University of Wollongong
Research Online
University of Wollongong Thesis Collection
University of Wollongong Thesis Collections
1991
Gravity flowrate of bulk solids from mass flow bins
Zhi Hong Gu
University of Wollongong
Recommended Citation
Gu, Zhi Hong, Gravity flowrate of bulk solids from mass flow bins, Doctor of Philosophy thesis, Department of Mechanical
Engineering, University of Wollongong, 1991. http://ro.uow.edu.au/theses/2095
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
GRAVITY FLOWRATE OF BULK SOLIDS
FROM MASS FLOW BINS
A thesis submitted in fulfilment of the requirements
for the award of the degree of
Doctor of Philosophy
from
The University of Wollongong
by
Zhi Hong G u
B.Sc. ( NEUT ), M.Sc. ( NEUT )
Department of Mechanical Engineering
1991
DECLARATION
This is to certify that the work presented in this thesis was carried out by the author
in the Department of Mechanical Engineering of the University of Wollongong and
has not been submitted for a degree to any other university or instituition.
Zhi H o n g G u
i
Acknowledgements
I would like to acknowledge my supervisor Professor P.C.Arnold, Head of
Department of Mechanical Engineering, The University of Wollongong, for his
grateful supervision, assistance, encouragement and other help during the period of
this study. I also wish to thank m y co-supervisor Dr. A. G. McLean, Senior
Lecturer in the Department of Mechanical Engineering of the University of
Wollongong for his helpful suggestions and supervision during m y last two years
of P h D candidature.
I am very grateful to the University of Wollongong for providing me a
Postgraduate Research Award for this study.
Acknowledgement is also made to all the staff of the department, especially to Mr
M . Wall and Mrs R. Hamlet. M a n y thanks are extended to the technical staff in the
Workshop and Bulk Solids Handling Laboratory in this department with whose
help and expertise the experimental apparatuses were constructed, in particular to
Mr. D. Cook, Mr. S. Rodd, Mr. S. Watkins, Mr. R.Young and Mr. K. Maywald.
I a m grateful to Mr. D. Jamieson for his help in using the computer systems.
Finally, special acknowledgement is made to my dear wife Jun Dan, my son Yu
Fan and m y parents for their encouragement and considerable help.
ii
ABSTRACT
The gravity flow of bulk solids from mass flow bins and hoppers is a subject of
considerable practical and theoretical interest M a n y published papers have brought
greater understanding of the flowrate of coarse particles. Research on finer
materials was insufficient due to imprecise predictions of the air pressure gradient at
the hopper outlet, which is a very important factor in predicting the flowrate of the
fine particles.
The work in this thesis develops theoretical models for predicting the interstitial air
pressure gradient and for predicting the flowrate of bulk solids from conical mass
flow bins. T h e theoretical models are based on the continuum mechanics theory.
T h e boundary conditions follow air pressure and bulk density continuity in a
vertical direction. A close agreement between theoretical results and experimental
results was obtained for predicting both flowrate and air pressure distribution. Both
theoretical and experimental results indicate that the flowrate of free flowing bulk
solids increases rapidly at first and then more gradually as the particle permeability
constant increases.
The use of 'permeability' enables the theories developed to be applied in describing
the flow behaviour of both coarse and fine particle mixtures as well as composite
particle size-distributed bulk solids. The use of consolidation-related bulk density
and permeability enables the theoretical models to be applied to both compressible
and incompressible materials.
iii
Results of both theoretical and experimental work on the effect of material
surcharge level on the flowrate indicate that for fine materials this effect is
significant
Based on the original theoretical model, simplified expressions for predicting the
dynamic deaeration coefficient K J
and a simplified flowrate model are presented.
Finally, the study is extended to examine, theoretically and experimentally, the use
of standpipes attached at the hopper outlet to increase the limiting gravitational
flowrate.
iv
Table of Contents
Acknowledgements i
Abstract
ii
Table of Contents
iv
List of Figures
ix
List of Plates
xvii
List of Tables
xviii
Nomenclature
xx
Chapter
1
Introduction
1
2 Literature Survey 7
2.1
Research on Flowrate without Air Retardation
2.2
Research on Flowrate with Air Retardation
7
20
3 Measurements of Bulk Solids Flow Properties 35
3.1
Introduction
35
3.2
Particle Size and Particle Size Distribution
37
3.3
Effective Angle of Internal Friction and Angle of Wall
Friction
39
3.4
Particle Density
41
3.5
Bulk Density
41
3.5.1
Bulk Density Measurement
44
3.5.2
Models Used to Fit Experimental Results
45
3.5.3
The Models Appropriate to the Bulk Solids
47
3.5.4
Application of the Bulk Density Equation
54
3.6
Permeability
61
3.6.1
Permeability Measurement
61
3.6.2
Models Used to Fit Experimental Results
63
The Prediction of Air Pressure Gradients in Mass Flow Bins 67
4.1
Introduction
67
4.2
The Bulk Density Distribution in Mass Flow Bins
69
4.3
Superficial Velocity of Air Relative to Particles
74
4.3.1
Flowrate of Air
74
4.3.2
Relative Velocity of Air to Powder
75
4.4
Air Pressure Gradients in Mass Flow Bins
76
4.5
Air Pressure Distributions in Mass Flow Bins
78
4.6
Boundary Conditions for Application of Air Pressure Model
80
4.7
Discussion
81
4.7.1
General Observations
81
4.7.2
Comparison of Theoretical Model with Experiments
85
Prediction of the Flowrate of Bulk Solids from Mass-Flow Bins 96
5.1
Introduction
5.2
Theoretical Model for Predicting the Flowrate
5.3
96
98
5.2.1
Equation of Motion
5.2.2
Equation of Continuity
100
5.2.3
Equation for Predicting the Flowrate
102
Application of Theoretical Model
98
104
5.3.1
The Effect of Surcharge Level on the Flowrate
104
5.3.2
The Effect of Outlet Size on the Flowrate
113
5.3.3
The Effect of Permeability Constant on the Flowrate
119
VI
5.4
Comparison of Theoretical Results with Experimental Results
5.4.1
121
Sensitivity Analysis of the Various Terms in
Theoretical Model
121
5.4.2 Comparison of Equation (5.17) with Other Flowrate
Models
6
Experimental Facilities and Test Bulk Materials
125
134
6.1
Introduction
134
6.2
Test Rig for Measuring Mass Flowrate
135
6.3
Measurement of the Flowrate
139
6.3.1
Calibration Procedure
139
6.3.2
Test Procedure
142
6.4
6.5
Processing the Flowrate Data Measured
6.4.1
Original Fitting
6.4.2
Problem of Initial Fitting and its Improvement
6.4.3
Typical Flowrate Measurements
143
144
149
Preparation of the Bulk Solids Mixtures
150
6.5.1
The Range of Different Sand Mixtures
150
6.5.2
Median Particle Size and Size Distribution of the Bulk
Materials
6.5.3
6.6
143
154
The Internal Friction Angle and the Wall Friction
Angle
157
6.5.4
Particle Density
159
6.5.5
Bulk Density and Permeability
160
Measurement of Air Pressure Distribution in Mass Flow Bins
161
7 The Effect of Permeability on the Flowrate 168
7.1
Introduction
168
7.2
Experiments Using Glass Beads
168
7.2.1
Preparation of Glass Bead Mixtures
170
7.2.2
Mixtures of Particles for Laboratory Measurements
173
7.2.3
Mixtures of Particles for Flowrate Measurements
175
7.2.4
The Observed Particle Flowrates from the Bins
177
7.3
The Effect of Particle Size and Size Distribution
180
7.4
Discussion
184
7.5
Criterion for Classifying Coarse and Fine Bulk Solids in
Terms of the Effect of Interstitial Air Pressure Gradients
189
The Effect of the Surcharge Level on the Flowrate 194
8.1
Experimental Results
194
8.2
Discussion
199
8.3
Problems Which M a y B e Caused by Increasing the Surcharge
8.4
Level
202
Summary
212
Simplification of the Model for Predicting the Particle Flowrate 217
9.1
9.2
Original Model for Estimating the Dynamic Deaeration
Coefficient K d e a
217
Simplification of the Original K ^ e a Model
219
9.2.1
Random Simulation of K ^ e a
219
9.2.2
Sensitivity Analysis
222
9.2.3
Simplification of K<j e a Equation by Optimization
Technique
9.2.4
Comparison of K ^ e a Using the Original Model and
the Simplified Model
9.3
224
The Simplification of the Flowrate Model
231
239
10
Strategies for Increasing Limiting Flowrates
243
10.1 General Possibilities for Increasing Limiting Flowrates
243
10.2 Experiments and the Observations on Standpipe
246
10.3 Theoretical Model for Bin - Standpipe Configuration
250
10.4 Theoretical Predictions and Discussion
256
11 Conclusions 262
12 Suggestions for Further Work 268
References 270
Appendices
I
Theoretical Analysis Section
1-1
The Young's Modulus Coefficient K y for Bulk Solids under
Uniaxial Test
1-2
283
Equivalent Major Principal Stresses in Hopper Region of a
Mass Flow Bin
1-3
283
287
Difference Caused by Considering Variable Bulk Density in
Walters' Equation for the Vertical Stress in Cylindrical
Section of the Bins
293
II Experimental Measurement Section 295
III
II-1
The Results of Particle Size Analysis by Laser Particle Sizer
II-2
Instantaneous Yield Loci Measured for All Test Materials
300
n-3
Wall Yield Loci Measured for All Test Materials
307
Publications While P h D Candidate
295
314
IX
List of Figures
Figure Title Page
3.1 Gravity Flow of Bulk Solids from a Mass-Flow Bin 36
3.2
Yield Locus and Internal Friction Angle
40
3.3
Wall Yield Locus and Wall Friction Angle
40
3.4
Jenike Compressibility Tester (Arnold et al. 1980^)
44
3.5
A Typical Bulk Density Curve of a Fine Sand
45
3.6
The Flowchart of the Program to Calculate Minimum Variances
48
3.7
Total M i n i m u m Variance of Different Bulk Density Models for 79
Tests
3.8
3.9
49
Comparison of Experimental Results with Those Fitted by Model 13
in L o w Stress Range
50
Replotting of Figure 3.7 b) with Modified Model 13
51
3.10 The Typical Fitting Results for Best Fitting Models with Zero Stress
Excluded
51
3.11 The Typical Fitting Results for Best Fitting Models with Zero Stress
Included
52
3.12 Worst Fitting of Models 9,14,16 and Modified Model 13
52
3.13 Best Fitting of Models 9,14,16 and Modified 13
53
3.14 Pressing of Bulk Material in T w o States
55
3.15 Under-Estimation of Case I for Bulk Density for Plane Flow Hopper
and Conical Hopper Compared with the Results of Case II
60
3.16 Jenike Permeability Tester (Arnold et al. 1980[3])
62
3.17 Comparison of Fitting Results for Three Permeability Models
65
3.18 Worst Fitting of Permeability Model 2
66
3.19 Best Fitting of Permeability Model 2
66
Figure
Title
Page
4.1 Regions Defined for a Mass Flow Bin 69
4.2
Stress Field in Hopper (McLean 1979 [58] )
4.3
A Plot of the Relation between T|mp, rj0 and T| m a x
4.4
Typical Distributions of Air Pressure and Bulk Density in a Bin (with
Surcharge)
4.5
91
92
Comparison of Theoretical Results with Head's Experimental Results
(Sand No. 6)
4.9
90
Comparison of Theoretical Results with Head's Experimental Results
(Sand No. 10)
4.8
83
Comparison of Theoretical Results with Crewdson et al.'s
Experimental Results
4.7
81
Comparison of Theoretical Results with Experimental Results for
Alumina
4.6
73
Predicted Stress Distributions of Alumina During Flow
4.10 Predicted Bulk Density Distributions of Alumina During R o w
93
94
95
5.1 Flow from Region IU of a Conical Mass-Flow Bin 98
5.2
Predicted Flowrate Varying with H/D Ratio for Alumina, P V C
Powder and Sugar
106
5.3
Predicted Flowrate Varying with H/D Ratio for Sand M l to Sand M 6
107
5.4
Predicted Flowrate Varying with H/D Ratio for Sand M D 1 to M D 4
108
5.5
Predicted Flowrate vs. H/D Ratio for Sand M D 3
109
5.6
Predicted Flowrate vs. H/D Ratio for Sand M 5
109
5.7
Predicted Flowrate vs. H/D Ratio for Sand M D 1
110
5.8
Predicted Flowrate vs. H/D Ratio for Sand M l
110
XI
Figure
Title
Page
5.9 Predicted Air Pressure Gradient at Hopper Outlet vs. H/D Ratio for
SandMD3
111
5.10 Predicted Air Pressure Gradient at Hopper Outlet vs. H / D Ratio for
SandM5
111
5.11 Predicted Air Pressure Gradient at Hopper Outlet vs. H / D Ratio for
SandMDl
112
5.12 Predicted Air Pressure Gradient at Hopper Outlet vs. H / D Ratio for
Sand M l
112
5.13 Periodic Flow Behaviour for Fine Material under Fluidisation
Condition
114
5.14 Predicted Flowrate vs. the Hopper Outlet Diameter for Sand M D 3
115
5.15 Predicted Flowrate vs. the Hopper Outlet Diameter for Sand M 5
115
5.16 Predicted Flowrate vs. the Hopper Outlet Diameter for Sand M D 1
116
5.17 Predicted Flowrate vs. the Hopper Outlet Diameter for Sand M l
116
5.18 Predicted Air Pressure Gradient at the Hopper Outlet vs. the Hopper
Outlet Diameter for Sand M D 3
117
5.19 Predicted Air Pressure Gradient at the Hopper Outlet vs. the Hopper
Outlet Diameter for Sand M 5
117
5.20 Predicted Air Pressure Gradient at the Hopper Outlet vs. the Hopper
Outlet Diameter for Sand M D 1
118
5.21 Predicted Air Pressure Gradient at the Hopper Outlet vs. the Hopper
Outlet Diameter for Sand M l
5.22 Predicted Flowrate for Varying Permeability Constant
118
120
5.23 Comparison of the Effects of Density Gradient, Stress Gradient and
Air Pressure Gradient at Hopper Outlet for Sand M D 3
123
Xll
Figure
Title
Page
5.24 Comparison of the Effects of Density Gradient, Stress Gradient and
Air Pressure Gradient at Hopper Outlet for Sand M 5
123
5.25 Comparison of the Effects of Density Gradient, Stress Gradient and
Air Pressure Gradient at Hopper Outlet for Sand M D 1
124
5.26 Comparison of the Effects of Density Gradient, Stress Gradient and
Air Pressure Gradient at Hopper Outlet for Sand M l
124
5.27 Comparison of Theoretical Results with Experimental Results (D 0 =
0.02 m & H/D = 0.07)
127
5.28 Comparison of Theoretical Results with Experimental Results (D 0 =
0.02 m & H/D = 1.5)
128
5.29 Comparison of Theoretical Results with Experimental Results (D 0 =
0.0445 m & H/D = 0.07)
129
5.30 Comparison of Theoretical Results with Experimental Results (D 0 =
0.0445 m & H/D = 1.5)
130
5.31 The Percentage Error Between the Flowrates Obtained by Current
Model and Brown's Theory for Coarse Bulk Solids
133
6.1 Mass Flowrate Qp v. H/D Ratio (Smith 1978[53]) 134
6.2
Schematic of the Double - Bin Apparatus
136
6.3
The Schematic of Bin Hanger
137
6.4
Transducer Circuitry
138
6.5
Instrumentation Schematic
139
6.6
Calibration Arrangement
140
6.7
Typical Calibration Recorder Output
141
6.8
Comparison of Indicated Load with Actual Load
142
xin
Figure
Title
Page
6.9 Flowrate Prediction for Four Test Runs at One Surcharge Level for
Coarse Sand (Sand M l )
145
6.10 The Flowrate of Every Run after Smoothing
147
6.11 Flowchart of Fitting Procedure
148
6.12 Comparison of Flowrates for Different Loads Measured on the
Double Bin Apparatus with 0.0445 m Outlet
149
6.13 Schematic of Large Screen Siever
152
6.14 The Frequency Distribution of Selected Sand Mixtures
153
6.15 Cumulative Size Distribution for Sugar, Alumina and P V C Powder
156
6.16 Cumulative Size Distribution for S and M 1 to M 7
156
6.17 Cumulative Size Distribution for Sand M D 1 to M D 4
157
6.18 Flowability Characteristic in Free Flowing Zone (According to
Jenike's Flowability Zones'- ••)
159
6.19 The Apparatus for Measuring Air Pressure Distribution
163
6.20 Large Test Bin Geometry and Pressure Tapping Locations
164
6.21 Schematic of the Surcharge Level Control System
167
7.1 Mixture Bulk Density Characteristics 173
7.2
Mixture Permeability Characteristics
174
7.3
Mixture Equivalent Surface Area Characteristics
174
7.4
The Flowrates vs. Permeability Constant
179
7.5
Flowrate vs. Median Particle Size at Surcharge Level H/D = 1.5
180
7.6
Measured Flowrate vs. Median Particle Size for Sand M l to M 5 and
7.7
S a n d M D l t o M D 4 ( H / D = 1.5)
183
Flowrate v. Permeability for Sand Mixtures
183
Figure
Title
Page
7.8 Predicted Negative Air Pressure Distribution for Bulk Solids (from
0.02 m outlet)
7.9
185
Predicted Negative Air Pressure Distribution for Bulk Solids (from
0.0445 m outlet)
dP
7.10 -«- at the Bin Outlet vs. Permeability Constant of Particles
185
186
7.11 Predicted Air Pressure Distribution for Sand M l (at H/D = 1.5)
187
7.12 Predicted Air Pressure Distribution for Sand M D 3 (at H/D = 1.5)
187
7.13 Negative Air Pressure Distribution (Willis[54])
188
7.14 The Critical Curve of Air Affected Flow in Terms of Particle
Diameter (Flowrate Data from Crewdson et al.' -fy
193
8.1 Measured Flowrate vs. H/D Ratio for PVC Powder and Sugar 196
8.2
Measured Flowrate vs. H/D Ratio for Sand M l to Sand M 5
197
8.3
Measured Flowrate vs. H/D Ratio for Sand M D 1 to Sand M D 4
198
8.4
Comparison of Experimental Results with Theoretical Results for
Sand Mixtures Flowing from 0.0445 m Outlet
8.5
Pseudo-Steady Flowrate Affected by Surcharge Level (from 0.02 m
Hopper Outlet)
8.6
208
Weight Curves for Discharge of 55 Lim Calcite (adopted from Miles
(1968)[44])
8.9
207
Mass Variation Plots for Sand M D 3 , Sand M 6 and Sand M 7 from
0.0445 m Outlet
8.8
206
Mass Variation Plots for Sand M D 3 , Sand M 6 and Sand M 7 from
0.02 m Outlet
8.7
204
209
Flowrate Variation Plots for Sand M 7 Discharging from 0.0445 m
Outlet
210
XV
Figure
Tide
Page
8.10 Variable Speed Belt Feeder 211
8.11 Flowrate Variations with Different Belt Feeder Velocities for Sand
M 7 from 0.0445 m Hopper Outlet
214-215
8.12 The General Description of Flowrate Affected by Surcharge Level
216
9.1 The flowchart of Uniform 1 -1 Generator 221
9.2
Total Variance for Every Component (5000 Samples)
223
9.3
Total Variance for Every Case (5000 Samples)
b,
b,
Typical Variations of K ^ with the — ratio and — 4 n r ratio
224
9.4
Po
Po
e
226
bi
9.5
9.6
9.7
9.8
Typical Variations of K d e a with sin(8) and
* n < ratio
p0(sin 8 ) U 1
Program Flowchart for Optimization Processing
Comparison of K d e a Predicted by Simplified Model 1 with by
230
Original Model
233
Comparison of K(j e a Predicted by Simplified Model 2 with by
Original Model
9.9
226
233
Comparison of K<i ea Predicted by Simplified Model 3 with by
Original Model
234
9.10 Comparison of Simplified Model 2 with Simplified Model 1
234
9.11 Comparison of Simplified Model 3 with Simplified Model 1
235
9.12 Comparison of Predicted Pressure Gradient at Outlet Based on
Simplified K^ea Model 1 with that on Original K^ea Model
235
9.13 Comparison of Predicted Pressure Gradient at Outlet Based on
Simplified K^ea Model 2 with that on Original K ^ a Model
236
9.14 Comparison of Predicted Pressure Gradient at Outlet Based on
Simplified K ^ Model 3 with that on Original K d e a Model
236
Figure
Title
Page
9.15 Comparison of Predicted Particle Flowrate Based on Simplified K ^
Model 1 with that on Original K^ea Model
237
9.16 Comparison of Predicted Particle Flowrate Based on Simplified Kjjea
Model 2 with that on Original K^ea Model
237
9.17 Comparison of Predicted Particle Flowrate Based on Simplified K^ea
Model 3 with that on Original K^ea Model
238
9.18 Comparison of Simplified Flowrate Model with Original Flowrate
Model For the Experimental Parameters Examined
242
10.1 Schematic of Double - Hopper Bin Arrangement 244
10.2
Standpipe Installed at the Outlet of the Test Bin
10.3
Flowrate Q p vs. the L / D 0 Ratio of Standpipe for Alumina
248
Discharging from 0.0445 m Outlet (Surcharge Level H = 0.31 m )
249
10.4
Typical Discharge Modes from a Standpipe
249
10.5
Flowrate vs. Feeder Belt Velocity for Different Clearances between
Feeder and Standpipe Outlet (standpipe L/D 0 = 5)
250
10.6
Four Regions for Bin - Standpipe Arrangement
251
10.7
Comparison of Theoretical with Experimental Results
10.8
Flowrate Enhancement Factor for Different Materials (X = 0.75)
257
10.9
Predicted Air Pressure at the Hopper Outlet (X = 0.75)
260
10.10 Predicted Air Pressure Gradient at the Hopper Outlet (X = 0.75)
256
260
10.11 Predicted Air Pressure Distribution in the Bin and the Standpipe for
Alumina (X = 0.75)
261
10.12 Predicted Air Pressure Distribution in the Bin and the Standpipe for
Sand M l (X = 0.75)
261
Figure
Tide
Page
A-I-1.1
A n Element of Bulk Solids Compacted by Three Stresses
284
A-I-1.2
Compressibility Tester
284
A-I-2.1
The Cylindrical Coordinates for the Plane Flow Hopper
288
A-I-2.2
The Spherical Coordinates for the Conical Hopper
289
A-II-2.1 Instantaneous Yield Loci for Alumina 300
A-H-2.2
Instantaneous Yield Loci for P V C Powder
300
A-E-2.3
Instantaneous Yield Loci for Sugar
301
A-H-2.4 to 2.10
Instantaneous Yield Loci for Sand M l to Sand M 7
A-II-2.11 to 2.14 Instantaneous Yield Loci for Sand M D 1 to Sand M D 4
301-304
305-306
A-H-3.1 Wall Yield Loci for Alumina 307
A-H-3.2
Wall Yield Loci for P V C Powder
307
A-H-3.3
Wall Yield Loci for Sugar
308
A-n-3.4 to 3.10
Wall Yield Loci for Sand M l to Sand M 7
A-H-3.11 to 3.14 Wall Yield Loci for Sand M D 1 to Sand M D 4
308-311
312-313
List of Plates
Plate Title Page
6.1 Test Rig for Measuring Particle Flowrate
10.1 Bin - Standpipe - Belt Feeder Arrangement
10.2 Standpipes Used in Experiments
* (-) opposite page
136(-)*
248(-)
248(-)
xviii
List of Tables
Table Title Page
3.1 Shear Test Methods and Their Comparison (based on Schwedes
1983 t92] and Schwedes et al.l990[93'94])
3.2 The Models for Bulk Density - Stress Relation
42
46
4.1 The Pressure Maxima and the Ratio between Two Extreme Pressures
for Different Sized Bulk Solids and Hopper Outlet Sizes
85
4.2 The Flow Properties, Bin/Hopper Geometry and Observed Flowrates
for the Experiments of Crewdson et al. and Head
89
5.1 The Comparison of Eqn (5.17) and Nedderman et al.'s Method with
Head and Crewdson et al.'s Experiments
126
6.1 Bin Details 137
6.2 The Calibration Mass Increments
140
6.3 The Range of Particle Size for River Sand Mixtures
151
6.4 The Median Particle Size of the Test Materials
155
6.5 The Shear Test Results for the Test Materials
158
6.6 Measured Particle Densities
159
6.7 Measured Bulk Density Properties
160
6.8 Measured Permeability Properties
161
7.1 Bulk Densities of Glass Bead Mixtures 170
bi v
7.2 The - ^ Ratio for the Bulk Solids Prepared in Section 6.5.5
171
Po
7.3 Comparison of the Three Mixtures Parameters
176
7.4 The Observations for Mixtures with Similar Bulk Density
177
Table
Title
Page
7.5 The Observations for Mixtures with Similar Permeability 177
7.6 The Observations for Mixtures with Similar Equivalent Surface Area
178
7.7 The Relation between Experimental Flowrate from 0.02 m Outlet and
Median Particle Size or Permeability Constant
182
7.8 The Relation between Experimental Flowrate from 0.0445 m Outlet
and Median Particle Size or Permeability Constant
182
8.1 Actual Surcharge Level of Solids in the Test Bin with 0.02 m Outlet 194
8.2 Actual Surcharge Level of Solids in the Test Bin with 0.0445 m Outlet
195
8.3 Comparison of the Results with Other Previous Works in Terms of the
Effect of Surcharge Level on Flowrate
203
8.4 The Classification of the Flow for All Sand Mixtures Used in
Experiments in Terms of the Effect of Material Surcharge Level
213
9.1 The X components in Simulation 220
9.2 Simplified K ^ Model Minimization Results
231
10.1 A Summary of the Results in Using Standpipes 258
A-I-3.1 The G Values for the Materials Used in Experiments 294
The Results of Particle Size Analysis by Laser Particle Sizer:
A-H-l.l
for Alumina
295
A-H-1.2
for P V C Powder
296
A-n-1.3 to 1.8
for Sand M l to Sand M 6
A-H-1.9
for Sand M D 1
296-299
299
XX
Nomenclature
a exponent used to relate permeability to consolidation stress
a**
equivalent surface area of particles (* 10 -12 m 2 )
a n , b ll' c ll
coefficients for flowrate model, eqn (5.17)
&22' b22
coefficients for simplified flowrate model, eqn (9.21)
A
cross sectional area of flow channel (m 2 )
Ay
cross sectional area of the hopper outlet (m 2 )
Ap
cross sectional area of permeability test cylinder (m 2 )
b
exponent used to relate bulk density to stress in eqn (2.7)
bi, b 2 , b 3
constants used to relate bulk density to stress in density equation
models
B
optimum parameter array in Chapter 3
B0
initial value of B in Chapter 3
Bf
constant in Beverloo's flowrate model, eqn (2.1)
c, cj, C2
constants
C
permeability of bulk solids, * 10 -9 ( m 4 N ^ s e c 1 )
CQ
permeability constant of bulk solids (permeability at lowest
compaction), * 10 -9 (m 4 N 4 sec 4 )
CCTi
critical permeability value classifying the coarse and fine material,
*10- 9 ( m 4 N 4 s e c 4 )
dso
median particle size (|im)
dg
equivalent particle diameter used by in eqns (2.9) to (2.11) (|im)
d«
particle size (Lim)
in Beverloo's model, eqn (2.1), the unit of dp is metre)
D
diameter of vertical section of the bin (m)
DQ
outlet diameter of the hopper (m)
bulk density ratio, as defined by f = —
Po
bulk density ratio in Regions I, II, III, respectively, defined by
eqns (9.2) to (9.6)
function f of x
first derivative of f with respect to x
second derivative of f with respect to x
flow factor for a converging channel, in eqn (2.3) and eqn (2.7)
'actual' flow factor for a flow situation, in eqn (2.3)
flowrate enhancement factor due to standpipe
acceleration due to gravity, gravitational acceleration (m/sec2)
1
dP
effective gravitational acceleration, g* = g +
( j-) __ (m/sec2)
Pout
°
variable defined by eqn (A-3.8)
vertical distance from the vertex of the flow channel (m)
vertical distance from the vertex of the flow channel to outlet of
hopper (m)
vertical distance from the vertex of the flow channel to the transition
of a hopper (m)
height where the stress is assumed to be zero (near the hopper outlet),
h* = r* cos a (m)
surcharge level used in eqn (2.7) (m)
material surcharge level, as defined as the height of bulk solids in
vertical section of the bin (m)
dimensionless surcharge level, defined in eqn (4.1) and eqn (10.1)
height of powder bed (m)
a function of the ratio of the volume of the arch to the perimeter of the
arch, in eqn (2.7)
k
constant in Beverloo s flowrate model, eqn (2.1)
kh k 2
constants
Kj , K2 , K3
constants
K^ea
• ^ •
J •
• • ,
^
l+sin8
ratio of major and minor principal stresses K =
1-sin 8
dynamic deaeration coefficient of bulk solids in the mass-flow bins
K$ea
dynamic deaeration coefficient evaluated by original K ^ model
K^Ja
dynamic deaeration coefficient evaluated by simplified K(j e a model
Ky
coefficient related to Young's modulus in Chapter 3 and Appendix I-1
Ks
constant, K s = ( g ° ) 1 / b 2
K^
variable used in Walters' stress theory
L
length of standpipe (m)
M
mass flowrate of particles (considering the sense of flow), M = - Q p
K
(kg/sec)
P
air pressure in voids between particles (Pa)
P0
atmospheric pressure, P Q = 0 (Pa)
Pj
air pressure at transition level (Pa)
Q
volumetric flowrate of air (m3/sec)
QQ
volumetric flowrate of air at top level of material in a bin (m3/sec)
Qp
flowrate of a bulk solid (kg/sec)
Qpujax
flowrate of a bulk solid unaffected by pressure gradient (kg/sec)
r
radial distance from the vertex of the flow channel (m)
r*
the distance from the vertex of the channel to where the stress is
assumed to be zero, r* = 0.95 TQ ~ 0.99 TQ (m)
r0
radial distance from the vertex to hopper outlet (m)
rj
radial distance from the vertex to the transition of a hopper (m)
u
relative velocity of air to the particles (m/sec)
U
variable defined by eqn (9.29)
V
particle velocity (m/sec)
V5
bulk solids volume after compaction (m 3 )
Vbo
bulk solids volume before compaction (m 3 )
Vbelt
velocity of feeder belt (m/sec)
V0
particle velocity at the hopper outlet (m/sec)
Vt
Wt
terminal velocity of a bulk solid, in Table 10.1 (m/sec)
4hiM
coefficient defined by W n =
*
(Pa)
° C0p07tD02k
net weight of particles (kg)
y
constant used in Walters' stress theory
W
n
0
Y=[yj, ..., y n ] variables used in optimization analysis in Chapter 9
Z
constant, Z = kdp (m)
AP
air pressure drop across by powder bed (Pa)
AP
air pressure difference caused by resistance of particles (Pa)
AP0
air pressure difference caused by resistance of distribution gauze of
permeability tester (Pa)
APt
total air pressure drop (Pa)
% S and % L
the percentage of small and large particles, respectively
Greek Symbols
a
hopper half angle (degree)
Tl
weighting factor of the influence of the pressure gradient on flowrate
8
effective angle of internal friction of bulk solids (degree)
e
voidage of bulk solids
En
voidage of bulk solids corresponding to P Q
voidage of bulk solids at one third of the w a y along the hopper wall
from the hopper outlet, used in eqn (2.8)
strain
dimensionless vertical distance from the vertex of hopper, defined in
eqn (4.1) and eqn (10.1)
standpipe effect coefficient
absolute viscosity of air (Pa .sec)
bulk density of particles (kg/m 3 )
bulk density of particles at lowest compaction (kg/m 3 )
bulk density of particles at transition section of a bin (kg/m 3 )
bulk density corresponding to an arbitrary stress o e , used in
eqn (2.7) (kg/m 3 )
density of interstitial fluid (kg/m 3 )
solids density of particles fl^g/m3)
principal stress (kPa)
stress (kPa)
major principal stress (kPa)
an arbitrary stress for bulk density equation used in eqn (2.7) (kPa)
mean stress in three dimensional field (kPa)
m e a n stress for compressibility tester (kPa)
equivalent major principal stress (kPa)
computed m e a n stress in plane hopper (kPa)
computed m e a n stress in conical hopper (kPa)
Poisson's ratio of bulk solids
Poisson's ratio of bulk solids at m i n i m u m voidage
friction angle between bulk solids and bin / hopper wall (degree)
Molerus internal friction angle, infigure3.2 (degree)
variance for optimization analysis in Chapters 3 and 9
CO
constant used in permeability model 1 (Section 3.6.2)
(VV)V
convective terms, e.g., x-component in rectangular coordinat
dWx dVx dWx
A
V_-!C*
x dx + V„y~3y~
5- + V , -r*
Subscripts
r
r component
6
B component
<D
O component
z
z component
out
at hopper outlet
mp
minimum pressure position
c
conical hopper
w
plane flow hopper (wedge hopper)
V
vertical section
max
maximum
min
minimum
sp
standpipe
Co-ordinate Svstems
r, 0, z
cylindrical co-ordinate system for •
r,e,o
spherical co-ordinate system for cc
1
Chapter 1 Introduction
Bins and hoppers are used in a wide range of industries, such as agriculture,
mining, chemical engineering, power plants, cement and food processing, where
most bulk solids storage, handling and transportation systems are applied. O n e of
the most important requirements is that the material should discharge smoothly and
continuously w h e n the outlet is opened. O n e practical problem involved in
designing the handling systems for particulate materials is the attainment of an
adequate flow of material and the control of the flow at some desired rate.
When bulk solids are allowed to flow out of a bin or hopper under gravity alone, it
flow pattern can be basically of two types: mass flow or funnel flow^1 \ With
mass flow, the hopper is sufficiently steep and smooth to cause flow of all the
solids in the bin without 'dead' regions occurring during discharge. B y contrast,
funnel flow occurs w h e n the hopper is not sufficiently steep and smooth to force
material to slide along the walls or when the outlet of a bin is not fully effective, due
to poor feeder or gate design. The bulk solids flow toward the outlet through a
vertical channel that forms within stagnant material powders. From the viewpoint of
processing, mass flow is preferred in making the bulk solids processing system
efficient, reliable, predictable and more easily controlled. The flow of solid particles
from mass flow bins or mass flow hoppers is, therefore, a subject of considerable
practical and theoretical interest.
During the past three decades, considerable advances have been made in the general
understanding of the flow behaviour of particulate solids^
' ' . Most
publications deal with theoretical and experimental work concerning the flow
2
patterns in the bins and the mass flowrates which depend on the bulk solid itself
and the bin/hopper geometry. Generally speaking, previous research on the
flowrate of bulk solids from m a s s flow bins or hoppers can be classified as
follows:
i) according to different bulk materials concerned:
coarse - cohesionless - incompressible - no effect of air pressure gradient
on the flowrate considered' " ^
fine - relatively cohesionless - compressible - retarding effect of air
pressure gradient on the flowrate considered™ " ^
ii) according to different bin/hopper configuration:
conical bnVhoppert9-13'15'23-25'27'28'32'36-38'49'50'55-58'60]
plane flow bi^oppert 2 2 ' 3 0 ' 3 1 ' 3 3 ' 3 4 ' 3 6 - 3 9 ' 5 1 - 5 4 ' 5 8 ^
flat-bottomed bint 1 0 - 1 2 ' 1 4 - 1 8 ^
iii) according to the models of calculating the flowrate:
empirical correlations™ '
theoretical analysis based on:
- fluid analogies[20,21]
- continuum mechanics theoryt 22 - 24 ' 27 - 28 ' 31 - 33 ' 36 " 39 ' 46 ' 49 ' 51 ' 52 '
54-56,58]
- m i n i m u m energy theory^ ^
- stochastic model 125 - 26 ' 1133
- kinematic model^ 29 ' 303
Finite Element Analysis (a few applications for coarse material; normally
associated with the determination of the stress field)™' *
3
iv) in terms of the method of solution used for the theoretical models
simplification by assumptions, boundary conditions'22'27'28'34'38'
46,47,49,50,52,55-57,60]
perturbation technique^24,31'32'
numerical analysis'3 , 7 , ^S4'*8*^]
Almost all the empirical correlations, Finite Element Analyses and a number of
theoretical equations at present available in the literature to predict the flowrate of
particles discharging from mass flow bins are applied only to coarse particles.
A m o n g these studies, the most successful are those obtained by Beverloo et alJ ^
B r o w n ^ 1 J, Johanson^ \ They are verified by subsequent research. However,
w h e n particles becomefiner,but still substantially cohesionless, their motions are
significantly different from those of coarse material. The equations from the studies
on coarse material often overpredict the flowrates for fine materials' ' ' 1 , a
situation which needs to be avoided when dealing with practical applications.
An important difference between the flow behaviour of coarse and fine materials is
that the flowrate of fine particles is affected significantly by the interstitial air
pressure resistance. S o m e attempts have been m a d e to consider this effect on the
flowrate^43"60^. O f these attempts, Nedderman and his co-workers have paid more
attention to and m a d e great progress in examining the flowrate affected by
interstitial air pressure gradients^ 50 ' 51 ' 56 ' 571 . Particularly, the research by
Nedderman et al. in 1 9 8 3 ^ is considered the most successful of their studies.
