igure 47 Fr e q u e n c y d o m a i n r e s u l t s for o p t i m u m c hut e

149
flow would normally be implemented by stepping the control valve to
the new position. Furthermore, experimental testing programmes
normally include step response testing, which by means of convolution,
can be the basis of system performance prediction.
7 *3 -1 Step Response for the Straight Chute
Figures 48 to 53 represent the integral of the impulse functions
for the straight chute, cross-comparing chute inclinations with initial
flow settings. These cross comparisons are particularly useful in
highlighting similarities and differences not immediately apparent
in the h(T) system weighting function curves.
In Figure 48 for a chute inclination of 0 = 35° at a 12mm datum
flow setting the step response indicates some initial oscillatory
tendencies. At a flow rate corresponding to a height to breadth
ratio of one, with a 25mm datum flow setting, the step response
shows a favourable rise-time characteristic with little oscillatory
tendency. The 38mm datum flow is relatively sluggish and would
represent significant out-of-specification mix.
In Figure 49 for a chute inclination of 0 = 40° the general
trends observed above again apply; the ’dip’ is much more pronounced
and further to the right indicating the longer transit time. The
significant difference being that the 38mm datum flow rate now
exhibits extremely oscillatory behaviour, and could lead to unstable
flow patterns. Again the 25mm datum flow rate corresponds to the
steepest rise-time with an attractive step response.
Figure 50 shows that for a chute inclination of 0 = 45° the
12mm datum flow has the most attractive step characteristic. The
25mm datum flow corresponds to minimum descent time but the levelling
NORMALISED AMPLITUDE
Figure 48 Step response straight chute 0 = 35°

F i g u r e 49
I I I 1 > I I ¡ I i l l 1 I » I 1 I 1 I i 1 f I « 1 I «- » J I ^ I 1 1 1 I f
= 38 M M
r~ n ~ r- T - r » r i i i i m t i -t t t t t m i i i T T T 't i r r n i t i n —
0.8 0.9 1.0 1.1
TIME ID SEC.
Step response straight chute 0 = 40°

U1
N3
Figure 50 Step response straight chute 0 = 45°

NORMALISED RMPLI7U0E
153
F igure 51 Step r e s p o n s e s t r a i g h t c h u t e 12 mm fl o w
NORMALISED RMPLITUDE
154
Figure 52 Step response straight chute 25 mm flow
Oí
Figure 53 Step response straight chute 38 mm flow

156
off half-way to the step indicates initiation of oscillatory
influences in the impulse response. The 38mm datum flow rate was
particularly troublesome and required extensive prefiltering to
reduce the very large oscillatory behaviour and should not strictly
be compared to the rest of these diagrams.
In Figure 51 for variable inclinations at an initial datum flow
rate of 12mm, the increasing prominence of the 'dip’ is clearly evident
for increasing chute inclination. For all inclinations the responses
indicate oscillatory behaviour which would not usually be very
attractive for design considerations.
In Figure 52 for an initial datum flow rate of 25mm the
increasing time delay with increasing chute inclination is clearly
indicated. The response for the 0 = 45° inclination indicates onset
of increasing oscillatory behaviour at mid step.
In Figure 53 for an initial datum flow rate of 38mm the
increasing oscillatory behaviour is evident and has implied
consequences in designing chutes with height to breadth ratios greater
than one. It should be noted that the curve for 45° inclination is the
same one as was mentioned earlier and because of the necessary pre
filtering should not really be considered here.
The results above clearly indicate the desirability of operating
straight chutes under transient conditions with a datum flow rate
corresponding to a height-to-breadth ratio of no more than one. For
a height-to-breadth ratio of 1.5 the step response was increasingly
oscillatory with increasing inclination angle, 0.

157
These equivalent step responses complement the impulse response
h(x) curves and transient characteristics such as rise-time, overshoot,
the ’dip’, and oscillatory behaviour becomes readily apparent. It is
suggested that any subsequent work in this area, or in other areas
using the P.R.B.S./cross-correIation identification technique, shouId
consider this type of equivalent step response. It is calculated by
integrating the system’s weighting function and provides a lucid
delineation of the system’s transient characteristics.
7*3*2 Step Responses for Chutes of Known Form
In Figure 54 step responses for curved chutes of known form as
compared previously are contrasted. The overall performance of the
optimum chute is clearly superior. It has the smoothest, steepest,
rise-time with the minimum time delay.
The rise-time measured as defined in Figure 6 for the optimum
chute is half that for the straight chute inclined such that the
end conditions are at the same co-ordinates. The oscillatory
behaviour of the parabolic chute and the sluggishness of the straight
chute are again evident.
7.3.3 Step Response for the Optimum Chute
The step responses shown in Figure 55 compare the optimum chute’s
performance for various initial flow settings. The increasing time
delay with increasing flow rate indicates the non-linear characteristics
of this chute profile. Significantly, the least oscillatory response
corresponds to the designed flow rate condition of 25mm datum flow
setting or 4.0 kg/min. The designed flow rate exhibits the minimum
’dip’ and minimum overshoot; cI ear Iy favourable criteria for a system
subjected to transient operation. The 12mm datum flow rate,

158
NORMRLISED AMPLITUDE
F igure 54 St ep r e s p o n s e c h u t e s of know n fo r m
15^
F i g u r e 55 Step response optimum chute

160
corresponding to -50% of the designed criterion has the next most
favourable rise-time and oscillatory characteristics. The increasing
oscillatorv behaviour with increasing initial datum flow setting
again implies that for 'fast* stable flow the design point ought to
be near the maximum anticipated capacity of the chute.
The results above for the equivalent step responses highlight
the usefulness of this technique and support the earlier conclusions
comparing chutes of known geometric form. The straight chute’s
oscillatorv behaviour when operating with a datum flow corresponding
to a heicht-to-breadth ratio greater than one has direct consequences
for the designer who wishes the chute to operate significantly in a
transient mode.
The overall performance of the optimum chute was clearly
superior under these transient conditions and, considered together
with its s ¡perior performance in achieving minimum transit time
under stable ’fast’ flow conditions provides the designer with a
clear ’benchmark’ or ’yardstick' for his chute design.
7.4 Predicting Flow T rans ients
As outlined in Section 5.2.1 the cross-correI ation function or
impulse function represents the dynamic model of the system under
test, in this case, t h e hopper-discharge chute system. At this stage
it would be informative to compare the actual step responses with those
obtainable by c o n v o l u t i o n with the derived system model.
Althouoh tne hopper-discharge chute system behaviour is strictly
non-linear in as much as the impulse response functions show varying
characteristics for variations in the datum flow, the assumption of
linearity for small disturbances about a particular control point is

161
quite satisfactory. Based on this assumption it is possible, via
linear systems theory, to compute the response of the system to any
prescribed motion of the flow controlling valve. This can be achieved
by using Equation A5.14 repeated here
y (0) = (x(0) H (0))AX
x(0) H(1) + x(1) H (0)

y(1 ) = AX
x(0) H(,j) + x(,j) H (0) j-1 '

y(j) = 2
+ Z x(i) H (j— 1) AX
i— 1
(A5.14)
j - 2, 3, ... m - 1
repeated
In the case of a step input x(j) will be constant throughout.
Using Equation A5.14 the computed step response for the straight 40°
chute is shown in Figure 56. It compares favourably with the measured
step response obtained experimentally as detailed in Section 6.4 and is
also shown in the same figure. The ’dip’ and the rise-time
characteristics are clearly evident.
A significant aspect of the flow behaviour is that a sudden
opening of the valve, causes an initial reduction in flow, following
the transport lag, before the flow finally increases to its new steady
state value.
It is possible then, using convolution techniques, to compute
the response to any input. The corollary, for predicting the required
input to yield a desired output, inverse convolution, has proved a
little more difficult to implement but offers certain attractions.
The numerical difficulties are associated with the use of Equation A5.15
repeated here
r J ’ 1 -
1
x(j) 2 H(j-k) x(k) (A5.15)
H(0)
AA k=0 repeated
j = 0, 1, ... m-1 .
162
F igure 56 Co m p a r i s o n of c o m p u t e d a n d m e a s u r e d step
RESPONSES FOR STRAIGHT INCLINED CHUTE
163
When H(0) is very small or zero the predicted input approaches
infinity; obviously an impractical possibility. Using Equation A3.15
and the impulse function model for the optimum chute for the
designed flow rate of 4.0 kg/min, the required control valve motion
to produce a step output response was computed.
Figure 57 shows the predicted valve motion to achieve a unit
step increase in flow above a given datum flow rate. The curve
shows that the control valve needs to be stepped open 60 milliseconds
before the required step response by an initial amount equal to
approximately three times this required step and then be closed down
to the target unit step according to the position-time history shown.
This ramping action occurs during a period of 210 milliseconds. The
shutting transient gives rise to a momentary increase in flow which is
compensated by the slight overshoot in the control valve position
shown before settling down to the new steady state value
corresponding to the unit step increase in flow rate.
Linear systems theory in the form of the convolution integral
has been shown to satisfactorily predict the transient performance
of the hopper-discharge chute system. If large step changes above a
datum flow rate are contemplated the technique needs to be used
with care as a quasi-linear system model is assumed. Within the
framework of this limitation however, the prediction of the transient
behaviour of the hopper chute system enables the controlled blending
of granular material to be effected 'with certainty'; out-of
specification product being clearly delineated.
It is possible now, using convolution techniques, to compute
the response of the system to any given input. While inverse
convolution techniques enable the required input to achieve a desired

164
---------------- S --------------------- — r~*—

• • .»*
........ ”
•
•
1 4 1 1
.1 ’.2 .3 .¿I
• ELAPSED TIME (SEC)
•
•
ft
•
•
'**••••• ft------------------------------------- ------
•ft •
FLOW CONTROL VALVE TIME-HISTORY
HOPPER
ORIFICE
INITIAL
DATUM PREDICTED
INITIAL
OPENING
TARGET
DATUM
PERTURBING VALVE MOVEMENT
Figure 57 Predicted flow transient

165
response to be determined. Further, should optimal control be the
objective, it is possible to achieve this by establishing an
appropriate performance index based on an error signal representing
the difference between the desired and the actual hopper chute
response. This performance index would then be maximised or
minimised as aDpropriate to the given problem formulation.
7.5 Flow Non-Linearities
"7.5.1 Bin Outlet Flow Resulting from Valve Perturbations
As discussed in Section 6.5 high speed cine photography was
used to examine the effects of the perturbing valve operation on the
grain flow from the bin orifice. Figure 37 and Plate 9 depict the
differences in grain stream profiles resulting from an opening
perturbing valve motion and a closing motion. For the opening
perturbation, that is a sudden increased opening of the valve, the
grain stream increased almost uniformly to the new steady state
value corresponding to the greater flow. However, for the closing
perturbation, that is a sudden decrease in the datum setting of the
valve, the grain stream is deflected or bent as shown. At the same
time the stream thickness in reducing to the new steady state value
has a small ’overshoot', in that the thickness first decreases to a
minimum value before the steady state condition is reached.
In an attempt to identify the cause of this asymmetrical
behaviour the perturbing valve end section was ground down to a fine
edge as shown in Figure 36 so that it presented the least possible
across-the-stream surface area. Additionally the perturbing stroke
was halved to a value of ±2mm of movement. Retesting revealed that the
asymmetry persisted, and is probably due to the re-organisation of the
grain forces within the hopper and discharge orifice resulting from
166
the horizontal shearing effects of the perturbing valve motion.
In order to pursue this asymmetric effect further a ’two
dimensional’ transparent hopper was constructed and the results are
discussed below.
In Figure 58 the positions of individual grains falling from
the bin have been plotted as described in Sections 6.5 for typical
opening and closing perturbations of the valve. Representative flow
contours at the bin orifice are plotted for the grain stream flow
immediately following the valve movement with successive positional
contours spaced five milliseconds apart.
From grain path records such as these grain particle velocities
at the bin orifice have been computed and four sets of velocity
profiles are shown in Figure 59. The four flow regimes depicted
(top to bottom) are:
1. Steady state flow, with the perturbing valve in innermost
position.
2. Transient flow immediately after the perturbing valve is moved
to the outermost position.
3. Steady state flow, with the perturbing valve in the outermost
pos it io n .
4. Transient flow immediately after the perturbing valve is moved
to the innermost position.
For each flow regime two profiles are shown, the short dashed
line profile representing the velocity distribution across the stream
5 milliseconds after exit from the bin outlet and the long dashed line
profile representing the velocity distribution across the stream after
falling for 20 milliseconds from the bin outlet. Some general

167
opening CLOSING
PERTURBATION PERTURBATION
F igure 58 Grain path s a nd flow p r o f i l e s at b in o u t l e t

RESULTING FROM VALVE PERTURBATIONS
168
.8
Q
LU
cn z
o LU
VELOCITY METRES/SEC
VELOCITY PROFILE AT EXIT

------- + 5 M S E C
---- +20 MSEC
VALVE MOTION
F igure 59 Grain s t r e a m v e l o c i t y p r o f i l e s at bin o u t l e t

169
characteristics are indicated.
1. For steady state flow the velocity profiles are symmetrical
whereas for the transient flow phases following the perturbing
valve movements, the velocity profiles are asymmetric.
2. The velocity variation across the stream is more marked for
the 5 mi IIi second reference Ii nes than for the 20 mi I Ii second
reference lines. In the latter case the profiles have much
’flatter’ characteristics as a result of the grains becoming
increasingly influenced by gravity and air resistance.
3. The average grain stream velocity is approximately 1 metre/sec
for all cases after the grains have fallen for 20 milliseoncds
with the control valve in one of its two datum positions.
4. Immediately following the closing perturbation the velocity
of the far side grain stream is approximately ]0% higher than
the near side grain stream, indicating a slight increase in
flow during the shutting transient.
High speed cine photography has enabled the behaviour of the
grain flow from the bin orifice to be determined before, during and
after a perturbing valve movement. The grain stream velocities
accompanying this perturbing valve movement have been clearly
deI ineated.
7.5.2 ’Two-Dimensional’ Hopper Results
As mentioned previously, the problem of the asymmetric
characteristics of the flow during perturbations of the flow control
valve were pursued by using a so-called ’two-dimensional’ hopper. It
is recognised that the narrow funnel bin contemplated introduces flow
distortions not experienced in a full depth hopper. However, it was

170
felt that any additional insight into the flow patterns immediately
above, through and out of the hopper would be of benefit in
interpreting the transient characteristics of the hopper orifice.
Visual inspection of the high speed film confirmed that the
valve motion influenced the grain flow well into the body of the
model hopper. After a transient perturbation this influence causes
the body flow velocity to momentarily oscillate. This was alluded to
by Parlour (Reference 14). Initially a closing perturbation causes
the body flow to stop completely for approximately 24 milliseconds at
a cross-section 50mm above the orifice and then to oscillate briefly.
Figure 60 shows a full size representation of the shear planes,
illustrating the shift in the right hand shear plane when the
perturbing valve assumes its two steady state positions. Relative
slip occurs between the two pairs of lines associated with the same
datum point at the orifice. As for the case discussed in Section 7.5.1,
the path traced by the individual grains was plotted for successive
frames and from these the velocity profiles across nominated horizontal
sections within the ’2-D’ hopper were calculated.
Four flow regimes are considered:
1. Before closing - with the perturbing valve in the open position.
2. At closing - with the perturbing valve moving to the closed
pos it io n .
3. After closing - with the perturbing valve in the closed position.
4. At opening - with the perturbing valve moving to the open position.
Figure 61 shows the velocity profiles at a cross-section 100mm
above the bin orifice. The approximate uniform velocity profile before
and after perturbation is clearly evident. The disturbance to this
grain movement due to valve perturbations shows a slight asymmetry

F igure 60 Sl i p planes for '2-D' model hopper
172
o
in
m m /sec
o
o
o
m
□ BEFORE CLOSING
O AT CLOSING
> AFTER CLOSING
O AT OPENING
VALVE OPEN
VALVE CLOSED
F igure 61 V e l oc ity p r of iles 100 mm a b o v e o r i f i c e

173
+o the right hand side or perturbing valve side. However, the
disturbances produce essentially complementary influences.
Figure 62 shows the velocity profiles at a cross-section 50mm
above the bin orifice and the effect of the valve motion is becoming
more pronounced, resulting in a drop in the average centre line
velocity of 44$ of the pre-perturbation centre line velocity. The
steady state velocity profile with the perturbing valve closed now
shows a flatter profile.
Figure 63 shows three of the velocity profiles at the bin
orifice to be essentially flat with a slight asymmetric influence on
the perturbing valve side, indicating that there is little relative
movement between grains across the stream flow. This is expected as
the grains are leaving the influence of the hopper and are becoming
predominantly influenced by gravity and air resistance. The
velocity profile corresponding to the closing perturbation has a
centre line velocity of 0.75 metres/second. For this the grain
stream nearest the perturbing valve experiences a drop in velocity
of 18$ while the far side grain stream velocity drops only 7$,
compared to the centre line velocity. It should be noted that the
shutting perturbation is again accompanied by a slight increase in
the average grain stream velocity. At the centre line this increase
is approximately 7$ of the pre-perturbation centre line velocity. At
opening the slight drop in grain stream velocity nearest the
perturbing valve is clearly evident and supports the description of
the 'dip1 in flow resulting from an opening step.
Taking Figures 61 to 63 together one can observe the increase
in velocity from the 100mm point to the bin orifice within the shearing
section of the body of the hopper. The average centre line velocities
being approximately 0.13, 0.27, and 0.7 metres/sec respectively.

174
o
O
m m /se c
O
O
CN
O
O
co
2 X FULL SIZE lu
UJ
(X
LU
O
VALVE OPEN
VALVE CLOSED
□ BEFORE CLOSING
O AT CLOSING
> AFTER CLOSING
O AT OPENING
F igure 62 Ve l o c i t y profiles 50 mm above orifice

17e)
LU
FULL SIZE LU
cr
LU
U
O
O
m
o
o
LD
m m / sec
C] O
O
r-
o
o
oo
□ BEFORE CLOSING
o AT CLOSING
A AFTER CLOSING
0 AT OPENING
F igure 63 Ve l o c i t y profiles at chute orifice

176
The !2-DT hopper results have shown that the flow behaviour
below the bin orifice is influenced by grain force redistribut ion
above the bin orifice during perturbing valve movement. High speed
film analysis is an effective technique for identifying the grain
flow patterns associated with the control valve movement. The
necessary film analysis to plot individual grain paths is time
consuming but not difficult.
Overall the transient analysis has confirmed that the hopper
discharge chute system can be modelled using quasi-l inear systems
theory. The influence of the non-linear characteristics of the flow
control valve can be minimised by considering small perturbations
about the datum flow. The P.R.B.S./cross-correI at ion identification
scheme implemented effectively ensured this small perturbation about
the operating point was used. With this constraint,Iinear systems
theory has effectively identified the hopper-discharge chute model.
By using convolution techniques the chute performance under
prescribed transient conditions has been successfully predicted.

177
CHAPTER 8 : SUMMARY OF CONCLUSIONS
Un iform FIow
The experimental investigation of the performance of hopper
discharge chutes operating in the ’fast’ flow mode under steady
uniform flow conditions has delineated the following:-
1. The work of earlier researchers, particularly Roberts and Parlour,
has been extended. The design of chutes incorporating the
limiting angle, 0f, has resulted in chute geometries suitable for
’fast’ stable optimum flow operation.
2. The lumped parameter model has been found to be adequate for
analysing the uniform flow of millet seed through chutes subject
to generalised drag forces.
3. For the model, the generalised drag force included a viscous
drag component with a coefficient of approximately 2 (y = 2.0).
4. Optimisation based on minimising the descent time and maximising
the exit velocity in a prescribed direction has been formulated
using the lumped parameter model. The design of chutes to
minimise the descent time has been solved.
5. The combination of Charlton’s polynomial approximation method and
the optimising algorithm of Fletcher and Powell has resulted in
an optimisation computer programme that readily solves the
minimisation formulation providing a design procedure incorporating
a sensitivity analysis and all the significant design parameters.
6. Even at the small scale tested the optimum chute geometry exhibited
favourable performance characteristics (minimum transit time) whilst
maintaining the desired ’fast' flow pattern compared to chutes

178
commonly used. In terms of the minimisation of the descent time
the chute profiles were ranked in their order of merit;
optimum, parabolic and straight inclined chutes. These
incremental improvements in performance represent potential
operating cost savings.
7. Overall the investigation has shown the potential of discharge
chutes as flow controlling devices in the gravity flow of bulk
granular solids. For steady flow operation it is possible to
design chute profiles to achieve prescribed optimum performance
by the use of direct mathematical methods.
Transient Flow
The experimental programme designed to identify the transient
characteristics of the hopper-discharge chute system disclosed the
foI lowing:-
1. The invesitation supports the results of Robert’s pilot study
indicating that the P.R.B.S./cross-correI at ion method of process
system identification can be effectively applied to the hopper
discharge system model.
2. Although the transient flow shows minor non-linear characteristics,
the results indicate that an assumed linear or quasi-I inear
formulation is satisfactory for reasonably small movements of the
flow control valve.
3. With the aid of spectral analysis it is possible to determine the
magnitude and phase characteristics of the system function, H(f),
enabling both the system bandwidth and the behaviour within that
bandwidth to be determined.
179
4. The transport lags inherent in the chute’s weighting function
curves provide a useful measure of the material transit time.
5. The optimum chute impulse response function shows the most
favourable rise-time performance with a smooth step transition
from the ’dip’ to the peak value; a favourable characteristic
for transient operation. In bI ending,this would minimise out-
of-specification product. It appears that in generating a chute
profile to minimise the descent time under uniform flow
conditions one obtains a chute that performs well under transient
conditions. This is probably due to the fact that the design is
based on energy considerations constrained by boundary forces
such that lumpiness or grain stream bunching is avoided to
sustain stable ’fast’ flow.
6. The transient results indicate that straight chutes should not be
used with a height to breadth ratio (as previously defined) greater
than one for stable transient operation.
7. Linear systems theory in the form of the convolution integral was
used to predict successfully the transient performance of the
hopper-discharge chute system. If large step changes above a
datum flow rate are contemplated the technique needs to be used
with care as a quasi-linear model is assumed. Within this framework,
however, knowledge of the transient behaviour of the hopper-chute
system during controlled blending of granular material enables an
operator to work ’with certainty’, minimising out-of-specification
product.
It is possible then,using convolution techniques, to compute the
response of the hopper-chute system to any given flow control valve

180
movement to achieve a desired flow can be determined. Further,
should optimal control be the objective, it is possible to achieve
this by establishing an appropriate performance index based on
an error signal representing the difference between the desired
and the actual hopper-chute response.
8. The transient results generally support those of the uniform
flow case in ranking the chutes in the same order of merit;
optimum, parabolic and straight inclined chute, particularly when
it is recognised that the optimum chute had significantly better
transit time characteristics under uniform flow conditions. On
the other hand, the results clearly show that the optimum chute
needs to be operated at no more than the designed flow rate for
stable flow.
9. High speed cine photography of the transient behaviour of the flow
control valve generally supported the assumption of a quasi-linear
model for the system identification indicating that the non-linearity
was relatively minor and the use of the linear systems theory
justified. This analysis revealed that the non-linearity was
not due to the perturbing valve end section but rather to the
dynamic reorganisation of grain forces within the body of the
hopper.
10. The overall performance of the optimum chute was clearly superior
under these transient conditions and considered together with its
superior performance in achieving minimum transit time under stable
’fast1 flow conditions provides the designer with a clear
’benchmark’ or ’yardstick’ for his chute design.

181
Suggestions for Further Work
The following general areas emerged as possible extensions
to this study and would broaden the scope for application of the
present findings:
1. An experimental investigation on large scale chutes ought to
be conducted to verify that the theoretical scale up analysis
presented is supported.
2. An investigation based on a distributed parameter model may
result in an analytical solution to the transient behaviour of
the hopper-chute system.
3. An experimental investigation using cohesive materials may
reveal that only minor modifications to the present model would
enable the designer to achieve optimum chute profiles for
cohesive materials.
182
CHAPTER 9 : REFERENCES AND BIBLIOGRAPHY
1. HAWK, M.C. (Ed.) ’’Bulk Materials Handling - Vol.1".
University of Pittsburg 1971.
2. Official "Year Book" of Australia 1974, No.60.
CommonweaIth of Australia 1975.
3. ROBERTS, A.W. "The Dynamics of Granular Material Flow Through
Curved Chutes".
Mech. & Chem. Eng. Transactions, I.E.Aust., Vol. MC3, No. 2 (1967).
4. ROBERTS, A.W. et al. "Bulk Handling of Granular Material -
Some Recent Developments".
National Agricultural Machinery Conference Melbourne
July 14-15, 1971.
5. CHARLTON, W.H. and ROBERTS, A.W. "Gravity Flow of Granular
Materials: Analysis of Particle Transit Time". '
A.S.M.E. Paper 72-MH-33 Meeting Sept. 17-20, 1972.
6. ROBERTS, A.W. and ARNOLD, P.C. "Discharge-Chute Design for
Free-Flowing Granular Materials".
Transaction American Society of Agricultural Engineers General
Education, Vol.14, No.2. March-Apri I 1971 pp.304-308.
7. ROBERTS, A.W. "An Investigation of the Gravity Flow of
Noncohesive Materials Through Discharge Chutes".
Trans. A.S.M.E., JnI. of Engrg. for Industry, Vol. 91, Series B,

No. 2, May 1969. ......... ~
8. ROBERTS, A.W. et al. "Optimisation and Identification of Flow
of Bulk Granular Solids".
-' I.T.AjC. Symposium Automatic Control in Mining Materials and
Metal Processing. Sydney Aug. 13-17, 1973. Inst. Aust.
Publ.73/4 pp.156-161.
183
9. CHIARELLA, C. and CHARLTON, W.H. "Chute Profile for Minimum
Transit Time in Gravity Flow of Granular Material".
Journal of Agricultural Engineering Research Vo 1.17, No.2
1972 pp.149-153.
10. CHIARELLA, C. et al. "'Optimum Chute Profiles in Gravity Flow
of Granular Materials: A Discrete Segment Solution Method".

Trans. A.S.M.E., Jnl. of Engrg. for Industry, Vol. 97, Series B,
No. 1, February 1975. __________ _____
11. CHIARELLA, C. et al. "Gravity Flow of Granual Materials: Chute
Profiles for Minimum Transit Time".
Symposium on Solids and Slurry Flow and Handling in Chemical
Process Industries. American Institute of Chemical Engineers
Pittsburg June 1974.
12. CHARLTON, W.H. et al. "Gravity Flow of Granular Materials in
Chutes: Optimising Flow Properties".
Journal of Agricultural Engineering Research Vol.20 1975
p p .39-45.
13. CHARLTON, W.H. and ROBERTS, A.W. "Chute Profile for Maximum
Exit Velocity in Gravity Flow of Granular Material".
Journal of Agricultural Engineering Research Vol.15 No.3
1970 pp.292-294.
14. PARLOUR, R.P. "Mechanics of Granular Flow".
Unpublished Ph.D. Thesis, The University of New South Wales 1971
15. SARIDIS, G.N. "Comparison of Six On-Line Identification
AIgori thms".
Automatica Vo 1.10 No.1 Jan. 1974 pp.69-79.

‘ 64
16. ISERMANN, R. et al. ''Comparison of Six On-Line !cent i- icat icn
and Parameter Estimation Methods".
Automatica Vo I .10 N o .1 Jan. 1974 pp .51 — 103.
17. SASTRI, K.S. "Comparative Study of Correlation Schemes for
Linear System Identification".
Journal of Inst. Eng. (India) Electronics and Telecommun. Eng.
Div. Vol.54 Part ET2 Nov. 1973 pp.52-59.
18. ASTROM, K.J . and EYKHOFF, P. "System Identification: A Survey".
Automatica Vo I.7 No.2 March 1971 p p . 123-162.
19. HOROWITZ, I ,M. and SNAKED, U. "Superiority of Transfer Function
Over State Variable Methods in Linear Time - Invariant êedbacK
System Design".
I.E.E.E. Transactions Vo I.AC20 No.1 ~ e b . 1975 pp. 84-97.
20. SARIDIS, G.N. "Stockastic Approximation Methods for Identification
Band Control - A Survey".
I.E.E.E. Transactions Vo I.AC19 No.6 Dec. 1974 pp.796-809.
21. ROBERTS, A.W. and CHARLTON, W.H. "Application of Pseudo-Rardom
Test Signals and Cross-CorreI ation to the Identification of
Bulk Handling Plant Dynamic Characteristios".
Trans. A.S.M.E., Jnl . of Engrg. for Industry, Vol. 98, Series^sT

No. 4, 1976. _
22. CHODA, A. and WILLIS, A.H. "Flow Regimes in Inclined D u c t s ".
Transactions of the American Society of Agricultural Engineers
\ o l .10 N o . 1 1967 p p . 136-138.
23. iRAYFORD, R.S. "Development of the Tetrahedron Spreacer".
Proceedings Agricultural Encineerinc Conference .cl.2 Sve°ey,
N.S.W. Aug. 1974 pp.666-683.

185
24. KUESTER, J.L. and MIZE, J.H. "Optimization Techniques with
Fortran".
McGraw-Hill Book Co. 1973.
25. ASTROM, K.J. and EYKHOFF, P. "System Identification: A Survey".
Automática Vo 1.7 No.2 March 1971 pp.123-162 (Repeated).
26. ISERMAN, R. et al. "Comparison of Six On-Line Identification
and Parameter Estimation Methods".
Automática Vol.10 No.1 Jan. 1974 pp.81-103 (Repeated).
27. SARIDIS, G.N. "Comparison of Six On-Line Identification
AIgori thms".
Automática Vol.10 No.1 Jan. 1974 pp.69-79 (Repeated).
28. SASTRI, K.S. "Comparative Study of Correlation Schemes for
Linear System Identification".
Journal Inst. Eng.(India) Electronics and Telecommun. Eng.
Div. Vol.54 Part ET2 Nov. 1973 pp.52-59 (Repeated).
29. MACLEOD, C.J. "Comparison of Methods of Parameter Estimation
Using Pseudo-Random Sequences".
Electronics Letters Vol.9 No.15 July 1973 pp.342-343.
30. CANTON I, A. "Improving Transient Response Calculations".
Aust. Comp. Journal Vol.3 No.4 Nov. 1971 pp.156-161.
31 . CUENOD, M. and SAGE, H.P. "Comparison of Some Methods Used for
Process Identification".
Automática Vol.4 1958 pp.235-269.
32. BRIGGS, P.A.N. et al. "Introduction to Statistical Indentification

Methods in Control Systems".
Control Vol.12 No.117 March 1968 pp.233-238.

186
33. NIKIFORUK, P.N. et al. ’’Cross-corre Iat ion Identification in
the Presence of Drift".
Automatica Vol.7 1971 pp.243-254.
34. PHUDKE, M.S. and WU, S.M. "Function and Noise Model of a
Blast Furnace from Closed-Loop Data".
I.E.E.E. Trans. Vol.AS19 No.6 Dec. 1974 pp.944-951.
35. JOZSA, R.O. "Feasibility of Fluctuation Analysis for Flotation
Plant Dynamics Identification".
A.M.I.R.A. Symposium on Optimisation and Control of Mineral
Processing Plants, Brisbane July 1974.
36. BALCOMB, J.D. et al. "A Cross-correI at ion Method for
Measuring the Impulse Response of Reactor Systems".
Nuclear Science and Engineering Vol.11 1961 pp.159-166.
37. MAY, R. AppI¡cation of Correlation Techniques for Process
Control in the Steel Industry".
Private Communication.
38. CUMMING, I.G. and MAY, R. "Experimental Methods in the Steel
Industry".
39, BRYANT, G.F. et al. "The Design and Tuning of Steel Mill
Control Systems".
40. BENDAT, J.S. and PIERSOL, B.A.G. "Random Data: Analysis and
Measurement Procedures".
Wiley-Interscience 1971.
187
41. DAVIES, W.D.T. "System Identification for Self-Adaptive Contro
Wi ley-Interscience 1970.
42. HARRIS, R.W. and LEDWIDGE, T.J. "Introduction to Noise
Ana Iys is" .
P. I.0.N . Ltd. (London) 1974.
43. STAPLETON, C.A. "Test Signals for Linear Identification".
Proceedings I.R.E.E. Aust. March 1971 pp.85-92.
44. D'ALTON, L.B. "Identification by Pseudo-Random Binary
Seguenc ing".
Control and Instrumentation Vol.4 No.3 March 1972 pp.34-35.
45. HOFFMAN, R. "Comments on Correlation Measurement and
Identification - A Novel Approach".
I.E.E.E. Trans. Automatic Control Vol.AC17 No.6 1972 pp.829-830
46. ANDERSON, G.W. et al. "Use of Cross-correI ation in an Adaptive
Control System".
National Elect. Conf. Vo 1.15 Oct. 1959 pp.34-45.
47. BARKER, H.A. "Reference Phase of Pseudorandom Signals".
Proc. I.E.E. Vo I.116 No.3 March 1969 pp.429-435.
48. STAPLETON, C.A. and HOLLO, P. "A Comprehensive Classification
of Pseudo-Random Binary Seguences Using Difference Sets".
Int. Journal of Control Vol.12 No.1 1970 pp.129-143.
49. GOODWIN, G.C. et al. "Correction of Errors in P.R.B.N.
Ident if icat ion".
Electronics Letters Vo 1.6 N o .13 June 1970 pp.411—413.
50. GOODWIN, G.C. "Optimal Input Signals for Nonlinear - System
Idenf if icat ion" .
Proc. I.E.E. Vo I.118 No.7 July 1971 pp.922-926.

188
51. LEVADI, V.S. "Design of Input Signals for Parameter Estimation".
I.E.E.E. Trans, on Automatic Control Vo I.AC11 No.2 April 1966
pp.205-211.
52. GOODWIN, G.C. et al. "Optimal Test Signal Design for Linear
S.I.S.O. System Identification".
Pub I. 72/5 Dept, of Computing and Control Imp. College London.
53. GOODWIN, G.C. et al. "Optimal Test Signal Design for Linear
SingIe-Input-SingIe-Output Closed Loop Identification".
Conf. on Comp. Aided Cont. Syst. Design Cambr. April 2-4, 1973.
54. GODFREY, K.R. "Theory and Application of Pseudo-Random Sequences".
Control Vo I.10 No.96 June 1966 pp.305-308.
55. DAVIES, W.D.T. "Generation and Properties of Maximum-Length
Sequences".
Control Vol.10 No.96 June 1966 pp.302-304.
56. LEARY, B.G. "Linear System Identification - Correlation
Techn iques".
Control Systems May 1969 pp.21-25.
57. LEARY, B.G. "Linear Systems Identification".
Control Systems Vo I.1 No.2 April-May 1968 pp.28-32.
58. BROWN, R.F. "System Identification by Cross-correI at ion Part 1".
Control Systems June 1969 pp.32-35.
Control Systems Aug. 1969 pp.37-40.
Control Systems Oct. 1969 pp.31-34.

189
61 . MONTAGNER, G.J . and Trott, G.W. "A General Programme for
System Identification and Analysis Using a Desk-Top Mini Computer".
Proc. Conf. on Computers in Engin. Syd. May 1974 pp.203-207.
62. DEAN, S.K. "Flow Meter for Granular Materials".
Engineering April 1955 pp.430-431.
63. NUGENT, C.W. et a I. "Impact Line Flowmeter".
Proc. Powtech 1971: International Power Technology and Bulk
Granular Solids Conference pp.87-90.
64. NOLTE, C.B. "Flow Meter for Solid Particulate Materials".
Advances in Instrumentation Vol.24 Part 1 1969 Paper No.69-527.
65. NOLTE, C.B. "Metering of Solid Materials with Reaction
Dev ices".
A.S.M.E. Trans. Paper 73-MH-1 Aug. 1974 pp.741-746.
66. HENDERSON, J.M. "Momentum Equations of Variable Mass System".
A.S.M.E. Paper 68-MH-7 Oct. 20-23, 1968.
67. HENDERSON, J.M. "A Mass Flowmeter for Granular Material".
I.S.A. Trans. Vol.5 No.1 Jan. 1966 pp.78-83.
68. CARMICHAEL, R.W. "Measurement of Granular Flow".
Uni. of N.S.W. Report No.1974/ER/2 Dec. 1974.
69. LAZENBY, B.D. and DAVIES, M.A.S. "Solutions to Monitoring Flow
of Particulate Solids".
Control and Instrumentation Vo I.5 No.1 Jan. 1973 pp.29-30.
70. RUSNAK, J.J. "A Linear Momentum Mass Flowmeter".
Advances in Instrumentation Vol.24 Part 1 1969 Paper No.69-529.
71 . HAMID, A. "Microwave DoppIer-Effect Flow Monitor".
I.E.E.E. Trans. Indust. Elect, and Contr. Instrum. May 1975
pp.224-228.
; 90
72. HURTY, W.C. and RUBINSTEIN, M.F. ’’Dynamics of Structures” .
Prentice-Hall, New Jersey, 1974.
73. COULTHARD, J. "Ultrasonic Cross-correI ation Flowmeters” .
Ultrasonics March 1973 pp.83-88.
74. OKI, K. et al. "A New Method for Evaluating the Size of
Moving Particles with a Fiber Optic Probe".
Power Technol. Vol.11 (1975) pp.51-57.
75. BIRCHENOUGH, A. and MASON, J.S. "Local Particle Velocity
Measurements with a Laser Anemometer in Upward Flowing Gas
Solid Suspens ion".
Power Technol. Vo 1.14 1976 pp. 139-152.
76. Hycam Instruction Manual
Red Lake Laboratories Inc. Santa Clare, California.
77. NOVA User Manual
Data General Corporation, Southboro, Massachusetts.
78. U n ivac-Fortran 5 User Manual
Sperry Rand Corporation, U.S.A.

CHAPTER 10: A P PE N D IC ES
10.1 Flowmeter for Granular Material
10.1.1 Introduction
The design of a suitable flowmeter has proved quite difficult.
The criteria to be met includes a meter capable of measuring the flow
of non-cohesive materials flowing in a small scale open channel chute
and producing an electrical output signal having relatively
insignificant in-meter time delay. The similar related problem of
measuring grain stream velocity under the same design criteria was
also considered. It is recognised that with an impulsive momentum
type transducer measuring (mv) a constant output could occur with a
changing mass flow rate. It was thought that if a suitable velocity
detector could be built, the mass flow rate would be determined by
dividing the output from an impulsive transducer by the velocity
■
i .e ., , mv.)
(m = —
v
For at least 70 years attempts have been made to take advantage
of the inherent benefits of a solids flow measuring device based on the
principle of interpreting momentum changes. Dean (Reference 62), in
the mid-Fifties, designed a pneumatic flowmeter shown in Figure A1.1
for granular materials with some success. In devices of this type
an impulse sensor is attached to a pivoted beam in a fashion similar
to a weighing scale. It can be seen that the falling material generates
a torque about the axis of the beam.
The angular motion imparted to the moment arm by the material
striking the sensor is opposed by a spring-damper system. This
rotation is measured in DeanTs device by a pneumatic sensor which, in
turn, gives an output signal that for small angular displacements would
be linear and proportional to the amount of material flowing. The

PRESSURE AIR SUPPLY
TO
RECORDER THROTTLE
MEASURING ^ irrl DAMPING BELLOWS
BELLOWS
— DAMPING COCK
BEAM
ZE
-----------
KNIFE EDGE n NOZZLE ANVI L A ir
COUNTE RWEIGHT
__ >iti>_____ a n I j L __
l---- - ------------- 1------- .- 1 -7^------ 1
RESERVOIR
F i g u r e Al.l D e a n 's flowmeter for granular materials

193
accuracy of this instrument is influenced by any build-up of material
on the plate and more fundamentally by the accuracy with which the
falling material is centred on the inclined plate. Clearly a grain
stream centred on A would register a different flow rate to that
grain stream centred on B.
Most of these problems have been overcome by Nugent et a I
(Reference 63) by designing a flowmeter that, although it uses the
same principle as Dean’s flowmeter, has the pivot axis vertical
rather than horizontal. Figure A1.2 shows his model, incorporating
a Bendix pivot as a frictionless bearing. The Bendix pivot consists
of a number of leaf springs arranged so that they will always be under
tension, and require a constant bending force which results in a
constant friction pivot point that is virtually hysterisis free.
Nolte (References 64, 65) has been associated with the design
of a number of devices operating on the principle of measuring the
reaction to the material flow after it has fallen a known height.
His refinement in ensuring that the sensor moved only vertically and
hence became insensitive to the direction of entry of the material
resulted in a number of commercial devices being produced and are
typified by the device shown in Figure A1.3. In these devices the
material is passed through a set of baffles to remove any initial
vertical velocity. The material then falls a set height within the
meter and impinges on a conical sensor supported on springs and
constrained to move in the vertical direction only. The vertical
displacement of the sensor is detected by a displacement transducer.
Henderson (References 66, 67) proposed a flowmeter based on the
change in angular momentum principle, whereby material flowed into a
constant speed rotating vane system, as shown in Figure A1.4. The

— —
S
DISPLACEMENT TRANSDUCE R
Sf
BEAM COMPRESSION SPRING V)
.. \ ___ W
" V -- -___i
INCLINED PLATE FRICTIONLESS

^.BEARING DAMPING UNIT ^
b
RIGID FRAME
F i g u r e A1.2 Gr a n u l a r flowmeter of nugent et al
194
SPREADER
BELLOWS
SEAL
SPRING
HOUSING
L PNEUMATIC
TRANSDUCER
BOTTOM FLANGE
F i g u r e Al,3 N o l t e 's flowmeter for granular materials

196
change in angular momentum due to the granular material passing
through the meter, causes a torque in the supporting shaft which can
be measured by strain gauges attached to the shaft.
Material enters the meter and strikes the cone at the top of the
meter and is then directed into the top of the rotating vanes near
the axis of rotation. The angular rotation of the vanes causes the
material to slide radially along the vanes and be thrown into a
collecting hopper. The strain gauges are oriented so as to measure
torque only and ignore any bending moments which may be present.
As with most such systems measuring small toques from rotating shafts
with the usual high noise present due to the slip-ring subsystem,
considerable filtering is necessary. This filtering usually limits
the frequency response of the meter and can introduce a significant
time delay.
For the flowmeters analysed above the maximum to minimum flow
measuring range is limited and the in-meter transit time would make
them generally unsuitable for the present investigation. The
overall complexity of some of the devices also precluded their use in
this investigation.
The large range of devices designed for closed channeI operation
such as those mentioned by Carmichael (Reference 68) and those
mentioned in References 69 and 70 were reviewed to examine the
concepts involved in their operation and their possible use in an
open channel grain flowmeter.

ROTOR
ROTOR
IMPELLER
SYNCHRONOUS
MOTOR
F i g u r e Al.A H e n d e r s o n 's flowmeter for granular materials

198
10.1.2 FIowmeters/VeIocity Detectors Investigated
A number of flowmeter and velocity detector designs were
considered and some of the more promising devices were tested. The
desire to meet the design criteria of Section 10.1.1 with a relatively
simple and inexpensive transducer required considerable effort.
FIowmeters
A number of impulse type flowmeter transducers were tested.
Figure A1.5 shows a piezoelectric system measuring impact pressure
caused by the falling grain stream. Whilst this arrangement had
adequate sensitivity the associated charge amplifier was subject to
drift, requiring frequent recaI ibration and was thus not considered
suitable. The transducer’s sensitivity to moisture, requiring
special precautions for storage and the need for more than the usual
care in handling the signal cables when operating within the
grain stream, also- made it unsuitable in this application.
PIEZOELECTRI
TRANSDUCER
IMPACT
PLATE
Figure Al.5 Piezoelectric impulsive flowmeter system

199
Figure A1.6 illustrates a system based on a spring and
displacement transducer. The requirements for this transducer include
the ability to measure small flow fluctuations above a given datum
flow rate. A spring could be chosen to yield an adequate displacement
sensitivity on light loads, but for heavier loads the excessive
deflection away from the datum position resulted in a non-linear
response caused by the shift in the grain stream impact point.
F i g u r e Al.6 L inear displacement impulsive flowmeter system

200
Figure A1.7 depicts a measuring system based on a flapper valve
fitted with strain gauges to measure deflection. This design was
adopted and details follow ..in Section 10.1.3.

SWIVEL SUPPORT
F igure h i ,7 St r a i n gauge impulsive flowmeter system
Velocity Detectors
Figure A1.8 illustrates a visual velocity measuring system based
on cycling lights over a known distance and synchronising the
’t r a v e l l i n g ’ light beam to the grain stream flow. In the present
system the lights are cycled using a function generator to drive a
decade counter with each digit driving a separate light via switching
transistors. Some of the current problems with this system relate to
the reflection of light from the perspex chute, the randomness of the
201
grain flow and inability of switching the incandescent lamps on and
off rapidly. Modifications to this system as an ’o f f - l i n e ’ velocity
detector are being considered.
F igure A1.8 ' L i n e a r ' strobe velocity detector

202
Roberts (References 3, 7) showed that the velocity of the
grain stream at any given cross-section of the chute was a function of
the grain head provided the mass flow rate was constant
i .e.
_H_ = Vo
Ho V
Photodiodes arranged as shown in Figure A 1 .9 could be used to measure
the head and, knowing the inlet conditions and assuming uniform flow,
the velocity at that point could be calculated.
Figure Al.9 Photodiode height of grain stream sensor

203
Problems with light reflections off the perspex chute and with
the difficulty of packing enough photodiodes to make the height
measurement fine enough resulted in this method not being pursued.
Another possibility considered was the use of a very light
free-spinning wire paddle wheel photo-encoder as illustrated in
Figure A 1.10. This device would enable the surface and subsurface
velocities to be measured. Despite the requirements for low
friction bearings and low rotational inertia, together with the need
for a relatively simple electronic circuit, this device was a distinct
possibility. However, the DoppIer-effect meter described below
overshadowed this unit and further work on the wire paddle wheel' unit
was deferred.
Figure A1.10 Wire paddle wheel photo-encoder

204
Hamid (Reference 71) describes a recent development of a
microwave DoppIer-effeet flow monitor suitable for particulate solids.
The flowmeter utilises the principle of continuous wave Doppler radar
with results suggesting an integration or in-meter time lag of
approximately half a second for a grain velocity of 1 m/s. Further
investigation of this principle was undertaken and design details for
an ultrasonic unit are presented in Section 10.1.4.
The more conventional techniques using photographic methods
and static profile scales were also used successfully in this
experimental investigation.
10.1.3 Strain-Gauge Cantilever Flowmeter
In order to satisfy the conflicting design criteria of high
sensitivity and high natural frequency for the cantilever flowmeter
the structural parameters involved were investigated. On the one hand,
high sensitivity requires a flexible cantilever, but this implies
deflection away from the grain stream for a high datum flow rate
compared to a light load causing non-linear operation due to the shift
in the striking point. Also,a relatively flexible cantilever may have
its natural vibration frequency low enough to interfere with the
recording of step changes in flow.
The equation for the natural frequency of a cantilever from
Hurty (Reference 72, p.203) is,
con = <ei>n2 /^¡r
where £l = 1.875, 4.694, 7.855 for
n = 1, 2, 3 respectively.
The geometric parameters were selected to suit the sizes of
readily available bright metal strip so as to obtain an optimum

205
compromise between natural frequency and sensitivity.
The resulting flowmeter is shown in Figure A1.7 with Philips
PR 9833K/03 strain gauges fitted.
Calibration curves using a Tektronix Type 3066 bridge amplifier
are shown in Figure A 1 .11 . The measured natural frequency of 34Flz
compares favourably with a design target frequency of 37Hz. The
strain guage flowmeter was found to be simple in use, reliable in
operation and effectively met the design criteria.
10.1.4 Ultrasonic Velocity Detector
The necessity of implementing a non-contact velocity detector
has already been alluded to because of its many attractions. Various
devices have been described in the literature (References 69, 73, 74)
which rely for their operation on the Doppler shift principle. The
high equipment cost associated with both laser and microwave based
systems placed them outside the reach of this experimental study.
However the many advantages of a DoppIer-effeet meter were strong
incentives to find such a solution technique. The availability in
early 1967 of relatively inexpensive ultrasonic transducers
(approximately $7.00 a pair) was instrumental in the final
realisation of the ultrasonic velocity detector.
The Doppler principle is well known and the velocity of a
target is given by
.. = V(d fd (A 1 -1 )
i" 2foj cos0
where vf = target velocity
Vfjo = velocity of the emitted wave
fd = Doppler frequency
fa) = frequency of the emitted wave

FORCE (GM)
206
Figure Al.ll Calibration curves for the strain gauge flowmeter
0 = a n g l e b e tw een t h e c e n t r e line of th e e m it t e r ve
velocity di r a t i o n .
For a particular system Equation A 1 - 1 can be rewritten as
vt = Kfd (A i.2)
where K is a proportionality constant and is a useful measure of
sensitivity and resolution of the instrument.
wave Iength A =
fu)
For high sensitivity and resolution a small value of K is
requ ir e d .
B i r chenough’s (Reference 75) laser meter for measuring the
velocity of solids in gas-solid suspensions had a K value of
3.5 x 10 7 m / s e c 2 when using red light with a wave lencth of
7 x 10 7 m and assuming an incident angle 0 of 0°. On the other hand
Hamid's (Reference 71) microwave device had a K value of 1.43 x 10~2 m
s e c 2 which although far below those of the laser system has a
considerable cost advantage. In comparison, using the ultrasonic
transducers mentioned above, a device with a K value of 4.38 x 10~3 n/
se c 2 was implemented. This is a significant improvement on the
microwave device being better than three times more sensitive.
The prototype was built at a cost of less than S100 with the
assistance of the Electrical Engineering Department.
The testing programme included tests to measure:
1. the range
2. the dispersion of the ultrasonic beam
3. the effect of directing the beams by means of plastic tubes
4. the effect of air turbulence across the ultrasonic beam
5. the possibility of measuring velocity through certain materials

208
such as sponge rubber, felt and plastic sheeting.
Results of these tests are too lengthy to be presented here and will
be published at a later date.
A cal ibrating bench as shown in Frame 1 of Plate A1 was used
for most of these tests. Frame 2 shows the monitoring equipment
used and is essentially the same correlation and spectral analysis
gear used previously. Frame 3 shows the fitting of the plastic tubes
to the transducers and the opto-encoder used as a velocity reference.
Frame 4 shows the transducers in position to measure the velocity of
a rubber belt with wheat attached to its surface.
Plate A2 shows a second prototype being experimented with at
present. Its portabi Iity and digital readout are clearly evident
in Frame 1 where it is being used to measure the velocity of the same
wheat covered rubber belt. Frame 2 shows the layout of the printed
circuit boards. The use of a single multisided printed circuit
board should enable the device to be miniturised even further.
Overall the ultrasonic DoppIer-effect velocity meter as tested
was found to have definite protential as a low cost, non-contact
velocity detector. It was found extremely useful in measuring grain
stream velocities at orifice exit, at chute entry and chute exit
as part of the testing programme for the uniform flow of grain down
chutes.
10.2 FIowmeter Fi Iter
In order to record step responses using the strain gauge
fIowmeter, and recognising that this meter was extremely sensitive
to grain impingement, a suitable high frequency cut-off filter was
required between the meter and the ultraviolet recorder. As none was
available a second order maximaIly flat (Butterwot fh) fi Iter was

X-Y PLOTTER
PROTOTYPE
METER
SPECTRUM
DISPLAY CORRELATOR
• ©
CALIBRATING
MOTOR SPEED TEST RIG
CONTROLLER
Fr a m e 1 Ex p e r i m e n t a l rig Fr a m e 2 Fr e q u e n c y monitoring equipment
Fr a m e 3 Transducers without tubes Fr a m e 4 Tr a n s d u c e r s with tubes
P l a t e Al Ul t r a s o n i c velocity detector prototype

o n /o f f
SWITCH
slSgi
mmßm
LED READOUT
- .^>■.^ ■ , . , to.
¡ULTRASONIC TRANSDUCERS!
¡SWIVEL MOUNTED ssv !
?7^Ns^^wweTWi'* w * œ s
‘fe
■S I
••îsittWS» PC BOARDS
IN POSITION
ULTRASONIC
TRANSDUCERS
Í WHEAT ON"BELT
Fr a m e 1 Po r t a b l e meter in position Fr a m e 2 In t e r n a l view of meter
P l a t e A2 Po r t a b l e digital readout ultrasonic velocity transducer

designed using a readily available type 741 operational amplifier.
Figure A 1.12 shows the circuit of the filter with Ri, R2, Ci and C2
being selectable by a single rotary switch so as to give design cut-off
freguencies of 5, 10, 25 and 50H z . Using readily available
components, the actual cut-off freguencies of 5.2, 10.1, 29 and 52
were achieved. These were considered adeguate for the range of tests
to be conducted. This unit proved guite useful in a number of
applications related to the experimental work that reguired some
degree of filtering.
r2
FREQ R2 R2j r3 Cl c2
5 750 k 150 k ,27 .056
10 768 k 159 k .1 .0208
25 768 k 159 k ,047 .000828
50 768 k 159 k .022 .000414
F i g u r e A l , 12 F lo w meter f i l t e r
212
10.3 Camera Timing Lights
The two neon timing lights fitted on either side of the film
transport of the Hycam high speed camera were used to mark, or
expose, the edges of the film during a filming run and thus provide a
timing reference. Since no timing light generator was available the
simple circuit in Figure A 1.13 was used to provide this triggering
function. For the step response testing, particuIarIy , it was
useful to have the motion of the flow control gate, in this case the
perturbing gate, actually switch one of the timing lights on and off
to correspond to the opening and closing of the gate. With a 500Hz
square wave applied to the other timing light the film is marked into
one millisecond sections. From these timing marks the portion of the
film correctly exposed after the camera has reached operating or set
speed is easily seen as is the time interval between frames. As this
unit provides the timing reference for all subsequent film analysis
its accuracy and reliability are of paramount importance and frequent
calibration checks are called for.
F i g u r e Al.13 Ca m e r a timing light generator

213
10.4 Camera Technique
The Hycam is a high speed rotating prism camera accommodating
film spools up to 120 metres in length. The optical head used was
suitable for exposing 16mm full frame positive or negative movie film
and would accept any standard !C T mount lens. The quoted resolution
at the centre line was 68 lines/mm dropping to 56 lines/mm at the
edges.
The procedure adopted to photograph the front lighted
experiments was as follows. Firstly, an estimate is made of the
framing rate required to ’stop’ the motion to be photographed
minimising smear and a light meter reading of the average illumination
is taken. By using the camera performance curves of Figure 34 a
compromise between length of film and recording time at the required
framing rate is reached. The maximum aperture allowed by the prism
lens shutter system is f 3.3 so there is no point trying to expose at
an f stop lower than this. The exposure time for this camera is
calculated as follows
+ = _!--------- --------------- (Reference 76)

e 2.5 x pictures per second
For a given light level and film speed this exposure time
would correspond to a certain f stop.
The random motion of the grains necessitates a reasonable depth
of field so as to track grain motion. This, in turn, requires a
relatively high f stop setting which can only be obtained, all other
parameters held constantly increasing the film speed. The Ilford Mark V
negative film rated at 400 ASA was forced developed to double that ASA,
an improvement of one stop. The graininess of the emulsion hampered the
analysis of the projected image when the film was forced beyond this
value. It should be noted that for framing rates above about 1000
214
frames/second, the reciprocity characteristics of the film need to he
accounted for in determining the effective exposure. The use of a
spot type exposure light meter becomes essential when the field to be
photographed is too small for an averaging type meter. The
alternative of tria I-and-error estimation using an averaging type
meter is time consuming for satisfactory results.
10.5 Computer Programmes
10.5.1 System Identification Package
10.5.1.1 AIqori thms
The algorithms used were obtained by digitising the respective
signal analysis formulae derived in Section 5.2.
CorreI at ion
The correlation algorithm below performs all the correlation
caIcuI at ions
N-j + 1
R (jAT) = . E x (i)x (i+j ) (A5.1)
J N-j ¡=0
for j =0, 1, ... m-1
(note that R(0) = x 2 = mean square value)
Power Spectral Density
To analyse the power versus frequency relationships in the
signals under test, use is made of the fact that the power spectral
density function of an ergodic*

* random process is given by the Fourier
transform of its autocorrelation function. To provide smooth
estimates of the power spectrum use is made of Hanning’s window
*See Section 7.1

215
function as outlined by Bendat et at (Reference 40, p .318).
The discrete unique frequencies calculated are given by
f = kfc
(A5.2)
for k = 1, 2 ... m-1
where fc is the Nyquist folding frequency.
Leakage reduction resulting from the calculation of power
spectra is achieved using the Hanning lag weighting function,
D(r) = ?(1 + cos for r = 1,2 ... m-1

m '
= 0 for r £ m (A5.3)
The final digitised formula for the power spectral density becomes
m-1
G (k ) 2AT R(0) + 2 Z D(r)R(r) cos(-^-^-) (A5.4)
m
r= 1
for k = 1, 2 ... m-1
The cross spectral density function enables the system's
magnitude and phase versus frequency relationships to be determined.
These indicate the frequency response of the system, the presence
of any time delays and any phase shift. In a similar manner to the
single record spectrum, the following equations are derived by
Bendat et al (Reference 40, p.334) at the discrete frequencies
kfc
f
m
for k = 1, 2 ... m-1
The magn itude is
|Gxy(f)| = /C(k)2 + Q (k)2 (A5.5)
and the phase is
Q(k)
0xy(f) - tan 1 (A5.6)
C(k)
for k = 1, 2, ... m-1,

wne r e
m— 1
C(k) 2 AT A (0) + 2 I A (r ) 1)
r= 1
(Ac .
and
m- 1 ~]
4AT trr^
Q(k) Z (r) s in (A5.3)
m
r-1
The odd and even parts of the cross-corre Iat ion function M r ) anc
B(r) are given by
A(r) - A [Rx y (r) + R y x (r)] (45.9)
and
B(r) = i [RXy (r ) - Ry x (r )] (A5.10)
for r = 1 , 2, ... m - 1
Syster Computation
For a linear time invariant system the cross-correI ation function
for the input and output signals may oe expressed in terms c-f the
autocorrelation function of the input signal and the system weighting
function. Equation 5.5 is re-arranged and integration is approximated
by summation to yield the discrete impulse function,
HO)
Rxy(O)
Rx (0)AX
(A5. H )
J-l
h (;N 1 Rx v (j ) I Rx (j-k)H(K)
Rx (0) ' AX
L k=0
for j = 1, 2, ... m - 1 .
Alternatively, in matrix form it is
H = [Rx]_ 1 [Rxy]
AX
where Rx is the lower triangular autocorreI ation matrix.
An inaication of tne accuracy of the model estimation can be
gauged by comparison of the actual system output to that predicted by

217
convolution when the system is subjected to some prescribed input.
This corresponds to the problem of controlling the flow rate so as
to produce a prescribed flow output.
The output signal is given by
y(t) = /+x (A) H(t-A)dA (A5.13)

o
Limiting the number of computed points to the number of discrete
impulse function values previously calculated, and using the
midordinate method of calculating the common convoluted areas, the
digital form of the convolution routine becomes,
y (c) (x(O)H(O))AX
x(0)H(1) + x(1)H(0) (A5.14)

y( 1) AA
2
x(c)H(,j) + x(j)H(0) j-1 ‘

y (j ) + Z x(i)H(j-i) AA
2
i— 1 .
for j = 2, 3, ... m-1
The corollary for predicting the required input to yield a
desired output, inverse convolution, has proved to be more difficult
to implement. By rewriting Equation A5.13 and using the end value
method of calculating the common convoluted area, the following
expression, under certain condtions has yielded satisfactory results.
1 j-1
x(j) Z H(j-k) x(k) (A5.15)
H (0 ) AA
k=0
for j = 0, 1, ... m-1
The frequency response of the system under test can be found by
takinq 1he Lourier transform of the impulse function. Davies
(Reference 41, p .184) outlines the simpIications consonant with
pseudo-random binary signal testing, yielding a digital expression
for the Fourier transform of the impulse function.

218
N-1 N-1
G(ja)) = AT E H( iAT)cos(iwAT) jAT E H (iA T )s in (iwAT)
i=0 ¡=0
(A5.16)
In the programme package w Is varied from 0.1 to 1000 radians/second in
four IogorithmicaI Iy spaced decades.
10.5.1.2 User Experience
Once operational experience is gained with the system
identification package, some of the obvious advantages of the
assembler language level of programming make the package a powerful
data processing tool. The Data General User Manual (Reference 77)
was found useful here. The intimate interaction between operator and
hardware enables him, for example, to stop the process, examine or
change counters or programme variables, skip over or back to some
other part of the programme and then to continue processing by the
press of a switch. This facility of checking the recorded data
prior to lengthy data calculations can save considerable computer time.
This is particularly so when working with a system involving a number
of transducers, any one of which, should they become faulty, would
render all the results meaningless.
On the other hand, the very considerable disadvantage of
programming effort required to write programmes at this basic hardware
level needs careful consideration. To attempt to write software of
this nature on a newly installed system having only slow speed input
output reading and writing facilities is, in retrospect, very time
consumi ng.
The utility of the package programme is evidenced by the fact that
it has been used on many occasions for student demonstrations, data
logging and data reduction for student projects within the Department
of Mechanical Engineering.
219
10.5.1 .3 Programme Flowcharts and Listinq
The following is a brief description of the main subroutines used
by the system identification programme. Flowcharts for the more
complex subroutines are included with their listings.
SI GAN generate the operating system and by asking a series of
hierachial questions schedule the subroutine path.
SERV - contains all the device drivers; character input,
output and text writing routines.
ATODC - enables data logging using the analogue to digital
converter operating under programme control.
PLOT - an autoranging autoscaling plotting routine to present
results in graphical form on a standard teIetypewriter.
DATA IN - general data entry via any of the devices available.
Data is counted establishing N and M the correlation
counter set to N/10.
LIST - normalises the data; prints the mean; enables the
listing or the punching of paper tape (with parity);
or the plotting of data.
CORREL - general correlation routine for computing RXX, RYY, RXY,
etc.
SPECT - computes the power spectral density of the input and
output signals using the autocorre1ation functions and
a Hanning window for leakage reduction.
XSPECT - computes the cross-spectra 1 density function between input
and output signals using the correlation functions and a
Hanning window for leakage reduction.

220
IMPUL computes the system's impulse function.
CONVOL predicts an output based on the system' model sescrioec
by the impulse function.
INVERCON predicts an input based on the system mode! described ey
the impulse function.
FRESP computes a Bode diagram by taking the Fourier transform
of the impulse function.
MESSR start-up message, all running messages and storage array

221
F igure A5.1 SIGAN(1)- Ge n e r a t i o n of o p e r a t i n g r u n s t r e a m

F igure
A5,2
S IGAN(2)-
Operating
system
subroutine
ài
System Computation Signal Analysis Data Manipulation Data Collection

222
hierarchy
223
-00 1 SI GAM
; OFF/ON L I N E SIGNAL ANALYSIS...G J MOMTAGNER

J LAST UPDATE 9/8/75
; DATA A C C E P T E D F R O M T E L E T Y P E , H I G H S P E E D P A P E R
* T A P E R E A DER, C A R T R I F I L E , A TO D C O N V E R T E R .
5 N O R M A L I S E D DATA M A Y BE P L O T T E D OR LISTED.
; S I G N A L C O M P U T A T I O N S I N C L U D E A U T O C O R R E L A T I O N OF
I I N P U T OR OUTPUT, C R O S S C O R R E L A T I O N , P O W E R S P E C T R A L
; FUNCTIONS.
; SYS T E M C O M P U T A T I O N S I N C L U D E I M P U L S E F U N C T I O N
; C O N V O L U T I O N (TO P R E D I C T T H E OUTPUT)
; I N V E R S E C O N V O L U T I O N CTO P R E D I C T TH E INPUT)
; F R E Q U E N C Y R E S P O N S E . (B O D E D I A G R A M )
5 PRBS PA R A M E T E R S : -
I DT = T I M E D I S P L A C E M E N T ( D I G I T INTERVAL)
; S E L E C T N O I S E B A N D W I D T H TO S UIT S Y S T E M
I U S U A L L Y TEN T I M E S T H E S Y S T E M ' S B A N D W I D T H .
5 P R B S P E R I O D TO B E A P P R O X . 1 . 5 * S E T T L I N G TIME.
5 T C P R B S )=1.5*TCSETTLING)
; TC S E T T L I N G ) = T ( D E A D T I M E ) + 4 * T ( L O N G E S T
i T I M E CONST. OF I N T E REST)
; S E L E C T DT < 0 . 5 * T C S H O R T E S T T I M E CONST.
J OF I N T E R E S T )
I CORRELATION PARAMETERS:-
5 N = NO OF D ATA P O I N T S (SE L F COU N T I N G )
; M = NO OF C O R R E L A T I O N L A G S (SET TO N/10)
I DELAM = T I M E I N T E R V A L B E T W E E N L A G S
5 ( SET E Q U A L TO DT)
. TITL S I G A N ; G E N E R A T E O P E R A T I N G SY S T E M
. ENT W S A , G E T C , P U T C , . C R L F , . T X T T , .N E X T P , . R E S L T
. ENT T E M P O , T E M P 1 , C 1 2 , C I 5,N U M , N U M P 1 , J , M 1 , N
. ENT S U M , S U M P 1,R X X O , D E L A M , O N E , T W O , S T A R T
. ENT C 5 P 1,C 9 P 1, M S K , N L P 1» N 1 , I C H N , S R A T E , S C N T R
. ENT • PLOT, N L , C 9 , C 5 , W, WW, D, DD, K,M, T W NTY
. f:n t a m e s o , a m e s i , . g t c t , . g t c p , . g t c c ,p i
• ENT DEG, LNLOG, TEN
. E X TN USAR,TTIN,TTWR,RR,H,G,PLOT,GXY,,CON,FRESP
. EX TN RAY 1, RAY2, RAY 3, RAYA, RAY 5, RAY 6, RAY 7 , RAY 8
. F^XTN DEL, DATAIN, INPUT, OUTPUT, AUTOI , AUTOU, C R O S S
. EXTN MESO,MES1,MES2,MES3,MESA,MES5,MES6
. FXTN MES7,MES8,MES9,MEAN,A2DC,LIST1,LIST2,QUES
•EXTN TTYR,HSPTR,CARTF,A2DI,CRLF,TXTT
. EXTN N E X T P , R E S L T , N P A G E , H E A D , F I N T , F ENT
. p:x t n G T C T , G T C P , G T C C , INPOW, O U T P O W , X P O W
. EXTN C O N V,INCON,ICON,FREQ,STRTM
224
000 2 SI GAM
• ZREL
00000-177777 WSAS USAR % P O I N T E R TO U S E R A R E A
0 0 0 0 1 - 1 7 7 7 7 7 GETC: TTIN 0
9 P O I N T E R TO G E T A CHAR« R O U T I N E
0 0 0 0 2 - 1 7 7 7 7 7 PUTC: TTWR 0
9 P O I N T E R TO W R I T E A CHAR. R O U T I N E
0 0 0 0 3 - 0 0 0 0 0 l-.TTIN: TTIN 0
9 P O I N T E R S IN P A G E Z E R O TO R O U T I N E S
0 0 0 0 4 - 1 7 7 7 7 7 • RR: RR
0 0 0 0 5 - 1 7 7 7 7 7 .Hi H
0 0 0 0 6 - 1 7 7 7 7 7 • Gl G
0 0 0 0 7 - 1 7 7 7 7 7 . GXY: GXY
0 0 0 1 0 - 1 7 7 7 7 7 • QUESJ Q UES
0 0 0 1 1 - 1 7 7 7 7 7 •GTCTl GTCT
0 0 0 1 2 - 1 7 7 7 7 7 •GTCPI GTCP
0 0 0 1 3 - 1 7 7 7 7 7 .G T C C : G T C C
0 0 0 1 4 - 1 7 7 7 7 7 • TTYR: TTYR
0 0 0 1 5 - 1 7 7 7 7 7 • H SPTRi H S P T R
0 0 0 1 6 - 1 7 7 7 7 7 • CARTFl C A R T F
0 0 0 1 7 - 1 7 7 7 7 7 •A 2 D I 2 A2DI
0 0 0 2 0 - 1 7 7 7 7 7 . A2DC2 A2DC
0 0 0 2 1 - 1 7 7 7 7 7 •PLOT: PLOT
0 0 0 2 2 - 1 7 7 7 7 7 • CON: CON
0 0 0 2 3 - 1 7 7 7 7 7 . ICON; ICON
0 0 0 2 4 - 1 7 7 7 7 7 . FRESP: F R E S P
0 0 0 2 5 - 1 7 7 7 7 7 . D ATAIN 2 D ATAIN
0 0 0 2 6 - 1 7 7 7 7 7 .L S T 12 LISTI
00027-177777 •LST2; LIST2
0 0 0 3 0 - 1 7 7 7 7 7 •CRLF; CRLF
0 0 0 3 1 - 1 7 7 7 7 7 . TXTTl TXTT
00032-177777 •NEXTP2 NEXTP
0 0 0 3 3 - 1 7 7 7 7 7 • RESLT2 R E S L T
00034-177777 AMESO2 - 1 «
9 DATA SOURCE
00 0 3 5 - 177777 AM ESI 2 - 1 0
9 DATA L I S T I N G
0 0 0 3 6 - 1 7 7 7 7 7 AMES2I - 1 0
9 FREQUENCY RESPONSE
0 0 0 3 7 - 1 7 7 7 7 7 AMES3: - 1 0
9 C O N V O L U T I ON
0 0 0 4 0 - 1 7 7 7 7 7 AMES4: - 1
•
9 A U T O C O R . OF INPUT
0 0 0 4 1 - 1 7 7 7 7 7 AMES5: - 1 A U T O C O R . OF O U T P U T
0 0 0 4 2 - 1 7 7 7 7 7 AM E S 62 - 1 0
9 CROSSCORRELATION
0 0 0 4 3 - 1 7 7 7 7 7 A M E S 72 - 1 •
9 IMPULSE FUNCTION
0 0 0 4 4 - 1 7 7 7 7 7 AMES82 - 1 •
9 POWER SPECTRAL ROUTINES
•
0 0 0 4 5 - 1 7 7 7 7 0 NL: -10 9 NL
0 0 0 4 6 - 1 7 7 7 0 4 N L P 1: - 7 4
0 0 0 4 7 - 1 7 7 7 6 6 C9 2 - 12 p C9
0 0 0 5 0 - 1 7 7 7 6 6 C 9 P 1: - 12
•
0 0 0 5 1 - 1 7 7 7 7 3 C5: -5 9 C5
0 0 0 5 2 - 1 7 7 7 7 3 C 5 P 1: -5
l 'u u J O y ( . .1u \w*
m T E M P O 2 V-<
A 0
9 TEMPO
A A A C / i_ A A A A A A
V-' V ' v.» O H TEMPI:
A
V
0
9 TEMPI
A A A C C _ A A A A
\ Zj Zj V_< V.* 1
f Q
C 12: 12 9 C 12
v'uv.» J C v.»v-' v.' v.» 1 C 1 5: 15 0
9 C 15
1 NUMI MU M
A A C 7 ^ rs A A A A
V.I J f “ v.» V_' V.» V
_*V \w* 9
^1v • O v ' * V.» V.» V * V * V.» V.' N UMPI: V_»

AAA/- | A A A A A A
Js #
V'V^' V» O 1 "* V.» V.» L'
V• 9 J
A A A
v,» v v C ”>
t »
/ A
r
a A A A A A
v.* c» v.' v.' v.» M 1: 0
9 Ml
•AA A /* n
V. ' V. » *’' J * *
A A A A A A
V
,*KJ v1 M: V.'
0
9 M
A A
V.»
V.» V.»
."N /• /,
V ^
A A A
V, V »
•
.■> A A
v * v. N: A
V.'
0
9 N
A i-\
v.' \,i
|-> /•
D
C _ .A A A A
Z) <.-• ^
<*» 7
*
WS 7 0
9 W
0 0 0 6 6 - 0 0 0 0 1 0 ww: 10 0
9 WW
0 0 0 6 7 - 0 0 0 0 0 2 D: 2
0
9 D
0 0 0 7 0 - 0 0 0 0 0 3 DD: 3 9 DD
0 0 0 7 1 - 0 0 0 0 0 0 K: A
KJ
0
9 K
225
0003 SI GAM
0 0 0 7 2 - 0 0 0 1 7 7 M SK: 177 ; MSK
0 0 0 7 3 - 0 C 0 0 C C SUM* A
U ; sum
0 0 0 7 A - 0 0 0 0 0 0 S U M P 1* A
U
A
0 0 0 7 5 - 0 0 0 0 0 0 ICHM*
AS
0 0 0 7 6 - 0 0 0 0 0 0 SPATES V»
A A ^ 7 7Â A A A A A CPMTPf A
V_*
A A | A A ^ A A A A | C-DlTTt
ou 1 1 D rr 1 1 •HSPTR
A A 1A 1 _ A A A A A A
uu K* 1 uv u u u o Or\Y AY
AA t
\j I
A
V.» ; rxxo
UU 1 U C
A A A A w A A A A A A
\ A
U
0 0 1 0 3 - 0 0 0 0 0 0 DELAM: A
KJ ; DELAM
V.» U iU H
A
A A 1 A A A A A A A
U U L » V.* V.» i ARITHMETIC CONSTANTS
0 0 1 0 5 - 0 4 1 0 7 1 DEG: 57* 2 9 5 7 7 9 5 •
9 RADS. TO DEGS.
00106-045670
0 0 1 0 7 - 0 4 0 4 6 2 PI: 3*14159265
00110-041766
00 1 1 1 - 0 4 0 4 4 4 L N L O G S 2.30258509 •
LN TO LOG
00112- 1 53543
00113- 0 4 0 4 2 0 OME: 1.0 ; OME
uu 1 1 *4
A A 1 | /|_ A A A A A A
0 0 1 1 5 - 0 4 0 4 4 0 TWO: 2.0 ; two