Starting from the Beverloo empirical correlation, which w a s obtained for the
flowrate of coarse material from a flat-bottomed container, Nedderman et al.
developed semi-empirical equations to establish the relationship between the
4
flowrate for fine material and the air pressure gradients encountered by the particles.
However, the equations provided by Nedderman and his co-workers include one
variable A P (air pressure difference at the hopper outlet), which cannot be predicted
from their theories. Therefore, it is difficult to predict the flowrate completely from
their equations.
Ford and Davies in 1990^6 * concluded that' it is now clear that an air deficiency, at
the hopper outlet, appears for particles smaller than 500 |im and this deficiency
increases as the m e a n size decreases reducing the flow predicted by the Beverloo
equation', where the constant in the Beverloo equation is determined from
Nedderman et al.'s semi-empirical results concerning the effect of interstitial air
pressure gradients on the particle flow. In practice, it is difficult to apply the
Nedderman model to a range of actual materials. Firstly, a model for predicting the
air pressure gradient at the hopper outlet needs to be established. Secondly, using a
single particle diameter of 500 |im to delineate between coarse and fine particles is
inadequate. T h e bulk solids processed in industry often include a range of particle
sizes. M a n y researchers have found that the flow properties are affected by the
particle size distributicHi^63"703, indicating that a theory needs to be developed which
can be applied to materials with a distribution of particle sizes.
Parallel to Nedderman's work, the effect of interstitial air pressure gradients on the
flowrate of bulk solids has been also studied for several years by Arnold, M c L e a n
and their co-workers, at the University of Wollongong^ 52 ' 54,58 " 603 . A s a factor in
relating the particle flow to the air pressure gradient, the parameter 'permeability'
was introduced into their models which are based on Carleton's force balance
theoretical model' 49 ' or Johanson's actual flow factor model' 22 '. Since insufficient
experiments have been done to compare with their theoretical results, the theories
5
have not yet been published. However, the concept of using the parameter
'permeability' is most valuable, although m a n y researchers evade it due to its
sensitivity to voidage between particles. However, according to the Carman Kozeny equation, it is this voidage that relates the air pressure gradient under a
certain superficial velocity condition. T o simulate the air pressure gradient, the
variation in voidage needs to be faced, therefore, permeability can be used for
adequately predicting the air pressure gradient. In addition, permeability is a
characteristic of a particle mixture, not of one single particle. Since the effect of
particle size distribution on the flowrate is of practical interest (Thomson 1986' 3 ) ,
it seems that permeability is a useful parameter to describe the relationship between
pressure gradient and superficial velocity for size-distributed particles.
Another difference between fine particle flow and coarse particle flow relates to
effect that surcharge levels of bulk solids have on the flowrate from mass flow
bins. This effect has long been considered to be slight™. However, it is noted that
this conclusion was drawn from experiments on coarse materials. Considering a
extremely troublesome phenomenon of fine material - flooding, which occurs when
particles become veryfineor by the addition of very fine particles, say less than 40
\im (Lloyd and W e b b , 1987^ 653 ), the particles can suddenly discharge from a bin at
a very high flowrate in comparison with the normal steady state flowrate. The rapid
flowrate m a y approach that of an inviscid liquid (Rathbone, Nedderman and
Davidson, 1987 t 7 2 3 ) which is proportional to V 2 g ( h m a x - h 0 ) , namely, the
flowrate in this case varies with the surcharge level. It is believed that as particle
size varies from coarse towards fine, the effect of surcharge level cannot suddenly
jump from insignificant in the case of coarse material to significant in the flooding
case. There m a y be some transition phenomenon in-between. Willis (1978)' \ for
example, observed that the flowrate decreases as the surcharge level increases for
6
fine material ( d ^ = 1 4 0 ^im). It is necessary to find a appropriate description of the
effect of surcharge level on flowrate.
The work described in this thesis, aiming to predicting the flowrate of bulk solids
from conical mass flow bins, focuses on the following aspects:
i) modelling adequately such flow properties of bulk solids as bulk density
and permeability under condition of consolidation (Chapter 3);
ii) establishing a theoretical model to predict the interstitial air pressure
distribution (gradient) generated by particles during discharge from mass
flow bins together with some experimental verification (Chapter 4);
iii) establishing a theoretical model to predict the flowrate of bulk solids from
mass flow bins, which is based on continuum mechanics theory and
appliabletoboth coarse and fine materials (Chapter 5);
iv) exarnining the effects of permeability and surcharge level on the flowrate
both theoretically and experimentally (Chapter 5,7 and 8).
This thesis also includes two extension aspects:
v) simplifying the calculation of the dynamic deaeration coefficient by the
M o n t e Carlo simulation and optimization technique and providing a
simplified flowrate model to predict the particle flowrate from mass flow
bins (Chapter 9);
vi) examining experimentally and theoretically the strategy for increasing the
limiting flowrate offinepowders by using standpipes (Chapter 10).
7
Chapter 2
Literature Survey
Much work has been done on the gravity flowrates of bulk solids from hoppers and
bins since early this century. Initial work has been carried out empirically. With
more understanding of the discharge of bulk materials from these early works, the
attempts have been m a d e to predict discharge rates using theoretical models.
However, m u c h of the work has neglected the effect of air pressure gradients for
the convenience of modelling. These models, therefore, are only used for coarse
material. The effect of air drag has been taken into account in theoretical modelling
only after the work of Miles et al. (1968)^
3
and has become more widely
recognized. In addition, the Finite Element Method is also of interest for use in this
field for some particular purposes, such as calculating the stress field in the bin or
wall loads. The research reported in the literature on the flowrates of bulk solids
from hoppers and bins is summarized in two different categories: one concerns
flowrate without air retardation; the other includes that with air retardation.
2.1 Research on Flowrate without Air Retardation
The work in this category includes three groups: empirical work, theoretical
analyses and Finite Element Analyses.
a) Empirical Correlations
The first work on conical hoppers was studied by Deming and Mehring in 1929™3.
They assumed that the flowrate varied with such parameters as hopper angle,
diameter of orifice, the repose angle of bulk solids, particle size and bulk density of
8
the material. Their empirical correlation was based on dimensional analysis and
their experimental results. They claimed their correlation would predict the flowrate
of free flowing material above particle sizes of 200 mesh (74 jxm) with any density
and particle shape. However, it is noted that the smallest particle size found in their
experiments was 125 mesh (120 \s.m) for A m m o n i u m Phosphate crystals. The
flowrates of this material were measured only from a very small outlet (0.001 m ) ,
in which the effect of air pressure gradient on the flowrate of such fine particles
was not observed.
Newton, Dunham and Simpson in 1945^ 3 studying the flow of catalyst pellets
(2.54 - 5.08 m m in diameter), proposed that the flowrate varied with the orifice
diameter and the height of material level. The authors found orifice blocking when
the orifice diameter was less than six times the mean particle diameter. This limit
corresponds with that suggested by Langmaid and Rose in 1957' \ as an
insurance to prevent mechanical arching at the oudet
Franklin and Johanson in 1955^13 studied the flow of such granules as glass
beads, lead shot and puffed rice with particle diameter of 0.787~5.207 m m
discharging from a cylindrical bin with an outlet varying in size from 6.6cL to
34dp. They correlated discharge rate with orifice diameter, particle diameter,
panicle density and material friction. They reported no influence of material level on
the flowrates observed.
Fowler and Glastonbury in 1959[123 scrutinised the effects of changing orifice
shape for discharge from flat bottomed bins. Materials of particle size ranging from
270 \Lm to 3300 ^tm were tested. They found that flowrate was related to hydraulic
diameter of the orifice, m e a n particle size and shape factor for the material; the
9
effect of material head was found to be negligible.
Rose and Tanaka in 1959^ 3 investigated the effects of the variables of hopper
angle, material head, cohesive force, particle density, particle size, orifice diameter,
diameter of the cylindrical section, shape factor and the repose angle of bulk solids.
The variables were arranged into dimensionless groups and the effects of each
group on the discharge rate were found by experiment. They concluded that the
effect of material head on flowrate was insignificant based on the experiments for
steel balls of 1.32 m m diameter. Experimental discharge results for particle sizes
ranging from 112 \un to 910 |im were conducted to examine the effect of particle
size. Linear correlations between log (flowrate) and log (D/d - 3) were observed for
the particle sizes above 200 n m . For particles having diameters less than about 200
Jim, the flowrate decreased rapidly as particle size decreased. This was the first
reported observation of the retarding effects of air drag on flowrate. Unfortunately,
this retardation effect was explained as the effect of cohesive forces and functions
of cohesion were empirically determined; no adjustment for air drag was modelled.
Beverloo, Leniger and Van de Velde in 1961[143 investigated the flow of granular
materials through m a n y differently shaped orifices in flat bottomed bins. Based on
dimensional analysis, they proposed that discharge rate was proportional to
p V g D 0 2 - 5 . A correction factor Z = kcL for D 0 must be applied to ensure their
empirical correlation fit their experimental data. They presented the following
equation for flowrate
Qp = BfpVg(D0-kdp)2'5 (2.1)
where Bf and k are constants. Beverloo et al. found Bf=0.58 and suggested k=1.4.
This correlation has been found
successful by
many
subsequent
workers 150,51,56 ' 573 . Nedderman et al. noted in their review in 1982 143 that, though
normally attributed to Beverloo, similar ideas had been in use prior to Beverloo's
paper and indeed a correlation of this type had been proposed by Hagen as early as
1852[4]; Weighardt in 1952 [ 1 2 9 ] also gave the flowrate proportional to ( D Q - Z ) 2 5 ;
and B r o w n and Richards in 1960^43 presented the concept of the 'empty annulus',
where flowrate was also proportional to (Dft - Z ) 2 5 . The constant B f is dependent
on hopper geometry; Nedderman et al.'-43' recommended the range 0.55<Bf<0.65;
the constant k is related to particle shape and is differently valued for various
materials within the range l<k<3 t513 .
Bosley, Schofield and Shook in 1969^ 3 examined, in a photographic study, the
effects of hopper shape, particle size, particle density and hopper size on velocity
profiles in hopper discharge. In their investigations, only coarse particles (1-2.5
m m in particle size) were used to eliminate the effect of air pressure gradients. The
velocity profiles were found to depend primarily on hopper shape. They claimed
that m a x i m u m velocities agreed reasonably well with Brown's theoretical values
but a significant effect due to wall friction was observed.
Van Zuilichem and Van Egmond in 1974^163 studied the density behaviour of
flowing granular material using G a m m a - R a y Absorption techniques. They
observed that flowrate was proportional to the term: {a constant * (hydraulic
diameter of orifice)
}.
b) Theoretical Analyses
The initial significant theoretical analysis of the flowrate of free flowing bulk s
11
was probably the work of B r o w n and Richards; the papers in 1959^173 and 1961^183
and a book in 1970^ 193 reported their work. They used a 'minimum energy theory',
established initially by Brown, to describe the flowrate of coarse incompressible
bulk solids from bins and hoppers. T o obtain better correlations with experimental
results, they accounted for the effect of the 'statistically empty annulus' adjacent to
the aperture (a concept proposed by Wieghardt in 1952^ 1 2 9 3 ), to reduce the
predicted values of flowrate. In their work, the independence of material head was
also observed. T h e following equation was presented by B r o w n (1961 )^ 1 8 3 for
conical hoppers:
Qp =
P 7c(D 0 -Z)
2 5
- ^i
f
5
1 - cos a ^°'
V 2 sin a
It was found by m a n y researchers^ ^ ' '
(2.2)
)
3
that the flowrates computed by the
Brown's theory showed a good agreement with the experimental results for coarse
particles.
Meanwhile, a quite different theoretical approach was taken by Jenike and his coworkers in the late 1950's to early 1960's[1,2'74"763. In their analytical work, the
principles of soil mechanics and plasticity were applied to study the steady flow of
bulk solids. The bulk solid was treated as arigid-plasticmaterial obeying Coulomb
type yield criteria. B y neglecting the convective terms [ ( V V ) V ], the stress and the
velocity fields under steady state condition were uncoupled allowing the stress
distribution to be obtained first and then the boundary problem for velocities
solved.
Johanson in 1965' 3 developed a model for predicting flowrate which included the
effects of inertia in the equilibrium of a cohesive arch of uniform thickness. This
model is one of the most successful for predicting the flowrate of coarse cohesive
bulk solids. Johanson derived expressions for the steady flowrate for both
axisymmetric and plane flow hoppers. T h e following equation is the form for
conical hoppers:
-Mi--1
tan a ^
(2.3)
ffa J
Johanson's method involves the determination of the critical flow factor for arching
ff and the actual flow factor for the material under dynamic conditions ffa, which
are related to the flow properties of bulk solids and hopper geometry. E q n (2.3)
gives a good prediction of the flowrate for coarse cohesive material. Experiments
carried out by Johanson using several different bulk solids in both laboratory and
fieldtests,supported his theory. However, he found the experimental discrepancies
to be larger for finer materials, where the effect of the negative air pressure
gradients becomes more significant. Therefore, eqn (2.3) is not able to predict the
flowrate for fine material.
Jones and Davidson in 1965^ and McDougall and Evans in 1965^ 3 described
the flow of particles from an orifice by using the fluid analogy. They assumed that
Bernoulli's fluid energy equation could be employed as the particles being
discharged were analogous to a discharging fluid. In particular, Jones et al. studied
the flow of particles through orifice plates and shaped nozzles in the side of an air
fluidised bed. In their experiments the particle sizes of the bulk solids used ranged
from 85 to 350 urn. The nozzles were shaped so that there were no interparticle
pressure at all points in the nozzles, that is, the radius of cross-section r along the
axis of the nozzles x was of the form of r = CjA/
^
- where Cj and c 2 are
constants. Their results indicated that the solids flow under the incipient fluidisation
condition can be predicted by treating the fluidised solids as an inviscid liquid, in
which the flowrate w a s proportional to the square root of the material head.
McDougall et al. presented a theoretical equation for the flow of bulk solids from a
flat bottomed bin. They retained the air pressure difference across the outlet when
applying Bernoulli's theorem in their model and predicted the flowrate and the air
pressure as functions of such parameters as the material level in the bin and bulk
density. D u e to the difficulty of evaluating the air pressure difference, they
presented a simplified model which involved the parameters of the bulk solids
(particle internal friction angle and bulk density) and outlet size. However, the
simplified model did not indicate well the effect of the air pressure difference for
practical applications since the particle parameters internal friction angle and bulk
density can describe only the interaction of particles not the interaction between
particles and air. For example, the effect of the air pressure difference across the
outlet is significant for fine material and insignificant for coarse material; the
internal friction angle and bulk density can be applied for any material.
Savage investigated the flow of bulk solids from a hopper in 1965^ 3, based on
continuum mechanics theory and the Mohr-Coulomb yield criterion, in which the
velocity and the stress fields are coupled by considering the convective terms in the
theoretical m o d e l Compared with the B r o w n equation and the experimental results
of D e m i n g and Mehring^ 93 , Savage's model presented an overestimation of the
flowrate. H e extended the work in 1967 t243 and the modified model was solved by
using a perturbation analysis. H e claimed that the theory agreed reasonably well
with the D e m i n g and Mehring's experiments except for small values of a, where
Savage suggested that the theory be modified by incorporating a correction for wall
friction. This theory showed that the wall friction is a important parameter in
reducing the flowrate for narrow angled hoppers. Savage showed that the change in
flowrate due to wall friction could be more than 100%, whereas Nedderman et
alJ 3 concluded in the review paper that the change in flowrate was rarely more
than 1 0 % .
Mullins in 1972 - i974i25.26,H3] pr0p0Se(j an alternative approach, named
stochastic theory, to predict the velocity distribution in a hopper by modelling the
flow as the upward diffusion of voids by random processes. The basic concept was
the representation of convergent particle flow toward an open orifice, under
gravity, as equivalent to a counterflow of voids entering the orifice and migrating
upwards through the bed by a biased random flight as they are repeatedlyfilledby
particles moving d o w n from above. The appropriateness of the model was not
obvious. However, a continuous velocity distribution was predicted without
considering the stress field which indicated that it is not necessary for the particle
flow from hoppers to be considered as being driven by the stress field. This idea
was sustained by the kinematic model of Nedderman and Tiizun (1979)^ I
Davidson and Nedderman in 1973^273 derived an expression, known as the 'Hour
Glass Theory' based on continuum mechanics, to model the flowrate of
cohesionless material from a smooth walled hopper. They found the predicted
flowrate was about double the observed result and concluded that this difference
was caused by the effect of wall friction. They observed that the flowrate and stress
distribution were very insensitive to surcharge at the top of the hopper.
Williams in 1977[28] extended the theory of Davidson and Nedderman1273 to
include a solution for wall friction. The author produced two solutions to the
equations of m o m e n t u m and continuity to provide upper and lower limits. Since
these limits differed by only about 2 0 % , Williams claimed that it was not
necessary, for the prediction of discharge rates for design purposes, to obtain
complete solutions to the equations. Comparison of the theory with experiments
showed good agreement for coarse materials, (within 8 % ) , but over-predictions for
fine particles less than about 500 |im (for sand with mean particle size 250 Jim, the
mean of the upper and lower limits of predicted flowrate was about 3 0 % higher
than the measured flowrate). This discrepancy was attributed to the fact that
appreciable air pressure differences which developed in the airfillingthe void space
were experienced by these particles. These air pressure differences were not
considered in his model.
Nedderman and Tuziin in 1979™3 developed a kinematic model for the flow of
granular materials which, similar to Mullins' stochastic model' '
, 3
, m a d e no
reference to the stress distribution within the material. T h e experimental
measurements were m a d e on velocity distributions in a two-dimensional hopper
with Nedderman et al. claiming excellent agreement with the kinematic model. They
concluded that free-flowing granular materials discharging from hoppers are not
driven by the stress field, as has usually been assumed, but m o v e simply by the
particles in one layer slipping into the spaces vacated by the layer beneath. They
suggested this model solely for steady flow of a dilated material.
Drescher, Cousens and Bransby in 1978[3°3 presented a kinematically admissible
velocity field for the mass flow of granular material in a plane flow hopper. The
proposed solution was based on assumptions suggested by the flow pattern
observed during experimental work. In particular, they incorporated rupture
surfaces observed during flow into the solution. The material filling the hopper was
treated as plastic, though different flow rules were specified for different areas of
the material. Stresses were not considered because the static counterpart of the
kinematic solution did not contribute to the approach adopted. The main thrust of
the work was to provide a realistic mathematical model for understanding the
velocity fields observed in experiments, rather than a complete solution based on
simplifying assumptions to allow flowrates to be predicted. The predicted velocity
field for the mass flow of granular media through a plane converging hopper,
obtained from rather simple assumptions, agreed reasonably well with that
observed experimentally. A criterion was suggested to indicate the transition from
mass to funnel flow as the head of material in the hopper is reduced. However, this
criterion cannot be predicted theoretically and has no practical implication.
Nguyen, Brennen and Sabersky in 1979^ 3 presented an approximate solution to
the flow of a cohesionless granular material in a conical hopper. The bulk solid was
treated as a perfectly plastic continuum which satisfied the Mohr-Coulomb type
yield condition. They concluded that the method presented was appropriate for
small hopper angles with discrepancies from the experiments occurring at larger
hopper angles. They pointed out that the continuum model described the behaviour
of granular materials fairly well.
Kaza and Jackson in 1982t333 provided a power series solution for solving their
mathematical model instead of using a perturbation technique or introducing extra
assumptions to simplify the model. Their theoretical model was based on the
equations of continuity and m o m e n t u m balance for a cohesionless Coulomb
powder discharging from a wedge-shaped hopper. Compared to the perturbation
analysis of Brennen and Pearce in 1978 [ 3 1 3 , the power series solution was
expected, as claimed by Kaza et al., to solve the hopper discharge problem more
accurately over the range of values of hopper angles since it provided smaller
computed residues. Unfortunately, the flowrates calculated from this solution lay
further from the experimental data than those obtained from the Brennen-Pearce
model. The authors suggested some possibilities for the gap between theory and
experiment, such as the form of the boundary condition at the wall used m a y have
been incorrect; the experimental characterization of the materialtodetermine internal
friction angle and wall friction angle m a y have been in error or there was no
continuous solution which remained bounded at the lower traction-free curve and
approached the radial stress and velocity field high in the hopper.
Michalowski in 1984^343 described the flow of granular material in a plane flow
hopper in two stages, initial stage and advanced stage. H e concluded that a
theoretical description of the kinematics of advanced flow, based on the plane
plastic flow theory of incompressible material and coupled with the radial stress
field, described the real velocity field accurately, particularly in hoppers with
smooth walls. The description in his work gave results similar to those obtained by
Drescher, Cousens and Bransby in 1978 [303 . For the initial stage of flow, only part
of the material in the hopper experienced plastic deformation, which indicated that
the traditional plastic flow theory did not seem to provide a promising framework
for analysis in this stage of flow.
Pitman in 1986 and 1988[36,373 presented a mathematical analysis based on
continuum mechanics for incompressible granular material flowing from two- and
three-dimensional hoppers. A numerical analysis was presented to solve the
equations governing the flow of the particles. It was found that there was usually
some oscillation during the material flow. For mass flow hoppers, the oscillations
were slight. Moreover, the oscillations were more apparent in two dimensions
while stable three-dimensional flows showed little oscillation. However, for both
cases the computed stress field converged to the radial stress field, towards the
outlet of the hopper.
Jenike in 1987* 3 presented a theory which extended his previous theory
developed nearly thirty years ago. H e pointed out that the design method based on
his original theory is adequate for the design of mass-flow hoppers, but it predicts
incorrect channel flow angles in funnel flow. T h e correction presented was
obtained by replacing the modified Tresca yield pyramids of the original theory
with conical yield surfaces and relating the strain rates to the stresses by the Levi
flow rule which is that the stress deviatorics are proportional to the strain rate
deviatorics. The corrected theory provided a significantly larger channel flow angle
in funnel flow than the initial theory. The bulk solids considered were elasticplastic, frictional, cohesive and compressible.
Poldeman, Boom, de Hilster and Scott in 1987' 3 presented a mathematical model
for the solids velocity distribution in mass flow hoppers. T h e residence time
distribution measurement and flow visualisation studies were carried out in
experiments. B y comparing with the experimental results, the theoretical model
which w a s based on classical plasticity theory, was claimed by the authors to
provide a good prediction for the velocity distribution of free-flowing material in
mass flow hoppers. However, the model of Poldeman et al. cannot predict the
velocity distribution in a mass flow hopper completely as a constant in the model
must be determined by individual experiment
Ravi Prakash and Kesava Rao in 1988t393 studied the steady compressible flow of
cohesionless granular materials from a wedge-shaped hopper. Their continuum
model used the critical state theory of soil mechanics to generate the stress and
density profiles and allowed the discharge velocities to be obtained. A n
approximate expression for the discharge velocity was also presented, which
predicted discharge velocities to within 1 3 % of the numerical values from their
original model. The difference between the results from incompressible flow and
compressible flow was examined for two materials, glass beads and Sacremento
River sand. They observed that the effects of compressibility m a y safely be ignored
for glass beads, but not for Sacremento River sand. Incompressible models
predicted in some cases significandy higher discharge velocities than those obtained
by the incorporation of density variation. For instance, the incompressible model
overestimated discharge velocities for the above two materials by 2 % and 4 5 % ,
respectively. They pointed out that the discharge velocity was more sensitive to
density variations than to stress profiles.
c) Finite Element Analyses
With its development for modelling complex flow behaviour, the Finite Element
Method ( F E M ) seems to be an alternative method for use in the bulk solids gravity
flow area.
Probably, the first application of the FEM in this area was the work of Fried,
Carson and Park in 1976^783. This application focussed on simulating the nonlinear
flow problem of gas in a contact bed reactor. A s the solid particles m o v e d
downward by gravity, gases were forced to flow through the particles from several
particular points. Because of the complexity of the reactor geometry and the
nonlinearities involved, the F E M was used to predict the gas velocities and pressure
gradients near the gas inlets.
Recendy, several attempts to use F E M to compute the velocityfieldin flowing bulk
solids have been made, such as those by Haussler and Eibl (1984)^ 403 , Runesson
and Nilsson (1986) [41] , W u (1990) t79] . The F E M has several advantages including:
• modelling the flow of a bulk solid flowing from a bin with complex geometry;
• showing the stress and velocity profiles at any position in the bin;
• modelling the initial flow and steady flow for bulk solids.
O n c e the velocity profile is determined, the flowrate of bulk material from silos /
bins can be evaluated according to the velocity of particles at the outlet. However,
using F E M has some limitations such as:
• time consuming calculations;
• indicating only indirectly the relationship between particle velocity and particle
flow properties.
D u e to these limitations, most F E M applications have attempted only to predict the
stressfieldsin the bins or the wall load distributions; the prediction of the flowrate
of bulk solids from a regular mass flow bin does not warrant calculation by F E M .
2.2 Research on Flowrate with Air Retardation
As mentioned in Section 2.1 a), it is believed that the first appearance of the effe
air retardation on flowrate of fine material was in the work of Rose and Tanaka
(1959)^ 133 , although it did not c o m m e n d their attention at thattime.Bulsara, Zenz
and Eckert in 1964 [ 4 3 3 conducted some experimental work on improving the flow
of fine cohesive particles. They particularly aimed at rendering 'sticky* powders
free-flowing. They found that the flow of cohesive particles can be improved by
adding sufficient coarse particle into thefinepowders (e.g., adding sand with mean
particle size of 715 ^im into fly ash with size of 20 Jim or into Micro-Cel of 3 |Am)
or applying air pressure differentials to suck the solids out of the bin. This result
indicated that the effect of air pressure on the flow of fine cohesive particles was
considerable. Johanson in 19651-223 stated that drag effects could be very important
to the flow offinematerial but his model avoided including this effect by assuming
the material to be coarse. McDougall and Evans (1965) 1213 attempted to describe the
particle flow affected by air pressure at the outlet, however, their equation did not
take account of the interaction between particles and air.
The first major experimental work involved with air drag on flowrate was
conducted by Miles, Schofield and Valentin in 19681-443. They pointed out that
w h e n air drag forces cannot be neglected, the B r o w n and Johanson equations
overpredict the flowrates, with the over-estimation becoming more significant as
the particle size decreased. They presented experimental evidence to clarify some of
the processes which take place w h e n discharging fine powders. Flowrates of
coarse gravel, fine sand and calcite, from a hopperfittedwith replaceable cones
with different outlet size and wall angle, were recorded. Negative air pressures
were observed for the fine sand discharge but no noticeable change in pressure was
observed for gravel. For 55 fim calcite, the cycling between fast and slow flow was
observed (as shown in Figure 8.8). Their experimental observations also indicated
that both air injection into the flowing mass and the extension to the hopper oudet
by means of a standpipe had a significant effect on increasing the flowrate of fine
sand but no significant effect for the coarse gravel. However, the increase in
flowrate of particles by air injection was not proportional to the flowrate of air.
S o m e optimal case exists for particular material and bin geometry. The significance
of using a standpipe below the hopper outlet lies in creating a suction effect at the
hopper outlet to increase the flowrate. In their work the standpipe was found only
effective w h e n it was full of material. This phenomenon was also observed by later
workers 1 * 4 - 863 .
Holland, Miles, Schofield and Shook in 19691-463 attempted to explain theoretically
the phenomena observed in the above paper. They stated that the flowrates
computed by Brown's equation are remarkably accurate if the particles are coarse.
A s the particle size diminishes, however, the flowrates obtained by Brown's theory
invariably exceed the experimental values; the effects neglected m a y be described as
air drag forces and interparticle forces. The authors discussed the forces involved in
the flow of particles, inertia forces, gravitational forces, fluid drag forces,
interparticle pressure forces and interparticle shear forces. B y simplifying these
forces (neglecting the fluid drag force and the interactions between particles), the
authors derived a velocity equation which agrees exactly with the B r o w n equation
for coarse material. B y considering the air drag effects, Holland et al. presented a
relation between the air pressure gradient at the hopper outlet and the flowrate. The
agreement of the presented air pressure gradient at the hopper outlet with the
experimental data was claimed to be reasonably good. However, due to the
complexity of the solution and the lack of boundary conditions such as voidage
distribution in the hopper, no equation was presented to predict the flowrate of bulk
solids.
Shook, Carleton and Flain in 1970^473 proposed two correlation equations for
calculating the flowrate under the influence of air drag forces: one was based on the
Richardson-Zaki expression to estimate the drag force for hindered settling of
particles; an alternative approach was based on the Carman-Kozeny equation for
fluid flow to model fluid drag forces. Both equations required the determination of
porosity in the hopper and at the hopper oudet but no procedures were proposed for
these determinations.
McDougall and Knowles in 1969' 83 presented a systematic investigation of the
effect of pressure difference across an orifice on the flow from flat bottomed
bunkers. T h e experimental work w a s conducted with Mustard Seed, average
particle size of 2.2 m m , flowing from orifices ranging from 12.7 to 76.2 m m in
diameter. The experiments were carried out on four types of flow: gravity flow,
vacuum flow, pressure flow and counter-current flow. A m o n g these four types of
experiments, the gravity flow can be considered as no air pressure difference
existing across the orifice since the particle size was coarse. The three others
created air pressure differences: for the vacuum flow test a suction effect was built
up at the outlet to increase the flowrate; for the pressure flow test, a positive air
pressure gradient was created to force the particles to flow quickly. Both these flow
tests had a similar effect on the flowrate and, hence, the distinction between
'vacuum flow' and 'pressure flow' w a s not maintained throughout the
experiments. For the countercurrent flow test, an air current w a s forced
countercurrent to the solids flow. The results showed that the effect of the air
pressure difference was quite dramatic, even for the coarse material used in the
experiments. W h e n the adverse air pressure gradient was sufficiendy large, no flow
of bulk solids at all was observed. The work led to a general conclusion that air
pressure differences can be an important factor in the flow of solids from orifices.
With free flowing materials, it was apparent that useful increases of flowrate
resulted from pressurizing the hopper.
Carleton in 1972^493 carried out a force balance on the fluid solid system and used
the fluid drag coefficient on a single particle as the air drag model to predict the
flowrate of fine material (particle sizes below about 200 p m ) . H e developed the
following nonlinear equation:
Ill
4Vosina 15pJ\iJvJ
+
D0
T
Psdp3
=g
(2-4)
A nomograph was also presented to simplify the solution procedure of eqn (2.4).
The equation proposed for the velocity profile was claimedtogive good agreement
with experiment for free flowing materials except for hoppers with small orifices
(below about 2 c m in orifice size) where wall effects become predominant.
Crewdson, Ormond and Nedderman in 1977'503 investigated air retarded discharge
theoretically and experimentally for conical mass flow hoppers. The authors, in
their theoretical derivation, considered the effects of interstitial air pressure
gradients and compressibility on the flowrate offinematerials (particle size less
than 500 (Jm). A n adjustment was made to the Beverloo flowrate correlation to
account for interstitial air pressure gradients. However, due to the lack of a reliable
value of the compressibility index at low stress levels, only a qualitative
comparison of the theoretical analysis with the experimental observations was
reported.
Spink and Nedderman in 1978^ 3 reported a theoretical and experimental study,
based on the work of Spink 1976^803, observing the flowrate offineparticles (50
p m < d p < 500 p.m) from plane flow hoppers. The theoretical model yielded interrelated equations for particle flowrate and for the distributions of particle stress, bed
voidage and interstitial air pressure in the hopper. G o o d agreement between
experimental and theoretical voidage and air pressure distributions was found. The
experimental study on the air pressure distribution showed that a pressure minima
(negative) occurred near the hopper oudet and a visible pressure maxima (positive)
near the top surface of material was observed for relatively large outlet cases.
Furthermore, it was observed that the negative air pressure gradient increased with
increase in the hopper outlet and with decrease in particle size. For very fine
particles such as fine sand (with a particle size range from 63 to 90 p.m), unsteady
flow w a s observed. The theoretical and experimental studies on the flowrate
showed that the flowrate offinematerial (particle size range from about 100 to 500
|im) increased as its particle size increased. However, although they showed the
correct dependency on the particle size, the theoretical discharge rates were about
twice the measured values. For coarse material, from the concept of the 'empty
annulus*' 3, Spink et al. (1978)^
3
believed that the particle flowrate decreases as
particle size increases. However, this variation, as pointed out by Head (1979)^ 3,
is caused by orifice blocking and, therefore, is valid only for coarse material
flowing from a small outlet.
Smith in 1978^533 investigated the interactions of feeders with mass flow hoppers.
O n e interesting finding was the variation of the free fall flowrate at various H / D
ratios of the bulk solid (Shirley Phosphate with median particle size 140 p.m), as
shown in Figure 6.1. In particular, he found that for a hopper with surcharge
(H/D>0) the flowrate decreased as the surcharge level increased; for a hopper
without surcharge (H/D<0) the flowrate increased with an increase in surcharge
level. T h e decrease in flowrate for H / D > 0 was explained as the effect of the
counterflow of air through the hopper oudet For the cases H/D<0, there was no
adequate reason to explain the increase in flowrate. Nevertheless, this is the only
work found which reported the variation of the flowrate with the surcharge level.
Since Smith's results were observed under a variable material level during
discharging (i.e., the readings of the material height reported were taken after filling
the material in the test bin and before material was discharged), it seems that more
work needs to be done to examine the effect of a constant material surcharge level
on the flowrate.
Willis in 1978^ 5
3
conducted a series of measurements on the interstitial gas
pressures generated by flowing bulk solids in a plane flow variable geometry bin. It
was found that varying thefillrate, hopper geometry and type of material had a
significant effect on the interstitial gas pressures. It w a s claimed by Willis that the
experimental values compared favourably with the values predicted by McLean's
theory^583, which focused on the hopper section. However, Willis' work identified
a major deficiency in the theoretical treatment. In particular, the experimental
distributions indicated that the gas pressure at the transition is negative, suggesting
the bin be analysed as a whole, since the air pressure distribution boundary
conditions applying act at the oudet and top of the bin.
McLean's analysis in 1979^5 3 considered the continuity of the bulk solid and the
interstitial fluid, as well as the equation of motion including the effects of the
interstitial gas pressure gradients (for fine material cases). In the derivation of the
theory, consolidation-related bulk density and permeability equations were
introduced for modelling the flow of compressible simple bulk solids. T h e
predicted flowrates compared favourably with observed flowrates from an
experimental plane flow variable geometry bin. Additionally, he concluded that this
investigation provided initial estimates of the parameters for air injection techniques
which require a suitable pressure, quantity and location to improve the flow.
Head in 1979^553 investigated, theoretically and experimentally, the effects of the
particle size of granular materials and the effect of interstitial air pressure on the
flowrate as well as the effects of air injection technique on improving the flowrate.
It w a s found that the flowrate was affected by its particle size in following way: for
large sand particles (> 500 p m ) the flowrate decreased with increase in particle size
due to particle mechanical interlocking at the orifice; for fine sand particles (< 500
p m ) the flowrate decreased with decrease in particle size due to particle air drag and
compressibility. The study confirmed that for the sand less than 500 p m the
flowrates are affected by negative air pressure gradients at the oudet; the flowrate
can be varied by injecting air into the system and sensitive control of flowrate can
be achieved. The experimental observation on the effect of air injection indicated
that it should be located as near to the steady state negative pressure m i n i m u m
position as possible to obtain best control of air pressure drop across the orifice.
G o o d correlation for flowrate calculation, predicted using the models developed,
was claimed.
Nedderman, Tuziin and Thorpe in 1983' 3 presented theoretical and experimental
work on the effect of interstitial air pressure on the flow of granular materials from
hoppers. Significant effects were observed in air augmented discharge due to the
compressibility of the air. Based on the work of Harrison and Mushin (1979)' 3,
they proposed a theory for the case of low Reynold's number flow in a conical
hopper. The following equations resulted
1
~
^
(* 2K-3 AP \2
Qp = Qpmax[l+2K-rTp!^J
^
where
v5/2
_
rcpPp
vpmax4
(1+K)g
/
yj 2(2K-3)sina
The theory w a s also extended to higher Reynold's numbers and to cases where the
compressibility of the air was important. However, their theoretical model
overpredicted the measured flowrates. Some deficiencies in their theoretical model
which may cause this discrepancy were suggested, e.g., the assumption of smooth
hopper wall in theoretical analysis did not adequately represent the tested hopper
wall; the assumption that the stresses drop to zero on the spherical surface r=r0 was
indeed debatable. It was claimed that the agreement between the theory and the
experiments was better if the discharge rate was correlated using the orifice
pressure gradient as measured from the pressure profiles. To overcome the
overprediction of the theoretical model for higher Reynold's numbers, a 'discharge
coefficient' was introduced to their semi-empirical equation. According to their
experiments this coefficient was valued as 0.92. They suggested further work be
carried out to confirm this approach.