A A 1 | A A A A A A
1 ID Y U U V.» u
00 1 1 7 - 0 4 0 6 4 0 TEM:
A A $ A A _ A A A A A A
0 0 1 2 1 - 0 4 1 0 2 4 TWMTYS cvy • u
A O AN
U U 1C C
A A | O O — A A A A A A
V.»V^L*UV/V*
•N R F L
0 0 0 0 0 * 0 6 2 6 7 7 START: I0R S T •
0 RESET START
A A A A 1 f A / ; O A 7 7
L » U V-Mw» 1 Y D O u /f HALT •
0 WAIT # PRESS CONTINUE
A A A A A t
V. » VJ\. » V ' <o
A A
U U U U J C
O-*
JSR § • NEXTP
A A A A Q 9 A A ^ A O | _
u u u u O u u O u O 1 JSR § • TXTT •
0 STARTUP MESSAGE
0 0 0 0 A * 177777 STRTM
A A A A C
V.' V» S.» J
• A A A /i A O
U U M U J JMP . +3
A A A A / :
u u v ^ u C
t A A / ;A A
u u u u O
1 _
1 JSR § . TXTT
«
0 RESTART HEADING
00007*177777 HEAD
V.IUV.I
i r> #
1l'
a *1/ i « ^ ^ .
L 'O ^ U U V . ' LD A 3# WSA *
0 SET F O R M A T F O R
00011*020066- LDA c*ww : FLOATING POINT
00012*041521 STA 0# 121# 3
r \ r \ n <o 1
V. < U U IJ
Q A A 7
\J t\J C .\ J \J LDA 0#DD
00014*041522 STA 0#122# 3
000 15* 177777 FI NT
•
0 INI T A L I S E FLT* PT.
A 1 t 1 A/> 1 A _
JSR §.QUES 0 SET THE RUN S T R E A M BY
00 0 1 7 * 1 7 7 7 7 7 MES0 0 A S K I N G AL L R E L E V E N T
v.*\~i
A A A 9VA O
Q y.% A A , , ,
*4 1 sj \J LDA 3# PFT 0 QUESTIONS.
v_* ¿
L1 i I iO U U
A A A A 1 9 1 | T A A A
W ADD 0# 3
A A A
v.1u u
A A
c
9 A / l O A O / j .
c* H u u O * * STA 0 # AME S 0 0 DATA S O U R C E
V.»V.* c J t uu
A A A A Q
V.*
A A 7 / 1A A
f JSR §0.3
A A A O/ l• JSR § . TXTT • AMS. F0LL. QUES.
V. <v . *U u u O u J 11 » 0
00025*177777 MES2
i.>(.'v.'CO•0
A A r t A A £ A 1O _
uCu K 1 JSR § . Q UES
00027*177777 MES1
• DATA L I S T I N G
W
A A A
0 J uA •u^luyOJ
A / i A A O C « .
STA C#AM E S I 0
A A A A 1 1A A £ A (A
v_iv.<v.1O l y y D y I k JSR §• QUES
00032*177777 MES9
00033*040036- STA 0 . A M E S 2 0• F R E Q U E N C Y R E S P O N S E
00034*100015 COM# 0# 0# SNR
00035*000404 JMP • +4
00036*040040- STA 0# A M E S 4 0• AUT0C0R. OF INPUT
0 0 0 3 7 *0 4 0 0 4 2 - STA 0# A M E S 6 0• C R O S S C O R R E L A T I O N
0 0 0 4 0 *0 4 0 0 4 3 - STA 0# A M E S 7 0• I M P U L S E F U N C T I O N
226
0004 SI GAN
00041•006010- JSR 0.QUES
00042*177777 MES3
STA 0 * AM ES 3 « CONVOLUTION
»
00043*040037-
00044*024036- LDA 1# A M E S 2
00045*124015 COM# 1 » 1#SNR
O L ' u H D l u u u l j COM# 0*0# SNR
JMP • +5
A r t A C A • A A A / | A _
STA 0# A M E S 4 I A U T 0 C 0 R « OF INPUT
U O V.' 1 U H U U H C
A A A C
□
1 I A / ,A A / . A .
STA C*AMES6 #• C R O S S C O R R E L A T I O N
00052*040043- ST A 0* A M E S 7 •
# IMPULSE PONCTION
U O U D O U U U*J 1 d
A A A C ' l f A A A A I A
JMP PS
00054*124014 COM# 1#1# S Z R
U O V H 1 V» JMP
A A A C C • A A A / . « A
J O PS
A A / -A I A , .
rs r% c
o c » v.» J O
/L 1
u v D v i o JSR e.QUES
00057*177777 MES7
u u u D o W ^ U W 'i O
A A A £ A « A / . A A /. Q _
STA 0#AMES7 ; IMPULSE FUNCTION
A A A £ I « 1A A A 1 C
u c * v »0 1 i u u u i D C OM# 0 * 0 * SNR
u v . ' o o c u v - ' i ' ^ u J JMP PS
A A A £ A . A
v . ' u v .'00
/. A A / , A _
STA 0# A M E 5 4 #« AUT0 C 0 R . OF INPUT
00064*040042- STA 0*AMES6 #* C R O S S C O R R E L A T I O N
r » A r y / ;c
c» c» o O J
• a a /• a
u u O u
i
1v
A ^ n c
r j
•
« JSR @.QUES
00066*177777 M ES8
00067*040044- STA 0 # AMES8 • POWER SPECTRAL ROUTINES
#
A A A T A
0 C »0 i O
• J A A A
I V U V
1 C
1J COM# 0 * 0 # SNR
A A A T
0 0 0 f
1 f A A A /. | A
1 O O O H I C JMP PS1
A A A T A • A /i A A / . A _
v.# ^ K.» f r z L » *4 %.» v *4 L» STA 0* A M E S 4 #
•
A U T 0 C 0 R R . OF INPUT
00073*040041- STA 0 # AM ES 5 # A U T 0 C 0 R R . OF O U T P U T
00074*040042- STA 0 # AMES6 #• C R O S S C O R R E L A T I O N
00075*024043- LDA 1# A M E S 7
00076*124014 COM# 1* 1 # S Z R
A A A A A
V.« V.* I I
• A A A / i
O U V -'H C v.' JMP PS2
A A
O W
« A A f A /, A A
1 V.HJ U ^ U U
/| A _
t U STA 0# A M E S 4 « AUT0 C 0 R . OF INPUT
#
00101 *040042- STA 0# A M E S 6 5 CROSSCORRELATION
A A
V.* V.I
I A A *
1 V_i C .
A A A /, 1 C
t ' O O H 1 J JM P PS2
A A
i .1u
J A A
1 V.1O
I A A £ A | A
o y u w K *
. n C
" O
1
I
•
• JSR @ •Q U E S
00104*177777 MES5
00105*040041- STA 0* A M E S 5 ; AUT0 C 0 R . OF O U T P U T
00106*024043- LDA 1# A M E S 7 * IMPULSE FUNCTION
00107*124014 COM# 1*1# S Z R
A A
v_< O
t
1 1U
I <A t A A A /| A 7
LM -M -'H o I JMP PS2
«_<\ J I 1 1» t UA LA 'D
A A I I
o iu
£ A 1 A _
JSR @.QUES
00112*177777 MES4
00113*040040- STA C# A M E S 4 A U T0C0R. 0F INPUT
00114*006010- JSR 0 . Q UES
00115*177777 MES6
00116*040042- STA 0* A M E S 6 CROSSCORRELATION
001 17 *0 2 0 0 37-PS2J LDA 0# A M E S 3 CONVOLUTION
00120'024044- LDA 1# AMES8
00121*100015 COM# 0# C# SNR
00122*124014 COM# 1»1# SZ R
r \ r\ « O Q t r \ r \ r \ J \ r \ c\
v v 1 c J c*u JMP •+ 2
00124*000413 JMP . + 13
00125*006031- JSR @ • TXTT ; TIME DISPLACMENI
00126*177777 DEL
A A
\
j \
ji
I O T • ^ O n ,>
1 c / u c u u o
1 «
I LDA 0# G ETC
00 130* 0 2 4 0 0 3 LDA 1# • T T I N
0 0 131 * 0 4 4 0 0 1 STA 1»GETC
0 0 132* 177777 PENT
00133*120000 F DF C GET DELAM
227
0005 SIG AN
00134* 040103- FSTA 0# DEL AM
00135*100000 F EXT
001 3 6 * 0 4 0 0 0 1 - ST A 0# G E T C
00137*126120 ADCZL 1# 1
0 0 1 4 0 *0 2 0 0 3 4 - LDA 0# AM ESO ; DATA SOURCE
00141* 107014 ADD# 0#1 # SZR ; A TO D ?
00142*000403 JMP •+ 3 ; NO
00143*006020- JSR 0. A 2 D C ; YES» J U M P TO SUB.
00144*000413 JMP . + 13
00145*024035- LDA 19 A M E S 1
00146*020040- LDA 0# AME S 4
00147*030042- L DA 2» AME S 6
00150*150015 COM# 2 # 2 # SNR
00151*124014 COM# 1#1 * S Z R
UUIJC 1 C O I A A A / | / > Q
JMP • +3
00153*100015 COM# 0 * 0 » SNR ; AUTOCOR. OF INPUT ?
00154*000415 JMP NO ; NO
1J J
V. H. ’ Ou CÍD V.I JSR 0 . D A T A I N; INPUT D A T A C A L L
0 0156* 177777 RAY 1
00157*006032- JSR e.NEXTP
00160*006031- JSR 0.TXTT ; INPUT H E A D I N G
00 161 * 177777 INPUT
00162*006026- JSR a.LSTl
00163*000156* RAY 1
00164*020035- LDA 0# AMESI
00165*100015 COM# 0# 0# SN R ; LIST. A N D OR PLOT. ?
00166*000403 JMP NO
00167*006027- JSR «.LST2
Ars 1 7 A f « r 7 |
I / v.'
a a a
c ' v_' v j 1 O O RAY 1
00171 * 126120 M 0 : ADCZL 1# i ; S T A R T TO E X E C U T E THE
00172*020034- LDA 0 AM ESO ;
• RUNSTREAM
00173*107015 ADD# 0 # 1 SNR ; # DATA S O U R C E ?
00174*000413 JMP . + 13
00175*024035- LDA 1 AMESl »
00176*020041- LDA 0 # AMES5

00177*030042- LLA 2# A M E S 6
r \ .~ s <
O U 6
2
W
? | c
I JUV.»
1 C
1 ZJ COM# 2# SNR 2 »
00201 * 1 24014 COM# 1 1 SZR

# *
A A Q A n i
ov.' c
A A A /, A 7
v.» O H v.< O JMP . +3
Ars Q A n «
v/ v./ r . v.» O
1 A A A
1 c» u
• C
1 J COM# 0* 0* SNR ; AUTOCOR. OF O U T P U T ?
00204*000415 JMP N1
A A A A C t a A /■ A A C
JSR 0. d a t a i n ; O U T P U T D A T A C ALL
00206* 177777 RAY 2
00207*006032- JS R « N EX TP
.
A A A 1 A
V*V»£-I^*
f A A £ A 7
V. *V. » UU J
1 _
I JSR P.TXTT ; OUTPUT HEADING
0021 1 * 177777 OUTPUT
00212*006026- JSR 0.LST1
00213*000206* RAY 2
00214*020035- LD A 0 » AMES1
A A A
S»'V.* ¿
1 C
1 J
1 1 A A A
1 O V.» U
| C
I J COM# O p SNR ; LIST.
O p A N D OR PLOT. ?
A A A
VJ VJ
1 £
1 O
t A A A
s.» u
/i A 7
J JMP N1
00217*006027- JSR 0.LST2
f > n O O ' 1 •
L.» l < ez. C i L< V.I V.I V.I &
1 O
1 O
*
R AY2 *
00221 *0 2 0 0 6 3 - N 1 : LDA o *m ; SET P L O T T I N G C O U N T E R
00222*040062- STA 0» M 1
0 0 2 2 3 *0 2 0 0 4 0 - LDA O A M E S 4 ; A U T O C O R R E L A T I O N OF
p
00224*100005 COM 0# 0# SNR ; I N P U T D A T A CALL

0 0 2 2 5 *0 0 0 4 16 JMP N2
00226*006032- JS R 0 . N EX TP
228
0006 SI GAM
00227*006031- JSR » . TXTT
00230*177777 AÜT0I
00231*006004- JSR G.RR »«
J U M P TO C O R R E L A T I O N
00232*000170* RAY 1 »• R O U T I N E
00233*000232* RAY 1
00234*177777 RAY3
A A
UUCJ J
tO
A O A A / iA _
G tK 'n U LDA 0# A M E S 4
AAOOA •I
UUCÜÜ &A A / . A A
u u h u o MEG 0 #0
A A O O * 7 t | A A A A /.
COM 0 #0 # SZ R »•
PLOTTING ?
^ ^ O /I A • A A ‘A /»A A JM P N2
0 0 2 A 1 *0 0 6 0 2 1 - JSR #.PL0T
00242* 0 0 0 2 3 4 * RAY3
00243* 0 2 0 0 4 1 - M 2 J LDA 0» AM ES 5 # A U T O C O R R E L A T I O N OF
00244*100005 COM 0 * 0 # SNR » O U T P U T D A T A C A L L
00245*000416 JM P N3
««
ü u co/i A •Art
*40 u w O£rt»jn wJd JSR ».NEXTP
00 247 *0 0 6 0 3 1 - JSR ».TXTT
00250*177777 AUT0U
A r t oC 1 t r t r t / r r t f t
u u c J 1 uuOuu*i* JSR ». RR #■ J U M P TO C O R R E L A T I O N
rtrtrtCrtl
UUC J c
rt rt rtrtrtrt •
U L 'U C G U RAY2 #• R O U T I N E
rtrtrtC «
C 'U C J O
J« rtrtrtrtÇ O *
UV.H.ICJC RAY2
00254*177777 R AY4
00255*020041- LDA 0* AM ES 5
rt />
r t r t r t C Z 1 1 rtrt /irtrt
(j u c j D I u u *<u u MEG u * u
r t r t r t C 7 1 | rtrtrtrt
u u c j i Iw y u u *» COM 0 * 0 # SZR •
» PLOTTING ?
r t r t O i r t • r t rtrt /.rtrt
ui.ioOv.1 J JM P N3
r t r t r t i 1 * rtrt £ rt o 1 _
WV.ICU 1 v .'ü D w c l JSR ».PLOT
r t r t O A O * rtrtrtrtc/. •
c l ' c UC u u w g j *J RAY4
0 0263* 0 2 0 0 4 2 - M 3 : LDA 0 # AMES6 •
» CR0SSC0RREL. CALL
1A A A A C
COM 0# 0# S M R
00265*000416 JMP M4
rtrt O i £ • r t r t i r t O O «
l/ U tO O Du OG JSR ».NEXTP
rtrt O A 7 • o o A* * ^ t •
o u g D I v^yO uJI JSR ».TXTT
00270*177777 CROSS
00271*006004- JSR » . R R
r t r t r t T O 1 rtrtrt Q t 7 »
VJUG 1 G UUVJ G J O RAY2
r\ri 0 7 *1 • r t r t r t r t l l *
U C C I O 1/yv.iGJJ RAY 1
00274*177777 RAY 5
00275*020042- LDA 0# A M E S 6
rt rt
00276*100400 MEG U » U
00277*100004 COM 0 * 0 # SZR « PLOTTING ?

r t r t O r t r t l rtrtrt/|rt 7
U t' O u J u u u ^ u O JMP M4
rtrtrtrt « 1 r t r t / r t r t ]
u u Ou 1 UUUUG I JSR ».PLOT
r t r t r t r t r t l rt rt rt O 7 /< 1
UUOUG UUUG f RAY 5
m
0 0 303* 0 2 0 0 4 4 - M 4 : LDA 0# AMES8 9 POWER SPECTRAL
r t r t r t r t /> (
v .iu o u ‘ 1
« rtrtrtrtc
1u u u u O COM 0 * 0 » SNR 9 FUNCTION CALL
rtrtr tr tC I rtrtrt/ i£ 7
u u u u J uuu*4D # JMP N5
r t r t r t r t i l rtrtrtr t£ r t „
uu ou D u G t* u O O LD A 0#M •
* SET PLOT COUNTER
rtrt rtrt 7 • I rt rt /, rtrt
UUOU / 1 U U Mt' u N EG «rt rt»
SJ # VJ TO M-l
rt r t r t 1 rt • t r t r t r t r t r t
u u O 1u lu u u u u COM 0# 0
00311*040062- STA 0# M 1
rt r t r t I
u u O 1G u u u u OG JSR ».NEXTP
00313*006031- JSR » . TXTT
00314*177777 IMP0W
00315*006006- JSR ».G «
9 INP U T S P E C T R A L
00316*177777 R AY6 1 DENSITY
00317*000242* RAY 3
00320*020044- LDA G* AMES8
A A
003 2 1 * 100400 MEG V.' # %.«
229
0007 SIGAN
00322 1
1A AAA COM 0# 0# SZR s PLOTTING ?
UUJC J
a a q a o
000451 JMP N5
00324 A
w rtCw £ A A«
kj ¿1 JSR ©•PLOT
00325 uuoJlD RAY 6
00326 u
AA ÌO
l ' D u o c
_
JSR § . N EX TP
00327 ».»wOuO 1
A ft £ ft O 1 _
JS R ©.TXTT
A A AJuU
U U
QA 17 7 7 7 7 0UTP0W
00331 A
oA ur OAuAuZ
u_ * JSR #• G i OUTPUT SPECTRAL
A A ^JJC
nn 177777 RAY 7 • DENSITY
#
u «3*3O
A A Q A O AAAOA O • RAY4
00334 020044- LD A 0# A M E S 8
00335 1
1A A /iA A NE G A A
V» KJ o# o
00336 1
1U AU AAA
U U N>. COM 0*0# SZ R •
PLOTTING ?
00337 000435 JMP N5
O
v.* c* 3
O /i O
v.* ycOwcl" JSR 3» P L O T
00341 000332« RAY 7
00342 uoOuOd
ft ft £ ft n _
JSR S.NEXTP
f t f t £ftft/i_
00343 o u Ouu*! JSR f.RR
00344 U U U C fJ f t *7 ft «
RAY 1
00345 0002 7 2 * RAY 2
00346 A A A A A A
0 0 0 3 3 3
f
RAY4
00347 A A A A O 1
o o O o 31
.
JSR © • TXTT
A A O C A
u u o De# 177777 XP0W
00351 A A A A A AT
O O O O V.» f JSR ©.GXY # CROSS-SPECTRAL
00352 A A A A A A
o o o Oo
t
RAY5 9 DENSITY
A A A C O
u u O O J 0 0 0 346* R AY4
0 0 3 5 4 A A A 3A A C « O O O 3 RAY 6
00355 000341 * RAY7
0 0 3 5 6 AO O A A A C A
O 3 3 3
•
RAY 4
0 0 3 5 7 177777 RAY 8
A A A
020044-
A /:
v.' v.' 3 O 3 LDA 0# AM ESS
00361 1 oA oA * /4i Ao oA «
NEG ft
KJ 0
/>
lw<
0 0 3 6 2 i A A Ao Ao *4 ]
O O
/.
COM 0# 0# S Z R •
9 PLOTTING ?
0 0 3 6 3 000411 JMP N5
00364 A A A A A _ r
V. ' U v w / O c JSR ©.NEXTP
0 0 3 6 5 Ao Ao Ô Ao Ao 1 I .
JSR © •PLOT 9 PLOT MAGNITUDE
0 0 3 6 6 AO AO AV.* 3A C Co t 3 RAY4
0 0 3 6 7 oA Ao CO oA A3 oO — .
JSR ©.NEXTP
A A
^ • ^ .»0
A A A
9 ^ *
A A C A A |
o o Oo e I
_.
JSR @.PL0T •
9 PLOT PHASE
0037 1 A A A A C A «
o o o 3 3 f RAY8
0037 2 A0 O^ A0 A0 /0 ' 3A ^ LDA O.M 9 RESET M COUNTER
00373 040062- STA 0* M 1
00374 020043-N5: LDA 0» A M E S 7 9 I M P U L S E F U N C T I O N CALL
0 0 3 7 5 i1 A0 0A 0A 0A 3C COM 0.0# S N R •
9 R E Q U I R E S RXX A N D RXY
00376 000413 JMP N6
0 0 3 7 7 Ao oA o Ao A C3 . C
JSR e . H
A A /i A A A A A A |A
o o o 3 1 9
«
RAY 3
A A
V./ V_#
/. A
N»' 1
1 rs r\ r\ *\
o o o 3 3 o
C O #
RAY 5
A A /, A O
V»>^HUC 0 0 0 354* RAY 6
A A / .A A
\J KJ *-* \J O 020043- LD A 0# A M E S 7
y. r\ /, 1 A A
1o
/. A A
o o NEG «
0
A
%.<
00405 | A A A A /,
1 o o o o *4 COM 0 . 0 . SZR •
9 PLOTTING ?
00406 A A A /, A A
o o o *4 o O JMP N6
A A £. A O 1 «
o o O o ¿0 1 JSR ©.PLOT
00410 f A A A /.A A
o o o M V.» c RAY 6
004 1 1 0 2 0 0 3 7 - N 6 : LDA 0.AMES3 m
9 CONVOLUTION AND
0 0 4 1 2 |I Ao Ao Ao Ao C3 COM 0 . 0 . SNR 9 I N V E R S E CON VOL. CALL
004 1 3 0 0 0 4 3 3 JMP N7
004 14 A As>.i \DC A A O—
\j O d JSR ©.NEXTP
230
SIGAN
00415*006031 - JS R «•TXTT
00416*177777 CON V
00417*006022- JSR ««CON ; CONVOLUTION ROUTINE
00420*000410* RAYÓ
00421*000344* RAY1
00422*000355* RAY7
00423*020037- LDA 0# A M E S 3
00424*100400 NEG 0# 0
00425*100004 C0M A
kJ KJ0
0
A
J PLOTTING ?
rs a a o /. JMP
/ | O A * o **
N7
00427*006021- JSR «•PLOT
t»UU*lCd RAY7
/, « •
A A A r t Í A A
JSR «.NEXTP
uvy*4oc u y O u J I
n A y i O A ' A A / - n < » |
JSR « . TXTT
00433*177777 INC0N
00434*006023- JSR «•ICON *•
I N V E R S E CON. ROUTINE
00435*000420* RAY6
00436*000345* RAY2
00437*000371* RAY8
00440*020037- LDA 0# AM ES 3
00441*100400 NEG A
k.' kj
0
A
00442*100004 C0M 0#0#SZR • PLOTTING

0 ?
00443*000403 JMP N7
00444*006021- JSR «•PLOT
00445*000437* RAY8
00446* C20036-N7S LDA 0»AMES2 • FREQUENCY RESPONSE
#
00447*100005 C0M 0# 0 . S N R • CALL# R E Q U I R E S H
00450*000422 JMP N8
i1
A A / . C A A Z A A r t .
JSR «•NEXTP
00452*006031- JSR « • TXTT
00453*177777 FREQ
00454*306024- JSR «•FRESP
00455*000435* RAYÓ
00456*000430* RAY7
00457*000445* RAY8
a a /: #
/ , a A q a a Q A «
LDA 0# A M E S 2
00461*100400 NEG A
kJ kJ
0
/X
00462*100004 C0M A
O #
A C 7 D
O L.Í1
00463*000407 JMP N8
00464*006032- JSR «•NEXTP
00465*006021- JSR «•PLOT ♦
0 PLOT MAGNITUDE
00466*000456* RAY7
00467*006032- JSR «•NEXTP
00470*006021- JSR «•PLOT £ PLOT PHASE
00471*000457* RAY8
0 0 4 7 2* 0 2 0 0 0 3 - N 8 S LDA 0 # TTIN•
•
RESET FOR TELETYPE
00473*040001- STA 0.GETC •
0 IlN P U T
00474*002401 JMP «• ♦ 1
A A /, T C
U V- » H f J
• A A A A A A
U V . * U U V . » V.»
t
STA R T
A A A A A A •
. EN D S T A R T
231
-produce parity
-perform pagination
-generate a carriage return and a

CRLF: -line feed
advance paper to next page

NEXTP:
-format the results

RESLT: -print into columns and pages
-routine to ask questions

QUES:
-teletype non-echoing data entry

GTCT:
-high speed paper tape reader
GTCO:
-cartrifile
GTCC:
-initialise teletype
TTYR:
-initialise high speed paper tape reader
HSPTR:
-initialise cartrifile mag. tape uniy
CARTF:
-initialise analogue to digital converter
A2DI:
Figure A5.3 SERV- Device drivers

232
F i g u r e A5.A A2DI- A n a l o g u e to digital converter driver

233
0001 SERVI
.TI TL S E R V I C E
. EMT T T I N . T T W R # T X T T . C R L F » N E X T P # R E S L T » Q U E S
• EMT TTYR# CARTF # A 2 DI » GTCC» G T C P » G T C T # H S P T R
. E X T D . T X T T » •N E X T P » • R E S L T # N L » N L P 1 »MS K
. EX TD G ETC# PUTC# C 5» C 5 P 1# C 12# C 15# SCM TR
. E X T D A M E S O # I C H M # S R A T E # .G T C T # .G T C P # . G T C C
. EXTM M PAGE# SPCE3» R A Y 3 » F E M T # S T A R T »MES 10# M E S 11

. EXTM M E S 12
•M R E L
0 0 0 0 0 * 0 5 4 4 4 7 TTIN* STA 3# RI 3 9 INPUT A CHARACTER
t • /iA /
r\ r > r \ r> a q a
LDA 3# M S K
A A n A O * A / O /
c uu JO |lu
A
SKPDM TTI 9 I D L E T. T Y P E R E A D E R
A A A A A • A A A 9 7 7
V.» V.» KJ O L* / f f JMP . - 1 9 G E T CHARAC. A M D
/. 9 £
u v.»u u M
r * r\ r\
v D v O lv a a t a
DI AC 0» TTI
0 0 0 0 5 * 163401 AMD 3» 0 » SKP 9 CLEAR READER
0 0 0 0 6 * 054441 TTWR: STA 3# RI 3 99 OUTPUT A CHARACTER
00007 * 040444 STA 0» SAVO 99 S A V E AC0
00010*176400 SUB 3» 3 •
9 Z E R O AC3
00011*101222 M0VZR A
KJ 9
A
9 OC 6
%J
7 P
.tr 99 M O V E 1ST B I T INTO CAR R Y
00012*175400 IMC 3# 3 99 SET AC3 TO 1
00013*101024 M0VZ 0# 0# S Z R 99 ALL BITS CHECKED ?
u v -»v» 1 h rD 99
A A A | /. « A A A ^ ' j r
K J K jv .fi JMP •— 3 M0# GO B ACK
U U U l J ucv.»4*j O
A A A 1 e t A A A /,Q/-
LDA 0# SAVO 99 YES# L O A D THE M0.
A A A | r • 1 A fi A A A
V V V 1 U X X v* \J sJ KJ M0VS A A
C» # C* 99 S W A P IT TO UPP E R HALF
A A A 1 A f | A | 1 A A
KJKJK.>1 i 1V 1 1VU MO VL 0 »0 99 R E M O V E P A R I T Y BIT
VOUCU X
A A A A A f «*7fC.JCUU
O A A
M0VR 3# 3 99
M O V E P A R I T Y INTO CAR R Y
A A A A 1 « 1A | A A A 9
L'V» C lUl OUU1 M0VR *_»#
r\ a
KJ 9 A D D P A R I T Y TO WORD
A A A A A f J A |A A A 9
u u e e lui Ouv MO VS A
KJPA KJ 9 S WAP TO L O W E R HALF
00023*06351 1 SKPBZ TT0 99 I DLE T • T Y P E P R I N T E R
A A A O / i tA A A 7 7 7
fwf k j k j k j JMP .-1 ; G E T CHARAC. A M D
00025*06111i DO AS 0# TT0 •
9 S T A R T T Y P I N G IT
00026*034013$ LDA 3# C 12
00027*116415 SUB# 0# 3# S M R
v.»V»U O V» U Xuuun3>
A A A Q A f A l A A A / ,< £
ISZ ML 99 K E E P T R A C K OF
00031*002416 JM P @R13 99 N U M B E R OF L I M E S
A A A Q A f
U U U O u
A A / . A A C tt
U U ^ t U U J j ) LDA 3.MLPI 9 PRINTED
00033*054004S STA 3» ML
00034*034413 LD A 3# RI 3 9 SAVE REENTRANT
9
00035*054413 STA 3# RI 4 9 ADDRESS

00036*034413 LDA 3# RI 5
00037*054413 STA 3# RI 7
00040*050414 STA 2.SAV2 9 S A V E AC 2
00041*004414 JSR TXTT 9
9
D E L I N E A T E TOP
00042*177777 MPAGE 9
•
OF P A G E
00043*03041 1 LD A 2# SAV2 9 R E S T O R E AC 2
00044*034406 LDA 3# R 1 7
00045*054404 STA 3# RI 5
A A A
U \ .»U H O
f A A A /, A A
U U C H U C JMP 0R14
A A A
u u u h r u u u u u u alOi
A A A A A A
/, *7 § D i 7 • A
KJ
A A A C A f A A A A A A D 1 /■ f A
V.» U V^' J U V.» V» V» v.» U v.* A X H é V.»
A A A C | • A A A A A A D 1 C • A
V.» O I VJ %-* KJ v_» n X u i U
A A A C
v.*uu J o
O ® A A A A A A
u u u v.» u v.» n
D 1 7
I r
•
•
A
K.f
234
A A O O
U U U G SERVI
A A A C O A A A r t A A C A J J A , A
L*W Ut.'UV/ J H W u Z KJ
O A A C /
0 0 C C C 0 SAV2Z
i A
Ci
A ^^c:c
uu u DO 0 5 4 7 7 4 TXTT* STA 3,R15 9 TYPE A TEXT STRING
A A A C^
u u u D O 010773 ISZ Rl 5
A O A C 7
v » o u D / 031400 LDA 2,0,3
A A A £ A A A I A A A
u u u u u y c l C U U LDA 0,0,2 «
9 L O A D FIRST WORD
A A A il 1
o u u u I 034006S LDA 3,MSK «
» LOAD MASK
a a a / q
u u ^ ' O g 163400 AND 3,0 »
9 MASK LEFT BYTE
A A A £ A
u u u O o 004723 JSR TTWR «
9 P R I N T IT
A A A / /. A A 1A A A
v » u u 0*4 W C l o w LDA u, u* c 9 L O A D WORD
A A A ^C 1 A
1 11 Juw
A A A
M0VS 0,0 •
9 S W A P BYTEiS
A A A / /
u u u O O 034006$ LDA 3,MSK
A A A / 7
v.' U \.»0 f 1 63405 AND 3 , 0 , SNR i MASK RIGHT BYTE
A A A 7 A
v ^ v ^ v . * / V.* 00 2 7 6 1 JMP @R15
A A A 7
U U U f
1
1 004715 JSR TTWR «&
9 P R I N T IT
A A A 7 A
U U U /G 151400 INC 2,2 2 INCREMENT POINTER
A A A 7 O A A A 7 r r
v .»u v #O 4 O j JMP .-13
A A A 7 /|
u u v .» / *4 0 5 4 4 0 6 CRLF* STA 3, R3 2 CARRIAGE RETURN
A A A 7 C
u u u /D 020014$ LDA 0, C 15
A A A 7 / A A Z A A 1
u u v .» f O U U U O l u 3 JSR GPUTC
A A A 7 7
SJXJK J t i 020013$ LDA 0, Cl 2 m
0 LINE FEED
A A 1 A A A A £ A 1 A «
».»U 1 \.*O UV/ D U l u l JSR GPUTC
A A 1A «
UC# i V.» 1 00 2 4 0 1 JM P §R3
A A 1 A A A A A A A A A A . A
V U 1u c u u u u u u r \ j ! U
A A |A A •
\JU l u o 0 5 4 4 1 0 NEXTPJ STA 3,R2 , GO TO N E X T P A G E
A A | A i* A A A A 14
U J) LDA 0, C12
A A I A C
v. «c* 1 u D006010$ JSR 0PUTC
A A | A /“
u u 1 v/O 024004$ LDA 1,NL J> FILL CURRENT
A A 1A 7
u u l u f 1 24014 COM# 1 , 1 , S Z R ,• P A G E WITH
u u 1 l u 000774 JMP • -4 ; BLANK LINES
00111 0 0 4 7 6 3 JSR CRLF
0 0 1 1 2 0 02401 JMP 0R2
0 0 1 1 3 Au Au uA uA uA uA IDt Oc ** A
U
00 1 14 0 5 4 4 1 1 RESLT: STA 3, R4 2 RESULT FORMAT

0 0 1 1 5 Au Au £D Au Au |1 <tI •
JSR 9 . TXTT , THREE SPACES
001 16 177777 •
SPCE3 * BETWEEN EACH COLUMN
00117 010012$ •
ISZ C5P1 , K E E P T R A C K OF
A A | A A A A A / fl A / ,
X GU JMP . +4
00121 004753 JSR CRLF 5 NO. OF C O L U M N S

00122 024011$ LD A 1,C5 , PRINTED
00123 044012$ STA 1, C 5 P 1
00124 0 02401 JMP GR4
A A A A A A n / i « A
00125 n*4 • U
00126 0 5 4 4 2 5 OUESi STA 3, Rl 0 REQUEST PARAMETERS

00127 025400 LDA 1,0,3
A A « A A
L* 1 O u 044402 STA 1, • +2
00131 A A / A A 1i ®J> JSR T XTT ,• A S K THE Q U E S T I O N
A
00132 A A A A A A \J
00133 004424 Q 1 S JSR GT C T + 1 NON ECHO

0 0 1 3 4 024420 LDA 1, C60 0 IS R E S P O N S E
0 0 1 3 5 030420 LDA 2, C 6 3 ,• WI T H I N R A lN G L ?
0 0 1 3 6 122533 SUBZL# 1 , 0 , SNC
0 0 13 7 112532 SUBZL# 0 , 2 , SZC
0 0 1 4 0 000773 JMP Q1 ,• NO, WAIT * 0 R IT
0 0 14 1 1| UA JC AU As j AV.* MO V 0, 1 0 AC0 S A V E D
/ uA |
0 0 1 4 2 AV.MA.» D 1AV.' <T*T JSF 0PUTC 0 PRINT RESPONSE
0 0 1 4 3 004731 JSR CRLF
1 O 1A A A
00 144 i cluuu MOV 1,0
0 0 1 4 5 024407 LDA 1, C60 ,• R E T U R N ADDRESS
235
0003 SERVI
0 0 1 4 6 1 22400 SUB 1*0 9
P D E P E N D S ON
00 147 0 3 4 4 0 4 LDA 3* RI *
P RESPONSE
f\ r\ tco tO A ^ a a
U U 1J U 1C Ö U U U A DC 1*1
00151 11 Oc oQ u u u a a a
ADD 1*0
0 0 1 5 2 00 1 4 0 1 JMP 1*3
0 0 1 5 3 uA uA uA Au Au Au RI 1 A
U
0 0 1 5 4 AU AU AU AU Ui;UA C60: 60 ; ASCII 0

0 0 1 5 5 uA uA uA uA O£oA C63* 63 * ASC I I 3
0 0 1 5 6 0 6 0 1 1 0 GTCTJ NI0S TTI ; N O N ECHO F R O M
00157 054407 STA 3* RI 6 ; TELETYPE
•
a r\
CM_» I O w 063610
a * *
SKPDN TTI
00161 0 0 0 7 7 7 JMP • -1
00162 060610 DI AC 0* TTI

00163 034006$ LDA 3*MSK
0 0 1 6 4 163400 AND 3*0 •
* MASK LEFT BYTE
00165 002401 JM P @R16
0 0 1 6 6 UA UA UA UA AU AU R16S A
U
0 0 1 6 7 0 5 4 7 7 7 GTCPl STA 3* R16 •

* N O N ECHO F R O M H.S.
A A |* 7A
V U l 1V 060112 NI0S PTR «
* PAPER TAPE READER
00171 0 6 3 6 1 2 SKPDN PTR
00172 000777 JMP .-1
00173 060612 DI AC 0* P T R
00174 034006$ LDA 3*MSK
0 0 1 7 5 163400 AND 3*0 * MASK LEFT BYTE
0 0 1 7 6 AU AU AC ?f *fu
7A
JMP 3 R16
0 0 1 7 7 0 5 4 7 6 7 GTCCs STA 3* R1 6 •
* N O N ECHO F R O M
A O OA A
^ U 020453
C ÍK J K J LDA C*N0BYT •
* CARTRI F I L E
AA 1
¿VJ 1 1 0 1005 MOV 0*0*SNR
A A A^ O
uv/ ct; ¿ 004414 JSR FILBUF ; FILL THE BUFFER
A A OA Q
uu CU o 03 4 4 5 1 LDA 3* DABP •
» L O A D D A T A B Y T E PT.
AAAA /
UU GU^ 175220i
M0VZR 3*3 •
p W O R D A D D R E S S /2
A A AA C
u u cu 3 021400 LDA 0*0*3 •
# L O A D W ORD
A A AA Z
u u cu O 17 5 0 0 2 MOV 3*3* SZC i O D D OR E VEN B Y T E ?
A A AA 7 tA 1 A A A A A
# luíJVV M O VS U# U
9
P ODD* S W A P B Y T E S
A A A IA
VU\UíCI 034006$
KJ LDA 3*M S K 9
P EVEN* L O A D M A S K
00211 163400 AND 3*0 9
P MASK LEFT BYTE
00212 010442 ISZ D ABP 9
P INC* D A T A B Y T E PT*
00213 014440 DSZ N0BYT 9
P DEC. NO. OF B Y T E S
0 0 2 1 4 uA uA uA My,uAl« JMP .+ 1
0 0 2 1 5 002751 JMP 0R16

0 0 2 1 6 0 5 4 4 3 7 FILBUF: STA 3* R5 9
P FILL BUFFER
00217 0 5 0 4 3 7 STA 2*R6 P ROUTINE
A A AAA
V*u c c u 034432 LDA 3* BBUi A 9
P F I R S T A D D R E S S IN BUFF •
00221 175120 M0VZL 3*3 9

P R E S E T DATA B Y T E PT.
00222 054432 STA 3* DABP
AA001
uuccO 006451 JSR 9RREC 9
P E N D OF F I L E
00224 063077 HALT 9
P ERROR HALT
00225 063454 F U SKPBN CR T I 9
P START READING
U U 000417
c c u JMP OUT
002 2 7 0 6 3 6 5 4 SKPDN CRTI
A A
VU
A A A 7
A O A
vy u
¿G O V.» J * JMP
7 C
r FI •
P WAIT
00 23 060554
l DIAS 0* CRTI 9
P GET THE C H A R A C T E R
A A O O O
U L» C O C 034422 LDA 3* DABP P L O A D DATA B Y T E PT.
0 0 2 3 3 175220 M0VZR 3*3 9
P W O R D A D D R E S S /2
00234 031400 LDA 2*0*3 P L O A D WORD
0 0 2 3 5 1 75002 MOV 3* 3 * SZC 9
P O D D OR EVEN B Y T E ?
0 0 2 3 6 151300 M0VS 2*2 P ODD* SWAP B Y T E S
0 0 2 3 7 175002 MO V 3 * 3 * SZC P EVEN
00 240 143300 ADDS 2*0
236
0004 SERVI
00241*041400 STA A
9
A
KJ 9
'ì
ó
•
» S T O R E IN B U F F E R
00242*010412 ISZ DABP «
9 INC* D A T A B Y T E PT.
00243*010410 ISZ NOBYT •
0 INC. NO. OF B Y T E S
00244*000761 JMP FI
0 0 2 4 5 * 0 3 4 4 0 5 OUTs LDA 3# B B U F A 9 R E S E T D A T A B Y T E PT.
00246*175120 M0VZL 3*3
00247*054405 STA 3 . DABP •
9 S T O R E IN DATA
U V J G J U v j v ^ v O LDA 2 *R6 9 BYTE POINTER
00251*002404 JMP ©R5
0 0 2 5 2 * 1 7 7 7 7 7 BBUFA5 RAY3 i P O I N T E R IO S T O R A G E
0 0 2 5 3 * 0 0 0 0 0 0 N OBYT: V
0 0 2 5 4 * 0 0 0 0 0 0 DABP: A
U
A A A C C I n C * A
U U C J J V.» O
v A A A A a a n/r * A
UV.*C J U n o « V»
0 0 2 5 7 * 0 2 4 0 2 1 $TTYR: LDA l* • G T C T * T E L E T Y P E I N I T I A L I S A T I O N
A A A / A 9 / > O A / | A O
JMP • +2
0 0 2 6 1 *024022SHSPTR: LDA 1* • G T C P *• H I G H S P E E D P A P E R
00262*044007S STA 1 * GETC ; T A P E R E A D E R INITIAL.
O l A A « /lA A
u o g O j U U 1 JMP 0*3
0 0 2 6 4 * 0 5 4 7 0 2 CARTF: STA 3* R16 * C A R T R I F I L E INITIAL.
0 0 2 6 5 *0 2 4 0 2 3 S LDA 1* • G T C C
00266* 044007$ STA 1 # GETC
00267*006406 JSR ©GUNIT * S E T T A P E NO.
A A 0 7
V V G f
A
U
V A A i l /i A £
LA.» JS R ©GFILE «
S E T F I L E NO.
9
0027 1 * 126400 SUB 1* 1
00272*044761 STA 1*N O B Y T

A A 0 7 Q I
u u c f O
^ A O A 7 /)
u v .» c O • J JMP ©RI 6
00274*057534 RREC: 575 3 4 9 P O I N T E R TO G E T A REC.
00275*057307 GUNIT: 57307 «
9 P O I N T E R TO G E T A U NIT
00276*057341 GFILE: 57341 9 P O I N T E R TO G E T A F I L E
00277*054445 A2DI S STA 3* R7
JSR © • TXTT
A A Q A A « A A il A A f <T
V.# O V_* C» V.' v_> v.* I u»
00301 * 177777 MES10

0 0 3 0 2 * 1 7 7 7 7 7 A21: FENT 9 SET INPUT
A
A A 7 A
O
7 • I A A A A A
o 1 c* c* o v.* FDFC L*
00304*040441 FSTA 0* TEM 9 C H A N N E L NO.