McLean in 1984^603 modified the Johanson equation'23 by introducing an effective
1 dP
gravitational acceleration g* = g +
(-3-)
instead of g. It can be seen that a
Pout * r = r °
negative air pressure gradient at the hopper outlet retards the particle flow and a
positive air pressure gradient assists the particle flow. His work enabled the
Johanson's equation (2.3) to be used in both coarse and fine material cases, i.e.,
Qp = pDo j tan a ^
f fa j
where
for coarse material
f g
g*=i
S
+
1 ,dP,
— <dF>r=ro
Pout
for fine material
(2.6)
In this modification, McLean provided the following equation for the prediction of
the air pressure gradient at the hopper outlet.
1
(*£) =
QP
(JJLY ^ (( Wo) YRb" _ (1
^dr^Urn AoCoutPeUpeJ
l\DbffJ
Uf
b
hJ ,
(2.7)
provided the bulk density is described by
P = Pe( —)b
°e
Recently, McLean[1153 developed an alternative approach to eqn (2.7) by
introducing the voidage at one-third of the way along the hopper wall from the
hopper oudet e ^ ; the following equation resulted
fdPA Qp (eo-eifl)
I dr J ^ ~ A 0 C o u t p s
(l-eb)(l-cli3)
(2 8)
'
Ducker, Ducker and Nedderman in 1985t573 developed a theory to predict the
flowrate of bulk solid from an unventilated hopper. In this case, air must enter the
hopper through the orifice at a volumetric flowrate equaltothe volumetric outflow
of solid. This counter-current flow will set up an adverse pressure gradient which
will reduce the flow of solid even for such coarse material as Diakon, Kale Seed
and Mustard Seed with particle size of 0.662,1.63 and 2.10 m m respectively. For
sand with particle size of 285 p m , unsteady flow was observed from unventilated
hoppers and this was attributed to fluidisation caused by a high counter-current air
flow near the orifice. The theory was claimed to be in excellent agreement with
experimental results. Experiments on the same bulk materials were carried out to
obtain the flowrates from ventilated hoppers. The results showed a good agreement
with the Beverloo correlation, by setting the constant B f in eqn (2.1) as 0.58,0.57
and 0.66 for Kale Seed, Mustard Seed and Diakon respectively. For the 285 p m
Sand, flowrates less than the Beverloo prediction were observed due to the effect of
interstitial pressure gradients.
Arnold in 19861523 presented some attempts to predict the flowrate which were
based on the Carleton method eqn (2.4), using equivalent particle size d e instead of
particle size cL, i.e.,
\_ 2 4
4V?sina
D0
15
+
Pf 3 ji/ V 0 3
~5
1
=S
(2-9)
.3
Psde
In particular, two types of modifications were proposed:
Drag force modification: The drag coefficient on a single sphere was
replaced by that on a mass of particles by introducing the well known drag
modification expression of Richardson and Zaki 1/e
. Using the results
obtained by W e n and Yu^ 1 3, n-1=4.7, the flowrate was calculated by eqn (2.9)
and eqn (2.10).
de - e 2 ' 8 2 ^
(2.10)
Permeability modification: The equivalent particle size was estimated from
permeability factor C in terms of the Carman-Kozeny equation. In this case, the
flowrate was predicted by eqn (2.9) and eqn (2.11).
2
de =
180p f (l-e) C
1
2
(2.11)
3
e
W h e n studying the flow with air pressure retardation, a number of researchers have
examined two main techniques for increasing the limiting flowrate.
Air Injection - this technique has been reported in many publications^44,45'81"
83,116-121] ^
particular>
Papazoglou and Pyle in 1970 [81] , Altiner in 1983 [82] ,
de Jong and Hoelen in 1975™
3
investigated, both theoretically and
experimentally, the flow of fine particles from aerated conical hoppers, plane
flow hoppers and flat-bottomed bins. It was found that for low air flowrates the
effect of air injection on improving the particle flow increased with increase in
outlet size and with decrease in particle size. They also found that the effect of
air injection w a s different for different aeration methods. A m o n g these
researchers, Papazoglou et al. and Altiner studied the effect over relatively wider
range of air flowrates in their aerating system. They discovered that there was an
upper limittothe quantity of air that could usefully be injected. Beyond this, the
excess air fluidised the bed above the injection point and simply escaped
upwards without aiding the flow'43. Altiner reported that the limiting aeration
rates were different for different size particles, being less forfinerparticle. H e
also observed that thefinalm a x i m u m particle flowrate was the same for all sizes
with the same aerating system. Furthermore, he found that different m a x i m u m
particle flowrates were achieved with different aeration methods, air flow
distributed along the hopper wall near the outlet giving the higher m a x i m u m
particle flowrate than point air injection. Papazoglou et al.'s studies also
involved the flow behaviour of particles at high air flowrates. It was noted that
after the particle flowrate reached a m a x i m a the particle flowrate w a s
proportionaltominus the air flowrate. Papazoglou et al. explained this decrease
in particle flowrate with increasing air flow m a y be contributed to by significant
m o m e n t u m flux in the air phase leaving the injector pipe and to compressibility
effects. The experiments showed that the constant of proportionality was
dependent on both the system geometry and the particle diameter.
Standpipe - the use of a standpipe connected below the orifice of a container,
as an alternate technique to improve particle flow, has been reported by m a n y
researchers such as Miles, Schofield and Valentin (1968)^443; Yuasa and K u n o
(1971)[122]; M c D o u g a i i a n d puiien (1973) [123] ; de Jong (1975) [124] ; Leung and
his co-workers (1973 [125] , 1978 [ 1 2 6 ' 1 2 7 ] ); Ginestra, Rangachari and Jackson
(1980) 1843 ; Chen, Rangachari and Jackson (1984) [85] ; Knowlton, Mountziaris
and Jackson (1986)^ 3. O n e generalized result showed that the standpipe can
increase the flowrate but it is only effective when full of the particles. A m o n g
these researchers, Ginestra et al., Chen et al. and Knowlton et al. investigated
theoretically the gravity flow of bulk solids through a configuration which
included a hopper, a standpipe attached below the hopper outlet and a flow
control device at standpipe outlet
Ginestra et al. presented a one-dimensional theory which was based on a direct
integration of the equations of continuity and m o m e n t u m balance for the two
phases. The theory linked together equations describing the hopper, the
standpipe and the flow control orifice at the lower end of the standpipe. The
prediction indicated that the negative air pressure at the hopper outlet was
considerably greater than the pressures above and below the system, inducing a
vigorous flow of air downwards through the hopper thereby increasing the
particle flowrate due to the concurrent drag force on the particles. However, it
was claimed that the predictions of this theory did not agree with the
observations. Ginestra et al. believed that this discrepancy was caused by the
unrealistic simplifying assumptions such as the frictional forces between
particles and standpipe walls being assumed to be negligible.
Chen et al. extended the work of Ginestra et al. and presented a model by
relaxing some of the simplifying assumptions in Ginestra et al.'s theory, such as
considering frictional forces between particles and standpipe walls and treating
the gas as a compressible fluid. T h e predictions of the improved theory were
compared with measurements on a laboratory-scale standpipe and general
agreement of theoretical prediction with experimental results was obtained. The
research confirmed the fact that the use of a standpipe increases the particle
flowrate more significantiy for fine materials than for coarse materials, because
the suction effect built up at the hopper outlet is m u c h weaker in the case of the
coarse bulk solids. Furthermore, they also noted the existence of two different
particle flowrates under the same operating conditions in certain circumstances
(such as applying some amount of air pressure at standpipe outlet which was
greater than that at the top surface of material in the hopper). The upper value
showed an increased particle flowrate, while the lower value indicated that the
particle flowrate was reduced. In lower flowrate case, their predicted and
measured air pressure profiles indicated that a positive air pressure occurred at
the hopper oudet, so particle flow from the hopper was impeded rather than
aided by the air pressure gradient. Physically, it is believed that in this lower
flowrate case a relatively lean particle flow appeared in standpipe so that the
positive air pressure applied at the standpipe outlet could reach up to the hopper
outlet. This phenomenon supported the general result that the standpipe is only
effective when it is full of particles.
Based on the work of Chen et al, Knowlton et al. investigated the effect of
standpipe length on the flow of particles for a standpipe restricted by a
concentric orifice at its lower end. Comparing the theoretical results with the
experimental results, it was found that general agreement between theory and
observations was obtained. The research shown that the effect of standpipe
length on particle flowrate depended on the percentage of the opened crosssectional area of the standpipe oudet W h e n the open area was small (e.g., 5 0 %
or less), the particle flowrate was always smaller than that without the standpipe;
the standpipe length had almost no effect on the particleflowrate.Otherwise,
when the open area was large (e.g., 100%), the standpipe increased the flowrate
and the standpipe length did affect the particle flow. The longer the standpipe
length was, the more significant the effect became.
However, Ginestra et al., Chen et al. and some other researchers such as Yuasa
et al. studied a very long standpipe with the ratio between standpipe length and
its diameter more than 100. This kind of long standpipe cannot be widely
applied to the bulk solid handling system in industry. S o m e work needs to be
done to study the extend to which the flowrate is increased by a standpipe with
limited length.
Chapter 3
M e a s u r e m e n t s of Bulk Solids Flow Properties
3.1 Introduction
To model flow behaviour of bulk solids from mass flow bins, an essential step
involves the determination and measurement of the flow properties of the bulk
solids. The flow of the bulk solids from the bins depends on the geometry of the
bin (e.g., hopper angle, outlet size), particle characteristics (e.g., particle size,
particle size distribution, moisture content and particle shape) and the relationship
between particles and the bin (e.g., wall friction). Since it is difficult, at the present
time,toestablish a theoretical model which includes all these items, adequate flow
properties need to be determined.
Figure 3.1 summarised from studies in the literature how flowrate is related to
different flow properties. Generally speaking, these flow properties can be
collected into three groups of characteristics. The first group of characteristics is the
physical and geometric characteristics for both the bulk solid and the bin, the
second group of characteristics is the flow properties of the bulk solid which can be
determined from laboratory measurements, while the third group of characteristics
is usually predicted theoretically and/or measured in the laboratory from model
experiments.
Considerable research has been done to predict the flowrate of bulk materials from
mass-flow bins and hoppers using a range of bulk solid properties. S o m e
researchers have found that the effects of the angle of internal friction and the wall
friction angle seem to be small (Nedderman 1982)™3. Therefore, the flowrate was
calculated with knowledge of the particle size (normally median particle size), bulk
density and the bin geometry parameters of oudet size and half hopper angle.
First group
Particle Characteristics
Particle size
dp
Size distribution
Particle shape
Particle roughness
Moisture content
Particle density
o
r
s
Bin or Hopper
Geometry
Half hopper angle
Outlet size
Vertical section:
- diameter
- height
Ct
D0
D
H
- ^ ^ ^ ^
Bulk density
p
Second group Permeability
C
Internal friction angle
Flow function
Wall friction angle
1
Third group
i
( empirical research )
( theoretical research )
I
Flow factor
Stress / velocity field in bin or hopper
Interstitial air pressure gradient
Flow rate of particles from bin or hopper
Figure 3.1
Qp
Gravity Flow of Bulk Solids from a Mass-Flow Bin
A s the bulk solids become finer, the flow phenomenon becomes more complicated.
For example, the effects of particle size distribution, moisture content, cohesion of
the particles and the interstitial air pressure gradients become significant. In this
case, the median particle size is not sufficient to describe the size of solid mixtures;
the bulk density is not a constant, but varies with the compressibility of bulk
material; the permeability of bulk solids becomes an important parameter in
predicting the flowrate.
The solids used in this work are assumed to be fine and dry (or have a low and
constant moisture content). Nedderman (1985)^53 suggested that if flow does occur
the flowrate can be considered to be independent of the cohesion. Since this
independence has not been tested in systematic w a y in Nedderman's work, it is
necessary to assume which is that the effect of cohesion between particles
concerned in this work is negligible for mass flow conditions.
Hence, for this work the paramount flow properties of bulk solids are particle size,
particle size distribution, effective internal friction angle, particle density, bulk
density, permeability of particles and the friction angle between bulk material and
bin/hopper wall. In view of the importance of these parameters, the methods used
for their measurement are described in the following sections.
3.2 Particle Size and Particle Size Distribution
The size of a spherical particle is uniquely defined by its diameter. For a nonspherical particle the particle size is defined by an equivalent diameter which
depends on the method of measurement. There are several definitions of particle
size, such as volume diameter, surface diameter, surface volume diameter, drag
diameter, projected area diameter and sieve diameter
3
. Since the bulk solids
handled in practice are mixtures of a large number of particles which commonly are
not of uniform size but spread over a narrow or wide range of sizes, the properties
such as flowability, segregation, compressibility and permeability will be affected
by this particle size distribution, especially if there are fine particles present.
There are a number of methods used to determine the particle size and size
distribution of a particulate material. The methods can be briefly classified, mainly
based on Ramanujan et al/
3
and Jelinek™93, as mechanical methods (dry sieving,
airjet sieving), gravitation methods (gravity sedimentation, centrifugal
sedimentation), elutriation methods (gravity, centrifugal), optical methods (light
microscope, ultramicroscope, electron microscope), translational methods
(viscometry, permeametry), adsorption methods and laser methods.
The methods used in this work are dry sieving and laser particle sizing. The
Malvern Laser Diffraction Particle Sizer (2600 series) was used to measure the
particle size distribution for materials finer than 565 p m . The median particle size
used in this work was the volume percentile median diameter (volume diameter at
5 0 % point of the cumulative particle size distribution). The particle size in excess of
565 p m was measured by the dry sieving method.
Since only a small amount of sample is needed in the Laser Particle Sizer extreme
care must be taken to obtain a sample that represents the bulk material. Improper
sampling often introduces errors that are larger than the errors produced by the
instrument used for size analysis. In general, the Laser Particle Sizer can provide
more reliable and accurate particle analysis for fine materials with a narrow size
distribution.
In some situations where bulk solids with wide size distributions were analysed
and the effect of segregation could not be reduced satisfactorily, a combination of
sieving and laser particle sizing was used. In particular the sieving method
separated the whole particle mixture into several different size segments with
narrower size distribution, then the L P S was used to analyse each narrow size
segments. Finally the results were combined to obtain the results for the whole
mixture. In addition to using the standard sieve series some large area screens were
used for fine materials to achieve better results by increasing the sieving speed and
avoiding crowding due to the sieve loading.
It is necessary to mention that other particle characteristics such as particle shape
and particle roughness can also affect such flow properties of particles as internal
friction angle, bulk density and permeability. Since these flow properties will be
measured directly the results will implicidy include these effects.
3.3 Effective Angle of Internal Friction and Angle of Wall Friction
The state of stress at any point within the stress field can be described in the a, Tdiagram (o = compressive stress, x = shear stress) by a M o h r stress circle. The
yield locus is the yield criterion of the material at a corresponding bulk density
which depends cm the major consolidation stress aj.
Figure 3.2 shows a diagram of a typical yield loci of a bulk material. The Effective
Yield Locus, introduced by Jenike, is a straight line tangential to the stress circle at
steady state flow and passing through the origin of the normal stress, shear stress
axis. The angle between the E Y L and the a-axis is defined as the effective angle of
internal friction. For cohesive material, such interactions as friction and adhesion
between flowing particles have to be considered. Molerus™ ' 3 provided another
steady state flow yield locus which can be used for cohesive materials. The
intersection of the Molerus Yield Locus and the x - axis represents the cohesion in
the non-consolidated state. For cohesionless material, the Molerus Yield Locus and
die Jenike Effective Yield Locus coincide.
Molerus Yield Locus
ot
Figure 3.2
o
Yield Locus and Internal Friction Angle
Considering the bulk materials used in this work and the assumption that the
cohesion of the particles can be neglected (as discussed before), the effective angle
of internal friction 6 can confidendy be used as a parameter in models developed to
describe the flow of bulk solids from mass flow bins.
The wall friction angle describes the yield criterion of bulk solids at the bin or
hopper wall. It is determined by the material's Wall Yield Locus. A typical
linearized Wall Yield Locus is plotted in Figure 3.3.
Figure 3.3
Wall Yield Locus and Wall Friction Angle
The Effective Yield Locus and Wall Yield Locus are determined by shear tests.
There are a number of test methods available for measuring the shear properties of
particles. Table 3.1 gives a survey of shear test methods and a comparison of shear
testers. In this comparison the symbol " >/ " stands for " applicable ". The testers in
bold font can measure both internal friction angle and wall friction angle. From
Table 3.1, it is obvious that every method has its advantage. The use of a particular
method depends on the field of application. In bin design and flowrate analysis, the
Jenike shear tester offers a quick, reproducible and adequate method. It was used in
this work to obtain the two angles stated above. The test procedures were
conducted as detailed by Arnold et al.(1980)^ 3.
3.4 Particle Density
Particle density is mass divided by the particle volume. The particle volume of a
sample was obtained using a Beckman Air Comparison Pycnometer (Arnold et al.
1980^33). The mass of the sample was weighed on a Metder P C 4400 electronic
balance.
3.5 Bulk Density
Bulk density of a bulk solids is its mass divided by the volume of particles and
voids it occupies. It is not only an important conversion parameter from a
volumetric system to a mass system, but also a parameter to indicate the
compressibility of bulk material and to predict the voidage of solids during
processing.
Table 3.1
Shear Test Methods and Their Comparison
(based on Schwedes 1983[92] and Schwedes et al.l990[93,943)
Shear Testers
Direct
Indirect
- Shear Testers
/
Translational
Rotational
r-J—,
. 1 .
Comparison of \
Shear Testers
N
«
4>
H
&
u
s H
U
04)
0)
OS
a>
•8 I
co E>
II
r
u
**
«5
t
<n
4)
H
4>
h
09
0*
«
41
H
CO b03
03
0)
A
B CO
O
</> M
B
u
©
H OS
0)
1
on
4)
H
«-
8B9
CO
M
B
JE
1s 1
OS
PQ P
OJ
0)
*3
CO
fc
a
8
§
co DC
Angle of Internal Friction
Angle of Wall Friction
I
£I
Biaxial Triaxial
09
M
B
0>
U
\
V V
- V V V V - - - . .
Shear Strength,
Compressive Strength
VV--VV-----V
Stress Distribution
VVV---VVVVVV
Complete State of Stress - - V - -Time Consolidation
Complete State of
Deformation
- V V V V -
- - V - -VVVVVV-
Suitable for Elastic
Granular Materials
Suitable for Coarse
Material ( > 10 m m )
Suitable for Small Stress
( <lN/cm2 )
V V V V V V V
V
V
In the prediction of flowrate of coarse bulk solids from a bin / hopper, the bulk
density is generally considered as a constant parameter. Forfinematerials, the bulk
density cannot be regarded as constant; the compressibility of the material must be
taken into account in models to describe flow behaviour. In general, the value of
bulk density depends not only on the composition of the particles, such as particle
size, size distribution, moisture, shape and roughness, but also on its state of
compaction. Therefore, the specified bulk density, such as loose packed density,
dense packed density and tapped density, are used in an attempt to identify the bulk
density at different compaction conditions. In practice, especially in theoretical
analyses of solid handling, it is necessary to describe the bulk density of a certain
powder as a function of the extent of consolidation.
There are a number of relations to describe the variation between bulk density and
applied stress, as shown in Table 3.2. S o m e models, theoretically, cannot be
applied at zero or infinite stress. In fact, infinite stress in materials handling is
impossible. However, low stress conditions are frequendy experienced. Hence, it
is necessary to identify the best models which are both simple and sufficiently
accurate to describe the bulk density variations for bulk solids in the range of stress
from free external compaction to afiniteconsolidation.
The bulk density - stress equation, normally, is obtained by applying uniaxial
compaction in laboratory experiments. Here the stress is actually the major principal
stress. However, in practice, the powder is handled in three dimensional stress
systems. Furthermore present bulk density equations in practice utilize either the
m e a n stress and or the major principal stress. Use of the m e a n stress in these
equations, although convenient mathematically, m a y cause s o m e error. T h e
magnitude of this error is evaluated in Section 3.5.4.
44
3.5.1 Bulk Density Measurement
The Jenike Compressibility Tester (Arnold et al. 1980[33), as shown in Figure 3.4,
was applied to record the relation between the bulk density and applied stress.
Dial Indicator
Indicator Holder
^
Sample
Cover
Weight Carrier
&
Figure 3.4
Jenike Compressibility Tester (Arnold et al. 1980^33)
The tester cell (base of the tester) was 63.5mm in diameter and 19mm in depth. The
ratio of thickness of the specimen and inner diameter of the ring typically ranges
between 0.25 and 0.33. This measurement on this tester is a typical uniaxial
compression test (Svarovsky 1987™ 3 ). A typical bulk density - stress curve is
plotted in Figure 3.5.
The bulk density is a function of the consolidation. It increases with increasing
applied stress, varying rapidly at low stress and slowly at high stress.
p = P ( o- )
(3.1)
1400
Major Principal Stress (kPa)
Figure 3.5 A Typical Bulk Density Curve of a Fine Sand
3.5.2 Models Used to Fit Experimental Results
There are many models, as listed in Table 3.2, which can be used to describe the
density variation equation (3.1) for powders.
An algorithm for least-squares estimation of nonlinear parameters (Marquardt
1963 [1013 ) was used to determine the coefficients in these models and further to
find which models best fit the experimental results for the bulk solids examined.
The optimization process of fitting is stated as follows:
It is assumed that the experimental results are (o*j, pj) i = 1,2,..., n, one o
the models in Table 3.2 is p = Dj ( o, B ) j = 1, 2,..., 16, the problem is to
compute those estimates of the parameters B which will minimize the variance
^minJ = i I [ P i - P j ( ^ B ) ] 2
where B = (bj, b2,.... b m )
(j = 1, 2, ..., 16)
(m<n)
(3.2)
Table 3.2
Model
The Models for Bulk Density - Stress Relation
Author
a -->0
Expression
1
b1+b2lno
a ->oo
1
Balshin*
2
Smith*
3
Ballhansen*
4
Jones*
P= P
5
Athy*
P = Ps"(Ps-Po)e"blCT
Po
Ps
6
Heckel**
P = (Ps-p0)ebl-b2°
(Ps-P 0 )e bl
0
7
Halldin**
P = Ps-(Ps-Po)ebl-b2°
Ps-(Ps " Po) e b l Ps
8
Rim**!96!
P = (Ps - Po) e' bl °
Ps-P0
9
Nutting*
Po
10
Cooper*
P = P0eb^2
PsPO
p = p0+b1o1/3
Pabie^a
1+ b i e b 2 C T
S
(—^T~)
bj+b2lna
m
11
Kawakita[97]
12
JenikeM
Ps-(Ps-Po)We-b2/a
l+bio
P = PoPs
Ps+PobiCT
p = p 0 (1+20.89 o ) b l
13
Jenike^
14
- oo
0
Po
Psbi
1+ bl
oo
Ps
-
0
4*
P 0
0
oo
PsPO
b
Ps"(Ps-Po) i
Po
Ps
Po
oo
p = b2 ab2
0
oo
Johansonf98]
p = p 0 (1+ bl0-)b2
Po
oo
15
Schofieldt"]
p = p s (b 1 + b 2 hia)
-oo
oo
16
This work
p = b 1 o b 2 + b3
b3
oo
*<sited from Kawakita
**
I97!.
cited from German t10°].
After obtaining the optimum parameters B for all of the models in Table 3.2,
the total m i n i m u m variances ^ m j n ; (jI = 1, 2, .... 16) were compared to
identify the best fitting model.
The flowchart of the program is plotted in Figure 3.6.
3.5.3 The Models Appropriate to the Bulk Solids
79 experimental results for 22 sand mixtures, sugar, 2 alumina mixtures, PVC
powder, fine coal, Shirley Phosphate and glass beads were used to examine the
appropriate models for the bulk solids. Considering that the models 1,4 and 15 are
not suitable when the applied stress o approaches zero, the calculation for all of
models was carried out in two cases. O n e case examined the no-zero stress
situation, whereas the second examined the zero stress situation with the models 1,
4, and 15 excluded. The results of total m i n i m u m variance are plotted in Figure
3.7.
From Figure 3.7, it can be seen that for the bulk materials handled in a compacted
condition, the models 1, 4, 9, 13, 14, 15 and 16 provide the most accurate
prediction of the bulk density variations. For situations involving very low
compaction levels, Models 9, 14 and 16 give better density predictions. A s a
special case, it is noticed that the very large variance for model 13, Figure 3.7 b)
was caused by a near zero stress observation (Figure 3.8).
Let O
( i ) = 0.0
i = 1, 2
16
min
Input
initial experimental data
allowance error for calculation
m a x i m u m iteration times
Looping for every model i = 1,2,..., 16
Calculating the initial parameters B o
Calculating the variance <J>.
for current model
I
Calculating the gradients
Finding out the B by the algorithm for nonlinear parameters and m i n i m u m variance
3P
;
E = <D
mm , I
O
(i)=
min
Output
db
T~
O
(i)+E
min
I
(i), B,
O
min
j = 1, 2, ..., m
p. < a , B )
i
r
Calculating the values of p
by every model
Yes
Output thetotalvariance
O
(i)
i=l,2,..., 16
min
I
Stop the Program
Figure 3.6 The Flowchart of the Program to Calculate M i n i m u m Variances
1 2 3 4 5 6 7 8 9
10 1112 13 14 15 16
Model
a) N o Zero Stress Involved
109
c
B
e
o
o
§
•fi
>
g
1
108
107
106
105
104
103
1 2 3 4 5 6 7 8 9 1011 12 13 14 15 16
Model
b) With Zero Stress
Figure 3.7 Total Minimum Variance of Different Models for 79 Tests
1500 i
m
IL
1
100
r
°
M
&
--*--
500
Measured
Model 13
s
PQ
J
l
2
Stress (kPa)
Figure 3.8 Comparison of Experimental Results with Those Fitted
by Model 13 in L o w Stress Range
If it is assumed that piow is the bulk density at the lowest stress Oiow applied
experiments (0.37172 kPa - weight of the cover in Figure 3.4), then model 13 can
be modified to:
G= 0
Po
= <
n
_,_ (Plow-Pp)CT
p0+
bl0-b2
« —
CT
low
n^<r~
0<a<alow
(3.3)
O^Olow
Surprisingly, from the experimental data, the modified model 13 gave the same
fitting results as those from model 14 at each test point. With the modified model
13, Figure 3.7 b) replots as Figure 3.9.
The typical fitting results for the best models in each of the two cases are shown in
Figure 3.10 and Figure 3.11 (for sand with particle size in the range 90-106 pm).
•
c
E
—
e
I
"fi
>
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Model
Figure 3.9 Replotting of Figure 3.7 b) with Modified Model 13
1400
1
1
1
1
1
r
'
I
-!-
-
1380
A Measued
- ° - Model l
"""" Model 4
-*- Model 9
"+• Model 13
- ° - Model 14
-*- Model 15
-*- Model 16
1360
1340^
iitn
I
10
.
I
20
.
L
30
.
—
i
40
L
•
•
i-
50
Stress (kPa)
Figure 3.10 The Typical Fitting Results for Best Fitting Models
with Zero Stress Excluded
1450
20
30
40
50
Stress ( k P a )
Figure 3.11 The Typical Fitting Results for Best Fitting Models
with Zero Stress Included
Models 9, 14 and 16 as well as the modified model 13 (equation (3.3)) are best
fitting equations. As an indication of fitting accuracy, the worst fitting and best
fitting of these models are displayed in Figure 3.12 and Figure 3.13, respectively.
1600
Minimum Variance ( 8 points )
&
1400
ffl
13C
A
O Model 9
-O- Model 13 (14)
— - Model 16
Model 9
812.3555
Model 13 (14) 611.9850
Model 16
796.7709
10
20
Measured
30
40
50
Stress ( k P a )
Figure 3.12 Worst Fitting of Models 9,14,16 and Modified Model 13
1200
1
W>
1100
»9
s
s
Minimum Variance (8 points)
1000.
900
Model 9
Model 16
10
0~~
1.963542
1.868106
20
Measured
Model 9
Model 16
40
30
50
Stress (kPa)
a) Best Fitting of Models 9 and 16
1500
09
c
&
1300
1200
Minimum Variance (8 points) — O —
— ° —
2.216909
10
20
30
Measured
Model 13 (14)
40
50
Stress (kPa)
b) Best Fitting of Model 14 and Modified Model 13
Figure 3.13 Best Fitting of Models 9,14,16 and Modified 13
From Table 3.2, models 9, 14 and 16 provide p 0 at o = 0 as expected. W h e n o
tends to infinity, the bulk densities predicted also tend towards infinity.
Theoretically, the bulk density varies rapidly at low stress, slowly at high stress
and becomes almost constant at very high stress (assuming that the particles do not
fracture under high stress). The equation^
\ named model 17 here,
P =Pmax - (Pmax " Po> e '^^
provides a slighdy better fit to the results, in the range of the experimental stress
compared to that obtained using models 9, 14 and 16. However, compared with
these models, the necessity to use one more variable generates considerable
inconvenience. Based on the above evaluation, models 9, 14 and 16 are the
preferred bulk density expressions for practical applications.
3.5.4 Application of the Bulk Density Equation
As mentioned previously, the bulk density equation is obtained by one-dimensional
tests. However, bulk materials often are handled in a multi-dimension stress state.
This suggests that the application of the bulk density equation based on uniaxial
stresses m a y be of limited practical application.
a) General Considerations
It is assumed that the relation between strain { 6} and principal stress { O} for
bulk solids should satisfy
-D -X>
-o 1 -D { o)
_ -1> -1) 1 _
•
{e} =ie
1
(3.4)
Since the bulk solids are not linearly elastic, K y and Poisson's ratio o are not
constants but vary with bulk material voidage.
For the bulk solid in the uniaxial compressibility test, if density model 16 is used
the coefficient related to Young's modulus K y becomes (referring to eqn (A-1.10)
in Appendix 1-1)
A n approximation, to estimate the Poisson's ratio, is possible by introducing two
extreme pressing states, as shown in Figure 3.14.
VJ
'A
'/,
/?!!!!!j!!!!!j!j!^
i
^liiiiiiiiiiiiii^
A—T^Y/.
a). With High Voidage
Figure 3.14
\VA
b). With Low Voidage
Pressing of Bulk Material in T w o States
From Figure 3.14, it is considered that the compaction of bulk solids with high
voidage can be regarded as a one dimensional problem, in this case, v = 0. O n the
other hand, compaction of the material at very low voidage is considered a three
dimensional problem, \) = o 0 . Obviously all bulk materials handling situations fall
between these two states. Therefore, a simple way to estimate the Poisson's ratio is
to assume
v=\>0
60-e
-JL—e
0" e min
(3.5)
or
v.^-z^*.
P m a x " PO
Unfortunately, no publication on the Poisson's ratio for bulk solids has been
found. Values of the Poisson's ratio provided by Lambe and Whitman^ ^ are
\)0<0.3 for such materials as amphibolite, limestone, rock salt and steel; on^0.36
for ice and aluminum, for omax = (3 ~ 5) * Hr kPa.
If the density model 16 is applied, the relation between Poisson's ratio and stress,
eqn (3.5), can be rewritten as
/
t) = o0[ —— j * (3.6)
v "max /
For practical purposes, it is assumed that t) 0 = 0.3; o , ^ = 4 * 10 kPa.
b)
Application of the bulk density equation to mass flow bins
The work presented here attempts to enable the uniaxial measured data from the
laboratory to be applied to simulate the bulk density in different sections of a mass
flow bin.
Since the vertical movement of bulk material in the vertical section of the bin is
considered similar to that in the uniaxial laboratory measurement, it is assumed that
the vertical stress o*z in the vertical section can be direcdy usedtodetermine the bulk
density in this region, despite the fact that a z is not the major principal stress in
dynamic condition. That is:
in vertical section p = p0 + blv (o^)^ (3.7)
where blv = bx; b2v = b2 .
However, for the hopper section, many theoretical predictions for stress deal w
mean sn-ess[1»3»28»3l-33,39,58] j n ^
case> a n e q U i va i en t
major principal stress
needs to be estimated for predicting the bulk density. Based on the assumption that
the bulk density can be evaluated by the mean stress acting, the considerations
include the following steps:
• assume the computed mean stresses for different hoppers are:
<*r+ q 8 .
plane flow hopper mean stress omw =
conical hopper
mean stress
omc
a. + o e + o 4
= —*
j*
*;
convert the computed mean stressestoequivalent major principal stresses:
(referring to eqn (A-2.3) and eqn (A-2.4) in Appendix 1-2)
plane flow hopper ole = 2(1-D) o"mw;
3(1-D)
conical hopper
ale =
a^;
•
insert the equivalent major principal stresses into bulk density model, e.g.,
using model 16:
p = p0 + b! (o^)*2 (3.8)
• rewrite the bulk density equations as a function of the computed mean
stresses: (referring to eqn (A-2.5)toeqn (A-2.6) in Appendix 1-2)
p = p0 + b l w ( o
conical hopper
P = Po + b l c ( a ^ ) ^
m w
)
b 2 w
plane flow hopper
(3.9)
(3.10)
Once the experimental data are obtained on the Jenike Compressibility Tester
(Figure 3.4) then by using eqn (3.6), the eqns (3.9) and (3.10) can be easily
obtainedtogetherwith the eqn (3.7).
c)
Differences Resulting from Use of the Different Bulk Density Equations,
In this section the differences resulting from applying the bulk density equation to
two types of converging channel is made; it is assumed that the computed mean
stress (cr m w or o ^ ) has been obtained. The cases are considered as:
Case I: Simply using the computed mean stress (crmw or <*mc) in the bulk
density model 16 instead of the major principal stress;
Case II: Using the equivalent major principal stress, which is converted from
the computed mean stress (<J m w or a m c ) , in the bulk density model 16
and calculating the bulk density using eqn (3.9) or eqn (3.10).
It is assumed that the computed mean stress is in the range of 0 - 50 kPa.
Comparison of Case I with Case II has been carried out for the bulk materials
used in Section 3.5.3, the results indicate that the Case I gives a lower estimation
of the bulk density. Compared to Case II the percentage under-estimation from
Case I was 0.12% to 1.34% for a plane flow hopper and 0.17% to 2.05% for a
conical hopper, at 50 kPa computed mean stress; the corresponding absolute underestimation was 1.5 to 26 (kg/m3) for the plane flow hopper and 2.2 to 40 (kg/m3)
for the conical hopper. For the 79 bulk density measurements examined and in
comparison to the results obtained in Case II, the extent of the under-estimation
provided by Case I for the plane flow and the conical hoppers, in the range of
computed mean stress 0 to 50 kPa, is between lower bound and upper bound
plotted in Figure 3.15 in terms of absolute and percentage under-estimation,
respectively.
From Figure 3.15 a) and b), it can be seen that the under-estimation of bulk den
in the conical hopper is larger than that in the plane hopper. This follows from the
fact that the computed mean stress in plane flow hoppers is closer to the equivalent
major principal stress. Namely, for the same stress condition, the inequality
CT
le > amw > cmc
applies.
m
upper bound for conical hopper
^ 3 0
4>
2
IS
upper bound for plane flow hopper
t10
se
lower bound for conical hopper
lower bound for plane flow hopper
,e
\
*=*
10
20
30
40
50
Computed mean stress (kPa)
a) Absolute Under-Estimation
25
•
1
•
1
•
r
upper bound for conical hopper
upper bound for plane flow hopper
_
05 •
0.0
^
lower bound for conical hopper
lower bound for plane flow hopper
\
<-> r
1 i i i i i •
10
20
30
40
350
Computed Mean Stress (kPa)
b) Relative Under-Estimation
Figure 3.15 Under-Estimation of Case I for Bulk Density for Plane Flow
Hopper and Conical Hopper Compared with the Results of Case II
While the differences between Case I and Case II are small and either approach
could be used in practical situations, the Case II approach (using an equivalent
major principal stress) will be used in the theories developed in this work to predict
the air pressure distribution and the particle flowrate. This approach has been taken
as it is felt that the Case II approach is more realistic from a theoretical point of
view.
3.6 Permeability
The permeability C is a measure of the ability of a bulk material to allow air (or
other gases) to flow through under certain circumstances. Permeability is defined,
according to D'Arcy L a w , as a ratio of the superficial air velocity (relative to the
particles) and the air pressure gradient. T h e permeability is complex property
dependent on m a n y factors such as particle size, size distribution, shape, surface
roughness, moisture content and as well as consolidation condition.