00305*074440 FF IX TEM
n A n . - y /;
V_Mw» O O
0« a a
1 U V.» V* V.* V '
a a a
F EX T
00307*020437 LDA 0* TEM+l
00310*024437 LDA 1* C20 •
9 16 C H A N N E L S
00311*101113 M0VL# 0*0* SNC 9 IS IT WITHIN
00312*106512 SUBL# 0 * 1*SZC 9 RANGE ?
00313*000767 JMP A21 9 NO* WAIT F O R IT
0 0 3 1 4 * 040017? STA 0 * ICHN 9 YES* S T O R E IT
00315*061021 DO A 0* ADCV 0 SET TO F I R S T
A A
v'
Q | Zi f A A A A A 1 <T
I O v_» O v.» 1 J ' JSR © . T XTT 0 C H A N N E L NO.
00317* 177777 MESI 1
00 320* 0 0 0 3 0 2 * A22: F ENT 9
0 SET S A M P L I N G
A
«A «A Q O
v.» ^ < O r
1 • f 0A A A A
1 1 o v.» v.* v.» F DFC V.*
00322*040423 FSTA 0* TEM 9

0 RATE
00323*074422 FF IX TEM
.A Q O /, • i A A A A A
v.< v» j r ^ 1 v 1v_» v * v-' v_» F EXT
00325*020421 LDA 0* TEM+1
00 3 26* 100513 NEG L # 0*0* SNC 9 IS IT P O S I T I V E ?
00 327 * 0007 7 1 JMP A22 0 NO* WAIT FOR IT
00 3 30 * 0 4 0 0 2 0 S STA 0* S R A T E 0 YES* S T O R E IT
00 3 3 1 * 00 6001 ? JSR TXTT
00 3 3 2* 177777 MESI 2
0 0 3 3 3 *0 0 0 3 2 0 * 6 23: F ENT 0 SET NO. OF
237
0005 SERVI
00 334•1 P O O 00 FDFC A
V.» SAMPLES
00335*040410 FSTA 0* TEM
00336*074407 FFIX TEM
O O #
\JVJ
| o A A A A
f
l u u u u u FEXT
A A O / . A fA / | A £
LDA 0* T E M + 1
A A
00 341 * 10 1 4 0 0 INC L»# U
00342*040015$ STA 0# S C N T R S T O R E IN S A M P L E
00343*002401 JMP 0R7 COUNTER
A A Q /. /.
U U U U
t A A A A A A
U U U O R7i A
U
A A
U
A /|C fAU
U ^J
A A A A A
U U U U TEMJ A
KJ
A A A /| / » A A A A A A A
u u O h u \ .• c » V.» u u U
A A A /|7
^ ^ 0 ^ 4 /
fv.»
A A
u
A A A A
c C20: 20 16 C H A N N E L S
• END
Enter
( Subroutine^
Save return address

Get result format
Set plotting counter
I
Load plot pointer
I
Load first No.
Find max. and min.
Increment pointer
Decrement counter
Set range
to 6
Multiply by
10
Cal. scale factor
Cal.1Inti]«| Ml / .F.|
CaliINT2| = |MAX/S.F.+.95|
©
F igure A5.5 PLOT- A u t o r a n g ing /a u t o s c a l i n g plotter
239
COCI PLOT
. TITL PLOT
• ENT P L O T
. EXTN Y A X I S , S P C E 3 , F E N T
• E X T D W, D # M > O N E, T W O # • CRLF , • T X T T , C9* C 9 P 1

• E X T P WSA,NL» WW, DD,PUTC, TEN#MI
•N R E L
r\
y
a
y
a
K.' y
a
V.» 0 5 4 5 3 3 PLOT: STA 3# R7 ; PLOTTING ROUTINE
A A A A «
V.» 1 0340125 LDA 3 , WSA ; CHANGE FORMAT
A A A A A A A A t ff
V.» O y <z y y y 1 3> LD A o , w ; SPECI F I F A T I O N
A A A A O
V.» v.' v.» v.» O 0 41521 STA 0, 121,3
A A A A /»
M
A A A A
y <Zy y
A A
y <Z
C
T LD A 0, D
A A A A C
U U L» U J 041522 STA 0,122,3
A A A A /*
u u u u O
A A A A
y cZy y
O'"* C
c ,y J LDA o ,m i ; SE T P L O T T I N G C O U N T E R
A A A A H
V.» V,' / 040525 STA 0, C2í ;
A A A | A
V.» U V.» 1 U 032523 LDA 2 , @R7 5 P O I N T E R TO P L O T
A A A
y y y
1 1
11 177777 FENT SOURCE 9
A A A 1 A A Q | A A A
vy 1 Gi u G 1y y y FLDA r\
y 9
c\
y 9
rs
¿
A A A 1 A 1 A C A A A
y O y 1O i y y y y y FM0V 0, 1
i
r> r> r\
y y.» y 1
/>
H 1 11000 FM0 V 0,2
i c
r\ r^ r\
y uv.» 1 J 0 2 1 0 0 0 PLl: FLDA A
y 9
A
y 9
A
€Z
A A A
y~* y y
\
1O
r
122416 FSUB# i ,o , f s l e ; F I N D M A X I M U M
A A A 1 7
y \J y X /
1 A
1y
C A A A
y y y y FM0V 0, 1
A A A
y.' y.» y
O A
(Z y 112416 FSUB# 0 , 2 ,f s l e ; F I N D M I N I M U M
A A A A 1
L» V.» U G 1
l i t A A A
1 X X y y y» FM0V 0,2
A A A A A
y y y C .< Z
I A /, A A A
X y H y y y FIC2
f\ r\ r\ c\
u u v.» c O 014511 FDSZ C2
o a a a /i
000771 F JMP PLl
A A A A C
U V.' C ^ 044456 FSTA 1 , MAX ; S A V E MAX.
A A A
u u
A £
g D 050457 FSTA 2,MIN ; S A V E MIN.
A A A A T
y* y j y j CZ. i 146403 FSUB 2, i , f s n r ; MAX. > MIN. ?
A A A A A
\w' O 145000 FM0V 2,1 í NO
A A A
y y y
A
O
1
I 125003 F M0V i, i , f s n r ; IS R A N G E = 0 ?
A A A O
y y y O
A
CZ 024477 FLDA i, s i x ; YES, SET IT = TO 6
A A A A A
W OO a
y
r y r s r s c\ /» a
<z y y y *4 3) FLDA 0, O N E ; CALCULATE SCALE FACTOR
r* r\ r\ o
v_» y O
/*
Hr 030475 FLDA 2 , SIX
^ ^ A O C
v»> u v.» O J 1 3 2412 FSUB# i , 2,f s l t ; RANGE < 6 ?
v.»uOO
A A A A £ A A A /i 1 A
F JMP PL3 J YES
O / 034005SPL2:
A A A A A
w ^ FLDA 3 , TWO ; NO
A A A /, A
y» y 004412
y *-* y F JSR PL4 ; S. F. *2
000/11 0 3 4 4 6 6 FLDA 3, T W O 5 ; S. F . * 2* 5
00042 004410 F JSR PL4
A
y
A A y, A
y 034005$
y *H O FLDA 3 , TWO ; S* F • *2
00044 004406 F JS R PL4
000/15 A A yA T T/ G A
y y i F JMP PL 2 ; RE T U R N
0 0 0 4 6 0 3 4 0 1 7 $PL3: FLDA 3, TEN J L O A D TEN
O
A A A /» *7
u v v y M / 1 O y ¿Z y y F DI V 0 y
A
9 9• S.F./10
A A A C a
O o 164100 FMPY 3,1 i R A N G E * 10
/A A\
y y»y
C
y X
I
000764 F JMP PL2-2 ; RETURN
A A A
V.» U L» J G
c O
160100 PL4: FMPY 3,0 J S . F . * AC3
\w»
A A A C Q
y y o O 164200 F DI V 3, i ; RANGE/AC3
a a
v.» v.'D
a c y,
*-* 132415 FSUB# i,2,f s g e ; RANGE = < 6 ?
A A A C C
v_* y y
A A
y y
1 /| A A
X *4 y y F JMP y90A
A
9• NO, RE T U R N
000 56 103400 F RND A
y
A
9 y 9 »
YES, R O U N D S.F.
240
0002 PLOT
•
00057*040430 FSTA 0# SF 9 S T O R E IN S. F.
•
00060*034425 FLDA 3,MIN # L O A D MIN.
00061*114200 FDIV 0# 3 •
# D I V I D E IT BY S. F.
•
00062*054427 FSTA 3 # INT1 # S T O R E IN I N T E G E R 1
•
00063*074426 FFIX INT1 # T R U N C A T E INT. 1
00064*060425 FFL0 INT1
00065*034416 FLDA 3# M A X L O A D MAX.
00066*114200 FDI V 0,3 # D I V I D E IT BY S. F.
•
00067*030436 FLDA 2» P95 L O A D .95
A A r t 7 A | 1 C 7 A A A •
U U U 1U 1 3 / UOV.» FADD 2# 3 # A D D .95 -T O IT
00071*054422 FSTA 3# INT2 î S T O R E IN I N T E G E R 2
00072*074421 FFIX INT2 J T R U N C A T E INT. 2
00073*060420 F FLO INT2
00074*030415 FLDA 2 » INTl 5 L O A D MIN.
•
00075*034416 FLDA 3 # INT2 * L O A D MAX.
00076*156400 F SUB 2,3 # MAX.-MIN.
00077*030432 FLDA 2# SIX * L O A D SIX
00100*156411 FSUB# 2,3,FSGT; M A X . - M I N . =< 6 ?
I A 1 f /. /. A •
A A
V.» 1U 1
A A A
OV.» FJMP PL 5 # YES, GO ON
I A A 1 A A « A A •
A A
u u l u e U U 1 H v.»U FJMP 0,3 # NO, R ETURN
« A •
A A
U
A
1 U O
A A A A A
V.» V.'
A
U MAX: A
A A | A /, f A A A A A A A
V.» 1^ M U U V.» U U L» SJ
A A
U U
1 A
lu J
C f A A A A A A
V.» L* S J \ J \ J ^ MIN: A
u
A A |A ¿ |A A A A A A A
L» 1 V.' O V* L»
8A A A A A A
A A
u u
1 A 7
1u / u u u u u u SF: A
KJ
A A 1 A
u u 1 1 u u u u u u KJ
A A 1 « | VA
^ » ^ 1 1 1
A A A A A
V.» V.» U V.» u INTI: A
\y
A A 1 |A « A A A A A A A
U U 1 i d. \ JK JK A \ J K J y
A A I | n | A A A A A A
<.•\ J 1 1 O \J \ J vy \ j INT2: A
V.»
A A 1 1 /. I A A A A A A A
UU 1 1 H \y V.» V.» U KJ V/
A A « i A*
U U 1 l J \ J KJ U U U ISTRT: U
A A f | / f A A A A A A A
U U l 1 U V.» V.» U V.» U V.1
7 V A A A A A A
A A
uu 1 1 #
1 i
uu u uu u D DA T A : A
U
A A 1 A A 9 A A A A A A A
U U 1 C U U U U V.' U V V»'
00121*037424 r \j y O •
A
KJ *
A A C
KJ KJ Zj
00122*075341
1 0 7 « A /.A
A A
«y ty 1 y u *4 u
A A A
P5: 0- 5
A A | A 7, | A A A A A A
V.' 1 O M *y *y y \y vy vy
0 0 1 2 5 * 0 4 0 3 6 3 P95: 0 .95
00126*031463
0 0 1 2 7 * 0 4 0 4 5 0 TW05 : 2. 5
/A A | A A f A A A A A A
V.' V.' 1 O \y W U U U U
00131*040540 SIX: 6.0

A A | 7 O t A A A A A A
vy *y 1 O e u u u u u u
A A | 7 7 9 A A A A A »*> A
<y \ J 1 O O V.» u u u u u R7 : KJ
A A « 7 /, I A A A A A A A
V»' 1 O H U U U V.' U v.' C2: U
A A 1 7 c « A A A A /. A
V.» u 1 J J u u u u Hu C40 : 40
A A t 7 ¿ f A A A A CO
U V.' 1 O Ü V.» V.' U V.» J o C52: 52
A A 1 77 • A A A A C 7
UU 1 O í V.» U U u J O C53: 53
| y, • r\ r\ r\ r\ r c
a a
V.' \y 1 M V-'
a
\y V.' \y *y y y C55: 55
00141 *0 2 0 7 5 0 PL5: FLD A 0, INTl
00142*0300175 FLDA 2, TEN
0 0 1 4 3 ’ 140100 FMPY 2,0
A A
00144*103400 FRND xj 9 vy
*
00145*040750 F S T A 0 , ISTRT S E T L E F T M O S T Y
•
00146*074747 F F IX I STRT J A X I S VALU E
00147*060746 F FLO I STRT
r > A I c r> 1 <r > n n n / i
v.»v.i 1 J \ ' 1 F EXT
00151*006006? JS R 0 . CRLF
241
CC O 3 PLOT
00152 A A O A v
r
u m o ü j JSR e.CRLF
0 0 1 5 3 126120 ADCZL 1» 1 9 G E N E R A T E -2
0 0 1 5 4 127120 ADDZL 1* 1
«
9 -2*4
0 0 1 5 5 A A A A I1 1| * KJKJKJKJ FENT
00156 C24733 FL DA WINT1
0 0 1 5 7 AU GO U 7# 7O AO a
FLDA 0# SF
/>/> 1
i Ou 104100 FMPY 0# 1
00161 0 3 0 7 4 0 P L 6 : FLDA 2# P C C 5 R O U N D OFF THE
•
P
0 0 1 6 2 1 27405 FRND i # i #f s g e ; PRI N T E D SCALE

0 0 1 6 3 150400 FN EG 2# 2
0 0 1 6 4 147000 FADD 2b 1
0 0 1 6 5 144001 FFD C F 1
•
PRINT Y AXIS
P
0 0 1 6 6 146400 FSUB 2, 1 •
SCALE P
001 6 7 |1 A A A A A FEXT
U U i / v/ Ouu / 3)
A A £ A A »7
\J kj «JSR 3.TXTT
00171 177777 SPCE3
0 0 1 7 2 125405 INC 1»1 SNR 9
0 0 1 7 3 Au A A Oy , A £ JMP •♦ 6
0 0 1 7 4 AU AU AU 1| JC DC f FENT
00175 020712 FLDA 0» SF
0 0 1 7 6 1V 1uuu
f A 7 A A A
FADD 0» 1
001 7 7 r\r\ /Osr\ K »K J K J g F JMP PL6
A A A A A 1 A A A A A
U U Ü U U 1U U V . » V . » U FEXT
A A ¿A A £
uu cu 1 u u O u u ü J>
a a 1
JS R 3.CRLF
A A £ A A O <T « PRINT Y AXIS
O U OU U / «T JSR § • T XTT
A A O A O
UL» G L » C 9
UV.»CU O 177777
A A A A Q
Y AXIS
A A A A /
V V.' H. 0 3 2 7 2 7
KJ LDA 2» 9R7 •
P O I N T E R TO P L O T
rsr\ryr\c AU GA UA UA GA UA <T3>
K,* KJ U LDA 0* M 1 • SOURCE
P
V.» ¿GK.fO 0 4 0 7 2 6 STA 0# C2

v.» <Z # i1 c
A A 0 ^ 7
Ouuu
C A A A A
A DC Ib 1
A A A 1 A
U' V.» 1 V*
<-> 044011$ STA 1# C 9 P 1
0021 1 0 0 0 1 7 4 *P L 7 : FENT
0 0 2 1 2 Ou O i uuu
o 1 A A A
FLDA A
kj 9
A A
\j 9 ¿
m
9 LOAD FIRST POINT
00213 K ' H U U U 1 A /i A A A
FIC2
00214 024673 FLDA 1 # SF •
P S C A L E IT
002 1 5 Au OA uA uA 1| 7f J> FLDA 2* TEN
0 0 2 1 6 1 10100 FMPY C#2
00217 1 U7 UA QG UA UA 1
F DI V Ib 2
A A A A A
V.» C c u 024703 FLDA 1 P5 B 9 A D D (0*5) TO IT
00 221 11 O Q A A A O O ^ KJ FADD l 2 B
A A o o o
o \^' 050675
a, cz C, FSTA 2 b DDATA
A A O O O
\J ^ 074674
CZ. CZ,sJ FFIX DDATA 9 T R U N C A T E IT
00224 060673 F FL0 DDATA
A £ 7A
0 0 2 2 5 Av.»A \_» O fu FLDA 0 # ISTRT •
* C A L C U L A T E SPA C E S TO
0 0 2 2 6 030671 FLDA 2, DDATA P PLOTTING POINT
0 0 2 2 7 1 12400 FSUB 0* 2
A A A Q A
V_» V_» O V.»
Cj 050667 FSTA 2 # DDATA
00231 0 7 4 6 6 6 FFIX DDA T A
00 232 1 v.*uuuu | A A A A A
FEXT
0 0 2 3 3 145000 MOV 2b 1
0 0 2 3 4 A AV.' O£ AV.» AU 7í ^*T JSR @. TXTT
9
P LEAVE 3 SPACES
0 0 2 3 5 000171 * SPCE3
00236 I O lv/uu
| O 1 A A A
MO V Ib 2
00237 A A A 7 A A
V.» í V* v_» LDA 0 # C 53
00 240 0 1 0 0 1 1 $ I SZ C9P1
0024 1 020677 LDA 0, C55 TYPE EITHER A
9
00 242 0 0 6 0 1 6 $ JSR 0PU T C + OR A -

9
00 24 3 0 2 4 6 5 5 LDA 1# D D A T A + 1
0 0 2 4 4 125112 M0V L # 1 # Íb SZC
242
0004 PLOT
00245*126460 SUBC 1» 1
00246*124000 COM 1* 1
00247*020666 LDA 0# C40
00250*125405 INC 1 * 1*S N R
1 t A A A / . O Q
V.» U C J I V.* %J \J *4 \J O JM P • +3
00252* 0 0 6 0 1 6 5 JSR @PUTC •

9 S H I F T C A R R I A G E TO
00253*000774 JMP .-4 9 PLOTTING POINT
00254*020662 LDA 0# C52
00255*006016$ JS R epuTC •
9 T Y P E AN A S T E R I S K
^ A Q C / 1 A C
V.*v_* o O JSR 0.CRLF •
9 CARR. RETURN, L I N E
00257 *02 4 0 1 1 $ LDA 1* C 9 P 1 9 K E E P T R A C K OF
f r \ cr
u c u v.'
r\ r\
ocv.»v» I LDA 0# C9 9 X AXIS SPACING
00261 * 1 25005 MOV 1* U SNR
00262*040011$ STA 0* C9P1
r > r > 0 / ^ o t r\ r \ r \ r\ « o
u o c O o O G v . » U 1 v3 J LDA 0 NL•
00264*101400 INC A
\J »
A
KJ
00265*101415 INC# 0# 0* S N R 9 SUPPRESS NEW PAGE

00266*0140135 DSZ NL 9 ROUTINE
00267*014645 DSZ C2
1
A a 0 7 ^
U \.'G f V
t a a / > 7 a
u u v f d 1 JMP PL 7
00271*0340125 LD A 3, WSA 9 RESET FORMAT
00272*0200145 LDA 0 # ww •
# SPECIFICATION
00273*041521 STA 0 m 121*3
00274*0200155 LD A 0# DD
00275*041522 STA C m 122# 3
7
k.
0 7
» kjcz* 1
A
O • A A
o u Or uA wA / Ch JSR 9 m TXTT
•
9 P R I N T Y A XIS
A A 0 7 7 • A A A O A
V ' V.« O / / U U U C U
7
J
•
YAXIS
A A O A A • A O /, / A A
v.'u u v j O ^ u O O LDA 3# R7
00301*001401 JMP 1» 3 •
9 RE T U R N
. END
243
-* Get
Store data
Noy
Figure A5.6 ATODC- Analogue to digital converter

0001 ATOPC
• TI TL A T O P C O N V E R T E R
• ENT A2DC# INTP
• E X T D I CHN# SR A TE# • TX T T# N # M # N U M # N U M P 1
• E X T D SCNTR# TEN
• EXTN F EM T» RAY 1# RAY2# R A Y 3

rs A n a a f
v.» y y.' O 1 • L0C 1
A A A A t P» «■> P\ | ( T /. »
v.» y y 1 y y.' y 1 D *4 INTP
- ZREL
A A A A A _ A A A A A A T * A
y y y y y y y y y y y W U Cv 1 • VJ ; PAGE ZERO POINTERS
A A A A 1 — A A A A A A C A M T * A
y y.’ y.' i «.• y y y y y O L IV 1 Z VJ
A A A A 0 _ A A A A A A D I I C T l * A
y y y y G. y y y y y y i j U r P ' VJ J NO. OF WOR D S IN B U F F E R
A A A A A _ A A A A A A A
v.» y y y y m »_> y y y y VJ ; A D D R E S S O F FIR S T W ORD IN B U F F E R
A A A A A A A A A A A
M o u M M y y y y y y VJ ; A D D R E S S OF L A S T W O R D IN B U F F E R
00005-177777 RAY 3 ; MIN. A D D R E S S
A A A A £ _ A A A A A A A
U v* O y y y y y y VJ ; MAX. A D D R E S S
A A A A T _ A A A £ /, /,
y y y y / 420. ; MAX. NO . OF WORDS
a A A 1A A A A A A A PLÎ\T C • A
oo*.* i *.» <.<«.< u n N r J VJ
•NREL
0 0 0 0 0 * 0 5 4 5 0 2 A 2 DC: STA 3# SAV3
A A A A ( f J A A Ç A A
SUBZL
y y X
y k
y y G y G y VJ# VJ
A A t
* SET WAIT C O U N T E R
A A A A A f <A / . A A A A
y y y y G. y y y y y STA 0# V C N T ; TO 1 (SYNC.)
A A P A A I 1 1
l 1
A | /.A
M0 VOL
y y y yl y y *4 y
A
KJ 9
A
KJ # SET C L O C K TO
•
00004*061014 D0A 0# RTC ; 1 KHZ

00005*126120 ADCZL 1# 1
A |
A A A i
v.' H H u
/ , /. I A A A _
STA
y y y y y l y 1# CHNF ; SET C H A N N E L C O U N T E R
«A /AC\ 7 1 • 1
1 A A A A A ff
LDA
y y y y l y G y y y J> 0 # SCNTR ; SET S A M P L E C O U N T E R
MUM K' O *-« 1_<<w<I
•A(A/A«/Af<A/i<A<AA»|
STA 0# SCNT
00011*020465 LD A 0» PRAYi ; P O I N T E R TO X
00012*040471 STA 0 # INAD
00013*020464 LDA 0# P R A Y 2 ; P O I N T E R TO Y
00014*040470 STA 0# 0 U T A D
A A Ai
v.» V.'
r t dO\aj a Zj
1 Zj A
KJ k j KJ
A
LDA o # b u f p +3; RESET DATA
a A A | /
V.1V.' V.» I ü
« A A A A A * 7 .„
KJ \J KJ KJ KJ f LDA 2# BUFP+ 5 J BUFFER
A A A
V.»UO 1
1 7 9
# 1i
1 1 O A A A
sJ K.* K j KJ ADD 0# 2
A A A A A
V.* V.» V.» U
f A C A A A /;
V.» J U U U U STA 2# B U F P + 4
A A A A
V.» U v.» 1 I^
1 t A 1 /t A A
V.» U U O £ .
DSZ BUFP+4
0 0 0 2 2 * 101212 M0VR# 0#0 szc # # EVEN B Y T E ?
A A A A A «
K.I S^> d x J
1 A 1 / , A A
1 VJ 1 M V.» U INC ri
y *
r\
y
A A A A / , « A
v.' v_> v j ¿0*4
A A A O .
v.» H u v.» vj O STA 0# BUFP+ 1
A A A A C
V.» V.» V.» Ci J
9 A /. A A A / i .
STA 0# BUFP+ 2
A A A A / ;
U v.» u u O
f f A A /, A A
1 u c M u v . ' SUB P
y »
P
y
C
A A A A H
VJ VJ V.» f
f A /. A A A A -
KJ KJ KJ KJ d STA 0# BUFP
J
A A A A A 9 A A A
V.» V.» V.» V.' JJ C T
LDA 0# TM SK
VJL*L*O 1
A A A A 1 9 A
VJ
£O fi
O A A A
V»
A
MSK0 «
V*
.>
' • M A S K OUT AL L INTER.
#
A A A A A
V» K.» K.1 sJ d
fV.'V.</;
A A A A A
O U U O Î JSR @ • TXTT ; E X C E P T A TO D
A A A A A 9 A A A A C A f
KJ KJ K.» \ J \D K.> KJ KJ d Z J K J M ESS » A N D CLOCK
00034*060210 NIOC tti ; C L E A R T* T Y P E R E A D E R
00035*063610 SKPDN TTI J IDLE T . T Y P E R E A D E R
v»v»v»o O
A A A /■ uuvj f f f
A 9 A A A A A A
JMP .-1
V»v'VJO f «j O/u c l u
A A A A 7 A A A I A
NIOC tti ; C L E A R IT
00040*060177 IN TEN •
9 E N A B LE IN T E R R U P T S
00041 *0601 14 N I 0 S RTC ; S T A R T CLO C K
0 0 0 4 2 * 0 2 4 0 0 2 - GET: LD A 1 # BUI* P J L O A D NO. OF WORDS
245
0002 ATO DC
A /i A
U 1 25005 MOV 1 # 1 * S N R »• LAST ONE ?
00044 000442 JM P EMPTY •
# YES# B U F F E R E M P T Y
00045 014002- DSZ BUFP » NO# DEC* NO. 0F
0 0 0 4 6 A A A /• A t1 KJKJ KJ * A v J JMP • -**1 * W O R D S IN B U F F E R
A A A /l *7 A 7 A A A 7 __
' J U U 4* f v y O vy v y vy o LDA 2*BUFP+1
A A A C A A A | A A A
U U Ji.» u c 1 vy vj vj LDA 0# 0# 2 #
•
L OAD WORD
vy vy v y D 1 155400 INC 2 »3 # INC. A D D R E S S
A A A C A
vy J c 024006- LDA 1# B U F P + 4
a a
u u u
a c
J O
o
136015 A DC# 1*3# S N R #« AT MAX. ADDRESS ?
A A A
u u u
c /i
J*4 034005- LDA 3* BUFP«*3
A A A C C
O U M D J 054003- STA 3# B U F P * 1 # NEXT WORD ADDRESS
A A A
v » u u
C £
J O 034425 LD A 3 # INAD
7
a
u
a a c
J 9 151222 M0VZR 2 * 2 # S Z C #• EVEN A D D R E S S ?
A A A r A
034424 LD A 3* 0 U T A D #* NO# O U T P U T
A A A / 1
u u u O 1 101113 M0VL# 0*0# SNC j YES# INPUT. N E G A T I V E ?
A A A /! A
v» u u D c 126461 SUBC 1*1* SKP ; NO* SET UP M O S T
A A A £ A
o j j D O 1c O u U KJ A DC 1# 1 #
•
S I G N I F I A N T 16 B I T S
00064 045400 STA 1*0*3 ; STORE DOUBLE
A A A £ C
v .* u u O 041401 j STA 0*1*3 # PRECISION NUMBER
A A A £ £
VJ VJ VJ O O 175400 INC 3*3 •
# NEXT ADDRESS
A A A £ 7
VJ VJ VJ O f 175400 INC 3*3
A A A 7 A | A f A A O
V j V j VJ 9 VJ 1 v y 1 k j \j c MOV 0 * 0 * S Z C #• EVEN ADDRESS ?
A A A 7 1 A A A /| A Q •
U U U f I v y u O H v y o JMP •♦ 3 # NO* O U T P U T
A A A 7 O
UJV.» f c 054 4 1 1 STA 3# I N A D •
# YES# I N P U T
A A A 7 O
vy vy vy i O
r t A A / ,/ > Q
JMP • * 2
A A A 7 /|
v j v j vy i *4 054410 STA 3# O U T A D
v.»v.»u f 3 000745 JMP GET
A A A 7 £
U ^ » U / O 177777 PRAYlS RAY1 # P O I N T E R S TO I N P U T A R R A Y S
A A A 7 7
vyvyvy r f 177777 PRAY2: RAY2
A A 1A A A A A A A A T lT M t A
o u 1 vy u vyvyvyvyvyvy I IT J J vy
A A « A 1 A A A A A A A
v y v y X vy X KJ KJ KJ S J KJ KJ vy
A A 1 A A A
JV.» i u c 0 0 0 0 0 0 SAV3: vy
1A
A A
vy vy 1u O
O
0 0 0 0 0 0 INAD: A
vy
A
A A | A / .
I vy »4 0 0 0 0 0 0 OUTADl vy
• A
A A
vy vy 1 v»
C
J 1 7 7 5 7 3 TMSK* 177573
•
A A 1 A £
vy vy i vy D 0 2 0 0 0 1-EMPTY: LD A 0# SCN T # SAMPLE COUNTER
«
A A t A 7
vy vy 1 vy 9 101004 MOV 0 * 0 # SZR # A L L S A M P L E S TAKEN
A A 1 I A •
vyvy X 1 v» A A A 7 7 0
vy v/ vy 9 J c JMP GE T # NO
•
00111 vy *4 Avy vy A
A A £ <T
vy O o> STA 0# N U M # YES
•
00112 0 3 0 7 6 4 LDA 2# P R A Y 1 # P O I N T E R TO X
A A A A 1 A C
0 0 1 1 3 v y c v y v y I v i 3) LDA 0# S C N T R
0 0 1 1 4 11 Avy Av y/*4. Avy Av y NEG A A
v y # vy
0 0 1 1 5 11 vAy Av y vAy Av yAv y COM A

vy # vy
A
001 1 6 0 4 0 0 0 4 $ ST A 0#N # INPUT ARRAY

•
001 1 7 0 4 0 7 6 4 STA 0# I N A D # COUNTER
1A A /, A
A A
vy vy 1 c* \ j
A A A 7
v j h u u v j •
tf
J STA 0» N U M P 1
00121 177777 FENT
0 0 1 2 2 Avy £O AvyAvyAvy£ Of fJ> FFL0 NU M
0 0 123 Av O7 oA oAv A/ U£ CJ FLDA 2#NUM
0 0 124 02401 li F LDA 1# TEN
•
0 0 1 2 5 1X 7 A ' r\ r \ j v
a
c u v j FDI V 1*2 # SET M TO N / 10
00126 050752 ESTÀ 2# TEM
«
001 2 7 074751 FFIX TEM # TRUNCATE M
A A | O A •
v J vy 1 J u 06 4 7 4 7 FLD3 PRAY2 # P O I N T E R TO Y
00131 020421 FLDA 0# C2048 # B I N A R Y TO V O L T S F A C T O R
0 0 1 3 2 0 6 1 0 0 0 CON: FFL0 0*2 •
9 CONVERSION ROUTINE
0 0 1 3 3 Avy O7 XI Avy Avy Avy FLDA 2*0*2
0 0 1 3 4 1X 1i v*A Q A A
vy vy FDI V 0*2
00 135 vy D XA C 1 A A A
FSTA 2*0*2 *
# S T O R E IN INPUT AR R A Y
246
A A A
u v. » u ftTO DC
O
O
CO 136 0 6 1 4 0 0 FFLO 0*3

CO 137 0 3 1 4 0 0 FLDA 2* 0* 3
CO 140 1 I A
I l e #
A A A
GUV^ FDI V 0*2
OC 141 0 5 1 4 0 0 FSTA 2*0*3 •
0 S T O R E IN O U T P U T A R R A Y
00142 1 A / ( A A A
1 u h u u v ; FIC2
00143 1 | A A A A
1 1 U O U U FIC3
00144 014737 FDSZ I NAD i DEC. COUNTER
0 0 1 4 5 e*e* e# C O A ^ A 7
#
/; r
FJMP CON
00146 f A A A A A
1 k jk jk jk jk j FEXT
00147 024732 LD A 1* T E M + 1 0 LOAD INTEGER M
O F> 1 C 4
044005$
e# v-/ 1 D e » STA 1 *M •
0 S T O R E IN M C O U N T E R
00151 0 02731 JMP @SAV3
0 0 1 5 2 0 4 1 3 1 4 C 20481 204.8
0 0 1 5 3 146314
0 0 1 5 4 0 4 0 4 2 2 INTP: STA 0* SAVO •
0 INTERRUPT ROUTINE
00155 044422 STA 1 * SAVI •
0 S A V E A C C UMS.
00156 050422 STA 2* SAV2 0 0 TO 2
00157 0 6 3721 SKPDZ A DCV ; A TO D I N T E R R U P T ?
A A 1 Z A
000422
K JK J 1 Oe* JMP ADCT •
0 YES
00161 0 6 3 7 1 4 SKPDZ RTC •
0 CLOCK INTERRUPT ?
00162 000452 JMP TIME •
0 YES
00163 061477 INTA A
KJ
•
0 NEI T H E R * G E T D E V . C O D E
00164 024415 LDA 1* C L E R •
0 CLEAR INTERRUPT
0 0 1 6 5 le* 9 1 A *7 A A />
k j k J kJ ADD 0* 1
0 0 1 6 6 0 44401 STA 1*. + 1
0 0 167 r\r\r\
a /: a
K j O K j < Z s K j KJ Nf'QC A
KJ
A A
\ J kj
1 7
X 9
A
kj
A A A /. O
JMP DISM+1
00171 0 3 0 4 0 7 DISMS LD A 2* SAV2 •
0 DISMISS ROUTINE
00172 024405 LDA 1* SAVI •
0 RESTORE A C S •
001 7 3 020403 LDA 0* SAVO

00174 060177 INTEN •
0 ENABLE INTERRUPTS
00175 A A A A A A
V<U Gv;U\y JMP @0 •
0 RE T U R N
00176 A A A A A A C A T JA 9
j H V u *
A
e*
00177 A A A A A A C A TI 1 •
O H v l •
A
KJ
A A A A A A C A IIO * A
V.» V» C V.» U C» U V.» U V.» OH V C # V.*
A A A A
e; e» o e' X
|
0 6 0 2 0 0 CLERI NIOC A
KJ
A A
v.»u cu c
A A A
0 62721 ADCTl DI CP 0* ADC V
A A
u u cu O
A A Q
060121 NI OS A DCV
A A
e» V_'
A A /.
e» *4 0 2 4 0 0 2 - ADI s LDA 1 * BUFP 0 NO. OF WORDS IN BUF .
A A A A
e' V.' C U J
C A Q A A /> O _
e» 3 e' e» k j 9 LDA 2* B U F P + 5 ; MAX. NO OF WORDS
/-xa r^r\ s
v .»U Cu U 132415 SUB# 1*2* SNR 0« S I Z E OF E U F F E R ?
•
A A
V-'
A A 7
O V.» 9 063077 HALT 0 B U F F E R FULL* F A T A L
A A
UUG 1 U
A 1 A A 1 A A A O—
e» X <.> k j k j & I SZ BUFP
0 0 2 1 1
A O A A A / i ^
e» O e» e' e# *4 LDA 2* BU F P + 2 ; N E X T AVAIL. A D D R E S S
0 0 2 1 2
A /i 1 A A A
uH X u u o STA 0*0*2 0 STORE WORD
0 0 2 1 3 141400 INC 2*0 0 INC. A D D R E S S
00214 024006- LDA 1* B U F P + 4
00215 122015 ADC# 1*0* SNR 0 A T M A X . A D D R E S S ?
00216 A A A A A
e* o e» e' e* 3
C .
LD A 0* BUFP+ 3 0 Y E S * G ET M I N . A D D R E S S
00217 STA 0* B U F P + 2 0 S T O R E N E X T A V A I L . A D D .
A A
v.' e'
QO
CPG, e»
A 1 A A
e» 1 e»
1 A , .
X e' I SZ CHNF 0 I N P U T OR O U T P U T ?
•
00221 000406 JMP N EXT 0 INPUT
00 2 2 2 0 2 4 0 0 1$ LDA 1 * I Ch'N 0 O U T P U T
00 2 2 3 065021 D0A 1* A DCV 0 S E T A TO D M U L T I P L E X E R
00 224 024407 LDA 1 * CM 2 0 TO I N P U T C H A N N E L
00 2 2 5 044010- STA 1 * CHNF 0 RESE.T FLAG
00226 000743 JMP DI SM
00227 0 6 3 621 NFiXTl SKPDN ADC V 0 GET N EX 1 A TO L
0 0 2 30 000777 JMP . - 1 0 CON VERSI ON
247
0004 A TO r e
00 231 062621 * DI CC 0» ADCV
00232*000752 JM P ADI
00233* 177776 CM2: -2
0 0 2 3 4 * 0 6 0 1 1 4 TIME: NI0S RTC ♦
9TIMING ROUTINE
A A
V.1u c
R
O J
^
U
1 /i^
1 t v J u u
^
DSZ WCNT i DEC. WAIT C O U N T E R
A A 0 7 4
v.i v.i ¿o O D
t a
v.»
a a h
v.» f
o /:
O O JMP DISM+3 •
9N O T R E A D Y TO START
00237*060121 NI0S ADCV •
9S T A R T A TO D C O N V E R T E R
A A O/l
U V.»
A • A A A A A Aff
Ov.'v.'v.'cl) LDA 0 » SRATE •
9RESET WAIT COUNTER
A A A /| 1 f A /. A A A A _
v > C . *4 1 H V.» V.» V.» STA 0# W CNT
00242*014001- DSZ SCtNT i SAMPLING FINISHED ?
00243*000730 JMP DI S M + 2 •
9NO
00 2 4 4 * 0 60 221 NIOC ADCV 9YES» C L E A R A TO D
00245*060214 NIOC RTC 9CLEAR CLOCK
00246*020730 LDA 0# SAV0
A A O /, 7
f
• A A
V.»
A A A
CL K.9
A
JMP *»0 0 R ETURN
9
MESS: •TXT **< 15>< 12>< 1 2>T0 START» S T R I K E

A A A C A | A A C A f Q
L' U C J V./ UV.' J U 1 J
00251*052012
00252*020117
00253*052123
00254*051101
00255*026124
00256*051440
00257*051124
00260*045511
00261*020105
0 0 2 6 2 * 0 4 7 1 0 1 ANY KEY< 1 5>< l2> < 12>*'
00263*020131
00264*042513
00265*006531
00266*005012
/A O *7 t A A A A A A
V.' V.' o O / u U S.» V.» V.»
END
248
F igure A 5 ,7 DATAIN- Ge n e r a l d a t a e n t r y r o u t i n e
249
0001 DATAI
• TI TL DATAIN
• ENT DATAI N
• E X T D . C RL F * • T X T T , N # M , N U M , M U M P 1, TEMPO
•EXTD TEMPI,C15,WSA,AMES1,TEN
• EXTN F EM T, M E A N
-MREL
V» KJ KJ K.» 0 5 4 4 5 2 DATAIN2 STA 3, R5 5 D A T A R E A D IN,