3.6.1 Permeability Measurement
The permeability of solids can be measured on a packed bed or a fluidised bed of
particles. Considering the handling condition for gravity flow of bulk solids from a
mass flow bin, in which the particles are in contact with each other all thetimes,a
Jenike Permeability Tester (Arnold et al. 1980 [31 , as shown in Figure 3.16) was
used to find the permeability of a packed bed at different voidages (consolidationrelated permeability).
o
Air Pressure Gauge
-p
Air Pressure Gauge (P0)
/ Rotameter
^2'
Si
i
o o
•o
o
Y
&
M
•P
M
V,
•H
O
^
Air supply
4
r
iF2^
oo e
o o •H£
o V
•
• u
Test c linder
y
63.5 mm
H—=-:—H
dia
•r
Pi
PI
E-i
r-t
•ri
i ji8
^
Sampl 2
0 sample
04
<
Sample'
in * i\
<
o
H
Regulator
-AP t
•ri
+>
W
0
'"
•ri
Figure 3.16
Jenike Permeability Tester (Arnold et alS *)
The bulk density was obtained from the weight and volume of the bulk solid in the
test cylinder
P
" Ap Hp
(3.11)
The permeability was calculated from
-Sa-a
C =
A P Ap
(3.12)
For a certain flowrate Q of air through the packed bed in Figure 3.16, the pressure
drop APt includes the pressure drop due to the resistance to air flow of the particl
AP and that caused by the resistance of the distribution gauze AP0, i.e.,
APt = A P + A P 0
(3.13)
Substituting eqn (3.13) into eqn (3.12) gives
Hn
C = A-p£ APt - AP 0
(3.14)
3.6.2 Models UsedtoFit Experimental Results
As mentioned at the beginning of Section 3.6, the permeability of bulk solid
depends on particle size, particle size distribution, particle shape, particle roughness
and voidage. For the bulk solids with fine particles, the permeability of bulk solids
can be dramatically influenced by the appearance of the fine particles and the range
of the particle size distribution^64,66,100^. Therefore, an adequate permeability
model must be selectedtodescribe these factors.
Under the conditions in which D'Arcy law is valid, viz., the flow of air bet
packed particles consist of laminar, viscous flow without significant turbulence or
inertial energy losses, the models commonly used for permeability are in three
forms, i.e.,
model 1=
C-K^l-e)^)
model 2 :
C =
K,
ri04i
(Mahinda Samarasinhe et al.
)
(Jenike and Johanson[98'105'1061)
(1-e)'
model 3 : C = K 3
(Carman-Kozeny equation[87'107"109])
(1-e)'
where K j , K^, K 3 , a and co are constants.
Generally, model 3 is applicable to experiments on coarse materials with a narrow
particle size distribution (where the particle size d ^ can be readily identified) and in
an unconsolidated condition^108,871. Therefore, some limitations on the application
of model 3 are as follows:
• it cannot describe the flow of air through a consolidated particle bed
well^110^, namely, it cannot describe the consolidation-related permeability
well;
• it applies only over a limited range of porosities or to monosized particles
and it is not suitable for fine powders^ ™.
In comparison, Models 1 and 2 can be used for a greater range of particle mixtures
because they include one more variable to fit the experimental data. F r o m the
expressions, it can be seen that model 1 is identical to model 3 w h e n the index
number co is equal to 3. For instance, Mahinda Samarasinhe et alJ
•* observed in
their research that CO = 3.2 for crushed glass and co = 5.2 for N e w Liskeard clay.
In order to find the best fitting equation to the experimental data, a similar
mathematical treatment to that used for the bulk density models was employed. It is
necessary to mention that the 10 factor for permeability has been omitted from the
following model comparison.
The best permeability model was correlated to a total of 125 points of the 16
materials tested. T h e results for the total m i n i m u m variance for the 125
experimental data points are plotted in Figure 3.17.
65
Figure 3.17 Comparison of Fitting Results for Three Permeability Models
From Figure 3.17, it is evident that model 2 provides the best fitting model. As an
indication of fitting accuracy, the worst fitting and best fitting of model 2 are
displayed in Figure 3.18 and Figure 3.19, respectively.
In view of the accuracy of Model 2, this model will be used in this work. An
alternate expression for Model 2 is
-491
(3.15)
where CQ is a constant standing for the permeability at lowest compaction (without
external stress acting during permeability measurement)
Q) =
K2
( 1 - e 0 )5
21000
1
1
1
1
'
|
»
1
1
Bulk Material: Sugar
S u m of Minimum Variance of 6 points: 79514.0
T—E1"
19000 •
17000
15000
•*~~ Measured Value
"•""" Predicted Value
x
j.
13000
0.40
0.41
0.42
0.43
0.44
0.45
0.46
Voidage
Figure 3.18
250
i
I
Worst Fitting of Permeability Model 2
i
I
i
I
i
Bulk Material: Sand (0-425 p m )
S u m of Minimum Variance of 10 points : 157.43
200
150
100
"*—
50
0.40
0.42
0.44
0.46
Measured Value
Predicted Value
_L
0.48
Voidage
Figure 3.19
Best Fitting of Permeability Model 2
0.50
Chapter 4
T h e Prediction of Air Pressure Gradients in M a s s Flow Bins
4.1 Introduction
Analyses of the air pressure distribution in a mass flow bin have been conducted by
Crewdson et al.(1977)[50], Spink et al.(1978)[51], Head(1979) t55] , Nedderman et
al.(1983)t56' and McLean(1979)^ 5 -L In most cases, the pressure gradient is related
to the superficial velocity by use of evaluated parameters consistent with the
assumed actual stress field. The basis of these analyses is that the pressure gradient
is m a d e proportional to the superficial velocity of air relative to the particles by
employing either D'Arcy's law or the Carman-Kozeny equation. The majority of
these air pressure distribution predictions do not include the effect of height of
material in the vertical section of the bin. However, McLean (1979)^ ^ and McLean
et al. (1980)^ ^ provided the prediction of pressure in the hoppers with or without
surcharge.
Head and Crewdson et al. used a log function to describe the bulk density variation
in the hopper. Unfortunately such a representation of the density variation is not
valid for very small stresses. In the adoption of this density model, Crewdson et al.
deduced that the m i n i m u m air pressure in the hopper occurred constantly at
r=2.718282r0. However, this predicted position of m i n i m u m air pressure was not
confirmed by their experimental results. They observed that the position of
m i n i m u m air pressure were in the range between 1.7r0 and 1.8r0 for all
experiments.
In an attempt to offset the above deficiencies, the following mathematical model
attempts to include the effect of material head on the air pressure distributions in
bins. Particle dynamic deaeration phenomena are also considered. This model for
predicting the pressure distribution is based on the following considerations:
i. The stress near the oudet of hopper is a linear function of the vertical
distance measured from a point which is located between the apex of hopper
and the oudet, as shown in Figure 4.2.
ii. Bulk density and air pressure in the bin vary with the vertical distance
continuously.
iii. The flowrates of particles and air at any cross-section in a mass flow bin are
constant.
iv. The dynamic deaeration at the top surface of bulk material is proportional to
the flowrate of particles.
v. The mass flow bin is divided in three regions, as shown in Figure 4.1. The
principal stresses are assumed to be vertical in region I, radial in region III
while the region II is regarded as a transitional region.
The dimensionless depth values used in Figure 4.1 are defined as follows:
n
h
^lrnax-
hi
n
bo
h*
h
max
i
mp
(4.1)
69
Figure 4.1
Regions Defined for a Mass Flow Bin
4.2
The Bulk Density Distribution in Mass Flow Bins
i)
B u l k density distribution in region I
A s presented in Section 3.5, the bulk density o f a bulk solid is a function of the
extent of consolidation. T o evaluate the bulk density in the vertical section of the
bin, the stress level m u s t b e estimated. O n e of the typical theoretical analyses o f
stresses in the vertical section of the bin is the w o r k of Walters in 1 9 7 3 t l l l ] . H e
extended Walker's theory to predict the stresses in the bulk solids under both static
(initial fdling) a n d d y n a m i c (flow) conditions. Blight in 1 9 8 6 [ 1 1 2 ] presented a
m e t h o d to determine the stresses of fine p o w d e r s in silo. H e considered the effect
70
of the pore air pressure during material filling. However, he recommended that this
effect is very small especially for slow filling. It is acceptable that this effect on the
vertical stress can be ignored for steady state flow where there is continuously
filling while discharging material. Thus, once the effect of the pore air pressure on
the vertical stress is neglected, Blight's theory becomes similar to Walters' theory
but the lateral stress coefficient used in Blight's equations is not predictable.
Therefore, Walters' equation for the material under dynamic condition is preferred
to estimate the vertical stress distribution in current cases.
The equations presented by Walters and Blight were derived assuming that the bulk
density of the powder was a constant. However, it is believed in this thesis that the
bulk density varies with the stress. If this variation is considered in Walters'
equation, the derivation of an analytical equation for the vertical stress is not
apparent F r o m Appendix I - 3, it is found that this variation can be neglected at this
stage; the bulk density distribution in the vertical section can be approximated by
using Walters' vertical stress distribution under dynamic conditions^
•• (referring
to eqn (A-3.4) in Appendix I - 3), i.e.,
4K
°
(Tl)=
where
o QD
w hl (» __x
n
^Imax *\t
4K
[1_e
Kw =
tan <|> cos 8
2—
( 1 + sin S ) - 2 y sin 8
y = 3V[1'(1"c)3/2]
( tan<|) V
V tan8 )
]
(42)
The bulk density distribution, using eqn (3.7), becomes
4K w h,
P = Po + b l v [ 4 0 0 0 K ; ( l _ e
b2>
D
)
In fact b 2 v « l»then
^v «
p o b2
The above equation is rewritten as
b
P = Po + iv
r
4K w h t
Pogp
(l-e D
4000 K
w
(nmax- 1 !)
)
>2v
(4.3)
Hence, it follows that the bulk density at the transition level is
Pl = Po + bi v
ii)
PogD
4000 K. (l-e"
w
4K
w h l „, -.b2
n
D
)
(4.4)
Bulk density in region m
From the consideration i) mentioned in Section 4.1, as shown in Figure 4.2, th
stress in this region is described by:
O m c = ki (*1 -11*)
where rj* is suggested to be in the range of (0.95 ~ 0.99)rj0^58^, all theore
results concerning this model in this work are obtained by using
TI*=0.95TI0
From eqn (3.10), the bulk density in region in is
(4.5)
P = Po + bic K 3 (r) -Tj* ) 2c
The bulk density at T| =ri m p, is
P m p = Po + bi c K 3 ( limp - "H* )
iii)
(4.6)
2c
Bulk density in region II
The stress in region II is a transitional stress field in the upper part and b
equaltothe radial stress field in the lower part. For the hopper without surcharge,
Figure 4.2 a), it is assumed that die flow stress field is approximated by a parabolic
equation here, i.e.,
>mc = k 2 V ill - 11
In general, the stress distribution is assumed as
(
°mc- <
l/b2C
Pi ~ Po "
b
ic
)
+ WTii-,n
Therefore, the bulk density becomes
p = Po + bic
Pi-Po
»ic
l/b2c
+WTii-,n
.b2C
a) Stress Field in Hopper Without Surcharge
b) Stress Field in Hopper With Surcharge
Figure 4.2 Stress Field in Hopper (McLean 1979 [58])
O n assuming the bulk density varies continuously with the depth, at T)=T| mp ,
P=Pmp» then
1
p=p0+ blc <
fPi-po> °2C
I bic J
J_
JL
•
. 2C
b
r P m p - P o ^ c /"Pi-Po"\b2cl / 11 i~il
A bic J
I bic J J V ^ " ^ m p
4.3
(4.7)
Superficial Velocity of Air Relative to Particles
4.3.1 Flowrate of Air
Neglecting the change of air density in the bin, the absolute flowrate of air through
any section is constant, i.e.
M
Qi
where
+
i£j
=
Pi
Qi +
Mia
(4.8)
pj
Q ^ Q ; are the volumetric flowrates of air relative to the particles at
location i and j, respectively.
M ^ M j are the mass flowrate of particles at location i and j,
respectively.
Considering the sense of flow, M = - Q « , where
Q p is the absolute mass flowrate of particles.
' -' comes from the different sense between the particle flow
(downward) and height (upward is positive)
The subscripts i and j stand for arbitrary levels i and j, respectively,
in the bin.
Setting the reference level at the top level of material, then
Po
p
Since e = 1 - — — ;
Ps
Pi
and for mass flow, M Q = M j = M , then
The term Q0 stands for the dynamic deaeration condition at the top surface.
Assuming that
Qo = " Kdea — (4.10)
Po
where K<iea is a dimensionless coefficient (named dynamic deaeration
coefficient)
then the general expression for the relative air flowrate Q is
Q=£[('-K<iea>-£] (4.11)
4.3.2 Relative Velocity of Air to Powder
The superficial velocity of the air relative to the particles is the ratio betw
relative flowrate of air and cross-section area through which the air is passing, i.e.,
u
= A
In region I
u=
M
PoA>
[(l-Kdea)-y]
(4.12)
where A v is the cross-section area of the vertical section of bin
In region U and EI
u =
M
PoAc
[(1-Kdea)--^
(4.13)
where A c is the local hopper area
4.4
Air Pressure Gradients in Mass Flow Bins
The mathematical model for the pressure gradient is based on the D'Arcy law, viz.,
dP u
dh " " C
or
dP
dP.dn,
_u ,
dp ~ dh'dh " " C h i
i)
Li region I
Substituting eqn (4.12) into eqn (4.14) yields
^p.-_2lf-P_Yl_JL_rri-ir ^ p0"
dri " ColpoJ p 0 A v L U ***** "p~_
(4.14)
4h M
Let W 0 = ^ " "t ^ 2 a n d f = 7"» then
CoPo^Do^
p0
S-w.5'"[»-«*.)4]
or
g.w.fij'^-u-^',
(4.15)
where
f=l +
p
lv
Po
Pog°
(l-e
4000 Kw
4 K
w hl („
D
lTlmax
b
2v
_n\
1])
)
(4.16)
A similar method is usedtoobtain the pressure gradient in region II and region m .
ii)
In region II
S-w.^E^-O-Wf'l
(4.17)
where
f = —P
Po
with p determined from eqn (4.7)
Pmp is calculated by eqn (4.6)
(4.18)
iii)
In region III
^ = W„(^)2[fa-1-(l-Kdea)fa]
(4.19)
where
f -*Po
= i + _lC-^3(T| _ T1 *) b 2c
Po
4.5
(420)
Air Pressure Distributions in Mass Flow Bins
The air pressure distributions in each region is obtained by integrating the
gradients over the respective height ranges.
i) In region I (T|j < T\ < Tjmax)
2 T\
a 1
P-Pi = W0-| [ [f - -(l-Kdea)fa]di1 (4.21)
when r| = r\max, P = 0, then
~2 ilmax
-Pi = W 0 - | |
[fa-1-(l-Kdea)fa]dTi
where Pi is the pressure at the transition level and
f is determined by eqn (4.16)
(4.22)
79
ii)
In region II (ri mp <, T) <, r^ )
,TI [f^-d-Kdea)^]
P-Pmp = W 0 |
i
dn
•^mo
(n/Tin)
or
^-(l-Kd
P = Pmp + W 0 I
Imp (il/ilo)2
^'a
dri
(4.23)
where f is determined by eqn (4.18)
Pnm is the pressure at rj = r|mp
When 11=11!,? = ?! given by
fill
[f^-d-Kdea)^]
Pi - Pmp = W0 I 5 d^
mp
in)
(4 24)
'
(T|/T!O)
Inregionni (ri0 <> x\ <, ri mp )
< t A (•> v. \ r a< t
[f•a-1
" -(l-K d e a )f ]
P = W0[
J"
dr|
(il/ilo)'
ril
(4.25)
When rt = Timp, P = P m p given by
filmp
Pmp = W 0 J
V
Jl
lo
[fa"1-(l-Kdea)fa]
™
dti
dl/ifo)
where f is determined by eqn (4.20)
(4.26)
4.6
Boundary Conditions for Application of Air Pressure Model
The unknown variables in this mathematical model are K3, K d e a and r ) m p . Hence,
three independent boundary conditions are required to determine these variables.
1) Pressure Continuty
eqn (4.22) + eqn (4.24) + eqn (4.26) = 0 (4.27)
dP
2)Atr|=rimp,—
=0
( 1 - Kdea )-£m2- 1 = 0
Po
Kdea = 1 - ~2j- (4.28)
Pmp
3) Estimation of TJmp
The pressure minima point rj = rjmp is that separating the positive pressure gra
from the negative pressure gradient; it is also the point where the direction of
superficial velocity of the air relative to the particles switches.
It is assumed that rjmp depends on the bin geometry, mainly r|o and i]max. Figure
4.3 shows the relation between T]mp, rj0 and T)max observed in experiments. Since
there were insufficient data to identify T[mp more accurately, a simple relation
estimating Tjmp is used to avoid the effect from any particular experiment at th
stage, i.e.,
( l+ilmax'l
ilmp = ilo
*
V l+Tlo
(4.29)
>
4.0
3.0
o
1
• waiis
• Head
* Experiment (Chap. 6)
• Crewdson et al.
^^^
H
X
JW^
•
ex
E
^
r
2.0 -
B
X
X
-
^S*
.x'x
y^
1.0
1.0
•
i
2.0
•
(1+rj
v
3.0
)/(1 + TI A )
'max'
v
4.0
'0'
Figure 4.3 A Plot of the Relation between rj m p, T|0 and i l m ^
4.7
Discussion
4.7.1 General Observations
A s one significant basis of this air pressure gradient model, the eqn (4.8) describes
the motion of air between particles based on one kinematic consideration which is
that the absolute air motion in the bin equalstothe vector sum of relative air motion
to the particles and the absolute air motion with the particles. According to eqns
(4.8), (4.11) and (4.28), the assumed constant absolute air flowrate in the bin
equals M
|. Since the bulk density at the m i n i m u m pressure position
\Pmp Ps )
Pmp is always less than particle density p s , i.e., p m p<p s > ^ absolute air flow is
always in the same sense as the particle flow (downward). However, the variation
in the relative air flowrate which relates the air pressure gradient depends on the
magnitude of the local bulk density. Inserting eqn (4.28) into eqn (4.11), the
relative air flowrate is given by
Q = M(^i--I) (4.30)
It can be interpreted that the relative air flowrate is upward and downward as
P < P m p and p > p m p respectively. Consequently, the air pressure gradient either
retards or assists the particle flow depending on whether p < p m p or p > p m p
respectively.
A typical air pressure distribution in a mass flow bin is plotted in Figure 4.4,
having a pressure minima near the hopper oudet and a pressure maxima near the top
surface of the material. This pressure distribution has also been reported by
M c L e a n [ 5 8 ] , Spink [ 8 0 1 and Spink et al.[51]. The physical meaning of this
distribution can be explained easily with the current model. From eqn (4.28), the
extreme point of air pressure occurs at p = P m p - During material discharge, the
highest stress level appears at the transition section (Figure 4.9); the bulk density
increases as the particles flow towards the transition level from the top surface then
decreases as the particles discharge towards the hopper oudet. Figure 4.4 and
Figure 4.10 show that the bulk density increases rapidly near the top surface then
only slighdy in the remainder of the vertical section. Since a low value of bulk
density appears at both ends of bin and the m a x i m u m bulk density (Pmax) occurs at
the transition level for a hopper with surcharge or in the converging section for a
hopper without surcharge, p
m p
is located between low and m a x i m u m bulk
densities. There should be two extreme points which are located on either side of
the m a x i m u m bulk density position, as shown in Figure 4.4. Furthermore, since
83
the air pressure self-generated by the flowing particles is continuous, these two
extreme pressure values should be the maxima and minima values.
hopper
outlet
height above the hopper oudet
pressure
maxima
top surface
of material
Figure 4.4 Typical Distributions of Air Pressure and Bulk Density in a Bin
(with Surcharge)
The absolute value of the pressure maxima is very small compared with the absolute
value of the pressure minima; the magnitude of the pressure maxima depends on the
hopper oudet size and permeability of bulk solids. Table 4.1 shows the examples in
which the flows of Sand M D 3 (a low permeability bulk solid) and Sand M l (a high
permeability material) from both 0.02 m and 0.0445 m hopper outlets were
considered (the flow properties of two bulk materials and bin geometry are detailed
in Chapter 6). F r o m Table 4.1, it is apparent that the larger the hopper oudet size or
the lower the permeability of bulk material, the higher the pressure m a x i m a
becomes; the larger the oudet size, the greater the absolute ratios between pressure
m a x i m a P m a x and pressure minima Pmin- F r o m eqn (4.29), as the hopper oudet
size increases, the pressure rninima point moves toward the m a x i m u m bulk density
point; the pressure m a x i m a point, therefore, moves away from the top surface so
that the measurement of the pressure maxima can be less affected by the atmosphere
at the top surface of the material. Furthermore, the magnitude of air pressure
increases with an increase in the hopper oudet or with a decrease in permeability of
the particles (as discussed in Section 4.7.2). For these reasons the pressure maxima
is more visible in the experiments with the large hopper outlet and/or with low
permeability material. This fact can also evident in the experimental results
presented by S p i n k ^ and Spink et al.t51^ using plane flow hoppers.
From Figure 4.4, a negative air pressure gradient occurs in region III; a significant
negative air pressure gradient is generated near the hopper oudet. This is because
that the bulk density in this region p is less than p m p , from eqn (4.30), a negative
air pressure gradient is generated by flowing particles. Physically, the bulk density
in region HI decreases rapidly, due to dilation. The increase in dilation results in an
increase in negative pressure, causing an air flow opposite to the bulk solids flow,
to fill the increasing void volume.
Table 4.1 The Pressure Maxima and the Ratio between T w o Extreme Pressures
for Different Sized Bulk Solids and Hopper Oudet Sizes
Bulk Material
SandMD3
(with low
permeability)
Sand M l
(with high
rjermeabiHty)
Oudet Diameter
(m)
P
P
0.02
0.0055
max
p .
min
0.00047
0.0445
0.2000
0.00647
0.02
0.0005
0.00047
0.0445
0.0140
0.00539
max
(mmH20)
4.7.2 Comparison of Theoretical Model with Experiments
The theoretical results predicted by the current pressure model are now comp
with the experimental results obtained by
i)
experimental work (on alumina) detailed in Section 6.6;
ii)
Crewdson et al. (on three sand mixtures)'5™; and
iii) Head (on two of his sand mixtures flowing from hoppers with two
oudet sizes)1551.
For the bulk materials used by Crewdson et al. and Head, their flow properties
such as bulk density and permeability were estimated by comparing them to those
materials in Chapter 6 with similar range of particle size. The flow properties and
bin / hopper geometries for their experiments are listed in Table 4.2. The flowrates
Q p for Crewdson et al. were digitized from Fig.l in their papert501; the bulk
materials shown in 'referring material' in Table 4.2 are further detailed in Chapter 6;
the surcharge level for Crewdson et al. is assumed to be 0.01 (m) to simulate the
flow without surcharge; furthermore, the bj, b2 are used instead of the b l v , b 2 v
and b l c , b 2 c ; the estimation of the permeability constant C Q for those mixtures
which have two referring materials were initially from the relation which is assumed
to be proportional to square of d 5 0 particle size (following the Carman-Kozeny
correlation but neglecting the effect of the difference in porosities of two compared
materials), i.e.,
^0 = cl dso +c2
where
^ and C2 are constants.
e.g., to estimate the C0 for Crewdson et al.'s sand sized 212-300 pm, the
referring materials are Sand M 2 and M 3 with about 4360 and 2330 *10" 9
( M 4 N^Sec" 1 ) in permeability constant; 310 and 200 p m in median particle
size respectively. Therefore, the constants Cj and c 2 are determined as
0.0362 and 881.18 respectively. The median particle size d 5 0 for sand
212-300 p m is taken as 255 p m . Thus, the estimated permeability constant
CQ equals 3235 *10"9; take as 3250 *10" 9 ( M 4 N _ 1 Sec _ 1 ) in Table 4.2.
The comparisons of the theoretical and experimental air pressure distributions are
depicted in Figure 4.5 to Figure 4.8. The predicted pressure distributions closely
approximate the observed variations for sand mixtures and for alumina. The
theoretical results, which are similar to the experimental findings (the measurements
are detailed in Chapter 6), highlight the following:
i) Air pressure in a mass flow bin (negative air pressure gradient at the hopper
oudet) increases with increase of hopper outlet dimension (as shown in Figures 4.7
and 4.8).
Since the dynamic stress in the hopper is small, the difference between stresses in
the two bins with different outlet sizes is insignificant. Therefore, the differences
between the two bulk density variations and permeability distributions under
equilibrium conditions do not have a significant effect on the magnitude of air
pressure gradients. A s the outlet is increased, the particle velocity increases.
Consequendy, the increase in particle velocity relate to the increase in superficial
relative velocity between particles and air with a corresponding increase in air
pressure gradient. Therefore, the larger the hopper size, the greater the negative air
pressure in the bin, and further, the greater the adverse air pressure gradient at the
hopper oudet. This phenomenon can been also found from the experiments by
Head* 5 5 1 on conical hoppers; by Willis*541 and Spink* 801 on plane flow hoppers
(Figure 8.13 shows the Willis' experimental results).
ii) Air pressure increases with increase of surcharge level (as shown in Figure 4.5).
Physically, for a typical slice element of material in the bin, there are two types of
air flow in the element of particles: one is the flow due to air being expelled from
the element (relative motion of air); another air flow component is that due to drag
by the discharging particles (absolute air motion with the particles). A s discussed in
Section 4.7.1, the relative air velocity to particles is d o w n w a r d in the portion
between two pressure extreme points as p > p m p . T h e higher the surcharge level,
the longer this portion becomes. Therefore, the air pressure gradients generated in
this portion of the bin result in the air pressure being proportional to the distance
between two extreme pressure points and hence, the higher the surcharge level, the
greater the negative pressure produced.
iii) Air pressure increases with decrease in material permeability (as shown in
Figure 4.6 to Figure 4.8).
F r o m Section 4.5, it can be seen that the product C 0 p 0 is an important term to
indicate the negative air pressure. In particular, the negative air pressure increases
with the decrease in product C Q P Q . Furthermore, since the permeability indicates the
ability of the bulk solids to allow the air to flow through it, the permeability
constant C ft has greater effect on the pressure gradient than the bulk density
constant p 0 does. For a lower permeability powder flowing from a mass flow bin,
a counter-current air flow to fill the increasing void volume caused by particles
dilation in region D I is more difficult to generate requiring a higher negative air
pressure gradient
For a finer material with lower permeability, a higher pressure gradient generated
by powders results in more resistance to the flow of material in region in and a
decrease in flowrate. Furthermore, if the permeability of material becomes very
low, the resistance of air pressure gradient can cause difficulty in controlling the
flow of material due to the occurrence of flooding (more discussion on this topic
will be presented in Chapter 5 and Chapter 8).
iv) Effect of using blv and b2v instead of the blc and b^ in hopper on the predicted
air pressure is insignificant
A s mentioned in Chapter 3, the predicted difference in bulk density by using b l v
and b ^ instead of the b l c and b ^ in the hopper is small (less than 2.05% for the
tested material at 50kPa stress level). The error in the predicted bulk density and
permeability produced by application of these parameters will be further reduced
due to the low stress applying during flow. According to the calculation for alumina
used above, the over-estimation of pressure by using b l v and b 2 v instead of b l c
and b^c in pressure prediction model is less than 2 % .
1
a
00
8
3
o
en
SO
ON
ON
oo
-^
CO
8
O
s
NO
d
co oo
^^
CO
•a I
£
o
d
vo
•Sa
vo
vo
oo
,—<
t3
«o
»-H
CO
d
00
<s
s
«n
i-H
d
CO
rt-
CO
d
2
vo
•o
"*
O
o
d
d
CO
VO
CO
00
!Q
o
oo
d
r—t
8
d
3
CO
o
o
d
8 8
00
d
CO
CO
6
=L
S'a
1
gi
is
co
•O
CO
«-H
=t
o
1
•
co
i—1
8
VO
<*
ON
i—•
VO
o
d
a
8
•a CO
.2 *
o
o
§
^~v
*->
ON
<S
1
r-t
O
oo
d
1—t
o
00
CO
w>
en
o
VO
CO
CO
ON
OO
VO
d
8
ON
d
co
i—i
o
d
CO
©
CO
1—(
T-H
o
d
i
o
d
b
X!
GO
£
!
1
CO
8
2
VO
VO
d
CO
CO
6^
35
VO
8
E
CO
CO
a
=t
I
1
~
^—\
1
2.
x>
*
/"^
^
Q.
CO
/Cjisuoa^H
/CjmqB
-auusd
a
>-•
o
PQ
•™\
1Q
cd
•©-
?
a^
8
sa
/-S
o •8
Q
8
sOUPOIOar)
1
-a
«?
S
t-H
1
1
I
2 „
3
8 aa
* 38
3so co8 8m
• j
«o>
-O
vo
o
©
©
II II
©
\
H = 0.
H = 0.
i-3
mental
©
W5
& • °
\
so
©
!
\
t
t
\
ro
\
•A
a
I
SO
i
©
\
•\
u
•a
I
B<a
\
««
••\
d
£00
s
X
•
\
i
8
o
CO
©
"r\k
\
\
•
^>i
<Ml t
•
fl •• •
m
1 O 03
1
•TJ» •••
\
o coo
d
1
&»coo
A
o oo
jf
ID
mm
d
o
w
i—i
o
1
.
i
8
i
.
i
.
d
i
8 3 8
.
I
q
55
i
(J3JBM unu ) ainssay JI y
.
I
o
ri
.
I
o
oo
i
.
91
•J
t-;
a
d
09
3
<s
VO
d
T
o
s
111 a
CO CO CO
"1
d
8
CO
i
Tt
i
cs «o
e* cs
•n
-3
111
1
CO CO CO
CO
w
CO
c
/"•S
•<*
a
^
o
t<i)
*
e»
d
C
?
8
u
-a
=CS6
3
en
d
CO
-a
06
I
cs
o
s
CO
VO
i
E
q
©"•
CS
(jajBM UIUI ) amssay ity
92
s
09
/-s
a
09
0)
as
(U
00
T3
d
T3
(V
t&
= 0.02805
= 0.0304 (
Si
09
ss
11
1 I
tt
a09
CU
3.
^-v
a
"3 8
OS
*•»
C*i
iH
00
i
00
©•
CO
a
o
s
1
1
aua ©o 8dII
••*
Q
w
09
•
o
n
a
i
••
CO
•-*
u
o»
03
03
a
a *>
o
>w/
s
O
b
5
J3
00
<s
d
©
©W-l
•
*
"O
(jaiBM U I U I ) amssaid jny
r?
93
s
09
a
09
PCS
T3
0)
oo
d
a
•3
II
II
ss
H
as
3
a
0>
"0)C
aa
09
T3
03
0)
5?
0.0304 (
3
w a
0.02805
0.0304 (
09
00
0.02805
0)
09
II
II
ss
•
•
VO
00
iH
-a
a
•
03•
VO
i
1
CO
0
03
8
K
3
%
©
W
•a
CO
i§
&
•fl
CO
*-»
i—H
u
3
CO
s
Pi
•a
•a
8
«->
•a
I
i
cs
d
00
i
E
©
©^
•
*
cs
,o **o
^r I •o
I
(jaiBM unn) amssay Jiy
94
co *n
vo ©
II
II
xi
(50
3
( B ^ ) ssaus
95
cs
cs
a a
co m
vo ©
© ©
II II
X X
cs
•a
S
3
I
I
1
O
( rin/ib[) suig sqi ui uonnqinsiQ yCjxsuaa ^ g
Chapter 5
Prediction of the Flowrate of Bulk Solids from M a s s F l o w Bins
5.1 Introduction
The flowrate of bulk solids from mass-flow bins depends very much on the flow
conditions experienced by the particles near the outlet of the hopper. Considering
the dynamic balance of bulk solids flowing in the outlet region of a hopper, the
forces acting on the particles, for a cohesionless material, are as follows (based on
Carleton's classification^ ™ ) :
i) equivalent inertial force due to the acceleration of the particles
ii) gravity force
iii) resistance by interstitial air pressure gradients
iv) interaction between particles.
Under steady state conditions, the equivalent inertial force is related to the rate of
change in bulk density and cross-section along the flow stream. Most publications
treat the bulk solids as incompressible materials, whereas M c L e a n (1978) [ ],
considered compressible material in his research. Almost all publications consider
the resistance by interstitial air pressure gradients is negligible for coarse free
flowing materials discharging from large orifices. However, an increasing number
of researchers realise that the air pressure gradients do affect significantly the
flowrate of fine bulk solids. Unfortunately, there are different criteria to classify
bulk solids as coarse or fine material. For instance, Carleton (1972) [49] suggested
that the distinguishing criterion for particle size is about 200 (im. However, most
other researchers including Williams (1977) [28] , Crewdson et al. (1977) t 5 0 ] and
Spink (1978) [51] prefer 500 p m .
In comparison the present examination and mathematical model, for predicting the
flowrate, attempts to cover the complete practical bulk solids particle size range by
introducing permeability as a parameter to indicate the effect of adverse air pressure
gradients and by using consolidation-related bulk density to consider both
compressible and incompressible materials. The theoretical analyses indicate that for
the bulk solids under dynamic conditions in a mass flow bin, no matter if it is
compressible or incompressible material, the bulk density and stress gradients only
have insignificant effect on the flowrate. However, permeability and hence the air
pressure gradient generated exercises significant control on the flowrate, especially
for fine bulk solids.
Following assumptions have been considered in the derivation of the model:
i) The flow of bulk solids in the outlet region, as region III in Chapter 4, is
radial; however, it is assumed that the results obtained in Chapter 4 (based
on the consideration of particle flow through a horizontal cross-sectional
plane in region HI) for the air pressure gradient at the hopper outlet can be
applied without causing any significant error.
ii) The wall of the hopper is so smooth that the effect of wall friction can be
disregarded.
iii) The flow of bulk solids is continuous so that the model presented can be
based on continuum mechanics theory.
5.2
Theoretical Model for Predicting the Flowrate
5.2.1 Equation of Motion
The spherical coordinates (r, 6,0) are used for the axisymmetrical hopper shown
in Figure 5.1.
Figure 5.1 Flow from Region HI of a Conical Mass-Flow Bin
The equation of motion can be expressed in the form, according to Bird et al.
(1960)[128],
dVr 1 d,2 , o-e + a* dP /C1,
P V r i r = - - J ¥ ( r o - r )+ ^ ^
+
pgr-^
(5.1)
where gr is a gravitational acceleration in r direction (gr = - g cos 6)
Since the flow is converging, a 'passive type1 of stress distribution will prevail.
Assuming o r is the minor principal stress, o e is major principal stress and adopting
the Haar and van Karman hypothesis which states that the circumferential stress o*
is equal to the major principal stress^1,3,361, the relation between three principal
stresses is given by
oe = o0 =K or (52)
^ . .j L, .j [1,3,27,28,55,58,75] L t . e ,
There is considerable evidence
to suggest that the ratio of the
major and minor principal stresses is constant and can be expressed as a function of
the effective angle of internal friction of the material by,
K = -^i"L«
1 - sin 8
(5.3,
Substituting eqns (5.2), (5.3) into (5.1), and considering the effect of 0 (0 < a)
being small, the equation of motion (5.1) becomes
T7
dV r
do,-
2 (K-l )
dP
,.,.
According to the assumption i) in Section 5.1, the mean stress in region m
assumed in Section 4.2 ii), is rewritten as
'mc
= — - — ^
where r =
then
h x cos a
*- = — a — a r = Ki (r - r*)
; r* = -hj cos a
(5.5)
CT
CTr
3 ( r - r» )
" 2K + 1 K l
<5
and hence
da r
3
"dT~ " 2 K + 1K l
(5
Therefore, eqn (5.4) on substitution becomes
dV r
3Kt f 2 ( K - l ) ,
1
dP
=
^-dT 2KTTL
r
('"«*)-1J-P8--5"
(5
5.2.2 Equation of Continuity
For steady state flow in a mass flow bin, the mass flowrate of bulk solids is
Qp = A V r p
(5
and the flowrate gradient should be equal to zero, i.e.,
£-
(5.
Eqns (5.9) and (5.10) lead to
dA dVr dp
V
r p f + A p - ^ + A V r f =0
or
P ~dT
r
L dr
A dr J
(5
Since the cross-section area A is 2 TC r2 (1 - cos a ) for a conical hopper, then
dA
.
,,
N
-r- = 4rcr ( 1 - cos a )
or
J_dA
2
r
A dr
(5.12)
From eqns (3.10) and (5.5), the bulk density equation will be
fMr-r*)]b2c
P = Po + b l c [
IOOQ—J
then
dp
dF
=
b^
( r - r* ) ( p " P o )
(5.13)
Hence, on substituting eqn (5.13) into eqn (5.11) yields
dV r
2f
02c
. 2pl
or
dV r
b2c
~ ( f t pfJl L
(r-r*)
*]
( P - Po ) +
(5.
5.2.3 Equation for Predicting the Flowrate
Substituting eqn (5.14) into eqn (5.8) yields
(M<^<>-^¥
3K£i
r 22(K-1)
X f
(K-1),
I
dP
2K TT[
7 (r-**)-iJ-pg-5Since at the hopper oudet, r = r0, hence
r
Qp
-(
A y Pout
bfc
,
. , 2poutl
r ^ T ( P o «.-Po)+
3Kt 2(K
2
2K+1
1) 1
1
— J
p g
- ( -S)- ]- - -(f)r = r0
(5
The air pressure gradient at the outlet is
(dP\
_fdP)
where
cosa
is obtained from eqn (4.19)
Therefore
^ d W r = r0
"Q^jA 1 [ 1 -< 1 - K dca ) ^ l c o s a
(5
Equations (5.15) and (5.16) yield on substitution the following mathematical
equation for predicting the flowrate:
anQp +b11Qp-c11=0
(5.17)
where
ail=
( AoT^ut) [ p o u t U - r * ) ( P o u t - P o )+ i
bn =
^ut
[l-(l-Kdea)fout]cosa
CoPoAoPout
cn = g +
(2K ^ D P o u t L
r
° 'J
^
where
K =
1 + sin 5
1 - sin S
bicKsfdo-r"')^©:"!1^
fout=l +
K
3
Po
[
1000 cos a __ 1/hv.