A A A A 1
V.» V.» K.I K.» X 176400 SUB 3,3 #
p AMD STORED
A A A A A
K.» K.* V \J CZ 054005$ STA 3,NUM •
p
A A A A A
c* u u \ . » o 03 2 4 4 7 LDA 2,0R5 p STORAGE POINTER
A C A A
A A A A
KJ U
/.
U J O l »
« A
1 U
ff
ï STA 2 , TEMPI p LOADED HERE
A A A A
KJ V' KJ
C
J 126400 SUB 1» 1
r \ r \ r > r \ ¿r
K.J K .IK .9 X J O 044003$Dlî STA 1,M ♦
p START COUNTING
A A A A O
V.» L* K.» K.» f 044006$ STA 1 MUMP 1
, p NO- OF DATA P O I N T S
A A A
V.' V.» U
I A
I U 024445 LDA 1 C 1 15
,
A A A 1 1
KJ KJ K.9 X 1
A A A A % O C
LDA 2» VSA
A A A 1 A
V. » UL» 1 C 177777 FENT
A A A 1 Q
V*V.*V» 1 O 124000 D2s FDFC 1 p GE T THE CHARAC.
a a a ) /. 1 A A A A A
1 U U U U c» F EXT
A A A
V.» U O
1 C
1 J
A
v
o c A A 9
*3 D o o g LDA 3# 2, 2
A A A
K.» KJ K.*
1 /:
1 O 166415 SUE# 3# 1, SN R • IS IT AN M ?
p
A A A
U U V »
1 7
1 / 000417 JMP D4 p YES, L A S T D ATA
A A A A A
V.1V.» V»* 035001 D32 LDA 3, 1,2 •
NO, IS IT A
p
A A A
U U U
O
G
1
1 175004 MOV 3, 3, SZR p• B R E A K CHARAC- ?
A A A O O
K.» KJ K J C Z CZ
A A A /. A A
O U U H u O JMP • +3 • YES, S T O R E NO-
p
a a
V.» L*
a
U
a
G O
Q A A A A
o o o o
1 O
X g
•
F EMT
r% r\
v.' v_»
o /i
O *4
A A A A ZI *7
KJ K J KJ f O f F JMP D2 p NO, GET N E X T CHARAC-
A A A Oc;
O U G J
A A A A Q A
KJ K J K.» \ J C Z O
•
F EMT
A A A O A
v. î u u g D 064010$ FLD3 TEMPI
A A A
K.’ K.* KJ <Z
0 7
f 045400 FSTA 1,0,3 p S T O R E IN ARR A Y
A A A
u u v/ O
A A
X
«
X 1 A A A A
O O O O F IC 3 •
p INC. P O I N T E R
A A A A
U V
.*G» O
1
1
f
X
A 7 A A
O f KJ KJ
A
O
ff
^ FST3 TEMPI
A A A A A
KJ KJ KJ O ( Z X
1 A A A A A
K.I KJ KJ KJ KJ F EXT
A A A O T
KJ KJ KJ O 024003$ LD A 1 M , p I N C R E M E N T DATA
A A A A /,
O v.' v j O H 125400 IMC 1, 1 •
p COUNTER
A A A
«J U U
A C
U J v u v » / O I JMP DI
A A A A /:
o Kj o O 0 0 0 0 2 5 ’D4: FEMT
A A A A 7
V.# V.» V»* O /
A Z A A A C Ç
o O o o o D FFL0 NUM p C A L C U L A T E NO- OF
A A A /.A
KJ KJ K j KJ 030005$ FLDA 2, NUM •
p CORRELATIONS REQUIRED
v.»*41 0 2 4 0 1 4 $
A A A /l
c/
1
FL DA 1 TEM
,
A A A
V.» V.» *4 X /iA
tZ
1 A A A A A
J u g u v .» F DI V 1,2 •
p N/1C
A A A /i A
v.» U H J 050410 FSTA 2, TEM
00044 074407 FF IX TEM
000 4 5 X 1 A A A A A
V,' G» V.' G* FEXT
00046 024406 LDA 1 , TEM+ 1
STA 1, M • S T O R E IN M
A A A
KJ K^**-*# 0 4 4 0 0 4 $
/i 7 p
V.'UV.»Jv.» 0 3 4 4 0 2
A A A C A
LDA 3, R5
JMP 1,3 • RETURN
V» Jc»J 1 001401
A A A
V
C 1
p
0 0 0 5 P v.» u u nj•
•
A A A A A A
V.
» v.*
D C
V.»
A
000 5 3 0 0 0 0 0 0 TEM: A
K.»
A A A A A A A
000 54 V.KH»U G' K.9
*
A A A C C
V
.V.» -J 000 1 15 Cl 152 1 15 p ASCII M
F igure A5.8 LIST(1)- No r m a l i s e da t a
251
F i g u r e A5.9 L I S T (2)- D a t a listing and punching

252
0001 LIST
• T ITL L I S T
• ENT LI S T 1 # L I S T 2
. EXTD .CRLF#.TXTT#N#NUM*TEMPO#TEMPI#AMES1
• E X T D M# M 1* • R E S L T # •PLOT# C5* C 5 P 1
• EXTN F E N T # M E A N
•N R E L
r\ r\ r\ r\
y y y y 0 5 4 5 0 2 LISTI: STA 3# R5 ; CALCULATE MEAN
A A O A
y y y y
1
1
A A A A
y CZy y y O x
A A (T
LDA 0# N ; AND NORMALISE
A A A A A /| STA
y y y y CZ
A
U ^
A A A c ff
U U U j i 0 * TEMPO
%.»
A A A A
y y O 177777
Q
FENT
/»
A A A A
U V.» D» M 102400 FSUB 0#0
A A A
y y y j y Zj
A C
066475 FLD3 3R5 ; P O I N T E R TO A R R A Y
a r
v*'v.»
a a a
O 0 2 5 4 0 0 Dl: F LDA 1 * 0*3 ; LOAD NUMBER AND
A A A A 7 1
y uuu / 1 cOv»uu FADD 1 *0 ; A D D TO SUM
L'V.' 1 y 1I I y y y y
A A A 1 A | A A A A
FIC3 ♦ INC. P O I N T E R
f
A A A « 014005 S
1
y.»yty* 1 1 FDSZ TEMPO ; DEC. C O U N T E R
1
A A A
W % J I CZ 000774
A
FJMP Di
1
y y y 1 O
A A A
024004?
O
FLDA 1*NUM
i /. 1
X C Z y C Z y y^
y y y 1 *4 FDI V 1*0 ; CALC. M E A N
A A A I r
L»U U ID l UUUUV.»
1 A A A A A
F EXT
r \ r \ s rs r \ i ^
y y y 1 O y y y y y 1 J JS R @.CRLF
vJ y
A A A
y 17/
1
y y \ D y y cZ x
c JSR @ •TXTT ; PRINT HEADING
A A A A A
CZy \ m n MEAN
A A A A
y y cz 1
1 A A A A A
y y y y y sj
A § F ENT
A A A
y y y <Z CZ
A A
140001 FFDCF ^
A • PRINT MEAN
»
v.#y j y
A A A A A
r. O «
1 A A A A A
u u o v / u FEXT
A A A
y yyCZ**
A/i A A £
y y y y y 1X
A A | ff
JSR @.CRLF
A A A
y y y c z D
A C
uCuuu
A
LDA 0# N
L* V.»O
A A A il
CZ O v.#
A
M
A
^• A A C
O
C
X STA 0* TEMPO
r\ 7
A A A
v y y cZ t 032453 LDA 2#@R5 ; P O I N T E R TO ARR A Y
r\ r \ r \ a a
y y y O A A A |f
y y y y c z 1
A A
FENT
A A A
y y y
A|
o I A A C A A A
y CZ y y y y
H O«
U CZ * FLDA 1 * 0# 2 * LOAD NUMBER
A A A A A
y y y % j CZ 106400 FSUB o# i ; MEAN SUBTRACTED
V.»U V'O O
A A A A A
M Oy y y A C O A A
F STA i * o #2 ; STORE NORMALISED
v-'v' O /. 1
A A A A A / . A A A
FIC2 • DATA IN A R R A Y
p
r \ />
L'V .» O O 0 1 4 0 0 5 S
o c
FDSZ TEMPO ; DEC. C O U N T E R
u v.' O O / y y y f i y F JMP
A A A A
V.»
A A A 7 7 7
D2
v.' O 7/ Iy y y y y
A A A A 1 A A A A A
FEXT
L» /. 0 3 4 4 4 2
A A A A
LDA 3* R5
0004 1 001401 JMP i#3 ; RET U R N
000 4P 0 5 4 4 4 0 LIST2J STA 3* R5 ; LISTING OR PLOTTING
00043 031400 LDA 2*0*3 * P O I N T E R TO ARRAY
00044 A A A A
c z y y y y
A A
X LDA 0# N
0 0 045 /i / STA
A A A A
*-* y y y
ff
O X 0* TEMPI
0 0 046 0 2 4 0 14S LD A i* C5 ; RESET RESULT FORMAT
V'y y M/l*/
A A A 7 044015$ STA 1* C5P1
^ v.»a c
A
O ^ 024007S LDA 1* A M E S 1
aaac 1
y y y O 1 125004 MOV l* l# s z r ; DATA LISTING ?
^
y yn yr \ ZD
c cz
o 000414 JMP D4 ; NO* SKIP OVER L I S T I N G
t~> O <
"
cQ
rs r\
y y ZD y
r\
vî v .' y y J v ,'
>»Lnot
' JO • FENT
000 54 r D < »' v.< FLDA 1*0*2 J LOAD NUMBER
00055 144001 FFDCF l ; L I S T IT
•
1( r.>iHUL'V.1
y,n n n F I C2
00056
253
0002 LIST
/“% /*\ /> C *7 • 1n n n o r t
L.' L/ o D / * U U U U v . ' FEXT
A AA
v .» v . » u O u •050423 ST A 2 # SAV2 •
9 S A V E AC 2
A r i
o v-f O 1 •006012$ JSR © • RESLT
A ^ A / n
u u u D c •030421 LDA 2* S A V 2 #
•
R E S T O R E AC2
u u o U O •014006$ DSZ TEMPI
A A A / /, • <•><■> r> *
7¿*7
U U U U H u u u / O f JMP D3
A A A / C
U U O J *034415 LDA 3# R5
A A A / /
O O • 1 2 4 5 1 3 D4J NEGL# l#1#SNC • DATA PLOTTING ?
9
A A A / A
v / u ^ O # •001401 JM P 1,3 *
•
NO# E X I T
A A A H A • r t O A> *A t « % C
O U U f V.» u c u u It* « LDA 0#M ; YES# SET C O U N T E R
a a a *7t
V.*L»U / I •040011$ STA 0#M 1 ,
*
T O P L O T M OR N V A L U E S
A A A 7 A
yj\-> s** i •026410
o LDA 1 • @R5
A A A 7 A
U S J U # O •044402 STA I s +2
*
A A A 7 /,
U V J U / •006013$
*4
JSR ©•PLOT ,• P L O T THE A R R A Y
0 C 0 7 5 1A A A A A A A
U
7 /
A A A
v.»u u /O • 0 3 4 4 0 4 LDA 3# R5
A A A 77 *V/ IU J>
A A A A 1 A ff
LDA
K J W i i 0, M
A A « A A
u u i v u •040011$ STA 0.M1 s RESET M COUNTER
uu
A A 1 A
l u
1
•001401
1 JMP 1,3 ,• R E T U R N
A A 1 A A 1A A A A A A DC# A
u u l u c \J Kt Kf KJKJ %J> f \ D # U
A A
V-iU IU J w u w u w u jCAXIO#
| A Q ■ A A A A A A
H V c «
A
KJ
END
Figure A5.10 CORREL- General correlating routine
001 CORRL
• T ITL C O R R L
• ENT RR
• E X T D • R E SLT#N*M# J * C 5 * C 5 P l # TEMPO# TEMPI
• EXTN F ENT
• NREL
A A A A
*054461 RR: STA 3 * R9 9 CORRELATION ROUTINE
A A A f
V
_» Kj V» I *0240055 LDA 1*C5
•
9 R E S E T R E S U L T FORMAT
A A A A A
CL * 0 4 4 0 0 6 $ STA
K J KJ XJ
1* C 5 P 1
A A A A A
O * 0 3 5 4 0 2 •
v.* xj LDA 3* 2* 3 9 POINTER F OR STORAGE
A A A A /.
UUUO H *054010$ STA 3 TEMPI
m
A A A A C
s»' v.' v.' O * 1 5 2 4 0 1 S UB 2# 2 * S K P M J=0
/\ s s r \ r
\~J x.* \ j c' D * 1 5 1 4 0 0 RR1: m
INC 2 m 2 9 SET COUNTERS
A A A A *7 t A C A A A /i
W#I STA 2 m J
AAA 1A A A
v
.#V#u 1V.I * 102400 S UB X.9 9 XJ
AAA 1 |
UL»U 1 1 *040446 STA C#NJF
AAA I A
L»V'U 1 G * 0 3 4 0 0 2 $ LDA 3#N
•
\
»>o 1O • 1 5 6 4 0 0 SUP 2 3m 9 N-J
AAA 1 /• •
1M * 0 5 4 4 4 2 STA 3* N J
x j x j \j
9 STORE IN FLOAT. LIM I
AAA 1 r
c' x j i O * 0 5 4 4 4 3 STA 3*NJF+1
AAA 1£ •
tw
*v.* 1O *141120 M0VZL 2 0m 9 2 * J
AA A 17
S
-*\
^*X J i / * 026442 LDA 1# SR9
A A A AA t 1A 7 A A A •
V.'V/U CL L' i u / u u u ADD 0# 1 9 SET P O IN T E R
A O 1
^ U k .iL 1 *044007S STA 1 TEMPO
m 9 TO X C I + J )
AA A A A •A A A A A A e
V.' k.' U G C U LU U U O T LDA 0*M
AAA A Q
V.' V
.»V.' G O *034436 LDA 3#R9
A A A A/1
v.» ^ g: *4 •112415 SUB# C 2m SNR m 9 J=M ?
A A A A r
U U*U G J *001403 J MP 3 m 3 9 YES* E X IT
A 0/1
k JU U G U *031401 LDA 2 m I m 3 9 P O I N T E R TO D E L A Y E D
AAAAH
UUUG / * 1 7 7 77 7 FENT 9 S I G N A L LOADED HERE
A A A OA A /A
u u u j u *102400 F SUB XJ 9 XJ
A A A A 1
V.1V
-*kJ O i *064007$ FLD3 T E MP O 9 P O I N T E R TO R R C J )
A A A A A « A A 1A A A ODO•
v.*\
>»\
^*O u o i u u u n r \ cl • FLDA 2*0# 2 9 LOAD X C I )
A A A A A
V' vj OO *035400 FLDA 3*0*3 9 LOAD X C I + J )
a a a O/ i f 1A A kA A •
V.I \.i o M 1 /V
.« 1 G# F MP Y 3* 2 9 X C I)*X C I+J)
A A A A C
V U U*O J * 143000 FADD 2*0 9 A DD TO SUM
f 1A / .A A A
000 3 6 1 k.'*lVUkU FIC2 9 INC. POINTERS
9 1 1 AAAA
00037 l 1 UV.' k
^»V.' FIC3
A A A /, A
V V H 'H U*014416 Frsz NJ 9 DEC. COUNTER
0004 1 * 0007 7 1 F J MP RR2
00042 *060415 F FLO N JF
0 0 0 4 3 * 0 2 4 4 14 F L DA 1* N J F
00044 V1 1A A A A A
f.k.»f .\.»V FDI V 1* 0 9 CAL. RRCJ)
0004 5 *0640105 r L 13 tempi 9 LOAD STORAGE P O I N T E R
A A A •
0004 6 •041400 F STA ^ • 9 XJ 9 O 9 STORE I N R R C J )
r\
0 0 0 4 7 * 140001 F F DC F 9 P R IN T RRCJ)
A A A C A
v *U U Jv .' * 1 1 0 0 0 0 F.I C3 9 INC. P O I N T E R
9 «AA A A k A A
00051 v' K ' o FST3 TEMPI 9 STORE I T
• | AAAAA
00052 1 U U U V
.*k
.» F EX T
A A A r A
v*^' k.* 3 3 * 0 0 6 0 0 1 ? JSR £ • RESLT 9 F 0 RM A T TH E R E S UL T
00054 *030004? LDA 2* J 9 L O A D COUNTER
00055 •000731 JMP RR 1
f AA AA A A \T 1t /A
0 0 0 5 6 V■O \J»V* V* V-* V X J • V»
**\
0 0 0 57 • 0 0 0 0 0 0 N J F :
256
V >>.■O 2 CORRL
U V.»U v.» U U R9:

257
F igure A5.11 IMPUL- Sy s t e m impulse function

oooi ir* p u l
• TI TL IMPUL S E
• ENT H
• EX T P C 5# C 5 P 1 # • N EX T P # . I X T 1 # RXXO * J » K # M- » S Uh
. EX T L SUr* P I * T EN. PC » T E M P I , . R E S L T * D E L û M
• EXTN IM PUL» F EA3T
. N R EL
A A A A A
V'V.' V' V» 0 5 4 5 0 4 H: S T û 3» R I O P IMPULSE F UM C i l 0 M
A A A A 1
V.»V* V.» V.» 1 0 P400 15 LPA 1 »C5 P ROUTIME
A A A A A
V.»V» V.* V» C * 044002$ STA 1* C 5 P 1
A A A A Q A A /•A A A
V
.* \»* O L* vJ C* vj v.» O J ' JSR ® • N EXTP
/> r\ r\ ^ A A / A A /, (
V» U U U U H Î JSR e - TXTT •
P PRIMI HEADIMG
c
V.* V
. »*>
_» O 177777 IMPULSE
A A A A ^
177777 F EM T
A A A A T
V' V» V.» L' / 066475 F LT 3 «RIO P POINTER TO RXX
A A A I A
V. ' V' V.' 1 V' 021400 FLl A A A O
9 vj P vJ
A A A 1 1
V* V»V» 1 1 040005$ F STA 0» R X X 0
CCO 1 P 070014? F ST 3 TEMPI
0 0 0 1 3 J A A A A «A F EX T
0 0014 126400 SU E 1» 1
AA A 1 C
V.' V .» 1 D 044006? SIA 1» J P SET J=0
A A A 1 /■
\w ' V*V> 1 P 034466 L L 'A 3 » R 10
A A A 1»7
V.1V.' L' 1 / 025401 L P A 1» 1»3 P POINTER TO RXY
A A A A A
$»V» O r * v.» 044013? S TA 1» TEMPO
/A r\ r s o 1
o V* r . I 10 2 4 0 0 H 1 ï SUE P
A
VJ
A
VJ
000 2 P 0 3446 P L L A 3 » R 10
/•
0 0 0 P 3 V.' M A A A A 7
V.»v. »«j f J
<r
S T A 0» H P SET K =0
0 0024 0 4001 1$ S 1A 0 » S UM- P SET SUM TO 0
0 0 0 P 5 0 4 0 0 1 P? SIA 0» S U M P 1
000 P 6 0 3 1 4 0 2 M 'A 2» P» 3 P P O I N T E R TO Hirt)
0 0 0 P7 0 3 4 0 1 4 $ L PA 3» T E M P I P S E T P O I N T E R 10
A A A OA
V_* V..1\.* O V
.• 024006? L P A 1» J 1 R X X ( J - K ) W I T H n=
0 0 0 31 1 P 7 0 0 0 ADP 1 » 1
000 3 P 1 37000 A PI 1 » 3
000 3 3 0 5 4 0 1 4 ? STA 3 » 1 EM P 1
0 0 0 34 0 P 4 0 0 6 ?H2: L P A 1 » J
000 3 5 10 6 4 1 5 SUE# 0* 1» SMR P K = J ?
0 0 0 3 6 .A /A »A n A
V.* $»v.» M O V» JM P H3 P YES
A A A A A /; «
0 0 0 37 V*VJ V' \J V# V? r EM T •
P K< = J - 1
A A A /. A
V»V.» *-* * 064014? FLP3 T EMPI
Q •
000 4 1 0 P 1400 FLPA A
VJ 9
A
VJ » O P LOAD RXX(J-K)
0 0 0 4 P /A A C A A A
U 6 J U VJ O rLDA 1 » 0» 2 P LOAD H( K )
1
0 0 0 4 3 1O V.‘ V« VJ
,-N /l A A A
F IC2 P INC. K POINTER
000 4 4 014014? F US Z TEMPI P DEC. J PO IMI E R
00045 014014? F PS Z I EM P 1
00046 104100 FM P Y 0 » 1
0 0 0 47 0 2 0 0 1 1 ? FLDA 0» S U M
V'
«A «A A C A
V* VJ J u 1P 3 0 0 0 F Apr 1»0 P ADD TO S U M
000 5 1 0 4 0 0 11? F STA 0» S U M
O
6> <A *A C O
v» v.» sj 1
« A A A A A
VJ Vj V.* VJ VJ f f:x t
0 0 0 5 3 »i «**
/A 1 A A A O <T
V V VJ V I SZ K •
P I MC * K
00054 CP0007 ? LDA 0 »K
00 0 5 5 000757 JM F HP
0 0 0 5 6 0 0 0 0 37 *M 3 : F FM Í
259
0002 IMPUL
00057 *0 2 0 0 0 5 * FLDA A
1 ilAAv/
V} V V A
00060*064013* FLD3 TEMPO

00061*025400 FLDA 1*0*3 ; LOAD RXYCJ)
0 0 0 6 2 *0 3 0 0 1 6 $ FLDA 2* DEL A M ; R X Y ( J ) / D T
00063*144200 FDI V 2* 1
00064*110000 FIC3 ; IMG. A N D S T O R E
00065*070013* FST3 TEMPO ; POINTER
V
-/ OC u ü u u I 1 ^ FLDA 2* SUM
00067*146400 F SUB 2* i
A A A O A 1 ! A / . A A A
F DI V 0* 1 5 C A L C U L A T ^ HCJ)
A A ^ 7 1 < A /i C A A A
/ 1 V.» ^ J U <w» V.' F STA 1*0*2 î S T O R E IT
00072* 14400 1 F F DCF 1 ; P R I N T IT
A A A 7 A I I A A A A A
* O X t.» V.»C» C» V.» F EXT
74*006015* JSR
o o o o
G • RESLT
75*034010* LDA 3*M ; LOAD M c o u n t e r
76*0100065 I SZ J J INC- J
77'024006* LDA 1*U
00100*166414 SUB# 3*1* S Z R 5 J=M ? F I N I S H E D ?
O l o J W K . ’ I G yj JM P H1 i NO* C O N T I N U E
00102*034402 LDA 3* RIO J YES# EXIT
^»v.<K'O 1HuO
1 A Q I A A 1 / . A ' )
JM-P 3* 3
A A 1 A /, • A A A A A A n 1 A # A
u v 1 \.*h n l ^ t V.»
END
260
Enter
( Subroutine^
I
Save return address
Set result format
" '1 --------
Set pointers to
H. Y, X________
I
CalH(0)*X(0)*AX
2 1
¿X[H(l)*X(0)+H(0)»X(n
Set pointer to H(J)

X(J)
l ~
e^1X(0)*H(JHX(J)«HC0)
Add It to sum
ISet
Calc. Y (J)
Load X(K)
H(J-K)
1 -i
Store and print
Calc, summation part Y( J ) ___________ _
of Y(J)
1.... .... Inc. J
Increment K
F igure A5.12 C0NV0L- Co n v o l u t i o n

261
0001 CONVO
. T ITL C O N U O L U T I O N
.ENT CON
. EX TN h EN T
. E X TI C 5 , C 5 P 1 , J , K , M , S U K
. EX TP T EK P O ,T EM P 1, DEL AK , • RES L T
. N H EL
r»
3* R 1 2
a a
V.» V./ *054561 CON : STA ; CON V/OLUTI ON ROUTINE
r\ r\ r\ i
V'V.»V.» V *1 * 0 2 4 0 0 1 ? LPA 1* C 5
A A A A A
V »V.» V » CL • 0 4 4 0 0 2 $ STA 1 ,C5P1
A A
V.»
A A O
V.» V-l O * 0 2 1 4 0 0 LPA w» 9 AV' 9A O
A •
9 P O I N T E R T0 H
A A A A
K.» *025401 LD A u 1,3 ; P O I N T E R T0 X
#a r\ r\ r\ c LDA
v» v.' v.' Z) * 0 3 1 4 0 2 2 ,2,3 » P O I N T E R T0 Y
A A A A f
V 1 1.1 ' O * 0 4 0 5 6 0
. V V V. STA 0 , TP0 5 (TEMP. POINTERS T0 )
r\ rs r\ r\ »7 STA
'044560 1, T P 1
A A A 1 AN
1 V' * 0 5 0 5 6 0 STA 2, TP2
A A A I 1
XX'1 1 1 0 0 0 K0 V 0, 2
P\»A.“\ 1 Q F EN T
v.n.a' X c- * 1 7 7 7 7 7
A A A 1O
v'Vw'v.' X O * 0 6 4 5 5 4 PLI 3 TPI ; P O I N T E R 10 X
fA A I AAA P L LA r\ r\ r\ •
LOAD H C0 )
0 0 0 1 4 v.' Xu v . ' U ü \J S \J> » d. »
000 15 * 0 2 5 4 0 0 P LDA 1,0,3 * LOAD X C 0 )

A A A 1 A r\ u <-\
*w' v.' v.» X O *040546 F STA vu, n u
A A A Í 7
v.» v.» v. ' X / * 0 4 4 5 4 3 ESTA 1,X0
A A A Q <*S
V» ¿ '0v^> * 0 3 0 0 1 1 $ P LDA 2 , DELAM
000 PI *1 04 1 0 0 F MPY 0,1 ; H <0 ) * X ( 0 )
00022
• 1 O A
l O u
1 A A
i u u FMPY 1,2
00023
« 1 C A
X 03 v.» O
00
A A
F HLD 2,2
•
A A A
V
A /| 9 1 A
X
/ .A A A
*-* v_* v.» v_* F I C2 9 INC. POINTERS
f 1 1 A A A A
F I C3
00025 1 X U L ' V.» V.»
000 26 * 0 2 1 0 0 0 PLL A r\
9 KJ 9
0
d
•
9 LOAD HCl)
0 0 0 27 * 0 2 4 5 3 3 F L DA 1,xo
A A A Q A
v^» Ov. » * 1 2 0 1 0 0 P MPY 1,0 ; H ( 1) * X (0 )
A A A A 1
V.* V.' O X * 0 2 5 4 0 0 P LD A 1 , 0,3 ; LOAD X ( 1)
A A A A A
X* V.» O *034532 P L DA 3 , H0
A A A O Q
V- ' UV. ' O J * 1 3 4 1 0 0 P MPY 1,3 ; H( 0 ) *X ( 1)
0 0 0 34 * 1 6 3 0 0 0 F A LD 3,0
9 1 A A A A A a r\
^ a a t c
v,/ v>* v.» O X O u FHLV K.> 9
A A A
V 1 Q /
v_> C *0 2401 1$ FLDA 1 , DEL AM
/*>A A Q
v.' v ' v.» O
7*
f 1 2 0 100 P MPY 1
1 A
* O
•
A A A ¿J • «
X
1 A A A A
X V.» V* V.» V..» P I C3 9 I N C . AN L STORE
000 4 1 *0 70 5 2 6 PST3 t p i ; POINTER
0004 2 *064526 F LD3 TP2 ; LOAD P T . T0 Y
000 4 3 * 051400 P STA 2 , 0,3 î STORE IN Y C0 )
9 1 1A A A A
0 0 0 4 4 1 1 uv. »v. »v. » F I C3
r\ Q •
00045 *041400 P STA K.' 9 k.* 9 sJ 9 STORE IN Y ( 1)
0 0 0 4 6 * 150001 F P DCP 2
« 1 1A A A A
00047 X X V.»V
.»V
»* F I C3
•AP» A C «A
v » v_» v* 03 v_» * 0 7 0 5 2 0 F SI 3 TP2
« 1 A A A A A
0 0 0 51 X V.' V» V.»V.»V* P EXT
0 00 52 * 0 5 0 5 1 4 STA 2, TP0
v.1
A A A C O
.v *v.1 03 O • 0 0 6 0 1 2 $ JSH (i. H ESLT
« AAAA 1 O •
000 5 4 v* v » X F EN T
A
0 00 5 5 ' 1 4 0001 P P PCF U'
• i A A »AA A
0 0 0 56 i V
v' V V' V
.»VJ r EXT
262
0002 CON VO
00057 006012? JSR @ •RESLT
^ rs r\ £- r\ rs rs
V-'V.*V.»O V»* 102520 SUEZL O 9 v.»
»*>AO/ 11 r\ rs
101400 INC KJ 9 \J
v.»v.*v.'A
O OO 0 2 4 0 0 5 S C 1s LDA 1 *M
A
v.»u o OA oQ
A A A /.A A A 7
u M vjv.»vjO J STA 0# J
00064 034475 LDA 3* R 1 2 ; LO A D EXIT ADDRESS
000 6 5 1 0 6 4 15 SUP* 0 * 1 * SNR ; J =M ?
00066 001403 JMP 3*3 * YES* EXIT
A
v•v A.'A A 7/
v.»O 030477 LDA 2 * TP0 ; N O * S E T PT. TO H ( J )
r\ r\ r\ n r\
V.'V'V f v.» 000054• FENT
A'A
L L*»>7# 1 1 104000 F ICP
0007 P 064475 FLD3 tp l ; SET POINTER TO X(J)
A
L'V A.•V
A*7/ VAJ FLDA A 9 A\J 9A d • L O A D HCJ)
0 2 1 0 0 0 P
A.'A
v v.»A 7# */*i4 025400 FL DA i*o*3 ; L O A D XCJ)
rs r\ r\ ^ j c
l»v.» / O 1 1 0 0 0 0 F IC 3
v*l'v**f7O/- 070471 FST3 TP 1
A A A 79 7 9
V.'V'O 030463 F L DA 2 *X 0
AA V.'11VA.»O
A r\ r\ • H ( J ) * X ( 0)
140100 FMP Y d P kj p
AA u U» 11
1 A FLDA
030463 2* H O
A
V.' 1A 1A A
v.fA 144100 FMPY 2 * i ; HC0)*X<J)
A A 1
V.' 1 v.»O A Q 1 o a nn
FADE:
1 U / v-» 0 » 1
JV-N»Û 11 ,
v-
.\
«^/l 124300 FHLV \> 1
. a /-\i c:
V.*V» I V.»J 044006$ FSTA 1 9 SUM
A
V.»A V1 11A
L'C / jAAAAA F EXT
A A 1
V'V.» k» 9 A 7
050457 STA 2» T P 0
A
O AV.»1 111A
050007$ STA 2 9 TEMPO ; HCJ)
0 0 1 1 1 030450 L DA 2# HI 2
0 0 1 IP A
v'TOt ZJ v
A."A I 1
LDA 3*1*2 ; POINTER TO X
00113 054010? STA 3* T E M P 1
0 0 1 14 SUPZL A.»9 A
V O
10 P 5 2 0
00 1 15 0 2 4 0 0 3 ? C 2 : LDA 1 *U
00116 106415 SUP# 0*1* SNR ; K = J?

00117 000424 JMP C3 ; YES
0 0 1P0 a t a a 1 jV r»1 <r
3’ LEA 2 * TEMPI ; K < J- 1
0 0 1 PI 176520 SU&ZL 3* 3
00 1PP 17 5 4 0 0 INC 3* 3 ; AC3 = 2
0 0 123 1 7 3 0 0 0 A IT 3*2
0 0 1 0 4 A C A A 1A f STA 2 * T EM P 1 ; X (K )
00 1P5 A V.>
A A A A
vJ V.1V .> V* f J
(T LI, A 2* T E M P O
0 0 1 P 6 17 P 4 0 0 SUP 3* 2 * J-K
0 0 1 07 .A C Dv
.A <
_
>\ _
A,A -~7 (T
<v _>f J STA 2* T E M P O ; P 0 1 tNT E R TO H(J-K)
A A 1 Q r\
v 1 O vj A A .A A y A V
F EM T
0 0 131 0 6 4 0 1 0 ? FLD3 TEMPI ; L O A D PT. TO X C K )
F L DA *
- rs o • LOAD H(J-K)
0 0 1 3 P .A Q | (A»A«A n
»_•C. 1 O O V .* .» 9 d
v.' 9 k P
00 13 3 0 P 5 4 0 0 FLDA i*c*3 ; L O A D X (K )
0 0 1 34 104100 FMPY o*i ; X CK )* H ( J - K )
0 0 13 5 0 2 0 0 0 6 ? F L E'A 0* S U M
00 13 6 1 A 7 A A A
1U / V» F A DD 0 * 1
00 13 7 0 4 4 0 0 6 ? F STA 1* SUM
1#-># *\A A rs
0 0 1 40 .• V
I v .‘ K,> V.*Vw' F EX I
A A •
0 0 14 1 101400 TMC s^» P > INC. K
0 0 1 4 P #A «A'‘J ^ JMP C2
0 0 14 3 0 0 0 1 3 0 • 0 3 : •F F.N I
0 0 1 4 /i 0 6 4 4 P 4 FL1 3 TP2 ; POINTER 10 YCJ)
00 145 0 2 0 0 0 6 ? FLDA 0* S U M
oo i n e 024011? FL1 a 1* DELAM
0 0 1 47 1 0 4 1 0 0 FMPY 0 * 1
0 0 15 0 045400 FSTA 1*0*3 * STORE IT IN YCJ)

0 0 15 1 144001 F F 1 CF l ; PRINT IT
o a
V_' V» 3 COMVO
.A A 1 C, O • 1 I A A A A
V.' 4 ^ f- 1 i UU UU f IC3
OA
V
-*V
.» 153*070415 FST3 TP2
A A
U
1 ts /j • 1 A A A A A
* 4U V .»V.' u u FEXT
A A
L' L» 155*0060125 JS R @ RESLT
•
/AA
V* 156*0100035 ISZ J INC. J
o .a
V* V
»1 157 *0 2 0 0 0 3 5 LPA 0# J
/A/A
160*000702 JMP Cl
A /A
V_* V» 1 6 1 * 0 0 0 0 0 0 R 12: A
V.'
/AA 1 ¿10 9 A A A A A A Y'O • A
V-' V-' 1 C ¿0 V»V' ^'V.* \
»» A V* • V.'
.A /A 1 ¿ Q • A A A A A A A
V *V.» 1 V>O I'V* V»
'
/A/A 1 ¿./i 1 A A A A A A U A # A
K.*V.11 OM \
»»V.' V' Kj V» rl V* •
/A/A 1 ¿ C 9 A A A A A A A
V' V» 1 U J V'YV'V.» V.'Y V*
A A
V *v.• 166*000000 T P 0 : O
V* ; h
O iA
V.» 1 6 7 * 0 0 0 0 0 0 TP 1 : A
V» ; X
/A/A 1 T A f A A A A A A *T H A • A
V 1 í V.' V' V.» V» I r CL • V» > Y
. EM P
264
F igure A5.13 INVERCON- Inverse convolution

265
0001 IMVEH
.TI TL IN V E R S E C O N
• ENT ICON
. EX TW F ENT
. EX T P C 5 , C 5 P 1 , T E M P O , T E M P I , DEL AM, J , K , M , S U M
. FXTD SUM P I , . R E S L T
. MR F L
o o o o o 10 54533 icnws ST4 3 , R 12 J INVERSE COMVOLUTION
»-Nr\AA 1 r r\ /, »Arv t <T LPA
v.H» v.» v.» 1 v.» r 1 1 , C5 9 ROUTIW E
00002* 0^4002? STA 1 ,C5P1
00003*177777 F KW T
00004 *066477 •
FLD3 0 RI 2 9 P O I N T E R TO IMPULSE
r\ r\ 0
00005 *021400 FLPA \J 9 \j 9 O 9 FUNCTION H
r\ tj /*%
00006 *040473 F STA s.» * n ^
r\ r\ r\ r\ *7 » a n a a a y, ff
v•v»v_' v* / v.* / v» H r F ST3 TEMPI
A A A i A t 1A A A A A
V»V*V» 1 I F EXT
0 0 0 1 1 ’ 126400 SUF 1, 1 9 SET J=0
00012 *0 44006 $ STA 1, J
00013*034470 LPA 3 , R12
00014*025401 I.PA 1 ,L 3 9 P O I N T E R TO OUTPUT
000 15 ’ 0 4 4 0 0 3 ? STA 1, TEMPO 9 SIGNAL 1
A A
0 0 0 1 6 * 1 0 2 4 0 0 CO 1: SUF X
^>* V» 9 SET K=0
r\ r\ r\ \ <7 f r\ /x rs r\ n <r ST4
\_»v'v.' 1 1 Vw' f - 0>K
•
k»V. X.'Y0V_' 1 1- STA 0 , SUM 9 SET SUM TO 0
00021*040012? STA 0 , S UM P 1
000 22 * 0 344 6 1 L1A 3 , R 12
00023*031402 LPA 2,2, 3 9 P O I N T E R TO I N P Ü I
00024 *034004? LPA 3 * TEMP 1 9 SIGNAL X
00025* 024006? LPA li J 9 S E T P O I N T E R TO
•A Q « 1 Q7 A A A ALL 1* 1 H ( J - K ) wI TH K = 0
• .» u o C
jv 1 £6 / v.» x.» 9
A A A QH f 1 A O A A A ADF 1,3
VPV_*V*AA I L yJ f X.»
0 0 0 30 * 0 5 4 0 0 4 ? STA 3* T EMPI
0 0 0 3 1 * 0 2 4 0 0 6 ? CO2: LPA 1, J
00032*106415 SUF* 0*1 SWR , 9 K=J ?
A.A ,A Q Q f A A A /. Q A
#
•
„ V.'V.» J vJ X.' V' i-
-W* JKP CO 3 9 Y ES
*A/AA A A /. 9 A A A A «A A «
JH
v'L'v.' v* x^/v.» O F KW T J H< J - 1
00035 *064004? FLL 3 TEMPI
»A#AfA A r 9 <
AQ 1 /. #A#A r\ 0
ov.» ü C w CO i x^x
.* FL r A v.' 9 \,f 9 O 9 LOAD HCJ-K)
/A/A# AO *7 9 A A C »AO A
V»V. » ^ ' 0 I C
Z- D O V.' V
_* F L DA 1,0,2 9 LOAD X( K)
AAA y, .A 9 1 «A /, /A.A »A
V»X
^* X^' 1X ^
'*-•V .»V
.»V
** F I C2 9 INC. X POINTER
000 4 1 * 0 1 4 0 0 4 ? F LSZ T EMP 1 9 DEC. H POINTER
00042 *014004? FPSZ TEMPI
0 0 0 4 3 * 104 100 F MP Y 0* 1
0 0 0 4 4 * 0 2 0 0 1 1? F L PA 0 , S UM
00045 * 123000 F App L 0
00046*040011? F STA 0 , SUM
A A A /, T • X A A A A A
V •1»V » *-« f l v.»x.'v»\.'v.» F F.X T
0 0 0 5 0 ' 0 10 0 0 7 ? I SZ K 9 I NC • H
*
“x \f
200 5 1 * 0 2 0 0 0 7 ? LPA v.* 9 c\
000 52 *000757 JMP C 02