K
i=
3
^
h.
Kdea
b
Po f
1
^
" i-Kdea ic I Timp-Ti* J
rn cos a
Ti m p is determined by eqn (4.29) and r\* = 0 . 9 5 — ^
5.3
Application of Theoretical Model
The theoretical model (5.17) has been applied to predict the flowrates of the bulk
materials detailed in Chapter 6. A complete comparison of theoretical results with
the experimental results will be presented in Chapter 8. In this section, some
phenomena highlighted by the theoretical model are discussed.
5.3.1 The Effect of Surcharge Level on the Flowrate
The effect of surcharge level on the flowrate has been observed for the materials in
the current experiments. The predicted results for the effect of the surcharge level
examined in the experiments are depicted in Figures 5.2 to 5.4.
The theoretical results indicate that for coarse materials flowing from a small outlet
the effect of surcharge level is insignificant; whereas for fine material the effect can
be significant, depending on the outlet size of the bin. M o r e of this dependency is
discussed in Chapter 8 which provides more details of this interesting finding.
Furthermore, Chapter 8 will also compare these theoretical results with
experimental results.
In order to show theoretically that the effect of surcharge level depends on the
particle size of the mixture and the outlet size of the hopper, Sand M D 3 , Sand M 5 ,
Sand M D 1 and Sand M l (with permeability constants of 575.561, 1155.835,
2227.654 and 6517.522 *10" 9 M 4 N" 1 Sec"1, respectively) were used as different
particle-sized materials to check the above results over a wide range of surcharge
level and outlet size. In particular the range of outlet size varied between 0.010 ~
0.050 metre. T h e m a x i m u m outlet diameter was taken as 0.05 m from the
consideration D / D 0 » 3.0 (D=0.145 m ) . The highest surcharge level H / D = 12.0
was considered as the m a x i m u m extent of surcharge likely to be used in practice.
Predicted flowrates are shown in Figures 5.5 to 5.8.
From these results, it can be seen that for fine material (with low permeability
constant), such as Sand M D 3 , the effect of material level in the bin is significant
when it is flowing from the larger oudet In comparison for coarse material (Sand
M l ) the effect is insignificant This flowrate variation dependency is considered to
be due to the retarding influence of interstitial air pressure gradient, as expressed in
eqn (5.17). The predicted air pressure gradients at the hopper oudet corresponding
to these predicted flowrates are plotted in Figures 5.9 to 5.12. These figures show
that the interstitial air pressure gradient generated by the flow of fine material at the
outlet is greater than that for coarse material. For instance, Sand M D 3 experiences a
negative air pressure gradient about 6 - 1 6 times higher than Sand M l does in the
range of outlet sizes and surcharge levels considered.
It is also shown, from figures 5.5 to 5.8, that the rate of decrease in flowrate with
AO,
respect to material level, —rf-, decreases as the level increases. It is believed that as
A(g)
Ap
the surcharge level increases, the rate of increase in bulk density, ~ | j ~ , decreases
A(g)
AK^ea
and the rate of increase in dynamic deaeration coefficient, — g — , decreases (refer
A%)
to eqn (9.19)). This leads to a reduction in the rate of increase experienced by air
pressure gradient at the hopper outlet
A((f)r=r0)
p
(refer to term b
A(g)
u
in eqn
(5.17)), as shown in Figures 5.9 to 5.12. This reduction of increasing rate in
pressure gradient results direcdy in a reduction of rate at which the particle flowrate
decreases, as the surcharge level increases (refer to Figure 5.5 to 5.8).
0.12
• = c = s=i t=r
J
— —
0.10
o^
g
o
E
Legend
0.08 —
,™™r~
B
Alumina
• P V C Powder
• Sugar
0.06
0.04
1
2
3
H/D Ratio
a) Qp vs. H/D (from 0.02 m Outlet)
1.00
8
0.80
Q/O,
Legend
0.60
0
o
E
1
I
Alumina
• P V C Powder
• Sugar
0.40
0.20
2
3
4
H/D Ratio
b) Qp vs. H/D (from 0.0445 m Outlet)
Figure 5.2 Predicted Flowrate Varying with H/D Ratio for Alumina,
PVC Powder and Sugar
0.20
*
Material and Median
Particle Size
0.15
(Hm)
Sand Ml d50=370
Sand M2 d50=310
• SandM3d50=200
X SandM4d50=200
4 SandM5d50=155
0.10
£
0.05
Sand M 6 d 5 0 = 80
0.00
2
3
H/D Ratio
a) C^ vs. H/D (from 0.02 m Outlet)
15
Material and Median
Particle Size
i—I
8
(Hm)
1.0
SandMld50=370
rr**
i~—-"••
i
Sand M 2 (150=310
• SandM3d50=200
x Sand M 4 d 5 0 =200
SandM5d 5 0 =155
0.5
9
0.0
——e
©
2
-o
Sand M 6 d 5 0 = 80
3
H/D Ratio
b) Qp vs. H/D (from 0.0445 m Outlet)
Figure 5.3 Predicted Flowrate Varying with H/D Ratio for Sand M l to Sand M 6
0.16
Legend
H
SandMDl
• SandMD2
• SandMD3
* SandMD4
1
2
3
H/D Ratio
a) Qp vs. H/D (from 0.02 m Outlet)
1.20
Legend
a SandMDl
• SandMD2
• SandMD3
* SandMD4
H/D Ratio
b) <^> vs. H/D (from 0.0445 m Outlet)
Figure 5.4 Predicted Flowrate Varying with H/D Ratio for Sand M D 1 to M D 4
Oudet Size ( m )
Db=0.050
Do=0.0445
1^=0.040
%=0.035
Db=0.030
Do=0.025
Dfo=0.020
Db=0.015
D,j=0.010
6
H / D Ratio
8
Figure 5.5 Predicted Flowrate vs. H / D Ratio for Sand M D 3
Oudet Size ( m )
Eo=0.050
Do=0.0445
Eb=0.040
Eb=0.035
Efo=0.030
Do=0.025
Efr=0.020
Efr=0.015
D0=O.O1O
6
H / D Ratio
8
10
Figure 5.6 Predicted Flowrate vs. H / D Ratio for Sand M 5
Outlet Size ( m )
Eo=0.050
Do=0.0445
E^O.040
D^O.035
Db=0.030
D^O.025
Efo=0.020
1^=0.015
D(p0.010
H / D Ratio
Figure 5.7 Predicted Flowrate vs. H / D Ratio for Sand M D 1
2.0
•
r——'
T—
i
1—
1.81
—i 1
1.6
OutletSize ( m )
EQ=0.050
1.4
Do=0.0445
1?
E^j=0.040
1.0'
B
0.8
D(f0.025
06
E^f=0.020
0.4
k
0.0 *
i
C1
E^O.035
Efo=0.030
•
'
—*
*
iJtLx2
*4
1
X
6
— a —
l"K 1
8
10
!
l2
H/D Ratio
Figure 5.8 Predicted Flowrate vs. H / D Ratio for Sand M l
M
Do=0.015
Dg=0.010
-9500
Outlet Diameter (m)
i
-10500 -
§
E^O.010
Do=0.0l5
Eu=0.020
%=0.025
E„=0.030
-11500
CA
«3 -12500
D(F0.035
Dfo=O.040
Eu=0.0445
D<j=0.050
-13500
-14500
4
6
H / D Ratio
8
10
12
Figure 5.9 Predicted Air Pressure Gradient at Hopper Oudet
vs. H / D Ratio for Sand M D 3
-3000 r — •
1
•
i
4000
S
•
Outlet Diameter (m)
Eb=0.010
DpO.015
E^O.020
E^O.025
Eb=0.030
D(j=0.035
Eb=0.040
Eb=0.0445
D(j=0.050
-5000
-6000
•go
o CO
-7000 •
-8000 -9000
4
6
H / D Ratio
8
Figure 5.10 Predicted Air Pressure Gradient at Hopper Oudet
vs. H / D Ratio for Sand M 5
-1000
.
Oudet Diameter (m)
-x4
6
Eb=0.010
Do=0.015
Efr=0.020
Dfc=0.025
1^=0.030
D(j=0.035
E^O.040
Efo=0.0445
D(j=0.050
8
H/D Ratio
Figure 5.11 Predicted Air Pressure Gradient at Hopper Oudet
vs. H / D Ratio for Sand M D 1
-500
Oudet Diameter (m)
g
-1000
Eb=0.0l0
Do=0.015
Eb=0.020
Dfo=0.025
1^=0.030
D(j=0.035
E^=0.040
1^=0.0445
D(j=0.050
36
it
"gO -2000 -
4
6
H/D Ratio
Figure 5.12 Predicted Air Pressure Gradient at Hopper Oudet
vs. H / D Ratio for Sand M l
The generation of the air pressure gradient requires the flow of particles. It can be
seen, from eqn (4.19) and eqn (5.17), that the air pressure gradient at the hopper
oudet is proportional to the particle flowrate. However, it is also noted, from eqn
(4.19) and eqn (5.17), that the air pressure gradient increases with a decrease in
particle permeability, as discussed in Section 4.7.2. For bulk solids with very low
permeability, the very high pressure gradients generated near the oudet of a bin
reduce the flowrate dramatically and m a y cause difficulties to control the particle
flow, if, for example, fluidisation effects occur.
For sufficiendy fine material (with low cohesion), the air pressure gradient
generated at the hopper oudet by flowing particles m a y be great enough to fluidize
the particles at the hopper oudet T h e fluidisation results in more voidage between
the particles which releases the high pressure gradient and increases the flowrate
quickly (flooding) until the high air pressure gradient builds up again; the resulting
periodic flow behaviour is illustrated in Figure 5.13. This periodic flow behaviour
was observed in the experiments and will be detailed in Section 8.3.
However, the flowrate prediction model eqn (5.17) does not include this effect.
Further work needs to be done on the model to include the effect of the fluidisation.
5.3.2 The Effect of Oudet Size on the Flowrate
It is a clear that the mass flowrate of material increases with the increase in outlet
size. This variation is described by theoretical predictions shown in Figures 5.2 to
5.12. In addition, these figures show the effect of oudet size on the sensitivity of
the influence of surcharge level. For convenience of describing these effects,
Figures 5.5 to 5.12 can be replotted as flowrate or pressure gradient at outlet
against the oudet size, as shown in Figures 5.14 to 5.21.
From Figures 5.14 to 5.17, it can be seen that for a bulk solid flowing from a bin
the effect of the outlet size on flowrate is significant. They also show that •%?£•
increases with oudet size. Figures 5.18 to 5.21 illustrate that the negative air
pressure gradients at the hopper oudet increase with oudet size, indicating that the
increase in the pressure gradient is influenced by the increase in particle flowrate.
However, the influenced pressure gradient by particle flowrate will affect the
particle flowrate until a dynamic balance between them occurs. The effect of air
pressure gradient on flowrate is discussed in the sensitivity analysis of parameters
to flowrate reported in Section 5.4.1.
Build up of
the High Pressure Grad lent
at oudet
Release of
the High Pressure Gradient
at Oudet
•
t
Discharging
at L o w Flowrate
-•
Flooding
Fluidisation
Figure 5.13 Periodic Flow Behaviour for Fine Material
under Fluidisation Condition
:
0.80 I
0.01
•
1
r
0.02
0.03
0.04
0.05
Diameter of Oudet D 0 ( m )
Figure 5.14 Predicted Flowrate vs. the Hopper Oudet Diameter for Sand M D 3
0.01
0.02
0.03
Diameter of Oudet D Q
0.04
0.05
(m)
Figure 5.15 Predicted Flowrate vs. the Hopper Oudet Diameter for Sand M 5
0.00
0.01
0.02
0.03
0.04
0.05
Diameter of Oudet Dft (m)
Figure 5.16 Predicted Flowrate vs. the Hopper Oudet Diameter for Sand M D 1
8
*
o
O
E
1
0.02
0.03
0.04
0.05
Diameter of Oudet D0 (m)
Figure 5.17 Predicted Flowrate vs. the Hopper Oudet Diameter for Sand M 1
-9500
Surcharge Level
(H/D Ratio)
H/D=0.07
H/D=1.5
H/D=3.0
H/D=4.5
H/D=6.0
H/D=7.5
H/D=9.0
H/D=10.5
H/D=12.0
-14500
0.01
0.02
0.03
0.04
0.05
Oudet Diameter D 0 ( m )
Figure 5.18 Predicted Air Pressure Gradient at the Hopper Oudet
vs. the Hopper Oudet Diameter for Sand M D 3
-3000
8I
-4000
Surcharge Level
(H/D Ratio)
-5000
H/D=0.07
H/D=1.5
H/D=3.0
H/D=4.5
H/D=6.0
H/D=7.5
H/D=9.0
H/D=10.5
H/D=12.0
o
-6000 -
I
-7000 -
Z
-8000 -9000
0.01
0.02
0.03
0.04
0.05
Oudet Diameter D 0 ( m )
Figure 5.19 Predicted Air Pressure Gradient at the Hopper Oudet
vs. the Hopper Oudet Diameter for Sand M 5
-1000
14
rn <?
Surcharge Level
(H/D Ratio)
-2000
H/D=0.07
-3000
H/D=1.5
H/D=3.0
-4000 •»
H/D=4.5
H/D=6.0
-5000 •
H/D=7.5
H/D=9.0
2
-6000
cd
-7000
0.01
H/D=10.5
H/D=12.0
0.02
0.03
0.04
0.05
Oudet Diameter Dft ( m )
Figure 5.20 Predicted Air Pressure Gradient at the Hopper Oudet
vs. the Hopper Oudet Diameter for Sand M D 1
-500
Surcharge Level
(H/D Ratio)
c
U
-1000
1
H/D=0.07
a;
H/D=1.5
M
-1500 •
H/D=3.0
H/EM.5
H/D=6.0
H/D=7.5
-2000
H/D=9.0
•8*
H/D=10.5
-2500
0.01
H/D=12.0
0.02
0.03
0.04
0.05
Oudet Diameter D Q ( m )
Figure 5.21 Predicted Air Pressure Gradient at the Hopper Oudet
vs. the Hopper Oudet Diameter for Sand M l
5.3.3 The Effect of Permeability Constant on the Flowrate
Permeability can describe the relationship between interstitial air pressure gradient
and particle characteristics such as particle size distribution, particle shape and
particle roughness, as mentioned in Chapter 3. Since the negative air pressure
gradient plays an important role in retarding the material flow, especially for fine
material, the permeability can be used to display the influence of the particle
characteristics on the particle flowrate. F r o m the theoretical model presented in
Chapter 4 and Section 5.2, a zero stress was assumed to occur at r* = 0.95 r0, the
permeability at the hopper oudet is very close to the permeability constant at the
lowest compaction C n discussed by eqn (3.15). For convenience, a Q p - C 0
diagram is used to describe the variation of the particle flowrate for different bulk
solids.
For the sand mixtures used in experiments detailed in Chapter 6, the predicted
flowrates are illustrated in Figure 5.22. T h e theoretical results indicate that the
flowrate of material increases with an increase in permeability. T h e increasing rate
of flowrate reduces as the permeability constant of material increases. After a
particular value of permeability constant, the effect of negative air pressure gradient
on the particle flowrate can be considered as insignificant; in this case, the flowrate
of particles depends on its bulk density. T h e criterion for this particular value will
be discussed in Section 7.5.
These theoretical results were verified by experimental results, which are detailed in
Chapter 7 and Chapter 8.
120
o
o
o
o
o
o
o
oo
C/3
s
in
o
o
oo
• *
u
1
u
c u
CO
3
C/5 ^H
1
C
•4-»
a
rm>
o
o
o
^^\
>—i
55
*—/
1<u o
OH
»-H
*
•a
3
§
&
0-(
wo
'^
>
,o
<4-l
8
^>-
o
E
1
o
o
o
cs
K
•
(D3S/33[) °£) ajBiMOu pajorpay
o
5.4
Comparison of Theoretical Results with Experimental Results
5.4.1 Sensitivity Analysis of the Various Terms in Theoretical Model
In equation (5.17), the flowrate expression includes five terms:
i)
cross - section area gradient
ii)
gravity
g
iii)
bulk density gradient
(-p)r=r
• x
iv)
•,
A-
•
stress gradient
(7T~ ) r=r
,d<yr
2(K-
(-^
f
l)qrx
r
) T=TQ
and
v)
ah* pressure gradient
(-j-) r_r
It is obvious that the first two terms are essential for material flow from a
converging channel by gravity. The effect of terms iii), iv) and v) can be judged
qualitatively from Equation (5.17) as follows: consideration of the bulk density
gradient in eqn (5.17) increases the value of &n and then reduces the magnitude of
predicted flowrate, namely, the bulk density gradient resists the flow; consideration
of the stress gradient in eqn (5.17) increases the value of cn and then enlarges the
value of predicted flowrate, i.e., the stress gradient makes the flow increase; the
negative air pressure gradient plays a role in reducing the discharge rate. Most
researchers consider the effects of the bulk density gradient and stress gradient to be
insignificant whereas the effect of the air pressure gradient for fine material is
considered significant
A series of quantitative analyses were conducted to examine these effects. Sand
M D 3 , Sand M 5 , Sand M D 1 and Sand M l were used as the bulk solids samples. In
particular the effect of each individual term was examined by setting the relevant
term in eqn (5.17) to zero. It is only necessary to mention that for the analysis of
'disregarding the air pressure gradient', eqn (5.17) becomes
2 , 1 ,2 ^ 2_
r
o { An o
u
A
^m&x
"g
0 Pout
i.e., which yields
Qpmax - j P„ Do2"5 ^-f^- (5.18)
This equation is the same as that provided by Carleton (1972)[49J for coarse
material.
The results, as shown in Figures 5.23 to 5.26, verify that the effect of bulk density
gradient and stress gradient on the flowrate of bulk solids from the mass flow bin
can be disregarded, while the effect of the air pressure gradient m a y be significant.
The effect of the air pressure gradient on the flowrate becomes more significant
with decreasing material permeability.
"5b
M
ex
&
Q
0.01
0.02
0.03
0.04
0.05
Oudet Diameter D Q ( m )
Figure 5.23 Comparison of the Effects of Density Gradient, Stress Gradient
and Air Pressure Gradient at the Hopper Oudet for Sand MD3
3
M
Ci.
CT
ft
0.02
0.03
0.04
0.05
Oudet Diameter D 0 ( m )
Figure 5.24 Comparison of the Effects of Density Gradient, Stress Gradient
and Air Pressure Gradient at the Hopper Oudet for Sand M 5
I
PH
&
13
0.02
0.03
0.04
0.05
Oudet Diameter D 0 ( m )
Figure 5.25 Comparison of the Effects of Density Gradient, Stress Gradient
and Air Pressure Gradient at the Hopper Oudet for Sand M D 1
2.00
o.oo'
o.oi
0.02
0.03
0.04
0.05
Oudet Diameter D Q ( m )
Figure 5.26 Comparison of the Effects of Density Gradient, Stress Gradient
and Air Pressure Gradient at the Hopper Oudet for Sand M l
5.4.2 Comparison of Equation (5.17) with Other Flowrate Models
In this section, the theoretical results have been calculated to compare the
experimental results and the results predicted by published models for the materials
used in the experiments. In this comparison, the published theoretical models which
are considered the best existing models for fine material are as follows:
i) Nedderman, Ttiztin and Thorpe Method[56] [ eqn (2.5) ]
ii)
Original Carleton Method [ 4 9 ] [ eqn (2.4) ]
iii)
Carleton - Drag Force Modification^ [ eqn (2.9) and eqn (2.10) ]
iv)
Carleton - Permeability Modification^ [ eqn (2.9) and eqn (2.11) ]
v)
Modified Johanson Method^ 115 ^ [ eqn (2.6) and eqn (2.8) ]
In iii) to v), the estimation of the major principal stress was evaluated using the
procedure described by Arnold et al. (1980/ \ The bulk density and air pressure
gradient at the oudet in model i) were predicted by the present model developed in
Section 5.2. Figures 5.27 to 5.30 show the comparisons of the theoretical results
with experimental results for sand mixtures from the 0.02 m and 0.0445 m oudet at
H / D = 0.07 and H / D = 1.5 surcharge level, respectively.
In Figures 5.27 to 5.30, the comparisons show that equation (5.17) provides the
most accurate prediction relative to the other models examined. In particular the
worst fit occurs by the original Carleton method, especially for low permeability
materials. This is because eqn (2.4) only involves the median particle size and does
not include the effect of particle size distribution. This indicates that the median
particle size alone is not sufficient to describe fully the flow properties of the
particle mixture. In comparison the Carleton - Drag Force Modification predicts a
much lower flowrate. Hence it is considered that the introduction of the voidage
function e" 4 7 over-estimates the drag force, although it is well known that the drag
force acting on a particle in a dense solids-gas mixture is larger than that on a single
particle. For example, if voidage e is 0.5, the drag force will increase 26timesby
introducing e
.
On the other hand, the model of Nedderman et al.[561 provided an over-estimation
of flowrate. More comparisons of eqn (5.17) and Nedderman et al.'s method are
carried out with the experimental results obtained by Crewdson et alJ50J and
Head^551. The estimated flow properties of the bulk solids and bin geometry details
are presented in Table 4.2. The comparative results are listed in Table 5.1.
Table 5.1 The Comparison of Eqn (5.17) and Nedderman et al.'s Method with
Head and Crewdson et al.'s Experiments
Experimental
Flowrate
(kg/sec)
Predicted
Flowrate
by Eqn (5.17)
(kg/sec)
Predicted
Flowrate
by Eqn (2.5)
(kg/sec)
Crewdson et al.
Sand 355 - 422 ^ m
0.015
0.20
0.029
Crewdson et al.
Sand 212 - 300 ^ m
0.015
0.019
0.026
Crewdson et al.
Sand 90 -106 ^ m
0.010
0.011
0.015
Head-0.02805 m Oudet
Sand 186 - 268 ^ m
0.406
0.380
0.585
Head-0.0304 m Oudet
Sand 186 - 268 ^ m
0.477
0.462
0.712
Head-0.02805 m Oudet
Sand 86 -130 >im
0.230
0.186
0.283
Head-0.0304 m Oudet
Sand 86 -130 ^ m
0.274
0.221
0.337
case
127
co
oo
.2 ^
ON
J
e
~ ft
1 12 5 B
I
21
Sill
OoQ»
o
o
o
c
-*d
o
o
o
3
8
05
^^
*o
Ja
e24
VO
51m
S.
a«
Ed
9
g4)
•o
•Vo !1
£
"3
o
o
o
>n
o
o
o
^*"*
t*
*^
o
o
o
en
o
o
o
CM
o
o
o
Qi
§ 8
&5 CO
3 qT
ft
*
I
ft
9
Qi
o
!
O
1
•g
!
cs
»o
o
©
cO
d
8
d
o
d
( 03S/S5J) " £) 9JBJMOTJ SSBJ^
d
d
K
128
( D9S/S3[ )d o aiM^ou SSBJ^
129
d
n
S
o
il_
3
Qi
8
1
co -c
8
•
^
a I
•i o>
o
ft
PH
ft
I—<
23
3
Qi
I
I
1
•css
K
(03S/331)
£) ajBIAVOtl SSBJ^
130
ft
(03S/xb[) £) 3JBJM0TJ SSBJ^
The results in Table 5.1 confirm the over-prediction associated with Nedderman et
al.'s method. It is noted that Nedderman et al.'s results are always greate
those predicted by eqn (5.17), because of the following discrepancies betwee
two models.
Both models have one format, i.e.,
where Qpmax is the maximum flowrate, applied as (T~ ) ~ ^
g Pout ^ ™" h = r0
Tj is the weighting factor of the influence of the pressure gradient.
For the model of Nedderman et al.
n *-l0 D2.5|- 1 + K g 11/2
^>max - 4 P o u t M ) [ 2 ( 2 K - 3 ) sina j
and
r »
1 2 K
5.in8-l
2KO
" ! 1 + 3 sin 5
For Equation (5.17)
**
Qpmax** = Op™3* in equation (5.18)
and
r/* = 1
piy;
The ratio of the two maximum flowrates is
Q
Pmax
?A/
l
QD
** ~ V
^Pmax
1 + K
2 ( 2 K - 3 )
2
" /
.
=
*y 5 sin 8 - 1
(5.21)
For practical values of 8 the ratio expressed by eqn (5.21) will be greater than 1 and
Tj will be less than 1. For instance:
Q
*
Vvm
if 8 = 40°, then
±±
= 1.344
and
Tx* = 0.756
*
W = 1.633
and
r x * = 0.600
^Pmax
Q
if 8 = 30°, then
Pina
^Pmax
This means that Nedderman et al.'s model will over-estimate the m a x i m u m
flowrate and under-estimate the retarding effect of the adverse pressure gradient,
resulting the prediction of higher flowrate.
Furthermore, for coarse bulk solids, the flowrate predicted by eqn (5.18) is very
close to that predicted by Brown's theory, eqn (2.2), with the percentage error
described in eqn (5.22). The percentage error between the flowrates obtained by
two models for hopper half angle within the range from 5 to 30 is plotted in
Figure 5.31. This comparison indicates that the eqn (5.18) is reliable, since the
Brown equation is one of the most successful for coarse bulk solids (as mentioned
in Section 2.1).
the flowrate predicted by eqn (2.2) _ /
2
the flowrate predicted by eqn (5.18) ' V l + cos a
._ ..
ft
I
I
o
10
15
20
Hopper Half Angle a (degree)
Figure 5.31
30
The Percentage Error Between the Flowrates Obtained by
Current Model and Brown's Theory for Coarse Bulk Solids
The main conclusions can be drawn from above comparisons as follows:
• the flowrate prediction model, eqn (5.17), provides the most accurate
prediction relative to the other models examined;
•
by using the particle permeability as a parameter to account for the effect of
negative air pressure gradient at the hopper oudet on particle flow, the
current model, eqn (5.17), is valid for both fine and coarse bulk solids; for
the coarse bulk solids, the prediction of eqn (5.17) is almost same as
Brown's prediction;
•
eqn (5.17) includes the effect of compressibility of bulk solids on the
particle flow, hence, it is valid for both compressible and incompressible
bulk materials;
•
eqn (5.17) is only one model which describes the effect of material
surcharge level on the flowrate.
However, the numerical methods have to be used to solve the eqn (5.17). A
relatively simplified model will be proposed in Chapter 9 to predict the particle
flowrate from mass flow bins in semi-empirical fashion.
Chapter 6
Experimental Facilities and Test Bulk Materials
6.1 Introduction
To compare theoretically predicted flowrates to actual flowrates accurately, the
steady flowrates was necessary to be measured. These measurements should be
made noting that the manner in which bulk solids flow from a mass flow bin is
dependent on the geometric variables of the bin and on the flow properties of the
discharging material.
The first set of experiments in the study were carried out on a single hopper
apparatus. In these experiments, it was found that the flowrate changed during
discharging and it also changed with different levels. The same phenomenon was
observed in the experiments carried out by Smith in 1978^ \ as shown in Figure
6.1. These results raised the problem that the level of material perhaps affects the
flowrate, although m a n y researchers^4'11"14'17'19,24'13^ considered that the
flowrate is independent of the 'head' of the material.
2.2
-1
1—1
1
1— — | — " T
1
•
X
u
•5 1*
•
BX
B
' —r—
«
%
M
«
i
1.6
\A
1.2
X
K
B
•
X
X
-
Test 1
Test 2
Test 3
1.0
•
0.8
••
-2
Figure 6.1
•
•
A
0
--•— _J
L.
1
i
i_
L_
•
2
H/D Ratio
Mass Flowrate Q D v. H/D Ratio (Smith 1978[53])
T o avoid this effect further experiments were conducted with two hoppers, one
positioned above the other, the steady flowrates were measured by keeping the level
of material in the bottom hopper constant With this arrangement the effects of three
factors were studied: i) the effect of the material surcharge level on the flowrates by
measuring the flowrates at the different material levels; ii) the effect of oudet
diameter of the bottom hopper by using two different oudet sizes; iii) the effect of
median particle size and particle size distribution on the flowrate.
6.2 Test Rig for Measuring Mass Flowrate
The apparatus consisted essentially of two bins suspended by three force
transducers, as shown in Plate 6.1 and Figure 6.2. O f the two bins, one is called a
test bin, as depicted, and the other a storage bin. W h e n solid material in the test bin
flowed out, it was replaced by solids from the storage bin above. In other words,
the purpose of using this double-bin apparatus was to keep the level of material in
the test bin constant, allowing a steady flowrate to be measured with greater
accuracy.
Both bins had vertical sections made of perspex to allow visual monitoring of the
surcharge level during tests, whereas the hoppers or converging sections were
m a d e of polished galvanised steel. The bins were filled by manual material
transport. The location of the apparatus and the method for recirculating the
materials limited the capacity of the bins used. T w o test bins were m a d e with
different outlet sizes. The fixed height of the cylindrical section of the test bins was
0.4 m . A removable cylindrical perspex section was used to extend the height in the
test bin for higher level tests. With the help of a tapered standpipe (2 or 3 degree)
for lowest level measurements, the m a x i m u m difference in level obtainable on the
136(-)
136(-)
Plate 6.1 Test Rig for Measuring Particle Flowrate
bin hanger was 0.66 m , which was about 4.2 times the diameter of the cylindrical
section.
<M45(I.D.)
Filling funnel
Transducer
Storage bin
Bin hanger
(detailed in
Figure 6.3)
Test bin
Figure 6.2 Schematic of the Double - Bin Apparatus
The bin hanger consisted of three steel strips suspended vertically by steel wires
above and below, six turnbuckles at the ends of the wires and two bin holders, as
shown in Figure 6.3. T o vary the level of material in the test bin, the steel strips
m a d e with holes allowed the storage bin to be set at different levels. B y adjusting
the turnbuckles, the storage bin and test bin were aligned vertically using a spirit
level to ensure that the flow of bulk solids would be vertical.
S o m e principal dimensions of the bins used are given in Table 6.1.
137
To Transducers
4
Turnbuckles
Steel Strips
Steel Wire
Bin Holders
Turnbuckles
Figure 6.3 The Schematic of Bin Hanger
Table 6.1
Parameters
Bin Details
Oudet 1
Hopper half angle
a (degree)
Diameter of oudet
Do ( m )
Oudet 2
15
0.0445
0.02
Diameter of cylinder D
(m )
0.145
Range of surcharge H
(m)
0.0 ~ 0.66
138
In order to keep the level of bulk material in the test bin constant during discharge,
the flowrate of material from the storage bin must be greater than or equal to that
from the test bin.
The three force transducers used were Interface SM-50 transducers, each having a
m a x i m u m capacity of 50 lbs. The circuit arrangement for the three transducers is
shown in Figure 6.4. The chart recorder was a Y E W type 3066 pen recorder. The
instruments were connected in a simple w a y as shown in Figure 6.5.
The gate of the oudet from the test bin was hinged. This hinge was installed 0.15 m
away from the oudet, so that the gate had no effect on the flowrate during steady
discharge. The gate was m a d e of galvanised steel and built as light as possible to
minimize any inertia effects when opened.
0
I
m
I
A.C.
Figure 6.4
Transducer Circuitry
•o
Transducers
4TT
\
(
/•
A
_n/
-c_
-rif
-LZ
h\1
'
I-J
Transducer
Conditioning
Chart
Recorder
Unit
Figure 6.5
Instrumentation Schematic
T o minimize dust emission, skirts were mounted between the filling funnel and the
storage bin, and also between the storage bin and the test bin. Attention was paid to
ensuring that no tensile force existed in the skirt between the filling funnel and the
storage bin.
6.3 Measurement of the Flowrate
The chart recorder was used to record the mass of solids flowing from the test bin
versus discharge time. To get an accurate flowrate and reduce the effect of
experimental deviations, at least four runs were conducted for each mixture at each
level. At the beginning and end of the test, the recorder system (including
transducers and instruments) was calibrated to check the calibration constants.
6.3.1 Cahbration Procedure
The calibration procedure was carried out under the test bin as shown in Figure 6.6.
The total calibrating mass was 20 kg. The mass increments are listed in Table 6.2.
140
Table 6.2 The Calibration Mass Increments
Weight
No
Mass (kg)
0
Weight Hanger
0.500
1
10 lbs Weight
4.535
2
10 lbs Weight
4.535
3
10 lbs Weight
4.535
4
2 kgs Weight
1.955
5
2 kgs Weight
2.000
6
2 kgs Weight
1.940
Total
20.000
Mass
Bin Hanger
Test Bin
Bin Holder
Weight Hanger
Weights
Figure 6.6 Calibration Arrangement
At the beginning of the test, the calibration procedure was conducted as follows.
Firstiy, the scale was established by putting on the whole 2 0 kg weight and
adjusting the scale to give measurement pen deflection consistent with the scale
divisions. The linearity of the system was checked, as at the end of the tests, by
taking away the mass increments one by one. The typical diagram of the recorder
output is displayed in Figure 6.7. Figure 6.8 shows a typical calibration result for
the recorder system.
Mass Increments
to
20
No. 6
No. 5
No. 4
V)
93
2
No. 3
10
No. 2
No. 1
No.Or-1
0
t
Recorder Movement
a). At the Beginning of Test
b). At the End of Test
Figure 6.7 Typical Calibration Recorder Output
Actual Load ( k g )
Figure 6.8 Comparison of Indicated Load with Actual Load
6.3.2 Test Procedure
T o prevent the flowrate from being affected by different loading conditions, say
different filling speed, before recording the discharge process, the initial loading
conditions in the test hopper created by filling were destroyed by use of the
following test procedure:
a), bulk material was filled into the bins up to the required level via the filling
funnel,
b). the gate was opened to predischarge the material in the test bin until the head
of material in the storage bin was near to the oudet of the storage bin or until
the material first filled into the test bin flowed out,
c). refill the material,
d). for the higher levels, say H £ 0.45 m , steps b) and c) were repeated twice,
e). start the chart recorder and set the datum line,
f). open the oudet to start discharge,
g). stop the chart recorder w h e n the bins were empty, and then prepare for the
next run.
After taking the chart results from the chart recorder, a Talos digitizer was utilized
to read the data.
6.4 Processing the Flowrate Data Measured
For steady state flow, the tensile force on transducers stood for weight of material
in the bins, i.e., the curve plotted on the chart recorder was the weight curve of the
bulk solids discharging. T h e orthogonal polynomial method'-
\ which is a
method based on the least square analysis to fit efficientiy the experimental data,
was used to obtain the approximation of weight curve. With this method the degree
of polynomial was created by the program itself. The flowrate of material was then
evaluated by taking the time derivative of this polynomial.
6.4.1 Original Fitting
The polynomial in t used to fit the experimental data is:
n
W(t) = ^ p i t
i=0
i
+ E
where n the degree of polynomial
{Pi }
E
coefficients of every term
error term
(6.1)
Neglecting the error term, the weight discharging is:
n
W(t) = X P i l i
(6.2)
i=0
Differentiating eqn (6.2) with respect to t, the flowrate of the bulk material from the
test bin can be obtained, i.e.,
n-1
Q
P
=
-dT- = X a i t i
<6-3)
i=0
where
Oj = (i + 1 ) p i + 1
6.4.2 Problem of Initial Fitting and its Improvement
The error term E in eqn (6.1) is a combination of errors due to digitising and
numerical curve fitting. Since the fit processing is based on the least square
analysis, this error can be ignored for weight curve fitting. However, for evaluating
the flowrate, the differentiation m a y cause an increase in the magnitude of this
error. T h e flowrate curve obtained often appeared as a w a v y line. In some cases,
for example for coarse bulk material of which the discharging weight varied with
time linearly, as shown in Figure 6.9 a), the wave became rougher, as shown in
Figure 6.9 b). In particular the flowrate of the 4th run m o v e s very roughly. T o
obtain more accurate flowrate value, the initial fitting method needs to be improved.
It is noted that the error largely occurs at digitized points. It is acceptable to state
that the initial curve fitting procedure only gave a general variation of the flowrate
overtime.Furthermore the curve roughness was exasperated by the fact that if the
flowrate at tj goes up slighdy, then at some subsequent point it should go down.
21
If
1
1
1
1
1
I
1 "
1
18
""
15
c
'Eb
12
-
Run 1
Run 2
Run 3
Run 4
Xi
s
^y^
Xi
60
I
< ^
1
_1
_L
1
1
6
8
10
i
12
I
14
-
I
16
18
Time Recorded (sec)
Weight Curves Measured
a)
1.8
.3*
T — • — i — ' — i — • — i — • — i — • — i — • — i — * — r
1.6
bfc
run 4
tfSTSm,
1.0
I
.
I
.
4
j_
6
8
10
12
14
16
18
Time Recorded (sec)
b)
Flowrate Curves Obtained
Figure 6.9 Flowrate Prediction for Four Test Runs at O n e Surcharge Level
for Coarse Sand (Sand M l )
T o smooth out the variation in flowrate over time, cubic splines were introduced.