0 0 0 5 3 ' 0 0 0 0 3 4 * C0 3: F FWT
00 0 54 ’ 0 6400 3? FLP3 TEMPO
0 0 0 6 6 ’ 025400 F L PA 1,0,3 9 LOAD Y( J )
266
0008 IN VER
^ a c r 1 1 A A A A
\_t V.« Zj O 1 F I C 3
/n C *7 A 7 A A A A ff
V.I V.IV.I J / \J • v.* \ j y,t %J J> F S T 3 T E M P O
v^j V.# o O \»# 0 3 4 0 0 5 ? F L D A 3 * D E L A M

n A A £ 1 •
u o u O 1 1 6 4 2 0 0 F D I V 3 * 1 1 Y C J ) / D T
r\ /> /: o
vj v.»v.»Oc 0 3 0 0 1 1 $ F L D A 2 * S U M
A A A / Q
v.' O O 1 4 6 4 0 0 F S U B 2 , 1
/> r\ rs r /, A M A
V.» v.» v.» C ** 0 2 0 4 1 5 F L D A s.» 9 n v,'
/a r* /> /i c «
\»* k j v ' O D 1 0 4 2 0 0 F D I V 0 * 1 # U A L C U L A T E X C J )
A A A / / A y, C A A A •
V» O O U H J u KJ K.» F S T A 1 * 0 * 2 # S T O R E I T
•
0 0 0 6 7 1 4 4 0 0 1 F F D C F 1 9 P R I N T I T
A A A 7 A | A A A A A
V.» V.' V-* » v»* 1 V» V» V.» V.» V.* F E X T
A A A 7 $
V.' V' O / 1 0 0 6 0 1 3 $ J S R 0 . R E S L T
A A A 7 A
VJ \»' V-* / ¿L 0 3 4 0 1 0 $ L D A 3 * M
A A A 7 Q A 1 A A A £. C •
v* O f O KJ 1 V»»\.*uO J I S Z J 1 I N C . J
0 0 0 7 4 0 2 4 0 0 6 $ L D A l i J
A A A 7 C •
kj y*' \ j 1 Zj 1 6 6 4 1 4 S U B # 3 # 1 » S Z R » J = M ? F I N I S H E D
a A A 7 / A A A 7 A A •
S*'S*»KJ i O V.» I c u J M P C 0 1 9 N O * C O N T I N U E
A A A 7 7 •
U\.K* f / 0 3 4 4 0 4 L D A 3 * R 1 2 9 Y E S * E X I T
A A 1 A A
K » V 0 0 1 4 0 3 J M P 3 * 3
A A 1 A 1 A A A A A A M A • A
1 V.' 1 o u u u u u nvy • VJ
A A 1 A A A A A A A A A
l u c u u u u u u \j
A A 1 A A A A A A A A n 1 A # A
v_' \^» K ' O V.»U U U V . » C » n 1 G « \j
END
Enter A
C
z
Subrout ine y
T z l
Save return address
Set result format
T
„ , t ttR \. . Load 2 DT
Calc, cosi— )+ 1 Multiply by sum
*m •
I
„ , /ttRK \
Calc, cos-- Load storage pointer
* m /
I 1>
Load RR (R) Store and print
P.S.D.
Add product to sum Reset pointer to RR

(1)
_________ 1________ _
Increment R Increment I*
Figure A 5 .W SFECT- Power spectral den sity

268
A A A «
SPECT
• TITL SPECTRAL
• ENT G
• EXTN P E N T # S P C E 3
. E X T D C5» C 5 P l » T E M P l » M » K » D E L AM» RXXO*

• E X T D ONE» T W O » • RESLT» SUM
• E X T D • C R L F » . TXTT» PI
•N R E L
A A AAA
V.' KJ V
.» • 0 5 4 5 3 6 G: STA 3» RI 1 •
* POWER SPECTRAL DENSITY
AAAA 1
V-M.' 1 •024001$ LDA 1» C5 * ROUTINE
A A A A A
V» V' V
.» O •044002$ STA 1» C 5 P 1
AA AA Q
u L» V
.»v.* O • 0 2 1400 LDA A A
U #U #
7
Ü i POINTER FOR STORAGE
A A A A J , 1 A /«A A A A ff
V
.»U O U H u H \ Jv .» ü O J> STA 0 » TEMPI
A A A A C
V
.»V.»V
.» J •006014$ JSR CRLF
r\ r\ f\ r\ r
V.»c» y j O •006015$ JSR © T XTT
.
A A A A 7
V
_*# • 177777 SPCE3
A A A 1A
V
.»k j V
.» X V
.» • 102400 SUB A
U J
A
L» 5 SET C O U N T E R S
0001 1 ' 0 4 0 5 3 3 STA 0»MF # M FLOATING
AAA 1A
V
.»k j V
-*1r. •024004$ LDA l» M
AAA
1
1 A
O*044532 STA 1 »M F + 1
^ ^ ^ i y»
v
_*V» <.• 1 * A ’0 31401 LDA 2» 1» 3 •
P O I N T E R TO A U T O C O R
AAA 1 C
v.*ov.' D 1 • 177777 FENT •
9 RELATION ARRAY
r
UV.'L»1 D
A A A I
•060526 FFL0 MF
AAA
V
.»
1
i_*v,*1 r 7 •A AAA A
U C U U uO/! <
T
T F LDA 0» DELAM
A A A AA
L' V
.»V.' CO • 0 2 4 0 1 1 $
V» FLDA 1» TWO
AA A A
K.» CZ 1
• X1 O'"' 1rsr\
1
c . kj FMPY 1»0 1 2*DT «
AAA AA
•040515
K .' K.» K.r C m C . F STA 0» DT2
A A A A O
V
.» kj r> O •024521 FLDA 1»MF
A A A A
(Z>
>1
• 1 04100 FMPY 0» 1 Î 2*DT*M
r \ r > r \ r^s tz
K.» U U G J 1L»^
•A A A A | A
UGUW* FLDA 0» O N E
.1 ^
AA A A^
v
t •
c:O loo oo oc\<\ c\r\ a
EDI V 1
l
A
KJ
9 Î FREQ. S T E P HZ.
A A A Q7
V
.»V.' V
.» f • 140001 FFDCF A
KJ P R I N T F I R S T STEP
9
A A A 7 A
V
-' U v) V
_*
fo o l u u u
A A | A A A
F LDA 0 » 0» 2 LOAD FIRST AUTO
•
A A A A | V 1 A / .A A A
V
.»u u O X 1 FI C2 •
CORRELATION VALUE
#
A A A 7 A t A /.A A a 7
L» 1 #O
<
O H V.» U U f ^
r
FSTA 0» RXXO
A A A 7 7 « 1A A A A A
v.' v
_*O O 1 u u u u u FEXT
•006014$
A A A 7 /j
V
./ JSR ©• C R L F
a a a o c
•006014$
v 'U U O J JSR ©.CRLF
AAA 7 /
uuO O •006014$ JSR ©•CRLF
AAA 7 7
v.' V • 102400
.»v.» O f SUB A
9
A
SJ
V
.»V_
»V.» • 101400 Gl:
A A A /. /A
INC A A
U# U 9 S E T TO 1ST. FREQ.
00041 f A /, A A A C tt
U H V .IU U JÎ» ST A 0» K
A A A y, A
V' V •040505
.»C' ^ C STA 0» K F + 1
00043 •024004$ LDA 1» M
00044 •034472 LDA 3» RI 1 9 LOAD EXIT ADDRESS
0 0 0 4 5 • 122015 ADC# 1 » 0» SNR 9 K >M ?
«
00046 •001402 JMP 2» 3 * YES» EXIT
f\
0 0 0 4 7 • 10 2400 SUB A
V 9 C*
A A AC A
»
V V.» V
.» •040476 STA 0» KF 9 STORE IN FLOAT. K
00051 • 04047 1 STA 0» RF
r\
00052 • 10 1400 INC a
yj » yj
00053 • 040466 STA. A

*^< » rv
D
9 sf: t R=1
« 1C 1
000 54
A A A A
v_*
kJ v »»U X J F ENT
•
00055
A A A A A 7 <
T
V' <.• i J* FLDA A D Y V rt
*_» » rt A A i_»
269
Ar\r\ a
v.» c í SPECT
u v.O O 040013$ FSTA C* SUM ; SET SUM = R X X C O
r\ r\ r\ c
'JU U J
*7
/ 060467 F FL0 KF
r\ r\ r\ g r/ \
KJKJKJ O
1 A n n o «
1 UU U L»U FEXT
A A A i! 1
V
.»V.»O 1 1 2 2 4 1 5 G2 J SUB# 1*0» S N R J R > M ?
/>/ ■> r\ ¿l o
K J KJ %*J U ¿ 000434 JMP G3 *«
YES* CAL. G (K)
r* r> o
v.»v.»v.' O O 040460 STA 0* R F + 1 ♦# L O W P A R T OF R
A A A ¿ /i
X.' v.» O *4 000054* FENT
*
A A A i! C
v* v.* O O 060455 FFL0 RF FLOAT R
AAA / /
O O 020454 FLDA 0# RF •
LOAD R
r \ r \ /r * 7
v->O f 024455 FLDA 1*MF 9
•
LOAD M
A AA 7
O
A
030016$ FLDA 2# PI 9 L O A D PI
AA A 7 1
V' V.' # 1 140100 FMPY 2*0 #• P I * R
A A A 7 A
f 0
1
1
r \ r \ c \ r \ r\
C , \ j C* k j y j FDI V 1*0 9
•
PI*R/M
A A A 7
V
.»V.' V
O
.' / O 110060 FC0S 0*2 # C0SC P I * R / M )
•
A A A 7 /i
v 'c» o / *4 0 3 4 0 IOS FLDA 3» O N E
A A A 7 C
v-»uu r j
1
1
7 0 AAA
/ O l »v^v.* FADD 3*2 •
CC0SCPI*R/M>+1)
r\ r> r\* 7
vj v.» i
¿L
O 024450 FLDA 1» K F
A A A7 7 1 AA 1A A
V-'V'V' / / i Cv.» l u u F MPY 1*0 •
9 PI*R*K/M
A A 1A A
V.»V* l u u I
I A A A /! A
U U O V
.» O FC0S A
U #
A
L» 9 COSC P I * R * K / M )
AA
IV.» 1
1A 1 1 1 ^ 1 AA
1 l u lu u F MPY 0*2 I (C0SXC0S+1 )
•
AA 1A A
V.» l u c
A A C A A A
U ü JU U U FLDA 1*0*2 9
•
LOAD NEXT AUTO
AA
V.' 1O
1A A
O
1 A /. A A A
1 UHUUU FIC2 *• C O R R E L A T I O N V A L U E
A A I A /,
1 44100 FMPY 2» 1
AA
1 I AC
U J I
A A A A 1 A ff
OO V _* O FLDA 2» SUM
\w
M w* 1^0
A A 1A £ 1O A A A A
FADD 1*2 •
9 A D D TO SUM
uu lu 7#
A A 1A
050013$ FSTA 2» SUM
11
AA 1 1A
V-'O VJ 1 V.»\^
f A A A A A
FEXT
00111 102400 SUB A . A
Vj >» O
00112 040430 ST A 0» RF
00113 010426 ISZ R
00114 020425 LDA 0* R
001 1 5 000744 JMP G2
001 1 6 0 0 0 0 6 4 *G3í FEMT
00117 020420 FLDA 0» DT2 #
DIGIT INTERVAL
V»»V' I C V CAL. GCK)
A A 4 AA
-» 110100 F MPY 0*2 •
00121 064003$ F LD3 TEMPI

00122 051400 FSTA 2*0*3 9 S T O R E IN A R R A Y
00 123 X1
11A A A A
V U U U FIC3 #• INC. A N D S T O R E
00 124 A 7 A A A O ff
KJ i V.» O K.' OJ> FST3 TEMPI •
POINTER
001 2 5 1
1
c: ^
J u
^ ^ 1
u i FFD C F 2 •
P R I N T GCK)
00126 «
i
A A A A A
V.» v.» c* V
.»^ FEXT
00127 006012$ JSR e . R E S L T *• F O R M A T I T
AA 1 OA
V
.»U 1 J V
_» 030406 LDA 2» R1 1 i R E S E T P O I N T E R TO
00 131 1
A A IA A
u o l u u l LDA 2* 1*2 9
•
SECOND AUTOCOR
00 132 151400 IMC 2» 2 9 RELATION VALUE
00 13 3 151400 IMC 2» 2 1 INC. FREQ. C O U N T E R
00 134 V.» Vw»V,>v.» J X>
A AA AA C <
T
LDA 0* K A N D S T O R E IN K
0 0 1 3 5 U U U7 í o O7
A A A A
JMP G1
0 0 1 3 6 0 v.; v^* na l1 II «
A A A A A A
*
A
K-*
00137 V.» U V.» u u U V.»

f\OT O •
A A A A A A
le #
A
\J
A A A A A A
00140 U VJV.' W' W »
A
V
-*
00 14 1 UUV-'V., V-1'-1 nD •♦
A A A A A A A
*
00 142 c» V.» u v.' v.» D

A A A A A AÜ •
n r •
A
\J
A A A A A A A
00 143 O u X.» v< S»'
M L«
A A A A A A A
00 144 v,» v' v.* 11
vj r • KJ
A A A A A A A
00 145 V» V* V,' V
.» V-'
r* r * r\ r\ r\ L( L • A
00146 0 r\ r
v.' v.* U •
A A A A A A A
00 147 v./ V- '
^
END
270
F igure A5.15 XSPECT- C r o s s - s p e c t r a l density

271
OGGI X SPEC
• T I TL X S P E C T R A L
• EN T GXY# P 180
• EXTN FENT# S P C E 3
. EXTD C5#C5P1# T E M P O # T E M P I #M,K

• E X T D DELAM# ONE» • C R L F # • TXTT# • R E S L T
• E X T D TWO# TWNTY# P I # DEG# L N L O G
•NREL
AA A A A
y u L» c» V-» •054514 G X Y l STA 3# R20 #• C R O S S - S P E C T R A L
A A A A |
y %J U X•024001$ LDA i#C5 • DENSITY FUNCTION
#
A A A A A
K.I y y V.» C* • 0 4 4 0 0 2 $ STA 1# C 5 P 1
v.» u u u O • 0 2 5 4 0 4 LDA 1# 4# 3 # S T O R A G E PT. F O R
r \ c\ r\
A A A A /»
v.»o VJ v.' *4 • 0 3 1 4 0 5 LDA 2# 5# 3 #• S T O R A G E PT. F O R
A A A A r
v.» v.» u O •044003$ STA 1# T E M P O í MAGNITUDE
A A A A / f A C A A A /j <
o O P
STA 2 # TEMPI a
» PHASE
A A
A A A A 7
í• 102400 SUB y 9 y #a S E T H I G H M=C
* y Iy • 0 4 0 5 0 5 STA
A A A 1 A
u l 0#MF
r \ r s r\ \ «
u v.» u 1 i • 0 2 4 0 0 5 $ LDA 1#M
A A A 1O
v_#y \J X cL • 0 4 4 5 0 4 STA 1#MF+1 #a S E T L O W M
y j y j y j X O ’0 0 6 0 1 1 $
A A A 1 Q
JSR @.CRLF #a F O R M A T
r \ r \ c \ i/,
y y y X ^ •006011$ JSR § • CRLF
c» u u i J • 0 0 6 0 1 2 $
A A A i c
JSR 0.TXTT
A A A 1 r •17 77 7 7
y y y XO SPCE3
C» U U 1 í• 177777 FENT
A A A « *7
A A A AA
V-' V.»V.» <-* y
• A A A A A 7 <
V.»
r
1 41 FLDA 0# DEL AM p CAL. 1ST FREQ.
A A A A
y y
1
cí X • 0 2 4 0 1 4 $ FLDA 1 # TWO • A N D FREQ. STEP
p
2 * DT
A A A AO • 1 Q A 1A A
v.» y.* cz. r . X C*y l y y FMPY 1#0
•
p
A A A A A
V.» V_r O O •040511 FSTA 0# DT2
O a A O /i
O Nw» y d * * + •060471 FFL0 MF
n n / > nc:
y y y C* Zj •024470 FLDA 1#MF
a a a a A
O O U u D • 104100 FMPY 0# 1 p 2*DT*M
A A A 0 7
v.» U U G f
« A A A A
y cz y y
1A C
l u x F LDA 0# ON E
A A A O A • | A A A A A
V.' W J V.» l c u o y y FDI V 1#0 9 FREQ. S T E P HZ.
A A A O 1
y y y y X • 140001 FFDCF A
y
•
9 P R I N T F I R S T STEP
f
A A A 7 A
u u O c
$A A A A A
Xy y y y y F EXT
A A A 7 7
u v» u O O *006011$ JSR 0.CRLF
A A A o y,
•006011$ JSR @.CRLF
o v.» v.» O v) ’0 0 6 0 1 1 $
A A A 7 C
JSR @.CRLF
v.» y y y y • 0 3 4456
A A A Q /:
LDA 3# R20
A A A 7 7
^ V_r O 9 •021401 LDA 0 # 1# 3 •
9 P O I N T E R TO RYX
»V.' *-* V.' • 0 2 5 4 0 2
A A A /.
v.» Vw LDA 1» 2# 3 9 P O I N T E R TO ACR)
00041 •031403 L DA 2 # 3# 3 9 P O I N T E R TO BCR)
00042 • 0 3 5 4 0 0 LDA 3# 0# 3 0
9 P O I N T E R TO RXY
0 0 043 • 0 5 4 4 7 3 STA 3# TP0 9 RXY
0004 4 • 0 4 0 4 7 3 STA 0# TP 1 9 RYX
00045 • 04447 3 STA 1# TP2 0
9 A( R)
00046 •050473 STA 2# TP3 9 BCR)
00047 • 0 30467 L DA 2# TP0 9 LOAD POINTER TO
A A r
a
v.» V.» v .* O y • 1 0 2 4 0 0 SUB V.» # *.<
í*m t l
K » y y y 1 ’ 0 2 4 0 0 5 $ LDA 1#M
.A
v * O v.*
c O •106015 GX1 : A DC# Ó#1#SNR ; K=M?
00 0 5 3 •000^26 J MP GX 2
00054 •000017* ir E U T
00055 •021000 FLDA 2 ; RXYCK)
vuv
r>
' oG.
rv r\
XSPEC
v-' U V .O O 064461 FLD3 TP 1
c;7
v.»\.»U J f 025400 FLDA 1 0» 3
»
•
0 RYXCK)
r> r\ r\
v.»k.»v» O c*
1 1 AA AA
1 Xu u u u FI C3 •
# INC. A N D S T 0 R E
A A A /* <
k
,#V.» O 1 070456 FST3 TP1 •
POINTER
A A A / A
k. ' k j ^ i O d 1 uM uu u FIC2
A A A £ Q
v.»v' v.» O O
1 7 1A A A
l O 1 u u u FMOV 1,2
00064 1A 7 A A A
iu i u u V
.» FADD 0 1 , 0 2 * AC K )
142400
A A A/* c
uv'vÔ J FSUB 2, 0 9
0 2*BC K )
v» v
~ 064452
r> A A / /:
»D O FLD3 TP2
00067 0 4 5 4 0 0 F STA 1,0,3 0 S T O R E IN ACK)
A A A 7 A | IA A A A
V
.»V
.»V.’ # 1 1 \ .» u u u FI C3
0007 1 0 7 0 4 4 7 FST3 TP2
7
A A A
V
.» V 064447
A
.' / G. FLD3 TP 3
A A A 7 A
041400
U V 'U f J FSTA A
U l
A
ux
Q
O 0 S T 0 R E IN BCK)
A A A 7 /. | 1A A A A
T
w»V
.»V
.# / X XV U U U FIC3
o / ~
l.' ^
>* 7 c:
v.*
/ O 070444 FST3 TP3
A A A 7 /■ 1A A A A A
O vJ / O 1 U U U U V» FEXT
A A A 7 7
i / Xu XH u u INC A
V#
A
U
AA | A A
V
.» 1V.»V
_*
A A A 7 CO
u u u f Og JMP GX l
A A « A 1
v.» K» J 0 3 4 4 1 3 GX2: LD A 3, R20
A A 1A A
k' lu e 031402 LD A 2,2,3 •
0 P O I N T E R T0 ACR)
A A 1A A A A A A C / ,»
v' 1 O UUUU FENT
| r> r \ r y
^ V» 1 U H
a
u e
q
X L» vue* FLDA A A A
U X U X G 0 L O A D 2 * A C0 )
A A 1A C 1A A 7 A A
V.*u I u j X u u O u u FHLU A A
UX U
m
0 CALC. ACC)
A A 1 A /"
V
.» 1 O 124300 FHLV 1,1 9
0 CALC. ACM)
s\ «
a i
u u lu
*7
/ 040415 FSTA 0, A0 •
0 S T O R E IN AC0)
A A 1 1A
^ u» 1 X 044416 FSTA 1, AM 0 S T O R E IN ACM)
| A A A A A
0 0 1 1 1 X v.»^ vuo FEXT
0 0 1 1 2 102400 SUB A
0
A
KJ
001 13 0 0 0 4 2 7 JMP GX3

COI 14 R20: A A A A A A
V.' V
-*V.'
A
V.»
00115 MF:
A A A A A A
V.» V.' V
.»U
A
U
00116 A A A A A A
^ U»V.»V.' V_'
A
V
_*
001 17 KF : A A A A A A
V
-*Vj L' U U U
A
<.*
»A< A 4 or\ A A A A A A A
v.» 1 e v-
ì V
-» V
»*V
.»V.' O
00 1 2 1 A A A A A A
^ VJ V.I U»V
.» R ?
A
001 2 2 A A A A A A
V.' U U U V» RF: A
V
_»
00 123 A A A A A A
U U V.' V
.»V
-»
A
U
00124 A A A A A A
V.' V
.» V
.» A0: A
U
00 125 A A A A A A
VUV
.»V» V.»V
.»
A
V.'
00126 r\ r\ /
VJ V
a r\ rv
.*< J \ J VUv
_» AM: A
V.»
00127 A A A A A A
V
.»VUV.»V
-»V
.»^
A
VU
A A 1AA
y~* y..> 1 o v»
A A A A A A
V.»V
U SUM : 1
A
U
A A A A A A A
0 0 131 Vj W*V
.» VUV.»
00132 A A A A A A
V.' V' v
_» v„' V
.» SUM2: A
V.'
00 133 A A A A A A
V.' V
.»U» V
.»V
.»
A
V.*
00 34 1
A A A A A A
V.»VJ V.»V
.* DT2: A
V
-*
00 135 A A A A A A
V.» V.' K.J
A
U
00 36 1 OW .» U TPOs
A A A A A A
*V
.»O
A
V.* ; RXY ( TEMP. POINTER T0 )
00 1 37 OV.»V* TP :
A A A A A A
<.» V 1
A
V.* ; RYX
00140 -'K
-*C V'TP2:
A A A A A A
V U'
A
V
.» ; AC R)
00141 TP3:
A A A A A A
v_( vu y.* ^ v»
A
v
~ ' ; B(R)
0 0 142 101400 GX3: INC A
V.»0
A
KJ
•
0 SET T0 I S T . F R E Q .
00 143 040006$ STA 0, K 0 SET L O O P C 0 U N T E R S
00 144 040754 STA 0 K F+ 1
,
00 14 5 126400 SUB 1, 1
00 146 044754 STA 1, RF
00 14 7 044750 STA 1,KF
00 1 50 024005? LDA 1,M
273
0003 XSPEC
0 0 1 5 1 • 122015 ADC# 1# 0# SNR 9 K = M ? (FINI SHED)
00152*000541 JMP GX 6 9 Y E S , G 0 T0 P R I N T 0 U T
00153*000103* FENT i NO, C O N T I N U E
00154*060743 FFL0 KF
00155*126400 F SUB 1, 1
00156* 0 4 4 7 5 4 FSTA l* S U M 2 9
9 SET QCK) T0 0
00157* 0 2 0 7 4 5 FLDA 0# A0 9 L O A D ACO)
00160 * 034746 FL DA 3* AM •
9 L O A D ACM)
FEXT ♦
00162*101203 M0VR 0,0, SNC •

9 0 D D 0 R EVEN ?
00163 * 000404 JMP . +4 9 EVEN
00 16/1*000153* F ENT 9 0D D
00165*162400 F SUB 3# 0 9 AC0)-ACM)
u u 1 OD J FJMP .♦3
00167*000164* FENT
v.*v.» 1 r 1 Ov)v.»uv.» F ADD 3,0 9 ACO+ACM)
00171*040737 F STA 0# SUMI
_*V.* i m CL
A A f *7 0 * I A A A A A
V I U U V,» U F EXT
00173*034721 LDA 3, R20
00174*031402 LDA 2,2,3 •
R E S E T P O I N T E R T0 AC0)
o o
9
00175*035403 LDA 3,3,3 9

9 R E S E T P O I N T E R T0 BC0)
00176*054743 STA 3, TP3
00177*102520 SUBZL 0, 0 •
9 GENERATE 1
A A Q A A « A /i A 7 O 1
V* C1O V.* V-1 /c l STA 0, R •
9 SET R = 1
00201 * 106415 G X 4 3 SUB# C , 1, SNR 9 R=M ?
A AA A A A A /, /. A
A
KJ \J d» KJ d
f
JMP GX5 •
9 YES
r\ r\ r\r\ o • /,a 7 o a
V.HJCV.» -J
a
í cu STA 0 , RF+ 1 9 NO
A A A /, f A A A 1 / H f
u v .»c u * 4 u u u i D / FENT
A A A A C I A £ A 7 1 C
D uOu/ ID FFL0 RF
00206*020714 FLDA 0, RF
A A A A 7 f A A /. 7 A r
v.»uüv; / f v.i O F LDA 1,MF
A A A « A » A O A A 1 ¿ ff
U \.f G 1 U L» J U U 1 O «P FLDA 2, PI
0021 1 * 140100 FMPY 2,0 9 PI *R
1O • 1
UUG l O 1 G V.I G V.» U F DI V 1,0 •
9 PI* R/M
00213*024704 FLDA 1, KF
00214*120100 FMPY 1,0 9
9 PI*R*K/M
00215*104060 FC0S 0, 1
00216*110120 FSIN 0, 2
00217*064722 FLD3 TP3
A f fA/.AAA
A A A A
.* CZ d.
V l KJ *4 K.* KJ \J FIC2
A A AA 1 t « 1A A A A
V
.‘ V.» G G 1 1 \J U W‘V
j» i FIC3
A A Q Q n « A A I A A A
V
.» U U G G V.» G 1 U U ^ FLDA A A A
U# U# G 9 L O A D 2 * AC R )
A A O O Q • A O C /|A A
\Jk.» G G O U J J H U i.» FLDA 3,0,3 9 LOAD 2+ B C R )
00224* 104100 FMPY 0, 1 9
9 2 *A(R )*C 0S
\j CZ cZ Zj 1 / v.1 1 u i/ FMPY 3, 2 9 2*BCR)*SIN
00226*020702 FLDA 0 , SUMI
00227 *0 3 4 7 0 3 FLDA 3 , S UM 2
rt A Q 7 ^ »
O v_*
1 7 A
1 v.* »
o
v.» v.* FADD 0, 1
i
A A Q 7 1 • 17 7 A A A
V.'L' C O 1 1O u u u FADD 3, 2
00232*044676 FSTA 1,SUMI
00233*050677 FSTA 2 , SUM2
00234*070705 FST3 TP3
A A OQ • 1A A A A A
r. yj Zj 1 FEXT
00 236* 10 2400 SUB v»'9
A A
\j 9 RESET RF T0 0
002 3 7 *0 4 0 6 6 3 STA 0 , RF
0 0 2 4 0 *0 10661 I SZ R
00241 *0 2 0 6 6 0 LDA 0, R
00 242 *0 2 4 0 0 5 $ LDA UM
00243*000736 JMP GX4
274
XSPEC
0 0 2 4 4 0 0 0 204* GX5X FENT
00245 024667 F LDA 1 DT2
*
«
9 2*DT
00246 034662 FLDA 3 , SUM1
0 0 2 4 7 134100 F MPY 1*3 •
9 CALC. COO
/n
a a q
054660
c
FSTA 3# SUM 1
00251 03 0 6 6 1 FLDA 2 , SUM2
9
A A OÔ
U U G JC 1o u Iv .iv FMPY 1,2 9 CAL. Q(K>
00253 050657 FSTA 2 , SUM2
0 0 2 5 4 150100 FMPY 2,2 •
9 G(K>*Q<K.)
0 0 2 5 5 174100 F MPY 3,3 •
9 C ( K)*C(K)
002 5 6 1C7 AAA
FADD 2,3
00257 174240 FSQR 3,3
174020 FALG 3,3 9 LNCMAG.)
00261 0 2 0 0 1 5 $ FLDA 0 , TWNTY
0 0 2 6 2 114100 FMPY 0,3
•
9 20+LNCMAG.)
00263 A A A A «
FLDA 0 , L N l 0G
0 0 2 6 4 114200 9
FDIV 0,3 9 20*L0G(MAG. )
00265 064003$ FLD3 9
TEMPO 9 P O I N T E R TO GXY
00266 055400 FSTA 3 ,0,3 •
9 STORE THERE
00267 FIC3
rs a o 7
UV.»C # V
A
.» f A 7 A A A A C
U U V.' u u J) F ST3 TEMPO
00271 0 2 0 6 3 7 FLDA 0 , SUM 1
00272 024640 FLDA 1» SUM2
0 0 2 7 3 104200 FDIV 0 , 1
e
9 Q(K)/CCK)
0 0 2 7 4 124040 FATN 1# 1
0027 5 124400 FN EG Ì, 1 9 LAG NETWORK
0027 6 0 3 4 0 1 7 $ FLDA 3 , DEG
00277 164100 FMPY 3, 1 9 P H A S E IN D E G R E E S
A A QA A
V
.»u o u u
A
e*O
O h/!u u v .*_ A A A
FLDA 3# P 180 0
9 180 DEG. S H I F T
a a on 1 1 1
l»v.»Ou 1 1 1 A AA A
UUG FM0V o ,o ,f s l t ; QUAD. TERM NEG. ?
AA AA A
U U OU G 176400 F SUB 3# 3 •9 NO
A A AA A
u u Ju O lOr FADD 3* 1 •
9 YES
A A Q /» 0 6 4 0 0 4 $
A
FLD3 TEMPI
UU JU H 9 L O A D P O I N T E R TO QXY
A A AA C
V' V
.» J V 045400
,' J F STA 1,0, 3 9 STORE THERE
A A QA/
o u O t 'O 1I 1 I AA AA
O U U L» FIC3
V.' O V.»f
A A AA 7 A 7 A A A /, ff
U f J FST3 TEMPI
AAA I A | A A A A A
U V.' J 1 I FEXT
0031 1 A AAA A / ■(T
u g u u v .»0 3
> LDA 0, K
00312 AAA
v j u v .*D O u JMP GX3
00313 034601 GX6: LDA 3, R2C 9
9 ANSWER PRINTING
00314 031404 LDA 2,4 , 3 9 P O I N T E R TO GXY
00315 025405 LDA 1 5, 3
, 9 POINTER TO QXY
00316 A C A A A Q ff
O J O V . 'U u i1 STA 2, TEMPO
00317 044004$ STA 1,TEMPI
00 320 102400 SUB A A
00321 040006$GX7 : STA 0, K

00 3 22 024005$ LDA 1, M
00323 122415 SUB# 1,0, SN R 9 K=M ?
00 324 000414 JMP GX8
00325 A AAAA Of
v.«Ouvjv.» O J) LDA 2 , TEMPO
00 3 2 6 0 0 0 244* FENT
00327 A A
KJd 1A A A
1 V' KJ^ F LDA 0,0,2 9 LOAD GXY
A A AAA
u O Ov
.» 140001 FFDCF A
U
9
9 P R I N T IT
| A /,
00331 1 v,' M
A A A
V
-/V
-*v»' FIC2
1 A A A A A
00332 FEXT
00333
A C A A A A ff
v
_# O J> STA 2, TEMPO
00334 006013$ JSR RESLT
00335 a o a AA A C
J LDA 0, K
00 3 36 101400 INC 0, 0
273
CGC5 XSPEC
00337*000762 JMP GX7
0 0 3 4 0 * 0 0 6 0 1 1 SGX8: JSR 3.CRLF
00341*006011$ JSR §•CRLF
00342*006011 $ JSR e.CRLF
00343*020001$ LDA 0*C5 ; RESET RESULT
00344*040002$ STA 0 » C5P1
»
» FORMAT
00345*102400 SUB 0# 0
0 0 3 4 6 * 0 4 0 0 0 6 $GX9: STA C , K
00347*024005$ LDA 1#M
00350*036416 LDA 3,@PR20 •
> LOAD EXIT ADDRESS
00351 * 122415 SUB# 1# 0* SNR •
» FINISHED ?
V.H.*O J c yu 1 JMP 6 » 3
r'A Tt\'î» rtAAAA /.<r LDA 2 » TEMPI
00354*000326* F ENT
00355*021000 FLDA A A
SJ* KJ* d
A
L O A D QXY
00356*140001 FFDCF A
KJ
•
9 P R I N T IT
AA ATA f I A / ,A A A
F IC2
yuOOv.i 1o y u u y F EXT
00361*050004$ STA 2 »TEMPI
00362*006013$ JSR 0.RESLT
00363*020006$ LDA C*K
00364*101400 INC 0# 0
00365*000761 JMP GX9
0 0 3 6 6 * 0 0 0 1 1 4 *P R 20: R20
•Z R E L
0 0 0 0 0 - 1 4 1 2 6 4 P 180s -180.0
UVJUV.» 1
A A A A 1. A A A A A A
END
276
F igure A5.16 FRESP- Fr e qu ency response

001 FRESP
• TI TL FRE S P
.ENT FRESP
. EXTN FENT
. EX T P C 5 * C 5 P 1* T E M P O * T E M P I 9 D E L A M * . R E S L T * M
. EXTD P E G * L N L 0 G * T W N T Y * M 1ß P 1 8 0
. MREL
/> a #■> #
-\
V»V.»V* V
»' *0 54 5 6 4 FRESP: STA 3 * R 18 ß FREQUENCY R E S P O N S E
A A A A 1
V'V»»V.« V • 1 * 0 2 4 0 0 1 $ LPA 1*C5 ß S A V E R E S U L T F O R MA T
r\ r\ r \ g
*044564 STA 1* CSAV
A A A A A
c' o v.»v»O *024562 LPA 1*C3
A A A A
*044001$ STA 1 * C5
A A A A C
\
w 'V
.» V
.* J *044002$ STA 1* C 5 P 1
A A A A £ 0
*024561 LPA 1* C50 ß SET PLOTTING COUNTER
A A A A ?
*044013$ STA 1* M 1
A A A ! A
V.»V _
*V » 1V » * 0 2 5 4 0 1 LPA L L 3 ß OUTPUT POINTER G
A A A « 1
’ 031402 LDA 2* 2 9 3 ß OUTPUT POINTER PHI
AAA 1A
S^» V.'V-' 1 C ' 0 4 4 0 0 3 $ STA 1* TEMPO
A A A « A t A C A A A /. < r
v^'v'v*' I O V» KJ K.* *A I 1 STA 2 , TEMPI
I y, A A
v
.»v.' * 102400
1 *4 S UB SJ ß
AAA 1C
V
.* A0*040554
3 STA 0 *X •
ß SET FREQ. COUNTER
AAA « £
V-'V'V.f 1 C* ' 0 4 0 5 5 4 STA 0 *X + 1
AAA 1A
1 / *M i l l i TRI : F LNJT ß CAL. FREQ. RADS./SEC
A A A pA 0
*060551 FF L O X ß IN LOG. DECADES
A A A A « • A A A r r ,- N 0
U U V1' 1 V CZ. v.» Z3O FLPA Ci X ß IE. 0 W = 0 . 1E X P ( . 2 3 0 2 5 9 X )
AAA A A
\.'V' U C C '024555 F LDA 1*P23
A A A p ß
'104100 FKPY 0* 1
A A A p^j
* 124220 F EXP Ì 9 1
«/
"N
A A r^
V
.*V.' O 03 V.» CL D
3v.» FLDA 0 >P 1
A A
V»\»*V
A
O ^
» C. P '104100 FMPY 0* 1
.-Nr \ r \ r \ *7
v» v_' v' < 01/ * 0 4 4 5 5 2 FSIA LOW
AAA AA
v»*V.» O *0 2 0 5 4 3 F L FA 0 *P I 2
A AA Q 1
^'V-'V.'O I * 1 0 4 2 0 0 F FI V 0» 1 ß F R E G« IN HZ.
000 3 2 ’ 144001 F F PC F 1
A A A A A
v.» v» O O
t1 A A A A A
I V
.» V
.» V
_# F EX T
0 0 0 3 4 V^' t /* V»O
•A A /A0Â A
JSR P.RESLT
0 0 0 3 5 *0 2 2 5 2 7 LIA 0* PR18 ß POINTER 10 H
AAA A / 1
v_> ^*v 'O C *0 4 0 5 5 3 STA 0 , TP0
AAA O
v'v.a*vi / * 10 2 4 0 0 SUE
.
'J
KH V.'
ß
r\r\r\^
j. * 0 4 0 5 4 3
V'V.’ V.i
a
V
-‘ STA 0* J ß SET J=0
0 0 0 4 1 *0 4 0 5 4 3 STA 0 J+l
9
0 0 0 4 2 *0 4 0 5 4 3 STA 0 * S UM 1
0 0 0 4 3 *0 4 0 5 4 3 STA 0 * S UM 1 + 1
00044 * 040543 STA 0 * S UM2
00045 ’ 040543 STA 0 * S UM2 + 1
« A A A A 1 -7 9 ' T O O *
00046 V.* V.» V
.» 1 # i l \ CZ, • F EN T
A A A /, O
V
_*V.»V.* M f ’060534 F F LO U
A A A CA
l.' V .»J V * 0 2 4 5 3 1
*V F L DA 1* oy
000 5 1 *0 20 5 3 2 F L FA 0, J
9 A AA A A C < T
000 5 ? O v' 3 — F L DA 2* DEL AM'
A A A C O
V* V.» \^r ^3* ' * 1 0 4 100 FMPY 0* 1 ß J * W
000 54 * 130 100 F MPY 1 * 2 ß J * W* DT

0005 5 • 064534 FLD3 TP0 ß POINTER T O H C J )
A A Q
00056 *021400 F L DA K.’ ß ß O ß LOAD HCJ )
r\ r\ r\
V.» CÍ F R ES P
V.'^ U J / • 1 1A A A A
1 1U U U U ? IC3
V.»u u D u * 070531 FST3 TP0
a A i •
• 144060 FC0S 2# 1 J COS(J*W*DT)
A
u Au uA A O •
O ez • 104100 FMPY 0* 1 9 H(J)*COS(J*W*DT)
OOA / O
o w Do * 0 34 52 2 FLDA 3 * SUM 1
0 0 0 6 4 1u fu u u F A D D 3, 1
A
V»A A O
A CO •0 44 52 0 F STA 1* SUM 1
000 6 6 * 150120 FSIN 2# 2 9 SINC J*W*DT)
AA u A OA 7/ •
uv.» •110100 FMPY 0# 2 9 HC J ) * S I N ( J * W * D T )
A A A7 A
V/ •034517 FLDA 3* S U M 2
AA A 7 1 t 1T AAA A
uv.»u Í 1 1 /JUUU FADD 3*2
r\ r\ r\ ry c\
L' V-; J • 0 5 0 5 1 5 F STA 2* S U M 2
A A A 7 A
#O *074510 FFIX J
A A A 7 /, 9 J A A A A A
V.» i *4 FEXT
A A A H C tA OA C A 7
r O UCV.» J U / LD A 0* J+l
A A A 7 £,
V.'V>#V.* f O • 0 2 4 0 0 7 $ L DA 1*M
A A A 7 7 tA IA C A C
r r U 1U J U J ISZ J+ì
A A 1A A •
1u u * 1 0 6 4 1 4 SUB# 0*1 * SZR 9 J=M ?
AA 1A 1 •
\.»U 1 U 1 * 0 0 0 7 4 5 JMP TR2 9 NO
A A | A A 1 r t A r t A y (/> | •
1 v* FENT 9 YES
A A 1A O •
uu iu O ' 0 6 4 0 0 3 $ FLD3 TEMPO 9 P O I N T E R TO G
^ ^.» 1 •A A A C A I
V.»V 1 ^ /i u c u D u 1 FLDA 0* S U M I
^ A 1A C
U S.» 1 V.» J • 0 2 4 5 0 2 FLDA 1* SU M2
A A 1A / t 1A A 1A A A A
S^* 1 O 1O U 1U U FMPY 9 \J 9 C*C
A A «A 7 •
V.» 1 VJ / • 1 2 4 1 0 0 FMPY 1* 1 9 Q*Q
A A 1 1A V 1A 7 A A A •
1 I V.» 1 \J i U U U FADD 0* 1 9 C*C+Q*Q
001 1 1 • 124240 FSQR 1* 1
001 12 9 vA»»(Z,
A A A A C ff
\ j \ j Z) FLDA 0* DELAM
•
0 0 1 1 3 • 104100 FMPY 0* l 9 CAL. GCJ)
•
0 0 1 1 4 • 124020 FALG 1* 1 9 LNCMAG.>
0 0 1 1 5 *020012$ FLDA 0* TWNTY
001 16 • 104100 FMPY 0* 1 •
9 20*LN(MAG.)
00117 •020011$ FLDA 0* LNL0G
AA 1AA
o u I o u • 104200 F DI V 0* 1 9 20+L0GCMAG.)
00 12 1 •045400 F STA 1*0* 3 9 S T O R E IN ARRAY
0 0 1 2 2 • 144001 FF DCF 1 9 P R I N T IT
00 123 • J J A A A A FIC3
00 124 t Av.»71 A A^ AV»OO ff F ST3 TEMPO
00125 9 11 VA,'A VA.«A A FEXT
001 2 6 9 vA.‘AV.»£O Av.»Au U/■ J ff
JSR @ • RESLT
00 127 9 VA.»AV.»AV» 1| AV AA | F ENT
00 130 *020455 FLDA 0 * S UM 1
00 1 3 1 •0 2 4 4 5 6 FLDA 1 * S UM 2
0 0 1 3 2 • 104 200 F DI V 0* 1 9 Q/C
00 1 33 • 124040 FATN 1* 1 9 ARCTANC Q/C)
001 3 4 • 124400 FNEG 1* 1 9 LAG N E T W O R K
00 135 •034010$ FLDA 3* DEG
•
00 1 36 • 164100 FMPY 3* 1 9 P H A S E IN D E G R E E S
•
00 1 37 •034014$ FLDA 3 * P 18 0 9 180 DEG. S H I F T
00 140 •101002 EMO V 0 * 0 * F SL T • QUAD. TER?* NEG.
00 14 1 • 17 6400 FSUF 3* 3 9 NO
00 142 ’ 167000 FADD 3* 1 9 YES
•
00 14 3 * 064004$ F L'D3 TEMPI 9 P O I N T E R TO PHI
00 14 4 • 0 4 5 4 0 0 F STA 1*0*3 9 STO R E IN A R R A Y
00 14 5 • 144001 F F DCF 1 9 P R I N T IT
00 14 6 9 11 1i A AK.» VA.»AV.» F I C3
00 14 7 • 07 000 4 $ FST3 TFJ P 1
vj
00 1 50 07 4421 • FFIX X
9
00 15 1
4A A A A A
1 V
.» V
.» V* F EXT
279
003 FHE S P
00150* 0 0 6 0 0 6 $ JSR @ •RESLT
00153*010417 I sz X+l •
» IMG. 1c REG* COUNTER
00154*020416 LDA 0>X+1
00155*024412 L DA 1 , C5C •
» 4 DECADES ?
00156*106503 SUPL 0 * SNC
00157*000640 JMP TR1
00160*024406 L DA l , CSAV 9 RESET RESULT FORMAT
00161*0440015 STA 1,C5
00 1 6 2 * 0 3 4 4 0 2 LDA 3* R 18
00163*001403 JMP 3, 3 9 RETURN
r\
1
V'V.'
| s /, t
U H R 18:
r\ r\ r\ rv
V.'U
»A
00 165* 1777 75 C3s -3

A A 1 / /
1 O O CSAV:
| A A A A A A
vJ v ' L ' v . '
r\
V-*
A A
u u 1 »/7
| r
D C50:
f A A A A C A
Jv.» 50 •
9 O C T A L 50= DEC. 40
A A 1 7 A • A A A A A A
V>V-» 1 f V.' MM:
L» U U U U U
A
O
A A
171 1
1 |
X:
f A A A A A A
VJ V.» V,» V.» V7 V.*
A
A A 7/1
l
A
C
| A A A A A A
V.» V,» L ' V/
A
KJ
0 0 1 7 3 * 0 4 0 5 4 4 PI 2: 6.283185 ; two pi
00174*103754
0 0 1 7 5 * 0 4 0 0 3 1 PI J 0. 1
00 176* 1 14631
0 0 1 7 7 * 0 4 0 0 7 2 P23 5 0 . 2 3 0 259
00200* 17 1 100
AAAA
L* 1 V.» 0W:
1 | A A A A A A
V.' ^ \^f V»
r\
V.»
AA AA Q | A A A A A A A
U v» ¿Sv.» V U U V.» V.' V.'
A A Q/S O
V « V» c v ' O
f A A A A A A
V,»V.»V.» V . ' V ' V J: .1
A
V
A A A A /, • A A A A A A r\
V.'V_' ^ U V » V ' U V.7U Vw*
A A A A CT • A A A A A A
t>-' V.» O V.* V.» v/ «W
* SUM 1J A
V.'
AAAA/; • A A A A A A
v .»u V.« V,» V_* L* V.'
A A AA 7
/
« A A AAA A
v . ' * w * v.» »^* v.* SUMP: A
V'
AAA 1 A • A A A A A A ,A
W.' V» ¿T- 1 V.' V-» Vj V.'
A A A I
V' V.' c . 1 1
) f A A A A A A
V.» V-» TP0: A
EM t
280
10.5.2 Optimisation Programme
10.5.2.1 Alqori thm
The principal algorithm used was that of Fletcher and Powell and
was outlined in Section 3.4. The minimum of a multivariable non-linear
function subject to non-linear equality constraints is formulated as
follows
Minimise subject F(xi, X 2, .. x|\j)
to G|<(Xi , X 2 , •• X|\|) = 0
where k = 1, 2, 3 ... M
By Incorporating the constraints into a modified unconstrained
objective function, the method of Fletcher and Powell for the
unconstrained minimisation is then used. Inequality constraints are
treated by the use of slack variables with appropriate transformations.
The algorithm proceeds to define a new unconstrained objective function.

M M
$ = F - E Ak + B E G^2
k=1 k= 1
where A^ and B are constants. Having selected initial estimates for
the unknown variables. The numerical value of the function F, its
derivatives and the value of Gk are calculated. The Ak values
above are determined from
N M N
\ 9Gj 9Gk _ 3Gk 3F
E E E
A J3xj 3x; " 3xj 3x j
¡-1 j=1
for k = 1, 2, 3 .. . M
With an estimate of B a series of search directions and one
dimensional search steps are determined,and after the necessary
iteration to obtain the required convergence criteria Gk = 0 and the
function to be optimised is equal to <f>. (F = $ at convergence.)

281
10.5.2.2 User Experience
Some effort is required to formulate the polynomial
approximation algorithm mentioned in Section 3.3.2 into a form
amenable to the F Ietcher-PoweI I algorithm. However once this was
achieved and the programme package was tested on some problems whose
solution was known, confidence in the package was gained. The
resulting final programme package has proved computationally fast,
efficient, very stable numerically and with the added plotting
routines very useful for this investigation. The Univac-Fortran 5
User Manual (Reference 78) proved useful.
10.5.2.3 Programme Listing
The conclusions for the design of chute geometries under
uniform flow conditions suggest the use of this package programme.
Accordingly a listing of the optimisation programme follows.

c M»IN L I N E PROGRAM: CONMIN ALGORITHM BY FL ET CH E R - POWELL
C M I N I M I S E DESCENT TIME S U B J E C T TO CONSTRAINTS
C
I M P L I C I T DOUBLE P R E C I S I O N CA-H.O-Z )
REAL MU
DIMENSION X (25 I • 6 R ADF(25 ! • GI25>t CRADG(25l» H C2 5 I * F LAM(2 E î t S(2
1 5!
DIMENSION GOC 25 ) » GfU25>. U ( 5 0 ) t AC25I
COMMON IR E A D . I W R I T E t i S H l C H t NN. MM.XSC 11 0 0) * FS C . GSC 11 OO!. DE L DI c (1 00!
COMMON /CHUTE/ NV * N •NSTE Pt MC »XCOR»Y C O R . T A U . V l N l T t S T E P •MU»CI * C2* YRC
1 0! . XX t 251 • TF
E XTE RNAL LINK
C
C SETUP I N I T I A L VALUES AND PROGRAM PARAMETERS
C NV - NO OF V A R I A B L E S 1= DEPEND.* I N D E P E N D . )
C N - NO OF INDEPENDENT V A R IA B L E S
C H = NO OF CONSTRAINT EQUATIONS
C HC : KD OF OUTPUT PARAMETERS (YR PARAMETERS!
C X - CC E F c S • OF POLINOMIAL TO BEOPT IMISED
C F = FUNCTION TC BE OPTIMISED CDESCE NT T I H E j
C XCOR - END X COORDINATE
C YCCP - ELD Y COORDINATE
C DE L X = X V A R I A B L E CR IT E R IO N
C E P S = MAGN. CF AUX. FUNCTION P H I NEAR THE MIN.
C DE F X " X V A R I A B L E C R I T E R IO N DUE AN INCREASE OF E P S I N PHI
C DE LG - GRADIENT OF F AND G C R IT E R IO N
C MU - R E S IS T A N C E C O E F FI C IE N T IT ANG.D RAG F O R C E . V E L . DEPENDENT!
C TAU = COULOMB F R I C T I O N C O E F F I C I E N T ( V E L . DEPENDENT)
C V I N I T - I N I T I A L VELOCITY
C HB = HEIGHT TO BREADTH RATIO
C T- - L I M I T I N G THETA FOR FAST FLOU
C
T RE AD - 5
I URI TE -C
10 WRITF ( I W R I T E . 40)
READ ( I R F A D . 5 0 . E N D -30 ) NV . Mt MC •HB t V Ï N l T . MU • TA U. XC Gr .Y C CR. TF
Cl =0. 3 4 G * V I N I T * H B
C 2=0. 00 0G 9 0 9G *V IN IT *H B
D: L X - l .0-12
E P S - 1 . D -12
DE F X =1 .D- ti
D E L G = l. D - 6
N-NV-1
WPI TF ( I W R I T F . 7 0 )
WRITF ( I W R I T E . 90! N V . M. MC •HB. V IN IT . MU .T AU . Cl . C 2. X COR . Y CO R, Tc
NSTE P =4 0
S T E P t YCOR/NSTEP
c
C SET IN IT IA L ESTIMA TE S FOR POLINOMIAL COEFFS.
C
WRITE ( I W R I T E . 60!
READ ( I R E A D . 5 0 I ( X ( I ) » I = 1 > N )
C
C SET THE LAST C O E F F I C I E N T
C
283
SLMrO.
Z-YCOR**:
DC 20 I =1»N
SUM=SUH«XII)*Z/CI«1)
20 Z -Z* YCOR
T R U C z CXCOR-S UM 1 • t N V M l / Z
X I NV Ï -TRL' C
W R I T E C l W R I T E * 1 2 0 ) t 1 1 X C l ) » 1 = 1 # NVJ
I SWI CH - 0
C
CALL SCALER C N » H « X 1 1 . D - l 2 1 1 . D6 • 5 . D - 6 1 1 . D 2 » G0 » Gn I
C
c
W R I T E ( I W R I T E * 1 2 0 ) C I • X C l ) » 1 =1 « NV)
C A L L L I N K C 1 »X »F tC RADF .G t G R A D G )
C
CA L L CONK I N CL I NK t N * M» X . F t GR ADF t G t GRAD G 1 1 .D 2 * 1 . D 1 tH tF L A Mt DE L X* E P S •
1 DE F X * D E L G , 2 0 » 6 1 » 0 » I E R » S » G 0 » G N « W » A )
C
W R I T E C I W R I T E »801
I S W I CH -1
WRITE Cl W RI TE »110)
C A L L L I N K C 1 »X tF »G RADF tG » G R A D 6 I
VAL-YRCm
C
W R I T E C I W R I T E «1201 CI»XXCI)»I-1*NV)
WRITE C I W R I T E t l O O ) V I N I T t VAL t T F
I S Wl C H -0
G C TC 10
C
30 S T OP
C
40 FORMAT C * 1 T Y P E I N THE F O L L O W I N G DATA S E P A R A T E D BY A C O M M A * » / * N V ’ .
1 *» M t M C . H S » V I N I T » M U . T A U t XCOR t Y C O R V / )
50 F CRMAT C )
60 FORMAT C* T Y P E I N THE E S T I M A T E S OF THE N C O E F F I C I E N T S * t / * X U ) * . * #
I X 1 2 !...ETC*//)
70 F 0 RMA T t * * t / / / R X » * I Ni T I A L V A L U E S * )
80 F ORMAT C * 1 * » q X t *F I N AL V A L U E S * )
30 FORMAT C / / » * NV - * t I 2 t 5 X f * M - * » I 2 » 5 X » * M C z * . I 2 t 5 X f * H B z * t F 8 - 5 . * VI
l N l T z * t F 8 . 5 » 5 X . *MUz * » F 8 . 5 t 5 X t •T A U= * t F 8 . 5 t / * C l z * »F 1 1 . 8 . E X • * C2
2 - * .F 1 1 . 3» 5X» *XCOR- * F 8 . 5 . 5X t * Y C 0R= * t F 8. 5 . 5 X • *T HE TA F - * t F 1 0 « 3 / / )
100 FORMAT <//* I N I T I A L V E L O C I T Y z * , F l C U 5 / / * M I N . D E S C E N T TI ME z « , l p
1 E 1 5 . 3 / / * THETA F z * » 0 P F 1 0 . 2 )
110 F CRMAT ( / / t 3 X t *Y *t 15X t *W * t 15Xt * X * # 1 5 X *T * t 13X» *T HE TA * » 1 3 X t * V E L * t l 2 X
1 »*PAR.X*t8Xt*D IF.IN X *)
120 FORMAT ( / / * C O E F F I C I E N T S A RE * t q f 2 X • *X t • 1 1 1 » • ) z * , 1 PE 1 5 . 8 ) )
E ND
* 284
S L B R PUT I fi E C CMKI N i F N S # N * M * X * F t G R A D F » G * G R A D G ? B * R * H * F L A M f D E L X * E P S » D
1EF X » D E L G t l T » I P R . l N D * I E R t S t G O . G N * W » A 1
M A I N O P T I M I S A T I O N S U B R O U T I N E : P E R F O R MS OR C O O R D I N A T E S ALL CALCULAT
I C N S TO D E T E R M I N E THE S C A L E D C O E F F I C I E N T S OF THE
INDEPENDENT V A R I A B L E S .
I M P L I C I T DOU B L E P R E C I S I O N ( A - H t O - Z I
D I M E N S I O N X C 2 5 ) * G R A D F ( 2 5 Ï t G ( 2 5 1 t G R A D G C 2 5 ) t H ( 5 0 ) » F L A M ( 2 5 ) t St 2
151 t GO f 25 ) » GN ( 2 5 1 1 W t 5 0 ) » A I 2 5 1
CCMMON l R E A D » I W R I T E t I S W I C H t N N t M M * X S C C l O O ) » F S C » G S C t l O O Î t D E L D I F t 1 0 0 )
IN ITIA LIZE MATRIX H
IF fIN D -1 ) 10t20t20
10 C » L L M A I ’J P l H « 1 « N t S * GO» G Nt A LF A » D E L X » DE LG • W )
PRINT OUT INITIAL CONDITIONS
20 IP-IPR/10
I R=I P P - 1 C *1 P
I F I I P ) 4 0 t <i 0 « 3 0
30 W R I T E t I W R I TE 15 0 0 ) N , M »1 T • I P R» I ND »B t R t DE LX t E P S » OE F X »DE LG
W R I T E t l W R I T E *57 01 t I . X 11 ) » 1 =1 • N )
FUNCTIONS AND G R A D I E N T S AT INITIAL POINT
40 CALL ENS ( 2 t X t F » GRA DF » G » GRA DG )

IET=0
ICV--1
I E L =0
i n l =i
03TAIN NEW SET OF LAMBDA VALUES
I F 1 1 . G T . M ) GO TO GO
DO 50 U l . M
50 F L A M t l ) =0 . 0
GO C O N T I N U E
70 C A L L LAM3 t N * M • G R ADF * SR ADG * G» S t F LA Mt RtD E LG » W t A »1 Cv )
I E L - I E L ♦1
O B T A I N THE A U X I L I A R Y F U N C T I O N ( p H I ) t I T S GR A DI E NTtG N) t THE

MA G N l T U D E t GNA ) OF GN AND THE S U M l G S ) OF 3 * G * t I ) * * 2 F O R 1=1 TO M
L =1
80 P H I = F
I F l l . G T . M ) GO TO 100
DO 90 I =1 tM
90 P 3 I = P H I * G t I I * I B * G f I l - F L A M ! I ) )
100 I C K C G T = L
I - Ü C K C 3 T . L E . 1 ) GO TO 110
I F I T C K C G T . G E . 2 ) GO TO 310
03 TO 1110 » 3 1 0 ) • I CKCGT
1 1 0 . G M = 0.
285
I F I l . G T . N ) GO TO 150
DO m o 1 =1 , N
T=GRADF I I J
K=I
I F | l . G T . M ) GO TO 130
DO 120 J =1 . M
T =T* GRAD G ( K I *1 2 . * B « C ( J l - F L A M I J l l
120 K =K t N
130 G N I I ) =T
1 C| 0 G N A =GNA « T * T
150 GNA=SGRT!GNA)
GS - 0«
IF » l . G T . M ) GO TO 170
DO 160 1=1 •M
160 G S=GS«B « G I I I * G I I I
170 I C K CG T =L
IF I I C K C G T . L E . I I GO TO 180
IF I I C K C G T . G E . 2 I GO TO 350
GO TO 1 1 8 0 . 3 5 0 1 . I C K C G T
C
C PRINT OUT DATA FOR LAMBDA DETERMINATION
C .
180 I F I I P » 200 » 2 0 0 * 1 9 0
1 9 0 W R I T E 11 W R I T E »5201 I EL » I E T * I C V
W R I T E ( I WRI TE »5701 ( 1 » X( I I »1 =1 «N I
W R I T E ( I W R I T E « 5 1 0 ) < F L A Mi l l » 1 = 1 « Ml
W R I T E I I W R I T E * 530 I ( 6 1 1 1 . 1 = 1 . Ml
W R I T E f I W R I T E t 54 0 ) P H I « F « G S » G N A
W R I T E f I W R I T E *5701 11 « XSC11 I * 1 =1 «N I
C .
C D E T E R M I N E WHETHER CONV E R GE NC E HAS B E E N OB T A I N E D
C
200 I F I I C V - 1 ) 2 1 0 . 2 2 0 »230
210 I F I T N L ) 2 3 0 * 2 2 0 . 2 3 0
220 I E R= 0
GO TO 450
C
C LINEAR ITERATION
C
2 3 0 L =2
I NL = 1
I T N=0
I F I l . G T . N ) GO TO 250
00 240 1 = 1 «N
240 G O U I =GNI I I
250 CONTINUE
260 I F I I E T —I T J 2 8 0 . 2 7 0 * 2 7 0
2 7 0 I E R =1
GO TO 4 5 0
230 I F l l . G T . N ) GO TO 300
DC 2 90 1 = 1 . N
290 Wl I ) = x m
300 I E T = I E T * 1
ITN=TTN«1
I N L F = I NL
T NL =0
A LF A =1. 0
286
310 C fi LL LI NUN ( I N L . N . K F •W» S * AL F A * X » PH I . D E L X * 0 E F X . E P S )
l- 1 T N L - 1 I 34 0 . 330* 320
370 IER r?
GO TO 450
330 C A L L F N S t 1 • X t F . G R ADF * G » G R A O G )
G 3 TT 80
c
C 03TATN GN.GNA* AND GS
C
340 CA L L F NS ( 2 . X * F • G R A D F . G . GRADG)
G D TC 11C
C
C FIND INDEX INL FOR LINEAR ITERATIONS
C
350 INL=INL«2
IF (ALFA) 370.370.360
360 I F IGKA-DELG! 370.370.330
370 I N L -0
C
C PRINT OUT DATA FOR LINEAR ITERATION
C
330 IF (IP*IR ) 41 0 * i l l 0* 390
390 IF I M O D ( I E T t l R M 410.400•410
400 W R I T E ( I W R I T E . 5 5 0 ) I E T . I N L . A LF A , NF
W R I T E ( I W R I T E . 5701 ( I . X ( I I . I =1. N )
W R I T E ( I W R I T E . 5401 P H I . F . G S . G N A
C
C U P DAT E H AND 03TAIN NEW S
C
410 CALL MATUP t H . 3 . N •S t G Ot G N. A LF A » D EL X . DE L G . W J
C
C C O NV E R GE NC E QBTAINEO IF INLP=0 A ND I NL ” 0 ( I T N = 2 >
C
1= I I N L F * I N L ) 420.420.430
420 IER=0
GO TO 450
C
C 03TAIN NEW SET OF LAMOA VALUES OR CARRY OUT A FURTHER ITERATION
C
430 IF (IT N -1 ) 260.2S0.440
440 IF I I N L • ( K-M«1-ITN1f 70t70»260
C
C CCNCLUSICN
C
450 IF ( I P ) 490.490.460
460 W R I T E ( I W R I T E . 5 G 0 ) 1ER
K=1
1= t l . G T . N ) GO TO 480
D C 4 70 I - 1 . N
Wt I î - H ( K 1
470 K = K * N « 1 - I
4 3 0 C ONT I NUE
WRITE ( I W R I T E . 5 8 0 ) tW tI!«I= l*N >
490 P E T U R N
C
500 FORMAT ( / / ' I N I T I A L DATA FOR C O N S T R A I N E D MI N i M l Z A T 1 0 Nf » //• N = * . I 3 .
1 3 X . , P =* t I 3 . 3 X . » I T r * . I 4 . 3 X » t l P R = t . I 2 . 3 X . , l N D = » . I l / » D z ' . l p E l O ^ . E x *
28^
2 R - * t E 1 0 . 3 t 5 X t * D E L X - , #F1 0*3» 5X? * E P5 = • » E l 0 . 3 / 1 X » • D E F X - , t E l 0 . 3 i 5 X , • D

3F 16 = * . E 1 f . 3 )
510 F 3 RMA T f 1 2 X * ' L ANBDA = " , I 1P5E 16.« 7) )
520 FORMAT C / / * LAMBDA DE TE RMl NA T I ON* 11 «1* # A F T E R I T E R A T I O N * ? I 5 » 7 X * * I C V
1= M 1I
530 FORMAT t 1 7 X * * 0 = *» ( I P 5 E 1 6 • 7 M
5C,-° FORMAT \ % P H I = • • 1PE1 4 • 7 # 6 X ? * F = • • E l «1.7/1 X» *SU M OF B *3 • *2= • * E 1 U . 7 » G X
1* * KAGNI T I D E OF G R A D C P H I ) = * t E 1*1 • 7)
550 FORMAT 1 / * I T E R ATI ON •, 1 5 , 7X # • I N L = * f I I f GXf * ALF A = * t l PE 1 0. 3f 6Xf * NO OF
1 POI NTS = • »1 3 1
550 FORMAT t / / / » C O N S T R A I N E D MI NT M I ? A T 1 0 N C 0 M P L E T E D « , 7 X t * I E R = • , 1 1 I
570 FORMAT ( 1 X • f COE FF I C I E NTS A RE * » 3 C2 X » *X C * , 1 1, • ) z * , 1 pE 1 5 . 8 M
580 FORMAT 1« D I A G O N A L E L E M E N T S OF H = • , ( 1 P 5 E 1 2 . 3 ) 1
C
END
288
FUNCTION F UNF (X)

C
c evaluates parameters to be op t i mi s e d i ncludi ng
C Y COR . W f X C O R . T C K I N . T I M E ) t T H E TA » V E L
C
I M P L I C I T DOUBLE P R E C I S I O N ( A - H » 0 - Z l
COMMON I R E A D t l W R l T E . l S W l C H f N N t M M t X S C I l O O I t F s C f G S C I l O O l t D E L D I F d O O )
COMMON / C H U T E / NV « N « N S T E Pt MCt XCOR » Y C O R » T A U t V l N l T t S T E P » MU » C 1 t C 2 t YRt
1 91 t XX l 2 5 1 t T F
D I M E N S I O N X( 2 5 ) t X A R R A Y ( 5 0 l . Y ARRAY 150 1
RAD=57.29577951
I F I I S W I C H . E Q . 1» GC TO 30
C
C UP DATE IM T IA L C OE F F I C I E N T E S T I MA TE S
C
DC 10 I : l , N
10 X X I T » = Xt X I
c
C S : T THE LAST COEFFICIENT
C
S UM=0.
Z =YC OR * * 2
DO 20 1 =1 • N
S U K =S U M « X C I ) * Z / C I « 1)
20 Z= Z * Y C OR
TRUC =1 X C C R - S U M M C N V « l l / Z
XX I NV I =TRUC
C
C SET INITIAL VALUES OF Y C O R » Wt X C O R » T TO Z E RO
C
30 DO RO 1 =1 t MC
RO Y R I I ) =0.
IF I I S W I C M . E Q . 1 I WR I T E llWRlTEtGOI t Y Rt J I • J=1 * MC I
C
C SOLVE EQJATIONS FOR NSTE P INCREMENTS
C
DO 50 I =1 * N S T E P
C A L L RUNGE C MC t S T E P » Y ° I
H E T =F T H E T A I Y R ( 1 I I
THE T =A T A M T H E T I
C
C V=IVINIT**2«2(GY«W ll* « 0 .5
C
V EL= V IN IT**2«2.*(9.81*YR(ll«YR(2il
VE L =S QRT t A B S l V E L ) I
THE T 1 =THE T*RAD
C
C COMPARE CURVE TO P A P A B O L A
C
C =XC O R / I YCOR I • *2
P X = C * Y R ( 1 1**2
D I F F =Y R I 3 1- P X
I F f i s W I C H - . E 9 . 1 I WRI T E ( I W R I T E »6 01 I Y R I J ) • J= 1 » MC î , TH E T1 » VE L» P X, D I F
IF
X A R PA Y f ! » 1 ) = Y P ( 3)
289
Y A RR A Y ( ! ♦1 ! = -Y R ( 1 )
50 C ONTINUE
IFn sw iC H .C G .il call °LT c XARPAYf YARRA Y J
FüNFzYRfdl
RE TURN
C
60 F CRVAT ( 1X.1P8E16.8 1
C
E ND
SL'BP.CUTIhE FUNC (XtFtG)
DEFINES FUNCTION TO B E OPTIMISED AND C O N S T R A I N T S
I M P L I C I T DOUB L E P R E C I S I O N t A - H t O - Z l
DIMENSION X (2 5 ) t G(25 )
COMMON / C H U T E / NV » N » NSTE P »MC t XC OR t YCOR t T A L* t V I NI T * S T E P . MUt C I » C2 t YR t
1 9) t X X ! 251• TF
T- = LIM ITING THE TA FOR FAST FLOW IlN RADIANS}
TF A.zI F * 0 . 017*1532 925
TEST FOR BOUNDARY PENETRATION
M I N I I F t 0 . DO 1
EVALUATE FUNCTION VALUE. IF BOUNDARY PENETRATED APPLY PENALTY.
FrFUNF(X}-1.D20*U
CONSTRAINT EQUATIONS
THET-FTHETACYRC1 I I
THET-ATAMTHET )
G IDzTHET-TFA
RETURN
291
F fNCTÏQN F THE TA CY I
C
C EVALUATES THE F U N C T I C N FOR T HE TA AS A FUNCTION OF Y
C
I H P L I C I T DOUBLE P R E C I S I O N C A - H t Q - 2 )
CQMMON / : H U T E / N V . N f NST E P t HCt XCO Rt Y C O R» T A U t V I Nl T • S TE P • MU • C l * C 2# Y RI
I9Î.XX(25)tTF
S-O.
A X-Y
D3 10 I - l t N V
S =S * XX( I l * A X
10 A X - A X * Y
F THE TA - S
RETURN
C
END
292
SUBROUTINE DS l MQ ( A • B1 • 32 . N. KS I
C
C STIVES A SYSTEM OF SIMULTANEOUS LINEAR EQUATIONS
C
I M P L I C I T DO U B L E P R E C I S I O N I A - H . O - Z )
D I ME NS I C ^ A ( 2 5 ) . B 1 C 5 0 I * B 2 C 5 0 »
C
C F ORWARD S O L U T I O N
C
T C L - l . D -20
KS z 0
J*Jz-N
I F I I . G T - N 1 GO TO 120
DC 110 J z 1» N
JY z J* 1
Jv'z J 1
RIGAzO
I TzJJ-J
IF I J . G T . N I GO TO 30
DC 20 I z J , N
C
C SEARCH F OR MAXI MUM C O E F I C I E N T IN COLUMN
C
I J z l T ♦I
I c IA BSt3 IG A 1 -A BSlA (U m 10.20.20
10 B I G A z A ( I 1
IMAXzI
20 C O N T I N U E
30 C O N T I N U E
C
C TEST : 0 R PIVOT LESS THAN T O L E R A N C E (SINGULAR MATRIX!
C
I F ( A3 S t 91 GA l - T O L 1 <40.«40.50
«10 K S z l
RETURN
C
C INTERCHANGE ROWS IF NECESSARY
C
50 I I z J * N* t J - 2)
I T z I MAX-2
IF I J . G T . N J GO TO 70
DC GO K z J *N
I I z T 1 ♦N
I2 z Il.IT
SAVEzA i l l I
A I I I J z ( 121
AII2JZSAVE
C
C DIVIDE EQUATION 9Y LEADING COEFFICIENT
C
GO A! 1 1 1 z A C I I l / B I G A
70 SAVE z B I C I MAX 1
B1 I I M A X J z B l I J )
B 3IJ>=SA VE/ BIGA
S A V E z B 2 ! I MAX 1
0 211 rA X ) - B 2 ( J )
B 2 I J 1 - S A V E /BIGA
C
C ELIMINATE NE X T VARIABLE
C
I F ! J - Nl 8 0 * 1 3 0 * 8 0
80 I G S = N * < J - l J
00 13 0 I X = J Y * N
1 XJ - I QS« I X
IT= J-IX
I F t J Y . C T . N I G O T O 100
DO 90 J X r J Y . N
I X J X - N <I ¿ X - 1 M I X
JJX - IX JX*IT
SO A I I X J X ! = A ( I X J X ) - f A ( I X j > « A ( J J X ) >
1 00 B l I I X I = B 1 ( I X I - l Q l ( J ) « A l I X J l )
110 B 2 I I X > = B 2 ( Z X I - C B 2 ( J H A ( I X J ) I
120 C O N T I N U E
C
c BACK 5 OL UT I ON
c
130 IF IN-1) 1 7 0 . 1 7 0 » m o
140 NY -N -1
IT iN*N
IF 1 l . G T . K Y I GO TO 160
00 150 J =1 •NY
I A =1 T - J
1 3 -N- J
I C -N
00 150 K z 1 * J
BllIB)= £ l(IB!-A fIA > «B l(T C )
B2 I 19 I —3 2 C I B I - A f I A J * B 2 ( I C I
I A -I A-N
150 I C - I C - 1
160 C C N T I N U F
170 R E T U R N
C
END
294
SUBROUTINE MDW I XP • DY t Y 1
C
C E V A L U A T E S DW/OY t T A N ( T H E T A I t I C O S « THETA 1•V E L ) • •-1
C R E Q U I R E D FOR F I N D I N G Y C O R r W t X C G R t T
C I-E. D Y t l l t DY12 ) t D Y l 3)
C
R E A L MU
I M P L I C I T DOUBL E P R E C I S I O N l A - H t O - Z l
COMMON I R E A D t l W P l T E . I S W l C H t N N t M M . X S C I l O O ) f F s C t G S C l I O O ) f D E L D I F ( l O O )
COMMON / C H U T E / NV t N » NSTE Pt MC t XCOR r YCOR t TAUf V I Nl T* S TE P t MU» C 1 * C2 * YR <
1 9) t X X ( 2 5 ) * TF
DI MT N S I OF X P l M C 1 t D Y t N V 1
THET-FTHETAtY )
THET-ATAMTHET )
C S - C O S t THE T 1
SS-SINtTHET»
V E L 2 - V I N I T • •2 •2 • * ( 9 « 8 1 * Y « X P C 1 I I
C
C V - l V l N I T * * 2 * 2 t G Y « V I I I • • 0 •5
C
VEL-SQRTt ABSt VEL21I
s=o.
Z= 1 •
DC 10 I r l . N V
S - S * X X I I 1* 2
10 Z-Z< Y
TE R M - 9 . 8 1 « S S - V E L 2 * t C S * * 3 ) * S
TE RM-AB S t T E R M ) « T A U * t 1 . « C 1 / V E L « C 2 * V E L )
C
C C A L C U L A T E DW/DY
C T E RM = T A U ( l 4 C l / V « C 2 * V ) t ! G * S l N ( T H E T A ) - V * * 2 * C 0 S * * 3 ( T H E T A ) ! )
C DW/DY z - ( MU•V « T E R M » / C O S t T H E T A )
C
D Y t 1 ) - - t T E R M« MU * V E L ) / C S
DY t 2 ) -TAMt THF T )
D YI 3 >—1• / ( C S * V F L )
RETURN
C
END
295
SUBROUTINE CRADFG ( I F » X . F » GR ADF tG » GRA D G )

C
C CALCULATES NUMERICAL DERIVATIVES
C
I M P L I C I T DOUBLE P R E C I S I O N ( A - H t O - Z )
D I M E N S I O N X I 2 5 l # GR A D F ( 2 5 11 G I 2 5 1 . G R A D G C 2 5 I * GDCl 0 0 Î
COMMON l R E A D t I U R I T E t I S W I C H » N N t M M t X S C C l O O | t F S C t G s C l l O O I * D E L D I F ( I O O )
N-NN
MzMM
C
C FIND VALUES F CP F AND G.(FUNCTION AND C O N S T R A I N T S )
C
I F CM) 2 C #20 * 10
10 CALL FUNC ( X # F #G )
GO TO 30
20 C A L L F UNC ( X # F •G )
30 I F CI F -1 I M0#M0*50
MO R E T U R N
C
C FIND GRADIENTS OF F AND G
C
50 1= C l . G T . N ) GO TO 120
D0 1 1 0 I = 1 . N
DT -DE L DI F C I I
X ( I TzXC I M D T
I F ( M ) SO* 90 » 6 0
60 C A L L F UNC ( X #F D #GD )
Kz I
I F ( 1 •6 T . Ml GO TO 80
DO 70 J z 1 # M
G RAO G f K ) z ( G D ( J 1—G f J I I /DT
70 K z K ♦N
80 CONTINUE
GO TO 100
90 CA L L FUNC C X . F D f G l
100 XI I I z x ( I I - D T
110 G RAD F CI >z ( F D -F ) / DT
120 C ONT I N UE
RETURN
C
f NO
296
SUBROUTINE LAMB I N * M t =t GM»GV t S L t F L AM»R* DE LG * W• A» I C V Î