The basic concept is described as follows:
It is assumed that there is a smooth function, f(x), which is subject to
*n
i)
J [ f ( x ) ]
x
dx
is minimized
(6.4)
l
n
ii)
] £ [ f ( Xi ) - ( Q p )j f
i=l
<. Constant
(6.5)
Eqn (6.4) presents the smoothness of function f(x) and eqn (6.5) has a concept of
the least square method- in so far as finding the values of constants in the chosen
equation f(x) that minimize the s u m of the squared deviations of the observed
values from those predicted by the equation.
Eqn (6.5) can be expressed in terms of the standard deviation.
i=l
where d ( Q p ) j are smoothing factors.
Actually, eqn (6.4) and eqn (6.6) raise the problem of minimizing the functional
v[f( x )]. Using Variational Analysis, the reasonably smooth function consists of
several sectionalized cubic polynomials which have continuous f(x), f (x) and f (x)
at points of intersection (Reinsch 1967
[132]
).
In this program, the smoothing factors were taken as d( Q p )j = s (i=l, 2,..., n). If
smooth factor s = 0 or s is small, say under 0.001, or 0.005, the results will be th
same as before. If s is bigger, say 0.1, or 0.5 (depending on how rough the
original data were), the results after smoothing will be on a straight line. Usually
the flowrate curve was smoothed till the waves disappeared. Only for the runs
where the weight variations were virtually straight lines, was a large s used. In su
cases, the straight lines after smoothing were almost horizontal.
The flowrates after smoothing the original flowrates shown in Figure 6.9 b), are
plotted in Figure 6.10. The flowchart of the final fitting procedure is briefly show
in Figure 6.11.
» i • i • i—'—i • i •—i—' i • i « i
run 1
run 2
"'" run 3
run 4
1
"0
»
«
2
4
•
•
6
L-
1
8
I — 1 _ - J — 1 — • — 1 — • — 1 — 1 —
10
12
14
16
Time Recorded (sec)
Figure 6.10 The Flowrate of Every Run after Smoothing
18
148
Start Fitting Process)
Q
H
Input Initial Data
I
Plot the Weight Curve for Every Run
Calculate and Plot the M e a n Value of Weight
Original Fitting Process and Plot the Results
Select the Line
3
Input the Smooth Factor
Smooth Process and Plot the Results
Yes
Yes
Figure 6.11
Flowchart of Fitting Procedure
6.4.3 Typical Flowrate Measurements
The typical flowrate observations with river sand (minus 350 |im in particle size
with a median particle size of 2 0 0 u m ) are shown in Figure 6.12, for different
loading conditions.
For the loading condition called '8 kg loading' in Figure 6.12, the level of material
in the test bin reached right up to the outlet of the storage bin, that is, the
performance was the same as in the single bin apparatus. For the three other tests
(12kg, 16kg and 20kg loading), different levels of sand were attained within the
storage bin for each individual run.
Initial Loading effect
8
I
o.
O
i
Figure 6.12
Time T (second)
Comparison of Flowrates for Different Loads Measured
on the Double Bin Apparatus with 0.0445 m Oudet
The results of the experiments indicate that using a double bin apparatus enables a
steady flowrate to be observed for a significant period oftime.Further advantages
of theriginclude its ability to reduce the effect of filling velocity and storage time
on the resulting flowrate from the bin oudet.
The results of flowrate measurements will be displayed and discussed in Chapter 7
and Chapter 8.
6.5 Preparation of the Bulk Solids Mixtures
The bulk materials used for the experiments were river sand, alumina, PVC
powder, sugar and glass beads. Sampling techniques to obtain the most accurate
representative samples of the bulk solids, presented by Allen ( 1 9 8 1 ) ^ , were
employed.
6.5.1 The Range of Different Sand Mixtures
For river sand, several mixtures, graded into different ranges of particle size, we
produced by using a mechanical sieve shaker, as shown in Figure 6.13. The size
fractions produced are listed in Table 6.3. The nominal aperture sizes 180, 212,
300, 350 and 425 \itn were achieved by standard test sieves. In comparison the
sizes 98,154,223 and 328 |im were made from woven wire cloth.
In Table 6.3, sand mixtures MD1, MD2, MD3 and MD4 were specially made to
examine the effect of particle size distribution on the flowrate. The aim in preparing
these samples was to obtain the mixtures with the same median particle size but
different distribution. In particular Sand M D 1 had a narrow size distribution, while
the distribution of Sand M D 2 was wider than that of Sand M D 1 . The difference
between M D 2 and M D 4 was that Sand M D 2 had a uni-modal distribution and Sand
M D 4 a bi-modal distribution. Sand M D 3 and Sand M 7 had uni-modal and bi-modal
distributions, respectively (as shown in Figure 6.14). The samples were prepared
as follows:
Five initial sand mixtures were produced by grinding raw river sand and then
sieving it to provide:
Sand Mixture 1: 300-425 \un (same as Sand Ml)
Sand Mixture 2:223-328 (im
Sand Mixture 3 :154-223 Jim (same as Sand M D 1 )
Sand Mixture 4 : 98-154jim (same as Sand M 5 )
Sand Mixture 5 : 0 - 98 ^im (same as Sand M 6 )
Table 6.3
The Range of Particle Size for River Sand Mixtures
Sand Mixture Number
Range of Particle Size
dp (nm)
Sand M l
300 - 425
SandM2
212 - 350
SandM3
180-212
SandM4
0-350
SandM5
98 -154
SandM6
0- 98
SandM7
0-425
SandMDl
154 - 223
SandMD2
98 - 328
SandMD3
0-425
SandMD4
98 - 328
Frame
Filling funnel suspended on the frame
Sieve
screen
Sieve
holder
Electronically
controlled
vibrator
Buckets for collecting material
a) Basic Arrangement of Large Screen Siever
^+
^r
•I:::::-::
i
.1
i
1
•l:_:_
T - 4
b) W o v e n Wire Sieve Screen
Figure 6.13
c)
i
:
::i
4^-r
Sieve Holder
Schematic of Large Screen Siever
40
i
E
O
30
+
• 1—I
c_
CD
c=
cu
crr
c_
0
20
H
I
i—i—i—i
Sand
Sand
Sand
Sand
Sand
i f
-i
1—r—r-
MD1
MD2
MD3
MD4
M7
10
°io-
10^~
10
Particle Size ( micron )
-J-1-1-1 3
10 J
Figure 6.14 The Frequency Distribution of Selected Sand Mixtures
The sand mixtures with different distribution were produced by mixing some initial
sand mixtures together:
Sand MD1 was made of 100% Sand Mixture 3.
Sand M D 2 was made of Sand Mixture 2, 3 and 4, in approximately equal
proportions.
Sand M D 3 was made of Sand Mixture 1, 2, 3, 4 and 5, in approximately equal
proportions.
Sand M D 4 was made of about two thirds of Sand Mixture 2 and one third of
Mixture 4.
Sand M 7 was made of 4 1 % of Sand Mixture 1 and 5 9 % of Sand Mixture 5.
In order to minimize the error which was caused by particle segregation during
sampling for particle size measuring, especially for sand mixtures with wide
particle size distributions as Sand M D 2 , M D 3 , M D 4 and Sand M 7 , the material was
first sampled by riffle box, then classified by a number of sieves which were those
with 0,75, 90,106,125,150,180, 212, 250, 300, 355, 425 and 500 \im nominal
aperture sizes. It is assumed that there was no particle segregation during sampling
and measurement of the particle size of these sub-grouped materials. The median
particle sizes and particle size distributions of the mixtures were calculated by using
the following formula:
it
XMi(d|0)i
d50
~
i=l
i=1
(6.7)
11
Ad) =
IX f^)
i=l
it
i=l
where superscripts m and g stand for mixture and group respectively; dso for
median particle size; f(d) for frequency of band; M for mass of groups; n for the
number of groups.
6.5.2 Median Particle Size and Size Distribution of the Bulk Materials
There were three particle size analysis methods used.
i)
Malven Laser Particle Sizer (for alumina, P V C powder, Sand M l ~
M 6 and Sand M D 1 . The results are attached in Appendix II-1);
ii)
Mechanical Sieve Analysis (for sugar);
iii)
Combination of i) and ii) (for Sand M 7 , Sand M D 2 ~ M D 4 ) .
The median cumulative percentage diameter of the bulk materials measured in
experiments are tabulated in Table 6.4. The particle size distributions are depicted in
Figures 6.15 - 6.17.
Table 6.4 The Median Particle Size of the Test Materials
Bulk Materials
Alumina
P V C Powder
Sugar
Median Particle Size
d50 (jim)
100.2
127.2
784
Sand M l
370.8
SandM2
310.2
SandM3
197.6
SandM4
202.9
SandM5
155.7
SandM6
79.6
SandM7
113.0
SandMDl
201.6
SandMD2
201.4
SandMD3
197.3
SandMD4
199.1
Particle Size (Micron)
Sugar
"•
Alumina
— * —
P. V.C. Powder
Figure 6.15 Cumulative Size Distribution for Sugar, Alumina and P V C Powder
I
s
x>
1
•a
1
Particle Size ( Micron)
"0—
"•—
•»—
Sand-Mi
Sand-M2
Sand-M3
Figure 6.16
•+—
•«—
Sand-M4
Sand-M5
Sand-M6
Sand-M7
Cumulative Size Distribution for Sand M l ~ M 7
100
ft^ST
80
v ^
I
1
I
1
60
L-T
40
20
c -jSf**
10
10'
io-
Particle Size (Micron)
••—
-•—
Figure 6.17
Sand-MDl
Sand-MD2
••—
-°—
Sand-MD3
Sand-MD4
Cumulative Size Distribution for Sand M D 1 ~ M D 4
6.5.3 The Internal Friction Angle and the Wall Friction Angle
The internal friction angle and the wall friction angle for all materials were tested
the Jenike Direct Shear Tester method described by Arnold et al. (1980) [3] . The
wall materials tested were Perspex and Galvanised Steel. The shear testresultsare
attached in Appendix U-2 and Appendix II-3. Table 6.5 gives the results for low
consolidation conditions (al < 1 kPa).
The shear test results show that all of materials listed above are free flowing
materials according to Jenike's flowability zones'^, as plotted in Figure 6.18.
Table 6.5
Material
The Shear Test Results for the Test Materials
Angle of Internal
Angle of Wall Frictionty(degree)
Friction
5 (degree)
Perspex
Galvanised Steel
Alumina
36.5
22
21
P V C Powder
35.5
19
19
Sugar
43
26
23
Sand M l
37
25
25
SandM2
37
23
25
SandM3
37
25.5
25.5
SandM4
39
23.5
24.5
SandM5
38.5
29
28.5
SandM6
40
29
28.5
SandM7
40
27
26
SandMDl
36
26.5
27
SandMD2
41
27.5
27.5
SandMD3
39
27
27
SandMD4
40
28
27.5
--13-"
~
'-"
-"
—
—
2
4
6
8
10
12
Alumina
PVC Powder
Sugar
Sand Ml
SandM2
SandM3
SandM4
SandM5
SandM6
SandM7
SandMDl
Sand M D 2
SandMD3
SandMD4
14
Major Consolidation Stress (kPa)
Figure 6.18 Flowability Characteristic in Free Flowing Zone
(According to Jenike's Flowability Zones^)
6.5.4 Particle Density
Particle densities of all materials measured by the Beckman Air Pycnometer method
are given in Table 6.6.
Table 6.6
Bulk Solids
Material
Measured Particle Densities
Alumina
Glass
Beads
PVC
Powder
River
Sand
Sugar
4000
2520
1500
2700
1600
Particle Density
Ps
(kg/m3)
6.5.5 Bulk Density and Permeability
The bulk densities and permeabilities of all test materials measured by the me
described in Section 3.6 are displayed in Table 6.7 and Table 6.8, respectively.
Table 6.7 Measured Bulk Density Properties
Bulk Solids
Po
blv
3
b
2v
*lc
b
2c
Material
(kg/m )
Alumina
972.82
96.18
0.08508
103.82
0.08338
P V C Powder
593.16
47.26
0.10711
52.31
0.10498
Sugar
816.59
53.96
0.11817
60.51
0.11588
Sand M l
1329.25
125.15
0.05120
130.29
0.05029
SandM2
1309.36
127.33
0.05201
132.67
0.05108
SandM3
1264.70
151.29
0.05217
157.66
0.05124
SandM4
1318.44
183.64
0.05376
191.68
0.05278
SandM5
1155.32
129.76
0.05329
135.38
0.05233
SandM6
1031.01
249.14
0.07091
264.76
0.06952
SandM7
1333.31
256.58
0.06277
270.31
0.06157
SandMDl
1255.53
112.82
0.08362
121.580
0.08194
SandMD2
1281.21
201.61
0.05438
210.56
0.05339
SandMD3
1342.56
211.78
0.05100
220.43
0.05010
SandMD4
1236.40
166.51
0.05798
174.55
0.05690
Table 6.8 Measured Permeability Properties
6.6
Material
Permeability
Constant C Q
( *10- 9 M 4 N- 1 Sec" 1 )
a
Alumina
398.384
6.42186
P V C Powder
1566.698
9.06178
Sugar
21357.310
4.60027
Sand M l
6517.522
5.92424
SandM2
4353.567
5.48957
SandM3
2327.521
5.68797
SandM4
1535.968
5.85956
SandM5
1155.835
4.48150
SandM6
297.621
6.68156
SandM7
246.341
7.72660
SandMDl
2227.654
5.44753
SandMD2
1054.413
6.18770
SandMD3
575.561
7.44598
SandMD4
1335.804
4.50717
Measurement of Air Pressure Distribution in Mass Flow Bins
For fine materials, it is believed that the interstitial air pressure gradients provide
significant retarding influence on the flow from the bins. M a n y workers, such as
Miles et al. (1968) [44] , Miles (1970) [45] , Crewdson et al. (1977) [50] , Spink et al.
(1978) [ 5 1 ] , and Willis (1978) [ 5 4 ] and H e a d (1979) [55] , have reported that the
observed negative air pressure distribution is dependent upon the particle size of
material and the oudet size of the bin or hopper. According to the experimental
results of the current research described in Sections 6.2 - 6.4, the effect of the
surcharge level of fine material in a bin can be significant Arnold et al. (1989) [133]
stated that the decreased flowrate is due to the presence of adverse interstitial air
pressure gradients. In view of this importance it is worth examining the effect of
material level cm the air pressure gradients.
Since the test rig described in Section 6.2 was not big enough to measure the air
pressure distribution in the bin, a testrigbased also on the double bin model but
twice as big as that described in Section 6.2 was constructed. The structure of the
rig, as shown in Figure 6.19, consisted principally of:
i) Test Bin and Storage Bin
Perspex was used as the material for the vertical sections of the bins. The
converging section of the test bin was m a d e of steel which was machined then
galvanised. Figure 6.20 shows the geometry of the test bin and the location of the
air pressure tappings. The air pressure probes were set into the wall of the bin with
3 m m holes on the inner side of the test bin. G O R T E X cloth was used to cover the
inlet of the probe hole to prevent blockage. It was considered that the smooth side
of the G O R T E X , which was aligned on the inner wall surface, would not resist the
flow since the probe holes were very small compared with the bin geometry. Care
was taken to ensure that there was no leakage in the pressure measurement system.
The storage bin, with an external diameter 15 m m smaller than the inner diameter of
the test bin, was designed to be able to m o v e up and d o w n in the test bin. A n
adjustable gate was designed for the oudet of the storage bin. A skirt was mounted
between the two bins to reduce dust emission.
163
Ceiling
Ceiling
Ceiling
(RL 5800 m m )
Skip Controlled by Compressed Air
and Lifted by Fork Lift Truck
(RL 4280 m m )
Flexible Discharge Duct
Transducers
Surcharge Level Control System
(detailed in Figure 6.21)
Skip for
Collecting
Material
(RL 0.0)
Figure 6.19
The Apparatus for Measuring Air Pressure Distribution
164
0295 (LP.)
Figure 6.20 Large Test Bin Geometry and Pressure Tapping Locations
ii)
Surcharge Level Control System
The surcharge level control system, as illustrated in Figure 6.21, was designed to
lift the storage bin up and d o w n to any desired position in the test bin by use of a
winch. The balance plate of the system could be m o v e d up and d o w n vertically
guided by the rods of the bin hanger. A flexible discharge duct connected the fixed
inlet of therigto the movable storage bin.
iii) Flowrate and Air Pressure Measurement Systems
The flowrate was measured by recording the weight curve. The bin hanger was
constructed to ensure the total weight of the discharging material would be
suspended on three transducers. The transducers used were R S 250 kg load cells.
The load record system was identical to that described in Section 6.2. The
manometers with coloured water and video camera were used to measure and
record the air pressure profile at each probe point. The pressure readings were taken
under the steady state condition.
iv) The Facilities of Filling and Collecting the Bulk Material
The material was filled by skip, stool and fork lift truck (KOMATSU FG15-12C).
The geometry of the rig was limited by the height of the ceiling and the lifting
capacity of the fork lift truck. The m a x i m u m lifting height of the fork lift truck was
3.7 meters. A stool of 0.61 m effective height was used to extend the reach of the
skip. A belt conveyor and skip covered with a plastic sheet to reduce dust were
used to collect the material on discharge. The experiments were conducted for
alumina as the test bulk material. The load record system was calibrated by putting a
number of weights (totalling 100 kg) on the bin hanger.
A number of experiments have been carried out to measure the air pressure
distribution at two different material levels (0.05 m and 0.63 m ) . The experimental
results highlighted the following:
• use of the double - bin apparatus enabled steady air pressure
distribution in the test bin to be observed for a significant period of
time; the developed surcharge level control system on this enlarged test
rig easily adjusted the material level to anyrequiredposition in the test
bin to allow the measurement of air pressure as affected by the
surcharge levels;
• the experimental results indicated that the negative air pressure
generated by flowing particles increased with an increase in material
surcharge level; the experimentalresultswere displayed in Figure 4.5
and discussed in Section 4.7.2 together with the theoretical predictions;
• the measured minimum pressure position (hmp) agreed with the
assumption eqn (4.29)reasonablywell, as plotted in Figure 4.3.
However, due to time constraints, only limited experiments have been done to
examine the eqn (4.29). M o r e work is required to give further evidence for eqn
(4.29) by changing the test bin geometry, material surcharge level and the bulk
solids.
167
String
II
II
II
II
7
2
Storage Bin Holder
'
I
Balance Plate
J?
II ^
I
EZZE
9
Winch
Rods of the Bin Hanger
(three)
Test Bin Holder
Z
Figure 6.21 Schematic of the Surcharge Level Control System
Chapter 7
T h e Effect of Permeability on the Flowrate
7.1 Introduction
As illustrated in Section 5.4.1, the negative air pressure gradients have a significant
effect on reducing the flowrate of fine materials. M a n y researchers^44'45'50'51'54'55'
have reported that the magnitude of the negative pressure gradient increases as the
particle size decreases or the oudet size of the bin or hopper increases. T h e work
described in Section 4.7.2 indicated that the negative pressure gradient increases not
only with the decrease in particle size and with the increase in the outlet size but also
with an increase in material surcharge level. Furthermore, the predictions of
flowrate for Sand M D 1 to Sand M D 4 , as shown in Figure 5.4, indicated that the
flowrate also depends on the particle size distribution. The flow offinebulk solids
is so complicated that using one or two of material characteristics, say particle size
or particle size and bulk density, are not sufficient to describe the effect of particle
characteristics on flowrate. Current theoretical models for predicting the air pressure
gradient and particle flowrate employed the idea of McLean 1 5 8 ' 6 0 1 , Arnold [52] and
Arnold et al.[3] which related flowrate and pressure gradients through permeability
according to D'Arcy law. In the work reported in this chapter, particular attention is
paid to the relationship between flowrate and permeability of the flowing bulk
solids.
7.2 Experiments Using Glass Beads
This series of experiments was aimed at examining the effect of bulk density and
permeability on the flowrate of a range of bulk solids flowing from a model mass-
flow bins. Before analysing the experimental observations, it isfirstappropriate to
analytically predict the effect of bulk density and permeability on flowrate. This
analysis commences by examining the relationship between pressure gradients
across powder beds to the basic properties of the bed, using the Carman-Kozeny
equation:
A P = ( 1 - e )2 j ^
H*p
^ =
a^
c3
e
where
u
(7.D
e = 1- —
Ps
By comparison the D'Arcy law indicates
AP
1
Hp = C
u
(7-2)
It can be seen that the relationship between air pressure gradient and superficial
relative fluid velocity depends on the bulk density and on equivalent surface area or
on permeability. Hence, from a knowledge of the packing of particles, a particle
mixture can be made with a certain value of bulk density, equivalent surface area or
permeability by mixing different particle size components together in certain
proportions. However, bulk density and permeability are very dependent on the
state of compaction, as discussed in Sections 3.6 - 3.7 and shown in Section 6.5.5.
In order to reduce and neglect the effect of consolidation, the approximately
incompressible bulk solids, glass beads, were used in these experiments. A n
example of the 'manufacture' of mixture properties is illustrated in Table 7.1 which
declares the results of bulk density measured for two glass bead mixtures.
Table 7.1
Bulk Densities of Glass Bead Mixtures
Mixture
Po
b
lv
2v
Wv/Po
b
3
(kg/m )
Smallest glass bead Mixture
in experiments
djQ = 159 n m
1401.09
79.83
0.15649
0.057
O n e glass bead Mixture
in experiments
djQ = 325 Jim
1527.05
53.59
0.12423
0.0351
F r o m Table 7.1, it is apparent that the glass beads mixtures are relatively
incompressible. Normally, for glass beads which are typically particles with regular
shape and smooth surface, the compressibility of a mixture is inversely proportional
to the particle size. In this case, the ratio b jy/po of any mixture should be less than
that of smallest mixture. In particular, for all the glass bead mixtures tested, b jy/Po
< 0.057. It will be noted that b l v / p 0 is smaller for the glass bead mixtures than for
the bulk solids prepared in Section 6.5.5, as presented in Table 7.2. Furthermore,
since the magnitude of the consolidation stresses during mass flowing is small, the
effect of consolidation for glass beads, therefore, can be neglected
7.2.1 Preparation of Glass Bead Mixtures
The particulate mixtures used were specially prepared to exhibit particular values of
bulk density and permeability. Basically, three different sized glass beads were
used. The selected size ranges had the following nominal sieve apertures:
106-212 nm
212-355 urn
355-425 M-m
were referred to as small, medium and large particles,respectively.The small and
large particles were used to produce the binary mixtures to compare with the results
for medium particles. Here, the different particle size distributions were not
regarded as another parameter.
Table 7.2 The blv/p0 Ratio for the Bulk Solids Prepared in Section 6.5.5
Bulk Material
Wv/Po
Alumina
0.0989
P V C Powder
0.0797
Sugar
0.0661
Sand M l
0.0942
SandM2
0.0972
SandM3
0.1196
SandM4
0.1393
SandM5
0.1123
SandM6
0.2416
SandM7
0.1924
SandMDl
0.0898
SandMD2
0.1574
SandMD3
0.1577
SandMD4
0.1347
The bulk density and permeability were measured using the Jenike Compressibility
and Permeability Testersrespectively,as shown in Figure 3.5 and Figure 3.19. In
generating the selected mixture, each component of the mixture was weighed on a
Mettler P C 4400 electronic balance. The components were mixed together very
carefully in a plastic vessel to produce a uniform mixture and keep segregation to a
minimum. Also, attention was paid to filling carefully the test cylinder by spooning
the mixtures layer after layer to minimise the mixture segregation.
The bulk density and permeability were obtained from eqns (3.14) and (3.17):
W
and
0
C = SP.
A
p APt - AP 0
(3.17)
The equivalent surface area was determined from eqns (7.1) and (7.2), that is
a** = Mf ,
Psp2
,3 C (7.3)
( Ps " P ) 3
or
a** = \i{
~
E
3
C
Making up the mixtures of large and small particles required two steps. The first
step involved determining the mix of glass beads to produce the desired bulk
density, permeability and equivalent surface area characteristics of glass beads
(detailed in Section 7.2.2). The second step involved preparing the required
quantities of particle mixtures with a desired value of bulk density, permeability or
equivalent surface area for flowrate measurement, based on the mixes obtained in
the first step (detailed in Section 7.2.3).
7.2.2 Mixtures of Particles for Laboratory Measurements
A small amount of each mixture which fully filled the test cylinder of the
permeability tester (about 0.7 kg) was produced by carefully packing the particles to
obtain the selected bulk density, equivalent surface area or permeability
characteristics while noting the different components of small and large particles
(referred to as particle packing stage in Table 7.3). This allowed the determination
of the required percentage of small and large particle components with
approximately the same bulk density, equivalent surface area or permeability as the
m e d i u m particles. During these measurements, packing of the particles was
repeated four times for each mixture, and six air flowrate readings at each different
pressure drop for each packing were taken. The results are shown in Figures 7.1 to
7.3; the average value w a s taken from the 24 measured values, for each point
plotted.
3%S/97%L
87%S / 13%L
I
1503.9
0.0
Small
Figure 7.1
0.8
1.0
Large Medium
Large Solids Fraction
Mixture Bulk Density Characteristics
17%S / 83%L
r
8
CO
o
3393.45
U
I
•a
0.0
0.2
0.4
0.6
0.8
Small
1.0
Large Medium
Large Solids Fraction
Figure 7.2
Mixture Permeability Characteristics
27%S/73%L
r
6
CS
--*
331.9
i
o
*
*
0.0
Small
0.2
0.4
0.6
0.8
1.0
Large Medium
Large Solids Fraction
Figure 7.3 Mixture Equivalent Surface Area Characteristics
From Figures 7.1 to 7.3 it is seen that the bulk density characteristic is very typical
of that determined by other researchers^ 100 ' 134 -^]. Furthermore, the permeability
characteristic is very similar to those of German^ 100 ^, Leitzelement et alJ135^ and
Standish et alJ ^. The equivalent surface area can berelatedto the permeability as
indicated in eqn (7.3).
7.2.3 Mixtures of Particles for Flowrate Measurements
The large quantity of each mixture (about 10 kg) required for measuring the
flowrate of particles from bins was produced according to the percentage of
components determined from the experiments in the previous section (referred to as
mixture blending stage in Table 7.3).
The mixtures produced were as follows:
i. T w o mixtures having the same bulk density as the medium particles, see
Figure 7.1.
a. 3 % small and 9 7 % large particles
b. 8 7 % small and 1 3 % large particles
ii. One mixture having the same permeability as the medium particles,
Figure 7.2.
1 7 % small and 8 3 % large particles
iii. One mixture having the same equivalent surface area as the medium
particles, Figure 7.3.
2 7 % small and 7 3 % large particles
Precautions were taken to reduce the effect of segregation during measurement of
these parameters, including:
• the use of a riffler to obtain sub-samples for laboratory measurements;
• repeated packing of the particles at least six times for each mixture, and
taking of six air flowrate readings at each different pressure drop together
with the weight of particles and height of particles in the permeability tester
(Figure 3.19) for each packing; the average permeability was then calculated
from these 36 air flowrate observations and the average bulk density from
the 6 packing tests; the equivalent surface area was obtained using the
average bulk density and average permeability in eqn (7.3).
Table 7.3 shows the comparison of the results for various parameters measure
thefirststage (particle packing) and second stage (mixture blending).
Table 7.3 Comparison of the Three Mixtures Parameters
Particle Packing
Mixture Blending
a**
P
C
a**
3% S / 97% L 1501.94 5342.54
518
1496.99
6023.71
572
17% S / 83% L 1536.43 3367.06
379
1534.00 3395.14
379
27% S / 73% L 1556.89 2715.34
334
1552.73
2784.62
336
87% S / 13% L 1505.16 1565.20
154
1509.99
1590.12
159
1503.93 3393.45
332
Mixture
medium particles
P
C
7.2.4
The Observed Particle Flowrates from the Bins
The oudet sizes of the test bins D 0 tested were 0.0445 m and 0.02 m , respectively
(as described in Section 6.2). The experiment observations are listed in Tables 7.4
to 7.6.
Table 7.4
The Observations for Mixtures with Similar Bulk Density
Flowrate of Particles Q p ( kg / sec)
Bulk Density
Mixture
P
0.0445 m Oudet
0.02 m Oudet
3% S / 97% L
1496.99
1.732
0.248
87% S / 13% L
1509.99
1.599
0.234
medium particles
1503.93
1.706
0.246
Table 7.5
The Observations for Mixtures with Similar Permeability
Flowrate of Particles Q p (kg/sec)
Mixture
Permeability
C (*10'9)
0.0445 m Outlet
0.02 m Oudet
17% S / 83% L
3395.14
1.733
0.248
medium particles
3393.45
1.706
0.246
Table 7.6 The Observations for Mixtures with Similar Equivalent Surface Area
Mixture
Equivalent
Surface Area
Flowrate of Particles Q p (kg/sec)
a**(*10" 12 )
0.0445 m Oudet
0.02 m Oudet
27% S / 73% L
336
1.732
0.249
medium particles
332
1.706
0.246
From these Tables, it is apparent that for two different mixtures with same
permeability constant or equivalent surface area, the difference of their flowrates
were less than 1.6%. However, in the case of the mixtures with the same bulk
density the difference of their flowrates could exceed 7 % . This implies that for fine
bulk solids the bulk density cannot uniquely describe the effect of particle size
distribution on the flowrate of different mixtures of the same particulate materials.
In comparison, the parameter permeability can provide a more accurate description
of bulk solid flow behaviour. Figure 7.4 gives the variation in flowrate as a
function of permeability for two hopper oudet sizes. The flowrate increases with
the increase of permeability. The equivalent surface area, since it has a close
relationship to permeability as described in eqn (7.3), can be used as an alternate
parameter. However, in this thesis, the work is concentrated on investigating the
relationship between permeability and the flowrate offinebulk solids, since the
permeability directiy links the air pressure gradient with the superficial velocity
between air and particles.
0.30
—i—
.
0.28 -
i
•
i
i
•
0.26
0.24 0.22
0.20
•
t
r
^~-
•
.
.
-
0.18
0.16
0.14
0.12 0.10
1000
•
•
•
•
•
2000
3000
4000
5000
6000
7000
Permeability C (* 10'9 M 4 N" 1 Sec'1 )
a) From 0.02 m Oudet
1000
2000
3000
4000
5000
6000
Permeability C (* 10"9 M 4 N' 1 Sec"1 )
b) From 0.0445 m Oudet
Figure 7.4 The Flowrates vs. Permeability Constant
7000
Unfortunately, since there were insufficient glass beads available to allow mixtures
to be m a d e for measuring the flowrate within the low permeability region. In view
of this deficiency further experiments were conducted using sand mixtures.
7.3 The Effect of Particle Size and Size Distribution
To examine the effect of particle size on flowrate, further experiments were
conducted using sand mixtures Sand M l to Sand M 7 . The relationship between
flowrate and median particle size at level H / D = 1.5 is plotted in Figure 7.5. These
results indicate that for the mixtures which have quite different particle size flowing
from a large D Q / C L ratio outlet, where the effect of 'empty annulus1^ ^ can be
neglected, the flowrate increases with an increase in particle size. This variation of
flowrate was also reported by Crewdson et al.[50], Spink et al.[51], Head [ 5 5 ] and
Ducker etal.[57].
1.4
8
•5
O^
!P
1.2
1.0
0.8
B
0.0200 m outlet
• 0.0445 m oudet
0.6
i
4)
0.4
0.2
o.o
100
200
300
400
Median Particle Size d 5 0 ((im)
Figure 7.5 Flowrate vs. Median Particle Size at Surcharge Level H / D = 1.5
It is also shown, in Figure 7.5, that the median particle size is an important
parameter to compare the flowrates of bulk solids with narrow particle distributions
and different median particle sizes. However, the effect of wider particle size
distributions on the particle flowrate has not been reported by previous researches,
although m a n y researchers163"701 found that the flow properties of bulk solids are
affected by the particle size distribution. In this work, thisresearchwas investigated
using the sand mixtures Sand M D 1 to Sand M D 4 as the particle size distributed
materials. The Sand M D 1 to Sand M D 4 are the powders with the same median
particle size (d 50 = 200 pm) but with different particle size distributions. The results
are tabulated in Table 7.7 and Table 7.8. F r o m the Tables 7.7 and 7.8, it is seen
that the difference between the flowrates for the bulk solids with same median
particle size m a y be significant. Specifically, the flowrate of Sand M D 1 is about
twice as large as that of Sand M D 3 for both 0.02 m and 0.0445 m oudet. Figure
7.6 shows the variation of the flowrate with median particle size for Sand M l to M 5
and Sand M D 1 to M D 4 at surcharge level H / D = 1.5. F r o m the Figure 7.6, it is
clear that the median particle size cannot alone be used to predict accurately the
flowrate.
However, Tables 7.7 and 7.8 indicate that the permeability can provide a more
accurate prediction of the flowrate than that possible using the median particle size.
The significance of permeability is readily apparent in Figure 7.7 which depicts the
same results as those plotted in Figure 7.6, in which the Q p - C 0 diagram, as
illustrated in Section 5.3.3 and Section 5.4.2, is used.
182
3
oo
00
VO
VI
•a
0-
CO
o
13
B
"5a
o
en
oo
r»
oo
oo
en
VO
lO
M
>*^
CX
C
CS
1
•n
cs
00
VO
o
E
VO
s
ts
en
o
S
E
vo
cs
IT)
in
en
en
O
cs
VO
b
ON
CS
00
«—i
d
CS
s
m
o
1
I
a:
in
m
in
VO
in
Ov
00
8
CO
0>
«n
vo
r^
cs
cs
©CO S
vq
o •"
O
en
t-H
o
VO
»n
in
rm
vo en
«n
oo
VO
CS
in
en
en
cs
cs
en
i—i
•n
o
VO
»n
m'
i>
in
ts
e
w
g
d
cs
d
d
o
o
i
oo
o>
m
d
o
1
I
a
b
•n
d
d
d
E
d
ON
vo
•a-ao
en
8 8 Si
i-H
©v
t& s
VO
Si**
8
§
o
-0
en
o
cs
d
s
oo
•n
en
en
1.4
1
,
E1
1.2 -
a
*
*
•
•
1.0 •
0.8 -
B
»
0.6
B
\
•
0.4
SandMl-M5
SandMDl
SandMD2
SandMD3
SandMD4
from 0.0445 m Outlet
-
-
•
from 0.02 m Outlet
0.2
Q
B
-
i
0.0
100
200
300
400
Median Particle Size d 5Q (nm)
Figure 7.6 Measured Flowrate vs. Median Particle Size
for Sand M l - M 5 and Sand M D 1 - M D 4 (H/D = 1.5)
1.4
12
B
A
X
•
•
1.0
0.8
0.6
from 0.0445 m Outlet
04 I-
from 0.02 m Outlet
L
0.2 -
0.0
0
SandMl-M5
SandMDl
Sand MD2
SandMD3
SandMD4
1000
2000
3000
4000
5000
6000
Permeability Constant Co
*10' 9 ( M 4 N ^ S e c 4 )
Figure 7.7 Flowrate v. Permeability for Sand Mixtures
7000
7.4
Discussion
Permeability is a useful parameter to describe the flow behaviour of particles as it
provides a relationship between pressure gradient and air velocity which pass
through the particle packing. Permeability includes the effects of particle size,
particle size distribution, particle shape and roughness [64,66,87,95,100] j^ regard to
the effect of particle size, in general, the permeability of the particles increases with
an increase in the particle size (in particular it is proportional to the square of the
particle size [87*10°]). Whereas in regard to the effect of particle size distribution, the
permeability increases as the width of the particle size distribution decreases or as
the particle size distribution is narrowed without changing the median particle
sjze[66,ioo]
F r o m Figure 7.2 and the results of previous workers! 64 * 100 ' 135 * 137 ],
the permeability increases with the addition of large particles into a bed of small
particles. O n the other hand, permeability rapidly decreases as the small particles are
added into a bed of large particles. In comparison, permeability is fairly insensitive
to particle shape differences fee randomly packed structures in practical uset100!.
The flowrate of particles from a mass flow bin increases as its permeability
increases and remains constant or almost constant after permeability is greater than a
critical value Q n , as shown in Figures 7.4 and 7.7. This result w a s supported by
Spink et al. and Crewdson et al. and by the theoretical prediction displayed in
Section 5.3.3. For bulk solids with high permeability, the effect of interstitial air
pressure gradients can be neglected. However, at low permeability the flowrate
rapidly decreases as the permeability decreases. Figures 7.8 and 7.9 show some
theoretical results for air pressure distribution predicted by model detailed in
Sections 4.5 and 4.6. Figure 7.10 gives the general variation of negative air
pressure gradients at the bin oudet versus the permeability constant: the lower the
permeability, the higher the negative air pressure gradient at the hopper outlet
generated is. Furthermore, the lower the permeability, the greater the effect the
pressure gradients have in reducing flowrate.
0.500
0.400
- 0.300
0.200
1
•a
!
- 0.100
-20
•a
0.000
-10
Negative Air Pressure ( m m water)
Figure 7.8
Predicted Negative Air Pressure Distribution for Bulk Solids
(from 0.02 m oudet)
0.500
— a —
Sand M l
—•—
SandMDl
— " —
SandMD4
—•—
SandMD2
--•-- SandMD3
0.400
rS
0.300
•a
- 0.200
-40
Figure 7.9
C.