C
C COMPUTE S L AMBDA VALUES
C
I M P L I C I T DOUB L E P R E C I S I O N l A - H t O - Z l
D I P E N S I O f t F L A H ( 2 5 l t U C 5 0 M A I 2 5 I » S L C5 O I • G V I 5 0 J * H ISI2 5 ). F C5 O Ï •
1GMI501
C
C EVALUATE W l l l - F • G 111
C
A S r 0. 0
Qr O. 0
7-0*0
10 K -1
I F I 1 . GT. M 1 GO TO 50
DC 40 I r l . M
T- 0• 0
IF I 1•G T • ft) GO TO 30
DO 20 J - l t N
T =T«F I J ï *GMC K )
20 K - K ♦1
30 CONTINUE
40 Wl I ) - T
C
C EVALUATE A I M * 1 1 - 11♦J 1 - G ( I I t G l J J
C
50 K=l-N
IF I l . G T . P I GO TO 80
DC 70 I - 1 t M
K r K•N
L-K
DC 70 J = I *M
T-0.0
DO GO I i ; : l » N .
T - T ♦ GM( K 1 • GMl L I
L -L* 1
50 K =K ♦1
K =K-N
L L - M * I 1-1 ) ♦J
AILLM T
LL= M *fJ-ll*I
70 A I L L >=T
C
C E V A L U A T E RD AND Q
C - S =F - SUM t I “ 1 . M l OF F L A Ml I l ^ G 111
C - A G I ^ MAGNI TUDE CF G I I )
C - SGI = A 9 S ( S . G I I I I
C - R D - MA XIMUM V AL UE CF S G I / A G I / D E L G WITH RESPECT TO I
C - 0 =S U W OF DI AGONAL E L E M E N T S OF A
C
8 0 R3 r 0. 0
K =1
L -1
IF I l . G T . f t l GO TO 120
D3 110 I : l » H
297
SGIrW(3 1
IF C1 . GT• M ) GO TO 100
D C 90 J r 1
SSIrSGI-FLAMt J M A ( K )
90 K zK « 1
100 S G I r A B S f S G I l
AGIrSQRTfAC L I )
Q-Q*A C L )
LrL#M«l
T r A G I t DEL Q
TrSGI/T/DELG
1 10 R D z A M A X l t R D . T )
C
C COMPUTE S L I I I AND NEW V A L U E S FOR F L A M t I I
C - D r M A G N I T U D E OF CHANGE I N FLAM
C
120 Kzl
Q- Q *Z
I F f l . G T . M I GO TO l d O
DO 130 I r l i H
A CK ? r A ( K )« Q
SH Ilr-G V tll
130 K=K« K#1
1 «10 C O N T I N U E
CALL D S I F G ( AtW»SL t M t I E R )
IF C I E R ) 160*160*250
1 5 0 Z r 10. * Z * 1 . 0 - 1 2
GO TO 10
160 D r O . 0
IF f 1 . G T . M I GO TO 180
0 C 170 I = 1 * M
Zr Ul I J - FLAM! I I
FLA M fl)rw (I)
W CD rSLCIl
170 D rD ♦Z * Z
180 DrS QRT( D1
IF f l . G T . N ) GO TO 200
DO 190 I r 1 * N
KrI
s m iro .o
DC 190 J r 1 * M
SL I D r S L l I K U ( J I * GMt K )
190 K: K«N
C
C MODI FY SL C I ) I F NE C E S S A R Y
C - W r N O R M A L l S S D G R A D I E N T OF SUM OF G V ( I ) * » 2 1 I r l TO M)
C -SWrW»SL
C - S P r M A G N l T U D E OF COMPONENT OS SL P E R P E N D I C U L A R TO W
C - R r M A X I F U M ALLOWED VALUE CF S P / A 3 S C S W )
C
200 TrO. O
I F I 1 . GT. N ) GO TO 2<l0
DC 230 I : l , N
WI I )= 0.0
KrI
IF C1 . GT. M ) GO TO 220
DC 210 J -1 * M
298
W IIlzW d M G V lJM G M IK )
210 KzK*U
220 Z= W «I)♦ SlG N tl.D -20.U (Iil
WI I > - I
230 T =T * 2 • Z
240 T-SQRTCT)
SW = 0 . 0
I F t 1 •G T • K I 60 TG 260
DO 250 I - 1 t N
W I I l =WI I 1f l
250 SW=SW» W H ) » S L I I J
260 SW=SW*SÏGN( 1 . D - 2 0 . S W )
S P -0.0
I F I 1 . 6 T . K l GO TO 280
DO 270 1 = 1 tN
Z =S L 11 ) - S W * W C I I
SLITI=Z
270 SP=SP«Z*Z
280 SP=S3RTISP 1*1.0-20
T = A M I N 1 ( P * A B S ( S W I / S P » 1. »
IF 1 1 . G T . N I GO TO 300
D C 2 90 I = 1 . N
290 SL II1 = T *SL( I)«SW *W <I1
300 CONTI MUE
C
C CONVERGE »CE p A RAME TER
C
IF I I C V I 310*330*330
31 0 I I 00=0
D C 320 I =1*4
320 H I S I I I = 1 . D20
330 I C V =2
T = H I S < 11
DC 340 1 = 1 . 3
T =T * H Ï S C H 1 )
340 H I S I I I = H I S I 1* 1 I
HI S I 4 1=D
T=T/4.
IF IR D -1 0 0 .1 360*360*350
350 I 100=0
RE TURN
360 I 1 0 0 = I 1 0 0 « 1
I F I R D - 1 . 1 3 7 0 . 3 7 0 * 389
370 I C V = 0
R : TURN
380 I F ( 1 1 0 0 - 5 ) 4 1 0 . 3 9 0 . 3 9 0
390 I r 1 0 - T ) 4 1 0 * 4 0 0 * 4 0 0
400 I C V=1
410 R E T U R N
C
END
299
SUBROUTINE LlNMlN ( I ND t ND* NPT » B E G I N * S * A L F A » E ND » F N* DE LX * 0 E F X» E P S )

C
C PERFORMS ONE DIMENSIONAL SEARCH
C
I H P L I C I T DOUB L E P R E C I S I O N I A - H . O - Z 1
D I M E N S I OK SC 25 ) * F I 50 J . T C 5 0 >t B E G I N C 5 0 ) * ENDC50)
C
I F I I N D J 50 150 110
10 F I K K l z F N
N PT-NPT « 1
I C KCGT=N30T0
I F I I C K C G T . L E . 1 ) GC TC 80
1= ! I C KCGT « G E •6 I GO TO 380
GC TO C 8 0 . 9 0 » 1 3 0 * 1 7 0 . 310* 380 ) » ICKCGT
20 I F ( 1 . G T . N D ) GO TC 40
C
C GET FUNCTION VALUES
C
DO 30 I -1 • ND
30 E KDCI i - B E G I N I I M T I K K ) * S I I )
40 C O NT I N U E
RETURN
C
C CHANGE IP* AL F A CORRESPONDING TO DE L X AND DEFX
C
50 I ND - 1
NPT -0
Z=0. 0
I F U . G I . N D I GQ TO 70
DC GO I r 1 ,ND
GO Z =Z » S CI 1» S f I I
70 Z r S a R T I Z )
Z-AMAX1CZ.1.D-20)
DEL=DELX/Z
DE F zDEF X/ Z
C
C O B T A I N T H R E E P O I N T S * L* M AND N? WI T H L AND N ON O P P O S I T E SIDES OF
C Mt P C L ) AND F I N ) NOT L E S S THAN F I M) * AND THE D I STANCE S' CL TC M)
C AND ( M TO N) AT L E A S T D E F X
C - I N I T I A L AND RECOMMENDED P O I N T S
C
T t l ) =0. 0
KK -1
NG CT C - 1
G3 TO 20
80 F KEE F=F C 1 )
T(2)rALFA
K K -2
NG OTO-2
G C TC 20
C
C - FOI NTS F AND L
C
90 L=1
M- 2
300
IF «Ffl)-F<2>> 100t l lO t llO

100 M=1
L -2
110 I F 1A B S C T C M l - T C L 1 1 - D E F I 1 2 0 t l 5 0 t l 5 0
120 TIL»=TCP)«SIGW(DEF .T (L )- T C M l)
KK~L
NG0TC=3
GO TO 20
130 I F C F C L I - F C M I ) 1 R O . I 5 0 . 1 5 0
1 «I 0 I - L
LrK
Mr I
C
C -POINT N « D OU B L E STEPLENGTH EACH T I M E )
C
1 50 .0
2 - 1
Nz 3
160 Tl N l z T C M M Z * C T C M ) - T ( L ) )
KKrN
NGOTOr *i
G 0 TO 20
170 I F (F C N I- F C M I) 180*200.200
180 I - L
U H
MrN
Nr I
7 - 2 . 0
I F I ABSf T f H l - T I L I 1 - 1 . E 2 0 ) 160*190*190
190 I ND r 2
RETURN
C
C D E C R E A S E THE D I S T A N C E t L TO Nl TO L E S S THAN * I * D E F X . KEEPI NG THE
C DISTANCE? (L TO M) AND (M TC N ) AT L E A S T DE F X
C
200 NE Wr H
NBA D r 0
C
C -LET L 3E CLOSER TO M THAN I S N
C
210 I F t B B S « T C M I - T I L I l - A B S * TC M) -7 I N i l ) 230.230.220
220 IrL
UN
Nrl
Q '
C ! - E S T I M A T E THE P O S I T I O N OF THE Ml Ni ML' M POINT (NEW) FROM A PARABOLIC
C FIT FrA*3*T«C*T**2
C
230 T l r T C L ) - T CMl
T2rT(N)-T(M)
H lrA 3 S(FtLl-F(M ll/Tl
H 2 r A E S C FC NI -F C M I I / T 2
Cr I H 2 - H 1 1 / C T 2 - T 1 I
B zfH1*T2-H2*T1)/( T 2-T11
TC N E W ) r T ( M l - B / 2 • / CC«S I G N C 1 . D - 3 0 . C I 1
C
C - E ND CYCLE WHEN DISTANCE (L TO N) IS LESS THAN 4 »0 E F x
C
301
I' IM i.»O EFI 37 0 . 3 70» 40
C
C GElMETRi: AVE R A GE G c THE DISTANCES II TO Ml and (L TO Nl
2*»0 IF INBA3-2I 260*250.250

250 TINEWlzSGRTCT1*CT1-T2 I I
Tl N E W l z J f L I « S I G N I T I N E W I t T 2 l
NBADzO
c
c -NEW W I L L BE C L O S E R TC M THAN TO N OR L • I F NEW L I E S W I T H I N
c
D E F X OF H . CHANGE T C N E WJ SO THAT NEW L I E S B E T W E E N H AND N. AT A
c D I S T A N C E D E F X FROM M
c
2G0 I F f A B S f T C N E WI - T C K I l - DE F I 270.280 .280
27 0 TI N E W I - T I M M S I G N I D E F *721
c
c -LET NEW L I E B E T WE E N M AND N
c
280 I " I I T ( NEW l - T t M 1 ) * 1 2 ) 290.300.300
290 I = L
L-N
N=I
C
C -I MPROVE L.M AND N
C
300 KHzNEW
NG CTOz 5
GO TO 20
310 Z z A B S C T I N I - T I L I I
IF CFINEW I-FIM I) 320.330.330
320 I z L
L zM
Mz NE W
N E Wz I
GO TO 340
330 I zN’
Nz NE W
NEWzI
C
C -TEST THAT DISTANCE (L TO Ni DECREASED BT AT LEAST TE N PE R CENT
C
340 IF f ABSI T(N)-T( L I ) / Z - . 9 ) 350.360.360
350 NBADzO
GG TO 2 1 C
36 0 N B A D z N B A D M
GO TO 2 1 0
C
C OBTAIN THE FUNCTION VAL UE AT THE ESTIMATED Ml NU MU M P O I N T
C
370 KKzWFW
NG 0 TO z 6
GO TO 20
380 l z I F ( NEW1-FCM1I 4 0 0 . 4 0 0 . 3 9 0
390 NEWzM
400 ALFAz TINEWJ
F NzF f NE W I
IF 1 1 . GT. NO I GO TC 420
302
DC ¿4 10 I =1» NO
' n o E D I I ) =3 E G I N ( I ) « A LF A • S ( I )
'420 C O N T I N U E
C
C TEST WHETHER IMPROVEMENT IS SIGNIFICANT
C
I F f F K E E P - F N - E P S I *130 « *1*10 » *1*10
*130 I N D r - 2
RETURN
*4*40 1= I A B S l AL F A 1 - DE L I *130 • *450» *150
*150 I NDr -1
RETURN
C
END
StBPOUTIFE LINK C I F •X * F •G R A O F * 6 t G R A D G J
C
C APPLIES SCALING FACTORS TO F U N C T I O N AND INDEPENDENT VARIABLES
C
I M P L I C I T DOUBLE P R E C I S I O N C A - H . O - Z 1
D I M E N S I O N X I 2 51 . GR ADF I 25 11 -G 125 I * G R A D G I 2 5 1 * X R A W l l O O l
COMMON I R E A D t l W R U E t I S W I C H t N N » M M * X S C ( 1 0 0 J »F SC .G SC ( 1 00 J t DE L O I F ( 1 0 0 )
N=NN
K=HH
C
I F f l . G T . N ) GO TO 20
D3 10 1 = 1 . N
10 X R AWCI ) = m M X S C C I I
20 CONTINUE
IF IMJ EC.G0.30
30 C A L L GRADF G f I F « X R AW. c * G R A O F • G . G R A D G 1
I F I l . G T . M l GO TO 50
00 *10 J=1 v M
40 G l Jl=G( Jl/GSCC J )
50 CONTINUE
GO T P 70
SO CA L L GRADF G « I F t X R A W . F v G R A D F « G t G R A D G 1
70 F =F /F SC
I c IIF-11 140.140.80
80 IF f l . G T . K 1 GO TO 130
DO 120 1 = 1 . N
G R A D F ( I 1 =GRADF C I 1 / F S C * X S C f I 1
IF I M 1 120.1 2 0 .9 0
90 K=I
I" 11 . GT. M ) GC TO 110
DC 100 J =1 . M
GRADG ( K 1 = G R A D G I K I / G S C I J M X S C I I I
100 K =K * N
110 CONTINUE
120 CONTINUE
1 30 CONTINUE
140 RETURN
C
E NO
SUBROUTINE MA TUP I H* T ND * N*S » G Q • G N* A L F A * D î L X • DE LG * W I
C
C COMP UT E S SEARCH DIRECTIONS
C
I M P L I C I T DOUBLE P R E C I S I O N I A - H . O - Z J
D I HE N S I OK HC 50 11 S I 2 5 I * G Q C 2 5 I * G N Î 2 5 J * WC50)
GE N E R A T E I D E N T I T Y MATRIX
o n
- I D I A G O N A L E L E M E N T S ARE H f l l * H I N « i l * H t 2 * N ) * H I 3 *N-2 ) t ---- § H ( N* NJ )
I- 1Î N D - 2 I 10* 80* 120

10 K =1
I e t l . G T . N I GO TO 60
DC 50 I =1 tN
HI K î -1 • 0
NJ-N-I
TF I N J I 7 0 * 7 0 * 2 0
20 I F I l . G T . K J » 6 0 TO 40
03 30 J z l t N J
H J r K ♦J
30 H l K J ) - 0« 0
HQ CONTI NUE
50 H- K J * 1
60 C O N T I N U E
70 I F t I N D I 8 0 * 8 0 * 1 1 0
C
C DIRECTION OF STEEPEST DESCENT
C
80 IF 11.GT.N1 GO TO 100
OC 9 0 I =1 . N
90 S f I ) = - GDI I J
100 C O N T I N U E
110 RETURN
C
C UPDATE MATRI X H
C -COMP UT E G R A D I E N T CHANGE Y A n O S TORE IN GN
C
120 I F I I N D - 1*) 1 3 0 * 1 3 0 * 3 0 0
130 I F ( 1 . G T . N 1 GQ TO 150
DC 1 40 I =1*N
T =GNI I Î
G N I I } - T - G C CI )
1*»0 G O I T Î - T
C
C -COMP UTE -I • Y AND STORE IN G ill TO Gt N1
C
1 50 CONTINUE
I F f I . G T . N Ï 60 TO 210
D3 200 I - 1 • N
T - 0. 0
X- I
IF I l . G T . N) GO TO 190
DO 130 J - l * N
T r T * G N( J M H I K )
1= I J - I 1 1 6 0 * 1 7 0 * 1 7 0
305
icn k =k ^ : - j
Gn TC l a o
170 K =K« 1
130 C 3 NTT NUE
190 CONTINUE
200 W II)=T
C
C -COMP UTE POSITION CHANGE Z AND S T O R E IN G ( N* 1 ) TO Gl2*N)
C
210 K=N
IF l l . G T . K ) GO TO 230
00 220 1 =1 »N
K =K» 1
22 0 Wl K 1=ALF A * S t I 1
C
C -COMPUTE Q-YtH.Yt P =Y . Z AND R=SQRTI7*Z)
C
230 R-0.0
Q-0. 0
P =0 . 0
K -N
IF l l . G T . N ) GO TO 250
DO 240 I =1 . N
K- K ♦1
R=R« WCK ) * W( K )
Q= Q «G N (I ) * W( I )
240 P = P « G N i I 1 *WC K )
250 R - S Q R T I R )
C
C -LEAVE H UNCHANGED IF P OR R ARE VERY SMALL
C
IF 1 R- DE L X ! 3 0 0 . 3 0 0 . 2 E 0
260 IF I P - R ♦DE LG J 300* 3 0 0 . 2 7 0
C
C - C C MP U T F NEW H
C
270 K:l
I e « 1 . GT. N ) GO TO 290
DO 280 1 =1. N
L I =N* I
DC 280 J =1 . N
L J =N» J
H t K J =HC K 1 * W ( L I ) * Wt L J l / P - K C I > « W t J * / Q
230 K= K ♦1
290 CONTINUE
C
C COMPUTE NEW D I R E C T I C N
C
300 I F ( I ND -4 ) 3 1 0 . 3 8 0 . 3 1 0
310 I F I 1 « GT• N ) GO TO 370
DC 3 E 0 1 =1 . N
T =0 • 0
K=I
I F f l . G T . N ) GO TO 350
DC 340 J =1 * N
T=T*G0(J1«H(KI
IF I J - I » 320.330.330
306
3 20 Kr K♦ N- J
GC TP 3 0
330 K-KM
340 CONTINUE
350 CONTINUE
360 SII*r-T
370 CONTINUE
380 RETURN
E ND
SUBROUTINE PLT (XARRA YtY AR RA YJ
I M P L I C I T OOUBLE P R E C I S I O N ( A - H . C - Z )
REAL XXRAY.YYRAY
C
C SUBROUTINE TO P L O T RESULTS
C
COMMON I R E AD» I U R I T E » I S W I C H . N N t M M f X S C l l O O ) f F s C f G S C t l O O ) » D E L D I r i l O O )
CCKMON / C H U T E / NV » N r N S T E P » MC »XCOR t Y C O R » T A U » V I N l T » S T E P » M U * C 1» C 2 • Y R i
1 9 ) • X X 1 251 * TF
0 I HE K S I O K X A R R A Y I 5 0 ) . YARR AY f 50 I * X X p A Y ( 5 0 ) t YY RA Y i 50 I
1= I I N T C Q . G T • 0) GO TO 10
CALL START
NSTFPlzNSTEP«!
XXRA Y l N S T E P 1 * I ) z 0
XX RA Y f NST E P l * 2 l = «8<l6 6
Y Y R A Y f N S T E P 1 « 1 1=0
YYRAY<NSTEP1«2I=.8*|66
10 I K 7 E Q=I NTE Q* 1
DO 20 I -1» N S T E P 1
XXRA Y ( I ) =SNGLC X A R R A Y ( I I )
20 Y Y R A Y t l ) - S N G L l Y A R R A Y ( I ) « 5 « )
CALL L I N E ( X X R A Y » Y Y R A Y * N S T E P 1 » 1 » I N T E Q 1
I F < l M T E 3 . E Q . i l CALL AX E N 1 5 . 9 05 5» . 59 OS • 1 . 1 8 1 2 *2 t 0. . 1 . 1 1 5»
1»X A X I S ( HE TRE S ) * » 5 . 9 0 55 * . 590 6 » 1 . 1 8 1 2 » 2 » - 5 . » I •»15 t
2 *Y A X I S ( M E T R E S ) t t - . 0 5 » - . l l
RETURN
C
E NO
308
SUS ROUTI NE RUNGE I MC » ST E P* Y R )

C
C 4ÏH ORDER RUNGE KUTTA ROUTINE
no
EVALUATES NEXT INCREMENT FOR YCOR>U*XCCR»T»THETAtVEL
I M P L I C I T DOUBLE P R E C I S I O N C A - H . C - Z J
D I M E N S I O N DY t 3 1 t Y R I M C I . X T f l O O ) . Y M i l O O Î . Y K t l O O Î
D I M E N S I O N X P C I O O Ï » Z K l l O O Î t WK CI O O ) * X K t l O O î
NQ - M C - 1
TWrYRi1 J
S T E P H - O . 5•STE P
T ? r T W ♦S TEPH
C
C EVALUATE RUNGE KUTTA COEFFICIENTS Z Kt WK * XK » YK
C
DC 10 I r l , K Q
10 XTIT)= YRII«1]
DO 20 I = 1 »NQ
20 XPlT)= XTfI 1
CALL F 4 DW t X P . D Y . T W I
DO 30 I Mr 1 »NQ
30 ZKIIM IrO Y(IM )
DO 4 0 I -1 » N9
40 X P tl I r X T C I M STEPH*ZK( T )
C A L L F4DW t X P » D Y * T Z I
DC SO I r l »NQ
50 WKI I J r D Y t I I
DC S 0 I r l »NQ
GO X P i I l r X T I I MSTEPH*WK!I1
C A L L F4DW I X P » D Y » T Z 1
DD 70 I r l » NQ
70 X K 11 î r D Yf I î
DO 80 I r l » NQ
80 XFtI > rXTfIM ST EP*XK fI 1
C A L L F 4DW CXP »D Y »TW« S T E P J
OC 90 I r 1 *NQ
90 YK! T 1 r DY ( I I
C
C SET NEW V A L U E S F OR W t X C O R » T T H E N Y CH R
C
DO 100 1 = 1 »NQ
T E R M r Z K ( I Î « 2 . * W K ( I )« 2 . « X K ( I ) *YKCI ï
100 YM IIîrXTCIî«STEP*TERM /6.
Y R 11 J r T W« STE P
DO 110 I r 1 »NQ
110 Y R 11 «11 r YMt I >
RETURN
C
END
309
SUBROUTINE SCALER CM • M* X t DE L * T OP * BOT * B* GRA DG1 « GR AD G2 )

C
c c a l c u l a t e s s c a l i n g f a c t o r s f o r f u n c t i o n and i n d e p e n d e n t v a r i a b l e s
c
I M P L I C I T DO U B L E P R E C I S I O N f A - H t Q- 2 1
D I F E N S I O A X I 25 1 • G R A D G 1 C 2 5 ) * G R A 0 G 2 I 2 5 ) * G I l O O ) » G p U O O i * G M l O O l ,
1 3RAD F1I100)* GRADF2I100)* ABGDllOO)
C OMMON I READ » I W R I T E t I S V J I C H » NNt MHt XSC ( 1 00 ) *F SC *G SC Cl 00 ) *DE L D I F ( 100 I
C
NN=N
MM=M
C
c FIND FIRST AND SECONO DERIVATIVES
C
DELIAS8RTC A B S ID E L II
DE L3 =DE L • 1. / 3 . )
FSC rl.0
IF I l . G T . N ) GO TO 20
DO 10 1 =1 t N
10 X S C I I ) =1 . 0
20 CONTINUE
IF I K ) 6 C *6 0 * 3 0
30 IF l l . G T . M ) GO TO 50
DC 1«0 J =1 *M
qo G scu izi.o
50 CONTINUE
60 C A L L L I N K 11 » X •F •GRAD=1 * G*GRADG1 1
IF l l . S T . N ) GO TO 110
DO 100 I = 1 tN
X I I ) =Xt I l « D E L 3
CALL L I N K t l * X . F P * G R A D F l * G P * G R A D G l )
X I I >=XCI ) - 2 . * D E L 3
C A L L L I N K I 1 # X t F M t G R A D F 1 t GMt GRADG11
X I I >=X( I ) « D E L 3
GRADF1 I I ) = I F P - F M ) / 2 . / D E L 3
G R A D F 21 I 1=1F P « F M - 2 . *F ) / D E l 3 / D E L 3
IF I M 1 100*100*70
70 K:I
I F t l . G T . M ) GO TO 90
DO 80 J : l » M
G R A O S H K l =«GPt J I - G M I J1 1-/2./DEL3
G R A D C 2 I K ) -C GP I J M G M C J ) - 2 « * G ( J ) ) / D E L 3 / D E L 3
80 K=K«N N '
90 CONTINUE
100 CONTINUE
110 CONTINUE
C
C FIND F 5C AND F I R S T A P P R C X I M A T I ON TO XSC (SCALE FACTORS)
r
F SC =ADS l F J
IF IM) 260*260*120
120 K =1
I F l l . G T . H ) GO TO 160
DC I S O J =1 * M
s =0 . o
I F ( l . G T . K l GO TO 140
03 130 1 = 1 *N
S=S*GRADG1< K ) * * 2
1 3 0 X= K ♦1
140 C O N T I N U E
150 A 3 G D I J l z S Q R T I S l
I GO A N=N
R N = 1 0 •• S3 R T t ANI
IF t l . G T . K ) GO TO 210
Ou 200 1 = 1 tN
F S1=ABSCGRADFIIII/F »
FS2=SQRTt A B S C G R A D F 2 C I l / F I )
S =0. 0
K=I
IF t l . G T . M GO TO 190
DO 1 SO J =1 t N
G F A T = A B S ( G R A D G 1 I K I I / ABGD( J I • TOP*RN/BOT
IF IGRAT-1.1 1 8 0 t l 8 0 t l 7 0
170 T = A 9 S ( G F A D G 2 C K l / G R A D G 1 C K I I
S=ANAX1IS*T)
180 K =K♦N
1 90 S =A N A X H F S 1 # F S 2 * S * 1 . / T 0 P )
S =A NI N I t S t l . / B O T I
2 0 0 x s c n 1 = 1 . /s
FIND G SC <G S C A L E )
210 K=1
I F I l . G T . M ) GO TO 250
DO 24 0 J =1 * M
S =0. 0
IF ( 1 . G T . N I GO TO 230
DO 220 1 =1 * N
T =G RADG1 ( K I * X S C <I I
S=S« T*T
220 K= K ♦1
230 CONTINUE
240 GSCIJ1=S3RTfS I
250 CONTINUF
260 I F 1 1 . G T . N I GO TO 320
FIND FINAL VALUES OF XSC AND DEL DI F
DO 310 1 = 1 *N
S =0 . 0
ic tN) 300*300*270
270 K=l
I e t l . G T . M ) GO TO 290
DO 280 J = 1 * M
TrGRADG1( K ) / G S C ( J )
S =S ♦T * T
280 K= K ♦N
293 S =2. *B * S
300 S=S*CRA0F2CIl/FSC
S=ANAX1( S t l . / T O P / T O P )
S=ANAXltSQRTtl./S)*BOT)
XSC II)= S
31 1
DE L DI F ( I l = X S C 11 1« D E L I
3 10 X I I ) : X ( I l/XSCCI )
3?0 CONTINUE
RF TURN
E ND
312
10.5.3 Step Responses
Often in system identification studies, as was the case here,
there arises the need to compute the step response of the system.
The hopper-chute system’s response to the stepped opening and shutting
of the flow control valve was investigated. The interpretation of the
cross-correlation and impulse function responses for various chute
geometries was greatly facilitated by integrating these to find their
equivalent step response. Transient characteristics such as rise-time,
overshoot and settling time becoming clearly evident and readily
interpretable from the equivalent step response.
A short programme using Simpson’s one-third rule was written
to calculate these step responses and it incorporated plotting
routines to present the results in graphical form. Since this
technique is well known and easily programmed, a listing is not
included here.
10.6 Check of Value of Viscous Drag
10.6.1 Viscous Drag Based on a = dv/dt
Continuing from Equation 4.4
f2 ---- — ------ = / dt (4.4)

1 (E - — - Fv ) 1 repeated
v
vdv______
= f-
(E - £ - Fv ) (Ev - c - Fv2 )
v v
vdv
-F
(v2 - I v + £ )
-i r(2v - ! + & dv
2F , 2 E v .c
(V -
F + F}
313
9 9
2 2E v ç E_. ,c__E_
(completing the square)
v - 2F F 2F + F 4F2
0 E
2V - F +
f-
2F v 2 - —E v ,fc— 2 Ev c
F F - T F
1 I / 2 E v , cN E r dv
Ô F ln (v - -p + — ) - f (4.5)
2F F F 2F2 2 E v .c
V F + F
Now the second term in 4.5 becomes
dv
2F‘ 2 2E v c
V - 2F + F
dv (completing the square)

/
2F2 J . E .2 ,c J 2
V " 2F F " 4F2
Let (jü = v -
2F
E _ c_
a2 -
4F2 F
dv
/-.
2F2 J E_ 2 _ ,_Ef_
(v ‘ 2F 4F2
Upon substitution
E f_ dv
2F2 w2 - a2
1
E < '
2F2 2a a) - a u) + a
[ In (a) — a ) - In (a) + a )]
4F2a
314
Back substitution into 4.4 yields
E v2
’1n (oo - a ) = AT
2F ln (V? " f V + f 4F2a
vi
Ev £ gv2
1 E 1n (oo - a )
In (v2- -f v + ~) -
2F F F 4F2a b (oo + a)
Vl
2 Ev . c. E ^1 n (oo + a )
2 p In (v - -p + — ) +
4F2a (oo - a)
Vl
1
2 p"
I
In
, 2 Ev
(v — ~
c.
+ —) + —
E In (a) + a )- v2
F 4F2a (oo - a)
"v 1
v2
,n (V 2 - | v + § t F + In
F F (co - a)
Vl
1v 2
co + a 4F2a 1
In
oo - a (v2 - Ev + £)J f AT
F F
vi
E Ev c\
2F
ico2 + a\ 4F2a v i 2 - F + F
In = AT
CM
oo2 - aj ooi + aJ - Ev + c
>
CM
F F/
Simplifying by collecting terms
(oo2 + a) (ooi - a)\4F2a F v i2 - Evi + c \ 2F

In = AT
(oo2 - a ) (ooi + a )/ F v 22 - E v 2 + c j
Taking the inverse or anti-loge
ooioo2 + a (ooi - oo2 ) - a2 1 4F2a F v i^ - Evi + c \ 2F AT

= e
ooioo2 + a(oo2 - ooi ) - a 2 F v 22 - Ev2 + c
(4.6)
315
Now substituting back the variable values
E
0)1 - Vi -
2F
E
0)2 = V2 -
2F
2 _ E2 £ _ E2 4F
4F2 " F 4F2 4F2
a = /E2 - 4Fc
Cons ider
0)i0)2-VV - EP-^V1+v2^+Ê2p"
1 2 2 2
and
E
0)2 ” 0)1 — V 2 “ - Vi - 2 p = v2 - Vi
2F
substituting into 4.6 yields

/
VlV2- p(Vi +V) + “ a (v - vi ) - E^y- cF
E
77
2F 2
E^2
2
VV - p-^V1+v2^+"ÿp
1 2 2
T
Y+a ^v2“ Vi) --L-
T1 + £
4F2 F
e
/ /
Viv2 - ^ (vi + V 2 ) - a (V2 - vi) + y

- —
VV - ^Vl +V2^+a (v2-Vi) +-p
1 2 2f
2F v i v 2 - E (v 1 + V 2 ) - 2Fa (V2 - Vi) + 2c

2Fv i v 2 - E (vi + V 2 ) + 2Fa (V2 - Vi) + 2c
And final Iy
E
4F2a
2FVlv2 - E (vi + V 2 ) - 2Fa (v 2 - Vj) + 2c
2F v i v 2 - E (vi + V 2 ) - 2Fa (V2 - vi) + 2c
_L
2F
Evi2 - Evi + c (t2 - t i )
e (4.7)
^Fv!2 - Ev 2 + c
316
10.6.2 Viscous Drag Based on a = v d v / d s
Continuing from Equation 4.9

v2dv
= -Fds (4.9)
, 2 , c Ev
repeated
Now the L.H.S. of Equation 4.9 yields
(.,2 , c Ev Ev c. ,
(v + p - — + — - F ) dv
( V 2 + £. -
Ev
F
= 1 + dv
— (2v - i) - £ + I L \
2F v f’ F 2F2
1 + dv
(v2 + c _ Ex
’ /
Substituting into Equation 4.9 yields
E E c F‘2
1 + — {2v - pr) - p" + 2 F 2"
dv = -Fds (4.10)
Ev
(v2 + j
integrat ing
^ E i /v2 + c V2 /
EV^ + dv
V 2F ln F - T 2F2 /
\ J _ Vl Ev
_ (v2 +
F F '
=-F(s2-si) (4.11)
Right hand term of Equation 4.11 becomes
c ly_____ dv
2F2 F J, o .c Ev which is similar to R.H.S. of
(vz + ^ -
Equation 4.5.
Iett ing
a =
4F2
317
and
a) = v -
2F
the right hand term of Equation 4.11 becomes
afa)
2F2 F i (a)2 - a 2 )
E2 c dfa)
2F2 F 1 1 1
2a o)-a o)+a
u
1. /.E2 c \
2a 2p2 ^r"l (fa)-a ) — In ( (aH-a ) ]
1 / E2 c\ (fa)-a )
2a [ 2F2 " F In
(fa)+a )
Equation 4.11 becomes on substitution
V + ^ ln(v2 + f -
V2 c i . /fa)-a v2
2a 2F2 FÌ
F / 1niI
fa)+a,
Vl Vl
= -F( s 2 - Si)
i.e.,
c I . /faj-a V2
+ 1 il
v + w ,n r + f F 2a 2F2 F lrW
VI
=-F(s2-si) (4.12)
Applying the limits of integration 4.11 becomes
Ev, \-^r 1 /E2

v22 + f - 2F (fa)2-a) (fa)j+a) \2a 2F2
In
Ev;
\vi2 + f (o)2+a ) (fa)i-a )
=-F(s2-Si) - (v2-vi) (4.13)

318
Taking the anti loge of Equation 4.13 yields
FV22 + c - Ev? // . _\ / . \\ 2a 2F2

Fv + c - Evi
e F (s 2 “ sj) - (V2 “ Vi)
_ e_F(s2 - si)
e(V2“ Vi)
i.e
L J_ E2 _ c
2p / , x\ 2 a 2F2 F
( V 2 “ V i ) | FV 2 2 + C - EV2 o)2-a /o)i+a
Fv i 2 + c - 0)2+a (a)i-a
- e“ F(s2 - Si) (4.14)
Now
(o)2~a)(ooi+a) _ o)ia)2-a (g)i-g)2 )-a2

(u)2+a)(o)i-a) a)i0)2+a (o)i-o)2 )-a2
Similarity of expression between Equations 4.7 to 4.14
OJi—OJ2 - V i “ V2
OJ1C02 V 1V2 - J p (V1+V2) + - ^ 2
a)1a)2-a (u)i-o)2 )-a:

becomes
0)i0)2+a (032-0)2 )-a2
/ /
E E2 E2
viv2 - 2F (vi+V2) + /i?2- a (Vl-v2) - w
F^ c
V1V2 - 2 f ivi+v2) + “ a(vi-v2 ) - ,JpT + f
_ 2Fv i v 2 - E ( v i + v 2 ) - 2F a ( v i - v 2 ) + 2c
2FV1V2 - E ( v 1+ v 2) + 2Fa( vi ~V2) + 2c
Substitution into Equation 4.14 yields

E
(v2-vi) Fv 22 + c - Ev? 2F
F v i 2 + C - Evi
_L I t
2Fv i v 2 - E(v1+y2) - 2Fa(y 1-vo) + 2c 2a 2F^
2Fv i V2 - E (v 1+v 2 ) - 2Fa(vi-v2) + 2c
- e-F(s2 - Si) (4.15)

319
10.7 Pub Ii cat ions
During the course of this research programme the opportunity of
presenting the progressive results at conferences, etc., was
considered appropriate and the following co-authored papers were
written.
MONTAGNER, G.J. and TROTT, G.W. "A General Programme for System
Identification and Analysis Using a Desk-Top Mini Computer” .
Proc. Conf. on Computers in Engin. Sydney May 1974 pp.203-207.
ROBERTS, A.W. and MONTAGNER, G.J. ’’Solids in a Hopper-Di scharge
Chute System” .
Symp. on Solids and Slurry Flow and Handling in the Chemical Process
Industries. Amer. Inst, of Chem. Eng. Pittsburgh, June 1974.
ROBERTS, A.W. et a I . ’’Bulk Handling of Granular Materials: Some
Dynamic Identification and Optimisation Studies” .
Agricul. Eng. Conf. Uni. of N.S.W. Vo I.1 Aug. 1974 pp.422-444.
MONTAGNER, G.J. et al. "Optimisation Identification and Control
Applied to Gravity Flow of Grain in a Storage Bin-Discharge Chute
System” .
IFAC Symp. on Auto. Control for Agricul. Uni. Saskat. June 1974
pp.D6 1-15.
ROBERTS, A.W. and MONTAGNER, G.J. ’’Flow in a Hopper-Discharge Chute
System” .
Chem. Eng. Progress Vol.71 No.2 Feb. 1975 pp.71-72.

320
The author’s involvement in the Creative Design Competition
conducted as part of the subject Design I offered by the Department
of Mechanical Engineering at the University of Wollongong resulted
in the following co-authored papers.
MONTAGNER, G.J. et al. ’’Student Design Contest".
Machine Design Vol46 No.18 July 1974 pp.20-22.
MONTAGNER, G.J. et al. "Closing University-Industry-Community
Gap Through Conceptual Design in First Year Engineering Courses".
Proc. Annual Eng. Conf. Hobart Feb. 1975 pp.305-310.

Identification and Optimisation of Hopper-Discharge Chute Systems

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Identification and Optimisation of Hopper-Discharge Chute Systems

Uploaded by

Copyright:

Available Formats

University of Wollongong

Identification and optimisation of hopper-

SYSTEMS FOR BULK GRANULAR MATERIALS

G.J . MONTAGNER, B.E.(N.S.W.),

Thesis submitted for the Degree

Department of Mechanical Engineering,

The University of Wollongong.

LIST OF PRINCIPAL SYMBOLS 16

1.2 Outline of the Problem 19

1.3 Review of Previous Research 20

1.3.1 Granular Materials 20

1.3.2 System Identification 21

1.4 Scope of this Research 22

PART 1: UNIFORM FLOW 24

CHAPTER 2: GRANULAR MATERIAL FLOW DYNAMICS 25

2.2 Description of the Apparatus 26

2.3.2 Lumped Parameter 36

2.3.3 Statistical Evidence 40

2.4 Uniform Flow 43

2.4.4 Equation of Motion 47

3.2 Model Formulation 53

3.2.1 Minimum Descent Time 53

3.2.2 Maximum Exit Velocity 55

3.2.3 Additional Constraints 56

3.3 Methods of Solution 59

3.3.2 Polynomial Approximation 59

3.4 FIetcher-PoweI I Algorithm 61

3.4.2 Programming Considerations 63

CHAPTER 4: RESULTS AND DISCUSSION - UNIFORM FLOW 64

4.1 Comparison of Model Results with Theoretical 64

4.2 Check on the Value of Viscous Drag 66

4.3 Comparison of Model with Chutes ofOther Profiles 70

4.4 Sensitivity Analysis of Model Parameters 72

4.5 Theoretical Scale up Analysis 80

4.6 Design Procedure 94

PART 2: TRANSIENT FLOW 97

CHAPTER 5: TRANSIENT FLOW 98

5.1 System Identification 98

5.2 P.R.B.S./Cross-correI at ion Method 100

5.2.1 Theoretical Considerations 100

5.2.2 Hardware Implementation 105

5.2.3 Parameter Selection 110

5.3 Programme Development 111

5.3.1 Genera I 111

5.3.2 On-Line NOVA Package 112

CHAPTER 6 : CORRELATION TESTS 117

6.1 Genera I 117

6.2 Apparatus 117

6.3 Procedure 120

6.4 Step Responses ' 121

6.5 High Speed Cine Photography 123

6.6 ’Two-Dimensional’ Transparent Hopper 128

CHAPTER 7: RESULTS AND DISCUSSION - TRANSIENT FLOW 131

7.1 Transient Behaviour of Straight InclinedChutes 131

7.1.1 Impulse Function for Straight Inclined Chutes 132

7.1.2 Bode Diagram for StraightInclined Chutes 136

7.2 Transient Behaviour of Curved Chutes ofKnown Form 140

7.2.1 Impulse Function Comparison of Straight Chute

with Curved Chutes ofKnown Form 141

7.2.2 Bode Diagrams for Curved Chutes 143

7.2.3 Comparison of the Optimum Chute at Varying

Flow Rates 145

7.2.4 Bode Diagrams for the Optimum Chute with