-30
a
- 0.100
-20
-10
0.000
Negative Air Pressure ( m m water)
Predicted Negative Air Pressure Distribution for Bulk Solids
( from 0.0445 m outlet)
i
•a
Permeability Constant C Q
* 1 0 " 9 ( M 4 N^Sec"1)
2000
'
4000
•
—••
8000
1
1
1
i
•§B
B
-10000 -
*
a
s
O
C
•
a
•
Surcharge Level H/D = 1.5
S
•5
6000
1
i
i
£
1
B
0.0200 m Oudet
• 0.0445 m Oudet
-20000
£
a
-30000
Figure 7.10
i
1
1
1
j _
i
dP
-«- at the Bin Oudet vs. Permeability Constant of Particles
A n examination of Figure 7.10 suggests that there is a critical value of the
permeability constant C below which C has a marked influence on flowrate.
Figures 7.4 and 7.7 also indicate that outlet size has an effect on the onset of the
influence of permeability. Terming the value of permeability constant below which
the permeability begins to influence flowrate as O n , then it can be seen from Figure
7.7 that the value of Q n for the sand mixtures is approximately 2000* 1 0 9 (M 4 N 1
sec1) for the 0.02 m oudet and 3500* 10*9 (M 4 N 1 sec1) for the 0.0445 m outlet.
This can be also explained by the effect of adverse air pressure gradients. In
particular, Figure 7.11 and Figure 7.12 show the predicted air pressure
distributions for Sand M l and Sand M D 3 , respectively. Figure 7.13 shows the
experimental results obtained by Willis^54-'; similar results were also observed by
H e a d ^ w h o examined the flow of fine sand from a conical hopper fitted with
variable orifice diameters. Applying their findings to the current experimental
investigation indicates that for the same bulk solids and material level, flow will be
retarded by a greater negative pressure gradient the larger the oudet size. This re
supports the observations on the effect of surcharge level detailed in Chapter 8.
1
-1
I
1
'
1
— — i
|
0.500
•
•
—
•
•0.400
a
- 0.300
3
e
— 0.0200 m Oudet
• — 0.0445 m Oudet
******/
'
- 0.200
- 0.100
-
_*
-5
i
_J
-4
i*
•"•
-3
I « ^ »
-2
ri.,
-1
n
,,
0.000
o
a
i
•a
$
Negative Air Pressure (mm water)
Figure 7.11 Predicted Air Pressure Distribution for Sand M l (at H/D = 1.5)
0.500
•*— 0.0200 m Oudet
•*— 0.0445 m Oudet
-40
-30
-20
-10
Negative Air Pressure (mm water)
Figure 7.12 Predicted Air Pressure Distribution for Sand M D 3 (at H/D = 1.5)
6
•r>a
0.8
2
06
i
0.4
•5
*>
ed
Width of the
Hopper Oudet
0.04 m
0.06 m
0.08 m
1.0
5
.s
PQ
vvvv
12
£bO
02
S3
0
• r*
O
20
40
60
80
100
Air Pressure (negative m m water)
Figure 7.13
Negative Air Pressure Distribution (Willis^)
(Wedge hopper, O p = 15 ° , Slot length 0.61 m )
Generally, the steady state flowrate of a particle mixture decreases with the addition
of smaller particles due to the decrease in the permeability and, further, the increase
of the effect of air pressure gradients. This conclusion is supported by Bird et
al.t63! and Arnold et al.t133]. B y contrast, M e m o n et alJ67l presented results which
give a reduction in flowrate with the addition of large particles into small particles.
It should be noted, however, that the conditions of their experiments were quite
different from those in this investigation. In particular:
• different materials (limestone for large particles and sand for small) were
used in their experiments which had different solids densities. In this
situation, the comparison of the effect of large particle mass fraction on
mass flowrate could provide different results.
their hopper outiets, in comparison to the size of large particles size, were
too small to avoid wall effect on the flowing particles (Do/d ranged between
1.85 ~ 11.78). Hence, under these conditions, mechanical hindrance of the
large particles during flow through the relatively small oudet dimension
would have been experienced.
On other hand, for bulk materials of very low permeability, for which no steady
state flowrate could be measured, whether the addition of coarser material could
improve the flow behaviour depends on the permeability of resultant mixture. In
particular, the flowrate should increase if the permeability of the mixture increases.
For example, the steady state flowrate of Sand M 6 with a permeability constant of
297.621*10"9 ( M 4 N" 1 Sec"1) could not be observed except at the lowest surcharge
level. However, if some coarser material was added to produce another mixture,
say Sand M D 3 with the higher permeability of 575.561 * 1 0 9 ( M 4 N^Sec" 1 ), steady
state flowrates were then obtained. If, however, the resulting mixture had a similar
or even lower permeability, e.g., the bimodal-distributed Sand M 7 for which the
permeability constant was 246.341 * 1 0 9 ( M 4 N ^ S e c " 1 ) , a steady state flowrate
could still not be attained.
7.5 Criterion for Classifying Coarse and Fine Bulk Solids in Terms of the Effect
of Interstitial Air Pressure Gradients
The flow behaviour of bulk solids in a mass flow bin is dependent not only on the
particles themselves but also on the bin or hopper geometry. A s an indication of the
relation between pressure gradient and superficial relative velocity, the parameter
permeability has been shown to be significant. It was demonstrated in the
experiments that the flowrate of the particles from a mass flow bin increases as the
permeability increases and remains constant after the permeability is greater than a
critical value. This fact indicates that from the point of view of the effect of
interstitial air pressure gradients, a material can be defined asfineor coarse material
depending on whether its permeability is smaller or greater than the critical
permeability. The critical permeability value Ccri depends on the bin geometry and
bulk solid itself. A s discussed in Section 7.4, the critical permeability value was
about 2000* 1 0 9 ( M 4 N^Sec" 1 ) for the 0.02 m oudet and 3500* 1 0 9 ( M 4 N 4 S e c _ 1 )
for the 0.0445 m oudet This indicates that for bulk material with a relative large
permeability the effect of negative air pressure gradients could become significant if
the bulk solid flows out of a bin with a large oudet.
For convenience, it is assumed that the critical permeability Ccn is proportional to
the constant flowrate as predicted by Beverloo's equation!14!, that is
Ccri=c1Qp + c2 (7.4)
Restating Beverloo's equation
Qp = Bfpvg (D0-kdp)5/2 (2.1)
where Bf is a constant dependent on hopper geometry and k is a constant
dependent on particle shape, usually 1 < k < 3 t51l
it can be seen that when Do » kdp then
Qp = BfPViD05/2 (7.5)
According to the above assumption, Qri becomes
Ccri= c1BfpVgD05/2+ C2 (7.6)
Using Qri = 2000 * 10-9
C c n = 3500 *10- 9
( M 4 N^Sec" 1 )
for D 0 = 0.02 m;
(M^Sec'1)
for D 0 = 0.0445 m ,
then eqn (7.6) can be approximated as
Qn = [( 0.85 p Vg D05/2 + 1.8) * 103] *10-9 (M4 N^Sec"1) (7.7)
Since the critical permeability varies with such hopper geometry as oudet size, the
criterion classifying fine and coarse bulk solids must be related to the hopper
geometry.
Taking the sand mixtures where median particle size and permeability are listed in
Table 6.4 and Table 6.8 respectively, as examples and assuming for convenience
that p = 1300 kg/m 3 , then
for D0 = 0.02 m, Ccri - 2000 *109 (M4 N^Sec-1); comparing with
Sand M 3 and Sand M D 1 suggests that Carleton's criterion of 200
uml 49 l is reasonable to materials with a narrow size range;
for D0 = 0.1 m, Qn * 13000 *109 (M4 N^Sec-1) indicating that the
criterion of 500 umt28*50*51*55'61*62] becomes more acceptable.
In summary any criterion delineating coarse and fine particles cannot be made
independent of hopper geometry factors.
Crewdson et alJ ^ measured the flowrate of eight materials with different particle
sizes from four different sized hoppers. From their results, a critical curve 1-1 can
be found to distinguish the air affected flow in terms of particle diameter, as shown
in Figure 7.14. Fitting a parabolic function to curve 1-1 gives:
where ^Crewis *e ^owrate in Crewdson et alJ50]
Due to lack of information, it is difficult to predict the permeability parameter from
only particle diameter. F r o m the Carman-Kozeny equation, the permeability is
proportional to the square of the particle diametert87*100^ i.e.,
C - d* (7.9)
Assuming that the effect of porosity variation of bulk mixtures on the permeability
is insignificant compared with the square of the particle diameter, the combining
equations (7.8) and (7.9) is
Qri x wCrcw
This indicates that the relationship assumed in eqn (7.4) is reasonable.
It is necessary to mention that the coefficients in eqn (7.7) were obtained from the
experimental results at low surcharge level and hence, the critical permeability
predicted by eqn (7.7) is the criterion for classifying coarse and fine bulk solids at
similar material surcharge level conditions. A critical permeability to classify
significant and insignificant effects of surcharge level on flowrate will be suggested
in Section 8.2.
3000
•
1
I
* D Q = 0.003930 m
A
D Q = 0.005145 m
* D Q = 0.006895 m
• Do = 0.009155 m
1
2500 -
•
•
/
ft
2000 "-
O
"S*
I
o/
O
-
.
/
E
o
VI
VI
2
0
A*
1500 -
a
y/
-
1000
X
500 -
-
A/
A
.
~+
1
^
f
•*.
"-*
'
200
400
600
800
Particle Diameter d p ( H m )
Figure 7.14 The Critical Curve of Air Affected Flow in Terms of Particle Diameter
(Flowrate Data from Crewdson et alJ ^)
Chapter 8
The Effect of the Surcharge Level on the Flowrate
8.1 Experimental Results
A number of experiments, described in Sections 6.2 ~ 6.4, were conducted to
examine the effect of surcharge level on the flowrate of bulk solids from mass flow
bins. The actual surcharge levels and the bulk solids used in the two test bins are
listed in Tables 8.1 and 8.2.
Table 8.1 The Actual Surcharge Level of Solids in the Test Bin
with 0.020 m Oudet
Bulk Solids
Level 1
Level 2
Level 3
Level 4
(m)
(m)
(m)
(m)
P V C Powder
0.010
0.220
0.480
0.590
Sugar
0.010
0.220
0.480
0.610
Sand M l
0.010
0.220
0.480
0.610
0.250
0.530
0.250
0.530
0.250
0.530
SandM2
SandM3
(no test)
SandM4
(no test)
SandM5
0.010
0.220
0.480
0.610
SandM6
0.010
0.220
0.480
0.600
SandM7
0.010
0.230
0.500
0.600
SandMDl
0.010
0.230
0.500
0.600
SandMD2
0.010
0.230
0.500
0.600
SandMD3
0.010
0.230
0.500
0.600
Sand M D 4
0.010
0.230
0.500
0.600
Table 8.2 The Actual Surcharge Level of Solids in the Test Bin
with 0.0445 m Oudet
Level 1
Level 2
Level 3
Level 4
(m)
(m)
(m)
(m)
P V C Powder
0.010
0.310
(no test)
0.570
Sugar
0.010
0.220
0.480
0.600
Sand M l
0.010
0.235
0.535
0.610
0.235
0.535
0.190
0.510
0.215
0.525
Bulk Solids
SandM2
SandM3
(no test)
SandM4
(no test)
SandM5
0.010
0.220
0.480
0.610
SandM6
0.010
0.220
(no test)
0.600
SandM7
0.010
0.270
0.480
0.630
SandMDl
0.010
0.270
0.530
0.660
SandMD2
0.010
0.270
0.530
0.660
SandMD3
0.010
0.270
0.530
0.660
SandMD4
0.010
0.270
0.530
0.660
In Table 8.1 and Table 8.2, level 1 to level 4 stand for the different material heights
in vertical section of the test bins (measured from transition section of the bins).
The steady state flowrates obtained, corresponding to the surcharge levels in
8.1 and Table 8.2, are shown in Figures 8.1 to 8.3.
0.12
8
V3
M
s /
0.10
-
oP22
S
0.08
• P V C Powder
• Sugar
£o
0.06
(H
3
Vi
t
E
0.04
2
3
H/D Ratio
a) Q p vs. H/D (from 0.020 m Outlet)
/—\
o
»
•ta
M
0.80
0.70
0.60
(X
at>
0.50
iQ
0.40
t3
e"S
§
</>
• P V C Powder
• Sugar
030
0.20
cd
i2
S
0.10
H/D Ratio
b) Q p vs. H/D (from 0.0445 m Outlet)
Figure 8.1 Measured Flowrate vs. H/D Ratio for P V C Powder and Sugar
0.20
8
a
§
1
0.18
n
0.16
Sand Ml
• SandM2
A
SandM3
H
SandM4
* SandM5
0.14
0.12
E
1
0.10
2
3
H/D Ratio
<u
s
a) Q p vs. H/D (from 0.020 m Outlet)
8
•a
H
Sand Ml
• SandM2
A
SandM3
B
SandM4
* SandM5
o
E
•a
1
I
2
3
H/D Ratio
b) Q p vs. H/D (from 0.0445 m Outlet)
Figure 8.2 Measured Flowrate vs. H/D Ratio for Sand M l to Sand M 5
0.18
/—»N
8
•
a
M
0.16
v—«"
oP
22
£
%
0.14
E
0.10
O
15
0.12
13
SandMDl
• SandMD2
• SandMD3
• SandMD4
0.08
VI
si
3£
s
0.06
2
3
H/D Ratio
a) Q p vs. H/D (from 0.020 m Outlet)
8
CM
"5ft
H
I
SandMDl
• SandMD2
• SandMD3
• SandMD4
b) Q p vs. H/D (from 0.0445 m Outlet)
Figure 8.3 Measured Flowrate vs. H/D Ratio for Sand M D 1 to Sand M D 4
8.2 Discussion
Figures 8.1 to 8.3 show the variation of mass flowrates with H / D ratio for different
bulk solids. In some instances, the differences in flowrate are due to the variations
in bulk density, e.g., the difference in the flowrates between Sand M l and Sugar
with bulk densities p 0 of 1330 kg/m 3 and 816 kg/m 3 , respectively.
The relationship between measured mass flowrate Qp and H/D ratio can be
considered as a linear function over the range of H / D ratio examined in the
experiments. For the fine materials tested, increasing the material level caused the
flowrate to decrease. This decreased flowrate is due to the presence of adverse
interstitial air pressure gradients which are k n o w n to be significant for fine
materials, as pointed out by Nedderman et al. (1982)[4], Arnold et al. ( 1 9 8 0 ) ^ and
also as predicted in Section 5.4.1. With increasing H / D ratio, the adverse air
pressure gradients occurring at the hopper outlet increase in magnitude, as
discussed in Section 4.7.2. Hence for fine materials subject to unhindered gravity
discharge from a bin, the lowest material level is associated with the highest
discharge rate.
From the experiments reported, it can be seen that the extent to which the mass
flowrate depends on the H / D ratio varies with the outlet size of the bin and the
particle size of the bulk solids. For the coarse solids the effect of level is
insignificant, while for the finer solids, the effect of level becomes significant,
especially w h e n the solids are discharged from the hopper with the larger outlet.
Specifically, for Sand M 5 discharged from the 0.0445 m outlet, the flowrate
reduced 2 2 . 5 % as the H / D varied from 0.07 to 4.2. In addition, the flowrate of
P V C powder from the 0.0445 m oudet reduced 52.5% as the H/D changed from
0.07 to 3.93. In comparison, w h e n Sand M 5 and P V C powder were discharged
from the 0.020 m hopper outlet, the flowrate decreased 13.5% and 20.9%
respectively, as the H/D varied from 0.07 to 3.93.
The extent to which the mass flowrate depends on the H/D ratio is also influenced
by the particle size distribution. Sand M D 1 to Sand M D 4 had the same median
particle size but different size distributions, however, from Figure 8.3 it is seen that
the effect of level on the flowrate is different for each of these solids. This confirms
that the use of median particle size alone is not sufficient to describe the size of
solids mixtures; permeability is a more suitable parameter to describe the flow
behaviour of the bulk solids, as discussed in Section 7.3. From Figure 8.3 b) it is
noted that the flowrate line of Sand M D 3 crosses the line of Sand M D 2 at high
surcharge level (H/D » 4). W h e n H/D < 4 the flowrate of Sand M D 3 is less than
that of Sand M D 2 while H / D > 4 the flowrate of Sand M D 3 is greater than that of
Sand M D 2 . It is believed that since the permeability of Sand M D 3 is very low,
w h e n Sand M D 3 is discharged from the hopper with 0.0445 m outlet at high
surcharge level, the negative air pressure gradient generated at the hopper outlet is
high enough to introduce the fluidisation effect at the hopper outlet, as discussed in
Section 4.7.2 and Section 5.3.1, resulting in the higher flowrate. M o r e details will
be presented in Section 8.3.
As described in Section 4.7.2, both theoretical and experimental results on air
pressures indicated that the negative air pressure increases with increasing
surcharge level; Willis ( 1 9 7 8 ) ^ found that the negative pressures depend on the
hopper oudet size; Crewdson et al. (1977)'-50-' provided evidence of the variation in
the negative air pressures for different sand mixtures; Head (1979)'- ^ and Spink et
al. (1978) L
J
examined the flow of fine sand mixtures from a conical hopper and
plane flow hopper fitted with variable orifice diameters respectively. In agreement
with the above findings, the present experimental observations indicate that the
negative pressure gradient retarding the flow of a fine bulk solid will increase with
increasing hopper oudet size or with increasing surcharge level.
A comparison of present experimental results with others found in the literature is
tabulated in Table 8.3. This comparison has been restricted to situations where the
bins had either conical hoppers or flat bottoms. From Table 8.3, it is obvious that
the effect of material level in the bin on the flowrate is insignificant for coarse
solids. T h e result of Beverloo^ 4 * for fine sand (which roughly corresponds to the
Sand M 5 ) is interesting. H e reports that the dependence of flowrate on material
level is insignificant. While the present results for the Sand M 5 show a significant
dependence on material level for both outlet sizes. It is felt that there are two
principal reasons for the difference, namely
i) Beverloo used outlet sizes of ranging from 2.5 to 10 mm compared
with 20 m m to 44.5 m m in the present work.
ii) Beverloo's bin had a flat bottom which caused funnel flow and
prevented the establishment of significant negative pressures in the
converging flow channel. Such a situation can be compared with a
hopper having semi-permeable walls.
A further comparison of experimental flowrates was made with theoretical
predictions generated using eqn (5.17), resulting in the general variations plotted in
the Q D - C 0 diagram, shown in Figure 8.4. The general variations m e a n the
variation obtained from dispersed data as shown in Figure 5.22, and Figures 5.27
to 5.30. In Figure 8.4 both experimental and theoretical results demonstrate that the
flowrate of fine material flowing from a mass flow bin decreases as the surcharge
level increases. This effect of material level depends on the oudet size of the bin and
particle size distribution. For material with higher permeability, the effect is
negligible. A s the permeability diminishes, the effect becomes more significant.
However, w h e n the permeability of particles becomes very small, no steady flow
occurs, as discussed in Section 8.3.
From Figure 8.4, it seems appropriate to delineate the conditions when material
level has a significant effect on flowrate using a similar concept to the critical
permeability discussed in Section 7.5. T h e value of C Q where material level is
insignificant is approximately 2C cri . This is confirmed by referring to Table 6.8
where it is seen that the Sand M l , the only sand whose flowrate was not affected
by material surcharge level in the experiments, had a permeability constant of about
6500 * 10' 9 ( M 4 N" 1 Sec"1).
8.3 Problems Which May Be Caused by Increasing the Surcharge Level
As mentioned in Section 5.3.1, for the low cohesive fine material with very low
permeability, the negative air pressure generated near the outlet of a mass flow bin
could be high enough to effectively fluidise the particles. The fluidised particles
contain m o r e voidage and provide s o m e relief to the high negative pressure
gradient. T h e reduction of the air pressure gradient results in an increase in particle
flowrate, and m a y even result in flooding. Provided the flooding is not too severe,
8
o
i
5
E
8
>
Xi
cs
O
"O
TJ
Q
&
•e-S
4-C
o
Cfl
s
T9S
o
d
I
oo
TJ
o
•S
NO ©
CO
4-1
W
u
3
i
CO
CM
CO
8
d
13
T)
1
co
B
B
cs
CO
O
odd
*r\ •*+ •*+
O O O
£
CO
I
o
VO
CO CO CO
!!§.
° 2 2
•-H O O N
N CI VI
I I I
I t I
W-J >/"> lO
cs" cs" CS
o
r-
>n
cs
§
00
CO
ro
CO
o
d
d
TJ TJ
CO CO
E E
E E
« « £
OH
co in oq
I-H cs vi
tt> H
!!>
co < CH
CO
3
cs
m
8
CO
CO
5
cs
o <o
H
~H O
o o
>n >n
cs
VI
ON
CS
J I H O
ON
o
r~
co
t
o
d
CO o o
cs
3
cs cs
A V
O o
8
i i )
22^
u-i
ON
cs
d
!
TJ TJ TJ
vn
a
i-H -H
odd
f-
o
l l i
m v©
CS CO
NO
cs
pn
odd
CO
ON
I
<*3
o
d
CQ
•* 00 ID Vfl
no CS CS ON ITi
2 ON -H O -H
ON — T — ro
CS
cs
00
o
CO
CO
o
«8
CO
cs
I
•<fr
.. CO
t
O
NO
CO
«i.s
« s s S §.1
«5 a .S
E o« E o CQ E o«
vx
m
oo
.1
o d d23
d
to « ^ co 2
c S S S S
ISESE
00
sss
o
oo
oo
d
(
oo
c-»
o
C C. j- ao Q
d
I
I
ON
CO
i-H
o
ON
d
CM
•g
fa
to
P
JS
I
3
«S £
E
© CQ
CQ
E
I"
CQ
•a a
132
W
ON
23 ON
»5
w
13 p
S2
B C*
I 2
> i-H
CQ
o
CO
ON
CO
«o
"S3 00
E
8;
co r*
I
204
•3
1
•
1
•
1
M
11
•
^1
o
•
•
b
d
Hi
•
II
^"E|Q
1 111
I
i
1 111
o
o
o
>o
m
i-H
1 11 1 1 1
II
d
y^x\a
o
1 \l\l 1 1
©
o
•
+->
©
en
4-i
II
• >n
•*"
OO
a
I-H
9
00
i-H
1
d £
WvW
m
o &
©
\\\w
CO
\\v^
ffijp > \ ^
Xi
ccj
II
J
1
OH
i-H
*
©
©
©
/E|Q
y/>
/
• *
1
<u o
"">
II
O
en
II
o
U
cs
ffi|Q
•
«o /
!—•
II
ffilQ
1
©
©
©
^^W^^^^v^^V.
5 X
dII '
•
ffi|Q
00
cs
©
(0QS/SS[)
©
" £ ) 9JttIAV0fcI SSBJM
©
<s
©
p
ft
the high negative pressure gradient gradually builds up again until fluidisation
occurs again; no steady state flow results. Increasing the surcharge level aggravates
this periodic phenomenon indicated in Figure 5.13, since a higher material level
results in a greater air pressure gradient being relieved by the fluidising effect. The
results of Sand M D 3 flowing from a 0.0445 m outlet at higher surcharge level were
slightly affected by this flooding phenomenon. For Sand M 6 and Sand M 7 with
lower permeability, this effect occurred even for flow from the 0.020 m outlet;
Figure 8.5 shows the effect of material level on the flowrate of Sand M 6 and Sand
M 7 flowing from the 0.020 m outlet. In this figure, the pseudo-steady flowrate
represents a m e a n flowrate for cases where periodic flow takes place regularly with
high frequency (see Section 5.3.1).
From Figure 8.5, it can be seen that for Sand M6 flowing from 0.020 m outlet the
variation of flowrate with the material level H / D ratio is similar to that for Sand
M D 3 flowing from 0.0445 m outlet, Figure 8.3; flowrate decreases at lower
surcharge levels then tends to increase at higher surcharge levels due to the
fluidisation effects. The only difference between these two cases is that when H/D
is greater than about 2.7 (point a, Figure 8.5) the flowrate of Sand M 6 is greater
than that at the lowest surcharge level, while the flowrate of Sand M D 3 at the
lowest surcharge level is the highest flowrate for the range of surcharge levels
tested. T h e difference in these trends can be explained as the fluidising effect at
higher material levels being more significant for Sand M 6 case. However, from
Figure 8.5, the variation of flowrate for Sand M 7 gives evidence of a typical
flooding phenomenon, the flowrate increases as the H / D ratio increases. Rathbone
et al. (1987) [72] suggested that the flowrate of fine powder during flooding can be
estimated by an inviscid flowrate model where the flowrate is proportional to the
square root of height of material in the bin (measured from the hopper outlet). T h e
inviscid flowrate fitted from the experimental data for Sand M7 is also plotted in
Figure 8.5 where it can be seen that the variation of flowrate with surcharge level
for Sand M7 is close to the inviscid flowrate curve.
More information observed on unsteady flow for Sand MD3, Sand M6 and Sand
M7 has been illustrated in Figure 8.6 and Figure 8.7, showing the mass variation
plots of Sand MD3, Sand M6 and Sand M7 flowing from both 0.020 m and
0.0445 m outlets. From these figures the flow of bulk solids is more unsteady
when the particles discharge from a larger hopper outlet, at a higher surcharge level
and/or as the permeability of the bulk solid decreases, confirming the considerable
effect of these three parameters on the particle flow.
0.07
"SB
0.06 -
ctt
O
E
OH
0.05 -
0.04
>N
*
CO
0.03 -
3
VI
0.02
H / D Ratio
Figure 8.5 Pseudo-Steady Flowrate Affected by Surcharge Level
(from 0.020 m Hopper Outlet)
207
0
At Level 1
'
•
'
•
•
•
•
•
•
•
'
•
•
( M 4 N" 1 Sec"1)
At Level 4
At Level 3
Sand M D 3
= 575.561 * 10"9
•
32.1
21.1
-/*,'*
•"W-l
17-1
jgr
1
tCU.I
gll.l "
" •••
m >••
S.I
-i
, At Level 4
X 4.1
H 3.1
*':i
:
0
Sand M 6
= 297.621 * 10" 9
( M 4 N" 1 Sec'1)
:
. Si.iee.isejeuS8 JIB J H ^ I I .4 SI d l l J H j N M
J- W i
TIHECSicondsl'
I ' I ' I ' I ' I ' t ' I '
Sand M 7
: 0 = 246.341 » 10"»
At Level 4
Figure 8.6 Mass Variation Plots for Sand M D 3 , Sand M 6 and Sand M 7 from 0.020 m Oudet
(in these plots, 1,2,3 and 4 represent different runs in experiments)
( M 4 N" 1 Sec"1)
208
0
Sand M D 3
= 575.561 * 10"9
(M 4 N"1 Sec"1)
Sand M 6
C 0 = 297.621 * 10-»
( M 4 N" 1 Sec"1)
TIHElS.eond.l"
Sand M 7
C 0 = 246.341 * 10"»
( M 4 N"1 Sec"1)
Figure 8.7 Mass Variation Plots for Sand M D 3 , Sand M 6 and Sand M 7 from 0.0445 m Outlet
(in these plots, 1,2,3 and 4 represent different runs in experiments)
Further observations were conducted on the flow behaviour of Sand M 7 to present
more details about the unsteady flow. In Figure 8.7, w h e n Sand M 7 discharged
from 0.0445 m outlet at levels 2 to 4, the slow - fast periodic flows are observed.
This phenomenon was also observed by Miles et al. (1968)^ \ as presented in
Figure 8.8. For periodic flow, the particle flowrate varies between two values,
Q l o w and Q^o^ Figure 8.9 shows the flowrate variations obtained from the mass
variations of Sand M 7 in Figure 8.7. From Figure 8.9, it is confirmed that the flow
of Sand M 7 w a s steady at the lowest level (at level 1, H / D = 0.07) with the
flowrate of Q i o w (approx 0.2 kg/sec) but unsteady at higher surcharge levels with
the flowrate varying between Q l o w and Qh}gh> where Q l o w is about 0.2 kg/sec and
Qjjlgjj is about 0.6-0.7 kg/sec. This indicates that Q l o w is the flowrate unaffected by
fluidisation and Qjugjj is the flooding flowrate.
i
i
i
i
i
i
i
i
CONTEN
BOO
-
cc
UJ
\7-b cm ORIFICE
g; 400
WEIGHT
o
u_
X
o
^ c ^ J ^ 2 a n ORF1CE
o
\
t
20
40
i
60
i
80
i
IOO
i
120
i—
HO
i ^-H
160
TIME(J)
Figure 8.8 Weight Curves for Discharge of 55 |im Calcite
(adopted from Miles et al. (1968)[44])
210
• i • i • i • i • i • i • i • i • i • i • i • i'
i • i •
*n
i/»
CB—.
"o
*•
<n
<MUJ
<u
3
oE
m
5
O
oB
CM
.3
t/»
•
w»
i • i •
i •! • i
'
( 3»./8n) 3 1 H u
i
i
•
i
bO
C
• i-H
two
S3
Xi
o
vt
['••/8D 31HU M01J
M01J
oo
CM
4—1
I ' I ' I ' I ' I ' I ' 1 '-l-'-v^Ji
1
I ' I 'IS
| • | • | ' | • | ' | ' | ' | ' I ' I ' I ' I ' I ' I '
O
5
§
•a
cci
"S
>
i
5
O
o
<
^
E
, I • I • I • I • I . I . I . I • I • I • I • I ./I • I
. n r j - o ^ o r . j ^ . ^ r g
(3e»/6if)
3 i uy
MOTj
• I
• I • I
• I
• I •
I-
CJ»
(3»./8^|)
o>
r~
UD
I .1 I I I I
»/»«»ff**vJ
3ioy MOId
ON
00
3
bfl
E
In order to diminish the fluctuations during flow and obtain a steady flowrate, a
small variable speed belt feeder, as illustrated in Figure 8.10, w a s designed to
control the attainable steady flow. A tachometer was used to measure the belt
velocity. The distance between the belt and outlet of the test bin was maintained at
10 mm.
a) Belt Geometry
b) Belt Feeder Configuration
Figure 8.10 Variable Speed Belt Feeder
The observations were carried out to focus on the effect of feeder belt velocity on
flow behaviour of Sand M 7 from 0.0445 m hopper outlet. The flowrate variations
obtained for different belt velocities are plotted in Figure 8.11 a) to f). These results
show that reducing the belt speed reduces the extent of unstable flow. In particular
when the belt velocity was lower than a critical value, steady flow occurred. For the
flow conditions observed, this critical value was 0.2 kg/sec which was the value of
Q l o w in Figure 8.7. This suggests that Q l o w is the m a x i m u m attainable steady
flowrate for this particular bin configuration and material.
8.4 Summary
Summarising all the effects of surcharge level on the flowrate, Figure 8.12 gives a
general description in which the bulk solids have been grouped into three regions
A, B and C. The flowrate of a bulk solid with permeability constant in Region C is
independent of the surcharge level, while the flowrate of bulk solids with
permeability constants in Regions A and B will be affected by the material
surcharge level. In particular the flow of a bulk solid in Region B is steady and the
flowrate decreases with an increase in surcharge level; the flow of a bulk solids in
Region A is unsteady and the flooding phenomenon m a y occur, i.e., the flowrate
increases as the surcharge level increases. The b - b line is a criterion to decide
whether the effect of surcharge level is significant or not. This criterion was
assumed to be 2C c r i in Section 8.2. The a - a line is a criterion to distinguish
whether the flow is steady or unsteady depending on the surcharge level. These
two critical lines depend on bin geometry and particle characteristics as mentioned
in Chapter 3. According to the observations discussed in this chapter, the
classification of the flow for all sand mixtures in Figure 8.12 is listed in Table 8.4.
However, to describe the flow in Region A or determine the criterion a-a line
quantitatively, further work is required.
Table 8.4 The Classification of the Flow for All Sand Mixtures Used
in Experiments in Terms of the Effect of Material Surcharge Level
Permeability
Constant C Q
*10-9
In Flowing from
In Flowing from
0.020 m Outlet
0.0445 m Outlet
(N^N^Sec" 1 )
Case
Case
SandM7
246.341
Region A
SandM6
297.621
Region A
Sand Mixture
Region A
close to a-a line
SandMD3
Region A
575.561
close to a-a line
SandMD2
1054.413
SandM5
1155.835
SandMD4
1335.804
SandM4
1535.968
SandMDl
2227.654
SandM3
2327.521
SandM2
4353.567
Sand M l
6517.522
Region B
Region B
Region C
Region C
I
•
I
'
1
r
No Feeder
Qi ow
Z.
120
80
160
200
240
Time (second)
a) Without Feeder
1.0
0.6
lt + 9
£ °-4
02
1
1
•
1
1
1
'
1
r
I
I
Vbelt = 0.638 m/sec
&*> 0.8
8
1
T
•
9
iffiQrf
1 0.0 0
SJ
40
80
120
^low
I
160
200
240
Time (second)
b) With Belt Feeder Velocity of 0.638 m/sec
80
120
160
Time (second)
c) With Belt Feeder Velocity of 0.367 m/sec
240
120
160
240
Time (second)
d) With Belt Feeder Velocity of 0.247 m/sec
1.0
—1
1
1—
0.8 -
v
1
1
belt = ° -
200
1
1
1
1
-
"^sec
0.6
0.4 -
..J^l
02
0.0
-
Qlow
«
40
1
80
...
_!__..
120
.J
1
160
1
200
240
200
240
Time (second)
e) With Belt Feeder Velocity of 0.200 m/sec
1.0
0.8
Vbelt = 0.193 (m/sec)
0.6
Vbelt = 0.108 (m/sec)
0.4
40
80
120
160
Time (second)
f) With Belt Feeder Velocities of 0.193 and 0.108 m/sec
Figure 8.11 Flowrate Variations with Different Belt Feeder Velocities
for Sand M 7 from 0.0445 m Hopper Outlet
216
g
bfl
1
CO
X)
1
.8
i
u
VI
a
8
•xl
1
.£2
o
§
•0
f
C«
Q
o
cN
(D3S/Sjf) <IQ ajBlMOJJ SSBp^
217
Chapter 9
Simplification of the Model for Predicting the Particle Flowrate
From Chapters 4 and 5, the solution of the model for predicting the flowrate of
particles from a mass flow bin involves a series of integrations which have to be
solved by numerical methods. It is necessary to simplify the solution procedure of
the model for practical uses. This can be done by concentrating on the simplification
of the model for the dynamic deaeration coefficient Kdea.
9.1 Original Model for Estimating the Dynamic Deaeration Coefficient K^a
In Section 4.1, it is assumed that the pressure distribution is continuous in the
whole of the bin and hence from equation (4.27)
a-1
T W „_i
1o I tfl ~ d-Kdea)
9 f
f
J
ili
f\i [fiT - U - l W f r J
1J *l + f ~ ~2 dTl
a
•'Hmp
(Tl/rjo)
fnmp[f3 a " 1 -(l-K d e a )f3 a ]
X
dri = 0
(9.1)
2
(Tl/Tlo)
where
^•^{^"-"^^"V- ™
>2v
218
bip K o
bop
(9.4)
f3=l + ^ 2 ( T 1 - T l * ) 2 C
Po
From eqn (4.6)
P m p = Po + b i c K 3 Olmp-Tl*) 2c
then
blc K 3 =
P m p ~ Po
^
(Tlmp - "H*)
also from eqn (4.28)
1
" Kdea -
Po
Pmp
then
Kd,ea
Po
P m p - Po - i . K d
ea
Hence, on substitution {2 an(* h m &¥& (9-3) and (9.4) become
f2=1+ ££P°H
2
Po
fa=l +
where
r ( Po ^ e a ^ 1
L IP1-P0 !-KdeaJ JV
Kd,
Niea
f T|-Tl*
— T|* ^>2 C
1-KdeaUmp-^J
px is determined by eqn (4.4)
rjmp is estimated by eqn (4.29)
/
T ^
^\\-^\TL"Tlmp
>2C
(9.5)
(9.6)
9.2
Simplification of the Original K^jea Model
9.2.1 Random Simulation of K^ea
As illustrated in Section 9.1, the mathematical model for evaluating the dynamic
deaeration coefficient K d e a . as expressed in eqns (9.1), (9.2), (9.5) and (9.6), is
highly nonlinear. K^ea is an implicit function of 12 variables, i.e.,
K
dea = f(a» blv> b2v, blc, b2c, 5, <|>, p0, D0, D, H, a) (9.7)
where blv, b2v and blc, b2c are not absolutely independent of each other, cf,
Section 4.7.2, where it was shown that the effect of using b l v and b 2 v instead of
b l c and b 2 c in the hopper on the predicted air pressure is insignificant. In these
calculations, the results also indicate an insignificant effect on the Kd e a . Therefore,
eqn (9.7) becomes a function with 10 variables, i.e.,
K
dea = f( *, W, b2, 5, <|>, p0, D0, D, H, a) (9.8)
In order to simplify the expression for estimating K^ea*
a
specific stochastic
simulation, called the Monte Carlo simulation, is used. The Monte Carlo method is
a technique using random or pseudorandom numbers for solution of the model.
R a n d o m numbers are essentially independent random variables uniformly
distributed over the unit interval [ 0, 1 ] (Rubinstein 1981) [138J . Hence, eqn (9.8)
can be in the form shown in equation (9.9) by introducing the independent variables
X listed in Table 5.1.
Kdea = f ( x )
(9.9)
T o simulate the coefficient K^ea m o r e practically, wider ranges for all variables
were selected. From the flow properties of the materials used in the experiments
detailed in Chapter 6, the ranges of some variables for these materials were:
a:
4.5-9.1
bx:
50-264
b2:
0.05-0.12
p0:
593 - 1342
The ranges for these variables in this simulation were selected to be wider than
these ranges, as listed in Table 9.1.
Table 9.1 The X components in Simulation
X
Components
Corresponding Lower Bound Upper Bound Arbitrary Initial
x
x
Random
Parameter
low
up
Number R N Q
x
l
a
4
10
0.17182818
x
2
bi
50
300
0.27182818
x
3
b2
0.01
0.15
0.37182818
x4
5
20
70
0.47182818
x
5
<t>/8
0.4
0.9
0.57182818
x
6
Po
500
1700
0.67182818
x
7
D0
0.002
0.1
0.77182818
x
8
»/D0
2
50
0.87182818
Xg
H/D
0
5
0.97182818
a
5
60
0.70732926
x
10
The samples were drawn from a uniform distribution on the unit interval [ 0,1 ].
(9.10)
*i " x ilow+ (xiup " "ilow^Lj
where
RN
- the random number
i
j
- component of variable (i = 1, 2,..., 10)
- jth sampling of the variables (j = 1,2,..., N s )
In this program take N s = 5000
A simple uniform 0-1 generator used to create RN- is illustrated in Figure 9.
When R N 4 • i = 0.0, RN: • i was replaced by an arbitrary random number to
avoid "no random number obtained". In fact, this case did not apply in the
simulation processing.
Yes
RNjj.! =3.1415926
~
»
R N T = RN.j.1*RNiJ.1
-J*-
R N T = 10.0 * R N T
Yes
RN.. <== Decimal Part of R N T
T
Figure 9.1 The Flowchart of Uniform 0 -1 Generator
9.2.2
Sensitivity Analysis
In eqn (9.9), there are 10 variables to determine K d e a . The following sensitivity
analysis was used to find the contribution of every variable to Kd e a -
The first part of this sensitivity analysis included:
• calculating the original coefficient K^ with all of variables varying in their
o w n range listed in Table 9.1 using the K d e a model in Section 9.1, then,
• setting one of the random numbers as constant (0.5 in this work) each time
to obtain relative dynamic deaeration coefficient (Kd e a)i f° r every
component of X, and
• comparing (Kdea)i ^^ Kdea by calculating their variance, i.e.,
j=Ns
(Total Variance^ = ^
[ (Kdea>i j - K$!a j ] 2
(9.11)
During these calculations, the random numbers were created in the same way for
calculating K ^ a
a
nd (Kd ea )i to ensure the results were comparable in eqn (9.11).
The physical meaning of eqn (9.11) is that a lower total variance for component i
indicates less error produced by setting component i as the constant value, or
alternatively, a higher total variance corresponds with a more sensitive variable.
The results of this analysis for every component of X are plotted in Figure 9.2.
F r o m this figure, the first six most sensitive variables to K d e a ^ e x2> x 6' x 3' x 7'
x 4 and Xg.
Serial Number of Component with Constant Value
Figure 9.2 Total Variance for Every Component (5000 Samples)
The second step was to set the four relatively insensitive parameters as constants
and then check the total variance caused by this change. The result is plotted in
Figure 9.3 (named case 11). These four constant values were taken as follows:
a = 7.0
a s R N t =0.5
$ = 0.65 5 (degree)
a s R N 5 =0.5
D=26.0D0
a s R N 8 =0.5
a = 20.0 (degree)
a s R N i n =0.27
From Figure 9.3, it is clear that keeping these four variables constant, the total
variance is larger than that from their individual cases but the first six most sensitive
variables to Kdea *xtstm< x2» x6» x3» x7» x 4 an£i x 9' w n ich means the parameters
b,, p 0 , b 2 , D 0 , 5 and H/D are more important to Kdea ma *i the others.
224
1
2
3
4
5
6
7
8
9
10
11
Serial Number of Component with Constant Value (or Case Number)
Figure 9.3 Total Variance for Every Case (5000 Samples)
9.2.3 Simplification of Kdea Equation by Optimization Technique
Knowing the most important variables affecting Kdea makes it possible to simplify
the complex model for Kdea- O n e °f the most effective methods for this purpose is
the use of an optimization technique.
a) Assumed Explicit Models for Kdea
To assume a reasonably explicit model for Kdea> ^ is essential to judge how the
important variables bj, p 0 , b 2 , D Q , 5 and H/D contribute to the coefficient Kd e a i) Outlet diameter D Q is an essential parameter to the particle flow.
Theoretically, when D 0 = 0, there is no particle flow at all, then certainly
Kdea = 0- Th* s requirement should be included in the assumed explicit model
for Kdea-
ii) H / D ratio affects the air pressure gradient and further the particle flowrate.
From Figure 4.9, the bulk density at the minimum pressure position under
higher surcharge conditions is greater than that under lower surcharge
conditions. Therefore, from eqn (4.28), Kdea
can
be assumed to be
proportional to the H/D ratio. Considering Kdea * 0
as
H/D = 0, the assumed
explicit model Kdea includes the term [l+c1(H/D)C2], where Cj and c2 are
constants.
iii) The bulk solids (sand mixtures) used in the experiments detailed in Chapter 6
were used to examine the contribution of bulk density constants b1? b2,
p0 and internal friction angle 8. It was found that the VfO^.pn) ratio
b,
or — r r r ratio (examined for n= 1,2,3 and 4) provided an apparently
Poe^
b
i
predictable relation to Kdea- Figure 9.4 shows variations of Kdea with the —
Po
bi
ratio and
e
^r— ratio at surcharge level H/D = 1.5. However, no apparently
Po ^
predictable relation to K d e a for internal friction angle 8 or sin(8) was been
bi
found until considered together with the—ratio. Figure 9.5 demonstrates the
Po
b
variations of ICjea with sin(8) and
i
— T T T ratio.
P 0 (sin8r 1
Therefore, the following three explicit models have assumed to represent Kdea:
Model 1
v e /
i.e.,
( 1 V7
y
[ yq
«=^Ka fef —
(9.12)
ve *J
0.3
p
0.2 -
I
'
' — ~l
—1
1
•
-i
1
0.3
'
Ratio
o
•s
0,2
o
Xie^5
o.
-
- 0.1
__l
bi
rr
Xi
•
0.10
i
0.0
0.06
- 0.1 cN
0.08
0.12
K
Ratio
0.14
0.0
0.16
0.18
dea
bi
b,
Figure 9.4 Typical Variations of Kd e a with the - - ratio and — j g - rat10
1
Po Poe
050
T
>
T
r
«
1
>
1 0.3
0.48
0.2
0.46
to
1
CO
c
• fH
CO
• iH
0.1
042 r
CO
Ratio
01
[p 0 (Sin8) ]
040
038
0.06
0.08
0.10
J
0.12
u
0.0
0.14
0.16
0.18
K dea
Figure 9.5 Typical Variations of Kd e a with sin(8)and
— Q - T ratio
p0(sin 8) "
i
Model 2
K
Y2
.sun
d e a - Yi D o
yen
VPo e
J
i.e.,
x
K
d e a ~ Yi x 7
2 V3/
2x 3
^x6e
J
1
\y 4
{•d^T^y^
(9.13)
Model 3
.sim
K
d e a = yi
(
bi
lPoe
V2
f
\Y3
2b 2 J V sin 8
[••»(§n
i.e.,
/- i — \Y3
sim
Kdea = yi
JC2_^y2 4*i
2x-x
^x 6 e J ;
\ sin x 4 )
[1 + y 4 4 5 1
(9.14)
where x 2 , x 3 , x 4 , x 6 , x 7 and Xp are simulation variables
yi (i = 1, 2,..., n) are optimization variables
n = 7, 6 and 5 for Model 1, Model 2 and Model 3 respectively
b) Objective Function for Optimization Analysis
The total variance of using explicit model Kdea instead of implicit .
Ko^na is utilised
as the objective function for the optimization analysis, viz., a set of optimal val
are selected for variables Y=[ yh y2,..., yn ], to enable K^
to
best simulate K^a
for all of the N s samples. Basically the mathematical treatment is to minimize the
objective function 0 ( Y ) , i.e.,
minO(Y) (9.15)
where
j=Ns
c) Optimization Process and Results
Eqn (9.15) is an unconstrained optimization problem with 5-7 variables. The
objective function 0(Y) m a y not be continuous. T o solve this problem, the simplex
method of unconstained minimization devised by Spendley, Hext and Himsworth
(1962) [139] , and later improved by Nelder and M e a d (1965) [140] seems to be the
appropriate method because of its robust properties such as:
• simple and reliable
•
no requirement for continuous objective function
•
no calculation of the gradient needed for the implementation of the
algorithm
(Avriel in 1976[141] and Subrahmanyam in 1989[142]).
The simplex method of unconstrained minimization, called 'Flexible Polyhendron
Search* by Himelblau (1972) [143] , is briefly explained as follows:
Consider the minimization of the real function <D(Y), let Y 1 , Y2,..., Y n + 1 be the
points in R n that form a current simplex.
YGR"
Select the vertices at which the highest and lowest function values occur, i.e.,
points Y and Y1 be defined by
<D(Yh)=
max
<D(Yi)
(9.16)
O ( Y1)
(9.17)
i= 1,2,..., n+1
and
O (Y1) =
min
i= 1,2,..., n+1
Denote by Y the centroid of all the vertices of the simplex except Y , i.e.,
n+1
Y
=
IY1
n
- Y1
(9.18)
i=l
The strategy of the algorithm is to replace the vertex of the current simplex Y
which has the highest function value by a new and better point The replacement
of this point involves three types of steps: reflection, expansion and contraction
which are detailed by Nelder et al.[140], Avriel[141] and Himelblau[143].
Nelder et al. suggested that the calculation of the algorithm should be terminated
(
n
n
where
+
\
1
2
[ ( Yi)
X °
V2
E is a predetermined positive number
In this program, E = 10"
The program flowchart is plotted in Figure 9.6.
( ?)]2
(9.19)
- °
< E
I
Start of Program
|-
Createthe Random Number Nj
;
and Calculate X; ;
»> j
*> J
(i=l,2,...,6 andj = 1,2,...,NS)
Calculate initial Y
k
I
and <D(Y k ) (k = 1,2,.... n + 1)
Determine h, 1 and Calculate Y by Eqn (9.16) to Eqn (9.18)
Y* = Y + a ( Y
Reflection Step:
Calculate «D ( Y r )
Reflection Coefficient a = 1
-Yh)
I
Calculate Second Highest Function Value:
0(Y*)=max{0(Y i ),YUY n }
Yes
I
Expansion Step: Y *e = Y- + Y ( Y r - Y )
Calculate 0 ( Y e )
Expansion Coefficient Y = 2
Select
0( Y**) = min { 0 ( Y h ) , 0>( Y r ))
J
Contraction Step: Y C = Y + p (Y**- Y )
Calculate <&(YC)
Contraction Coefficient p = 0.5
Replace All Y 1 :
i
= i T •%-
( i = 1,2,.... n+1 )
N o ^--'inequality (9.19)^
is existing ?
Yes
End of Program
Figure 9.6 Program Flowchart for Optimization Processing
The results of the optimization applications for the three models (9.12), (9.13) and
(9.14) are listed in Table 9.2.
Table 9.2 Simplified Kdea Model Minimization Results
Optimization
Model 1
Model 2
Model 3
yi
0.6879528
0.690126
0.6631737
Y2
0.06456032
0.06396075
0.8408236
Y3
0.8421384
0.841538
0.1125652
Y4
0.08429366
0.08336415
0.01541624
Y5
0.01513928
0.01500192
0.883275
Y6
0.8753874
0.8770201
_
Y7
Minimum Total
1.610751
.
-
Variance
0.4070573
0.4074295
0.4182612
Variables
(for 5000 Samples)
9.2.4
Comparison of Kdea Using the Original Model and the Simplified Model
From the sections above, it is obvious that the three simplified models for Kdeaeqns (9.12) to (9.14), and Table 9.2 give a simulation and simplification of the
original model presented in Section 9.1 under the wider parameter ranges listed in
Table 9.1.
Comparison of Kdea
usin
g the original model (KJjea) and the simplified models
(Kdea) involves two aspects.
O n e covers the wider ranges for every parameter in Table 9.1. This part of the
work has been done by the optimization process. The values of m i n i m u m total
variance caused by the three simplified models for all 5000 samples, as listed in
Table 9.2, indicate the achievement of a quite satisfactory simplification for KdeaThis is evident from the fact that the m i n i m u m total variance implies that the
absolute average differences between original model and three simplified model
for every sample are 0.0001276,0.00012766 and 0.000129346 respectively.
Another comparison of Kdea and K ^ a is carried out for the sand mixtures used
in the experiments. The outlet diameters and H / D ratio considered are: D Q =
{0.01, 0.015, 0.02, 0.025, 0.03, 0.035, 0.04, 0.0445, 0.05} and H / D = {0.07,
1.5, 3.0, 4.5, 6.0, 7.5, 9.0, 10.5, 12.0}. The range of H / D ratio used here is
well in excess of the range stated in Table 9.1. The comparative results are shown
in Figures 9.7 to 9.9. In these figures, the results obtained by the simplified
models, eqns (9.12) to (9.14), show good agreement with those calculated by the
original model. In addition, comparisons of the simplified models 2 and 3 with
simplified model 1 are illustrated in Figure 9.10 and Figure 9.11. These figures
indicate that the three models predict similar values for Kd e a-
Further comparisons were carried out to examine the accuracy of the predicted
negative air pressure gradient at the hopper outlet and predicted particle flowrate by
using simplified K d e a models instead of using the original K d e a model for same
materials, outlet size and surcharge level used in above comparisons. The air
dP
pressure gradients at the hopper outlet (-^ ) _ and flowrate Q p are predicted from
eqns (5.16) and (5.17). The results are plotted in Figures 9.12 to 9.17. F r o m the
small deviations evident in these figures, it is obvious that the simplified K ( j e a
models can be used to predict both the air pressure gradient at the hopper outlet and
flowrate with very high accuracy.
0.20
"•'••
—
r
1
Total: 891 points
0.15 -
K
dea
•
\
0.10 *AT9*°
J jjpr*
i
0.05
0.05
0.10
K dga
0.20
0.15
Predicted by Original Model in Section 9.1
Figure 9.7 Comparison of Kdea Predicted by Simplified Model 1
with that Predicted by the Original Model
0.20
'
1—
—•
Total: 891 points
0.15 -
K
T
1
~
>"
j(^'
dea
-
o.io jjjFf"
—
0.05
0.05
.
— 1 _ _
0.10
__i
i
1
0.15
0.20
Kdea Predicted by Original Model in Section 9.1
Figure 9.8 Comparison of Kdea Predicted by Simplified Model 2
with that Predicted by the Original Model
0.20
Total: 891 points
a
K
0.15 -
S
dea
.
x7
1
•3
%*^r
^Jtpc%"
w
o.io -
I
r
I
0.05
0.05
jTjSBr
.
1
0.10
_^
.
_.
•
0.20
0.15
K^ga Predicted by Original Model in Section 9.1
Figure 9.9 Comparison of Kdea Predicted by Simplified Model 3
with that Predicted by the Original Model
0.20
•
0.18
I
1
1
•
1
'
1
-
i
—
i
—
i
i
—
•
i
—
-
•
Total: 891 points
•
0.16
K
0.14
dea
0.12
-
^ ^
11
0.10
0.08
• • ^ •
0.06
•
0.08
.
i
0.10
.
i
0.12
•
i
0.14
i
1 — i
0.16
" —
0.18
0.20
Kdea Predicted by Model 1
Figure 9.10 Comparison of Simplified Model 2 with Simplified Model 1
0.20
0.18 h
en
r
••
5
0.16 -
£
0.14 -
T
»"
Total: 891 points
K
dea
1
tS
0.12 0.10 -
.§
0.08 -
-4
0.06
0.06
0.08
0.10
0.12
0.14
0.16
K dea
Predicted by Model 1
0.18
0.20
Figure 9.11 Comparison of Simplified Model 3 with Simplified Model 1
1
CU
c
•a
Total: 891 points
,
^ooo
I
I
u
-
-8000
1 JKt
•
12000
•
x^
-16000
.
1
-12000
*
1
-8000
Prediction from Original K ^
•
-4000
Model (Pa/m)
Figure 9.12 Comparison of Predicted Pressure Gradient at Outlet
Based on Simplified Kdea Model 1 with that on Original Kdea Model
1
^
1
I
-4000 -
-8000 •
-12000 -
§
u
1
-16000
-16000
-12000
-8000
Prediction from Original K ^
-4000
Model (Pa/m)
Figure 9.13 Comparison of Predicted Pressure Gradient at Outlet
Based on Simplified Kdea Model 2 with that on Original Kdea Model
I
CM
—
CO
-4000 -
-8000 -
i
-12000 -
8
•a
o
-16000
-16000
-12000
-8000
Prediction from Original K ^
-4000
Model (Pa/m)
Figure 9.14 Comparison of Predicted Pressure Gradient at Outlet
Based on Simplified Kdea Model 3 with that on Original Kdea Model
2.0
1.8
1.6
1.4
•
- " — 1 — ' — I -
:
•
• — i — 'r
i
1
1
1
1
1
1 —~r
—•—i
-'—
Total: 891points
:
QP
''
^
1.2
1.0
0.8
0.6
0.4
0.2
*r.
0.0
0.0
i
0.2
. i
i
0.4
.
0.6
i
.
0.8
I
.
1.0
I
.
1.2
I
.
1.4
1
1
1.6
J
L- -
1.8
2.0
Prediction from Original K ^ Model ( kg/sec)
Figure 9.15 Comparison of Predicted Particle Flowrate Based on
Simplified Kdea Model 1 with that on Original Kdea Model
2.0
— I — 1 — l — l — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 — "I
1.8
m
1.2
1
-
Total: 891 points
1.6
14
'
"
-
•
QP
I | • | I | r-|-i | I
1.0
0.8
0.6
0.4
0.2
>\
0.0
0.0
i
0.2
. _i_
0.4
•
•
0.6
•
'
0.8
I
1
1.0
•
1
1.2
•
1
1.4
•—
1
1.6
1
J
1.8
2.0
Prediction from Original K dea Model (kg/sec)
Figure 9.16 Comparison of Predicted Particle Flowrate Based on
Simplified Kdea Model 2 with that on Original K d e a Model
2.0
3
CO
— I
1.8 h
1—l
1
1 — '
1 — • — I
•—i
r-
- i — | — i — i — i -
Total: 891 points
1.6 f-
QD
1.4
1.2
I
B
S
1.0
0.8
0.6
&
a
0.4 h
•S
u
02
a
0.0
0.2
I
0.4
i
L
0.6
J
0.8
i
I
1.0
u
1.2
1.4
1.6
1.8
2.0
Prediction from Original K jga Model (kg/sec)
Figure 9.17 Comparison of Predicted Particle Flowrate Based on
Simplified K d e a Model 3 with that on Original Kdea Model
In summary, all three simplified K d ^ models have the following main advantages:
• The simplified models provide accurate alternative predictions for 1^^. The
analysis conducted indicates that they can be used to confidently predict the
air pressure and flowrate of bulk solids from conical mass flow bins.
• The simplified models are reliable. Since these models were simulated and
simplified for a wider range of parameters presented in Section 9.2.1, the
models are applicable not only to the cases studied in the experiments, but
also to the cases with a wider range of the parameters.
Furthermore, it is noted from Figure 9.11 and Table 9.2 that although the error
predicted by Model 3 is a little larger than that predicted by the other two models, its
239
greater simplicity (only 5 constants) suggests this model be used, i.e., K^ga can be
accurately and conveniently evaluated using the eqn (9.20).
K^=0.6631737
bl
n
y>.8408236|' ^
2b,
\0.1125652
v sin 5 /
0.883275-
l+0.01541624[-jij
j
(9.20)
9.3
The Simplification of the Flowrate Model
Utilizing the simplifying assumptions justified previously, that is:
•
-sim
the simplified dynamic deaeration coefficient Kdea
(e<ln (9.20) is
employed);
•
the bulk density and stress gradients are insignificant, as indicated in Section
5.4.1, from the sensitivity analysis in flowrate model;
• blv and b^ can be used in the hopper section instead of blc and b2c (i.e.,
take bj = bjv ~ blc; b2 = b2v ~ b^).
the flowrate model, eqn (5.17) can be simplified to
a
2 2 Q p 2 + b22Qp-g = 0
(9.21)
where
22
*
f
1 f 2
I A Q p out ) r0
(9.22)
L
.a-1
out
(9.23)
fout=l + ^ ( T l 0 - T l * ) b 2
(9.24)
^^opXpout 11 " 0 "^^ 0111 ^ 080 '
where
K
3
T,S1III
K
L
d» Pof 1 Y^
1-KSS ^V^mp-^J
(9.25)
Substituting eqn (9.25) into eqn (9.24) yields
fout=
1+
Kctea f ^ o - n * b2
l-KgL^mp-^J
(9.26)
Inserting eqn (4.29) and TJ* = 0.95 T|0 into eqn (9.26), then
fout=l +
„sun
^dea
I-K3S
L
0.05(l+T]o)
lb2
0.05+ Timax-0.95 ri0
(9.27)
Rewritting fout in eqn (9.27) as a function of D, D 0 , H, a, results in
„sim
l
out l +
f„„,=
Niea TT
T-U
I-KS
(9.28)
where
0.05(1 +
U =
^ )
H
D0
1.05+ 2^-tan a-0.95-jj
(9.29)
From eqn (9.28), eqn (9.23) is rewritten as
„sim
fa_1
b22 =
r
l-(l-Ita)
out
CoPo^oPout
> cos a
i.e.,
fa"2
x
,
b22 =
out
„sim
(
.
TT
v
(9.30)
— KJea (1 - U ) cos a
Q)PoAo
D,0
into eqn (9.22), yields
Inserting rQ =
2 sin a
2
a22
"lA o P o f o u t J
4 sin a
D0
(9.31)
Finally, the simplified flowrate model eqn (9.21) is solved using eqn (9.20) and
eqns (9.28) to (9.31).
A comparison of above simplified flowrate model with the original model expressed
in eqn (5.17) is carried out for sand mixtures, Sand M l to Sand M 5 and Sand M D 1
to Sand M D 4 described in Chapter 6, discharging from 0.020 m and 0.0445 m
hopper outlets at surcharge levels H/D = 0.07, 1.5, 3.0 and 4.5. The results, as
depicted in Figure 9.18, indicate that the simplified flowrate model agrees well with
the original flowrate model. The simplified model can also predict the flowrate of
bulk solids (coarse,fine,compressible or incompressible) from mass flow bins
(with or without surcharge) and provides a simpler solution procedure (without
solving the integrations numerically). Therefore, for practical purposes the
simplified flowrate model should be preferred over the original flowrate model.
0.5
1.0
1.5
Flowrate Predicted by Simplified Model (kg/sec)
Figure 9.18 Comparison of Simplified Flowrate Model with Original Flowrate
Model For the Experimental Parameters Examined
Chapter 10
Strategies for Increasing Limiting Flowrates
10.1 General Possibilities for Increasing Limiting Flowrates
It is recognised that the flowrate of fine powder from mass flow bins can be orders
of magnitude less that the flowrate obtainable for coarse powders. A s analysis of
current data shows, the low flowrate of fine material is mainly caused by the
negative air pressure gradient at the outlet of a bin. Therefore, one of the most
important strategies for increasing limiting flowrate is to reduce the interstitial
pressure gradient, especially at the outlet. Various possibilities to achieve this
improvement are n o w proposed and discussed.
i) Low Surcharge Level (Double - Hopper Structure)
As detailed in Chapters 5 and 8, a fine powder free flowing from a mass flow bin,
whose permeability constant is located in region B in Figure 8.12, provides a
higher flowrate at lower surcharge level than at higher level. W h e n a large H/D
ratio is required for the storage capacity, a double hopper structure, as illustrated in
Figure 10.1, can increase the flowrate attainable. The upper hopper can be
designed with a smaller hopper angle and larger outlet size than those for the lower
mass flow hopper, the ventilate channel is used to keep the air pressure above the
top surface of material in lower hopper at atmospheric level, for any greater
negative air pressure in this area will cause a greater negative air pressure gradient
at outlet of the lower hopper. O n e of the limitations of this structure is the reduction
244
of storage capacity. A n increase in outlet diameter of the upper hopper can assist
reduce this loss of storage capacity but m a y diminish the effect of the doublehopper arrangement in improving the flowrate. Hence an optimal arrangement must
exist for a particular application. Another limitation is that this strategy cannot
increase the flowrate remarkably, compared with such techniques as air injection or
the use of a standpipe. For instance, for alumina discharging from 0.0445 m outlet
at H/D«2.1, the use of a standpipe of 0.212 m in length (L/D Q =5), can increase the
flowrate by about 2 0 0 % , as illustrated in Figure 10.3, while only about a 4 0 %
increase in flowrate can be achieved by discharging the material at a lower
surcharge level (H/D=0.07), as predicted in Figure 5.2.
Bulk material
surcharge level
Ventilate channel
Upper hopper
Lower hopper
Figure 10.1 Schematic of Double - Hopper Bin Arrangement
245
ii)
Addition of Coarse Material
From research in this work, as highlighted in Chapter 7, fine powder mixed with
coarse material flows more easily as long as the resultant mixture has a larger
permeability constant. Theoretically, the addition of coarse material m a y tend to
increase the flowrate; consistent with the fact that the flowability can be improved
by the addition of coarse material into a fine mixture (Prabhakara R a o et al.
1 9 7 1 [144] & B i r d et
^
1 9 7 6 [63])
Q u a n t i t a t i v e analysis on the addition of an
optimal amount of coarse material into a particular fine powder needs to be
developed in this case. For a large scale solids handling system, some additional
equipment would be needed to separate the coarse particles from the fine ones.
This requirement implies that this method can be only used for small scale systems.
In addition, this strategy cannot increase the flowrate substantially. If the cost of
additional facilities cannot match the benefits, this technique remains theoretical and
not of practical value.
iii) Air Injection
As summarised in Section 2.2, the use of this technique is well documented by
m a n y researchers such as de Jong (1969) [116] , Papazoglou et al. (1970/1971) [81] ,
Sutton et al. (1973) [117 ' 118] , de Lazzari (1981) [119] , Altiner (1983) [82] , Johanson
(1990) [ 1 2 0 ] and W e s et al. (1990) [121] . The negative air pressure gradient at the
hopper outlet generated by fine material, causes an inflow of air from the outlet
which retards the particle flow. T o overcome this inflow of air through the outlet, it
is possible to inject air at certain positions of the hopper to ehrninate the interstitial
air pressure gradient developed during flow. For successful air permeation it is
necessary to control the adverse air pressure gradients at the hopper outlet by
controlling both the pressure and rate of air permeated into the hopper. A s this
246
technique is well documented, the work presented in this chapter will not deal
further with this strategy.
iv) Use of Standpipe
Another technique, which can assist in increasing the flowrate of fine powders
without reducing the storage capacity in the bin and or requiring the addition of
coarse material or compressed air, is the use of standpipes. The simple addition of
a vertical non-converging pipe to the bottom of an existing mass-flow hopper
produces a pressure gradient in the standpipe and creates a vacuum at the hopper
outlet This suction effect will substantially increase the flowrate of bulk solids
from the hopper, as found in m a n y research studies such as Yuasa et al.
(1972) [ 1 2 2 ] , McDougall et al. (1973) [123] , de Jong (1975) [124] , Ginestra et al.
(1980) [ 8 4 ] , Chen et al. (1984) [85] , Knowlton et al. (1986) [ 8 6 ] and Johanson
(1990)t12°l. The effect of the standpipe is greater as the particle size becomes
smaller and the length of the standpipe becomes longer. In this chapter, some work
on the use of standpipes is reported.
10.2 Experiments and Observations on Standpipes
i) Test Apparatus
The experiments were carried out on a modified double-bin apparatus shown in
Figure 10.2, in which the standpipe was installed at the outlet of the test bin of the
double-bin apparatus detailed in Section 6.2. The belt feeder shown in Figure 8.10
was used for a long standpipe to obtain a steady flowrate, as shown in Plate 10.1.
The standpipes used in the experiments ranged in length-to-diameter ratio from 1 to
5, as illustrated in Plate 10.2. The joint between the standpipe and the outlet of the
247
hopper was sealed to avoid leaking. The measurements made were similar to those
without a standpipe as detailed previously (Chapter 6).
ii) Experimental Observations
The observations were made on the flowrate improvement attainable using alumina
as the bulk solid. Figure 10.3 summarises the results: the use of a standpipe
increased the flowrate remarkably; the flowrate increased with the length of the
standpipe. For instance, a standpipe with L/D 0 = 5 provided an increase in flowrate
of over 3 0 0 % for alumina.
iii) Problems When Using the Standpipe
As the higher flowrates were approached with the long standpipe, fluctuations in
the flowrate were observed. These fluctuations were explainable by two typical
discharge modes observed for material flowing through the standpipe: in one m o d e
the standpipe was partially filled with the bulk solids, while in the other m o d e the
standpipe was fullyfilledwith the solids, as shown in Figure 10.4. The higher
flowrate occurred only as long as the standpipe was full of material. Once it was
partially filled with the solids, the limiting flowrate was governed by the void in the
standpipe and a slower flowrate occurred. The transition state between these two
flow modes resulted in flow instability. This instability was suggested by Matsen
(1976)^ 145 ^ to be caused by bubble held stationary in the standpipe. Instability of
the flow is a c o m m o n phenomenon when using a standpipe. The same problems
have been also reported by Leung and Wilson (1973) [125] , Leung and Jones
(1978) [ 1 2 6 ] and Johanson (1990) tl20] . A jump phenomenon has been reported by
Chen et al. (1984) [85 ^ between the upper and the lower bound flowrate in some
situations, with the actual bounds differing by an order of magnitude.
248(-)
Plate 10.1 Bin - Standpipe - Belt Feeder Arrangement
Plate 10.2 Standpipes Used in Experiments
(L/D 0 = 1,2,3,4 and 5 from left)
248
Storage
Bin
Test Bin
Standpipe
Feeder
Q
f)
Figure 10.2 Standpipe Installed at the Outlet of the Test Bin
1
2
3
Ratio L / D Q of Standpipe
Figure 10.3 Flowrate Q p vs. the L/D n Ratio of Standpipe for Alumina
Discharging from 0.0445 m Outlet (Surcharge Level H = 0.31 m )
a) Partially Filled
b) Completely Filled
Figure 10.4 Typical Discharge Modes from a Standpipe
250
T o reduce the extent of these fluctuations and to ensure that the standpipe remained
full at all times a belt feeder was installed under the standpipe. With the belt feeder
installed, the m a x i m u m steady flowrate attained, using the long standpipe of
L / D 0 = 5 , w a s 0.728 kg/sec (Figure 10.5), which represents 9 7 . 4 % of the
m a x i m u m attainable flowrate for discharging from this test bin, as plotted in Figure
10.3. A s Johanson (1990)^ 120J pointed out, it is essential that the feeder controls
the flow slightly below the limiting rate as the higher flowrate occurs only as long
as the standpipe remains full of solids.
10.3 Theoretical Model for Bin - Standpipe Configuration
A theoretical model based on the previous models for predicting the pressure
gradients and the flowrate of bulk solids has been established to explain the
experimental observations for the standpipe.
i i i i i i i
M a x i m u m attainable flowrate with full standpipe
•*— h = 21 m m
•o— h = 2 8 m m
Flowrate without standpipe !
j
"o.O
•
i
1
1
•
1.0
2.0
Feeder Belt Velocity (m/sec)
1
3.0
1
4.0
Figure 10.5 Flowrate vs. Feeder Belt Velocity for Different Clearances between
Feeder and Standpipe Outlet (standpipe L / D 0 = 5)
251
The geometry of the mass flow bin with a standpipe installed is divided into four
regions, where regions I, II and HI are the same as those described in Section 4.1,
while the standpipe is considered as region IV, as shown in Figure 10.6. The same
definitions of dimensionless depth values presented in eqn (4.1) are used in this
model. In addition, the dimensionless depth values for ( h m p )Sp and h 0 S p in
Figure 10.6 are defined as
(Tlmp )sp -
( h m p )sp
hi
h
osp
_
TlOSp" ^
Figure 10.6 Four Regions for Bin - Standpipe Arrangement
(10.1)
252
The assumptions about the use of the standpipe, in addition to those in Chapter 4
and Chapter 5, are as follows:
• The standpipe remains full of bulk solid with the stress distribution in regi
IV being linear with respect to height, i.e.,
o-! = kxri +k2 ( TioSp < rj < rio ) (10.2)
where klt k 2 are constants.
• The minimum pressure position for bin-standpipe configuration (rimp )Sp is
located in the hopper section between T]mp and rig- The effect of suction by
the standpipe is assumed to be dependent on the length of the standpipe in
the form
("Hmp )sp ~ ^ m p " A ^ m p
where Anmp= (1 -e-xL/D°) (rimp-T]0)
i.e.,
/
x
- /1
(ilmp)sp =T1<>(1+
c
-A.L/pn "Hmax ~ ^o x
,^ n
>
1 + TJ 0
nn^
(103)
where A, is standpipe effect coefficient. (0 < X < 1).
This assumption means that the suction effect generated by the standpipe
reduces the magnitude of the negative pressure gradient at the outlet and is
based on following observations:
253
F r o m the theoretical flowrate model, eqn (5.17), the flowrate is decreased
by a negative air pressure gradient but increased by a positive air pressure
gradient. If the air pressure gradient at the hopper outlet was changed from
negative to positive due to the installed standpipe, the flowrate should
change suddenly from a low value to a high value for the flow without
standpipe and with standpipe respectively. However, from Figure 10.3, the
flowrate increased gradually as standpipe L / D Q ratio increases from zero.
This indicated that the reduction of negative air pressure gradient at the
hopper outlet depends on the standpipe length, which can be simulated by
setting the m i n i m u m pressure position in the hopper section between Tl m p
and rjo- E q n (10.3) describes that (n m p) S p equals T| m p for the flow without
a standpipe approaches Tjg with increasing standpipe length. This
assumption is acceptable w h e n studying the flow behaviour of bulk solids
from a standpipe with a limited length.
• The wall of the standpipe is so smooth that there is no resistance to the
flowing bulk solids. Hence, as long as the air pressure gradient at the
hopper outlet has been determined, the flowrate model, eqn (5.17), can be
directly used to evaluate the flowrate of bulk solids from bin - standpipe
configuration.
The pressure distributions in the four regions, similar to those illustrated in Chapte
4, are expressed as follows:
i) in region I (r^ < T| < Tlma*)
2
T]
P_P1 = W 0 - ^ [ [f a - 1 -(l-K d ea)f a ]dTi
Tlf -^l
(4.21)
254
when J\ = ri m a x , P = 0, then
2 MH13JC
"Pi = W
0
^
J
f
[f a - 1 -(l-K d ea)f a ]dTl
(4.22)
T|i Tli
where Pj is the pressure at the transition level and
f is determined by eqn (4.16)
ii) inregionH ((Timp)sp < T] < r^ )
/ x
f1! [f*-1-(l-Kdca)fa]
P-(Pmp)sp = W 0 l
P F
ATlmp)sp
2
dr,
(Ti/rio)
or
.11
[f^-d-Kdea)^]
P = (Pmp)sp + W 0 l
5
dn
•VimpJsp
dl/rjo)
(10.4)
where f is determined by eqn (4.18)
(pmp )sp is me pressure at r, = (Tjmp )sp
when T\ = r\l,P = Plt the pressure will be
Ml [fa^1-(l-Kdca)fa]
Pl-(Pmp)sp = W 0 |
J T
( lmp)sp
iii)
— — 2
(ri/rio)
dn
(10.5)
in region in ( Tj0 ^ rj ^ (Tlmp)sp )
? [f^-d-Kdea)^:
p-p O ut=w 0 |
a
r~~2
dT1
(10 6)
-
255
A s ri = (Ti m p ) s p , P = ( P m p ) s p , it follows that
frlnnO
f^^tf^-d-Kdea)^]
(Pmp)sp-Pout = W 0
where
iv)
losp
JT
- S
(Tl/Tio)2
1O
dri (10.7)
f is detennined by eqn (4.20)
in region IV ( rioSp ^ rj <r\0 )
J1
[
[fa"1-(l-Kdea)fa]dTl
•'Host)
P = W0
(10.8)
At the hopper outlet,
Pout = W
0
^o
I
[fa
^Hoso
1
-(l-Kdea)fa]dTl
(10.9)
Based on the assumption, eqn (10.2), and density continuity, f in eqn (10.9) is
determined by
f =i + ^
1 K
2C
-ri* J ' * Y ^^M - ^ P r v
(T,o_T|
- dealTlmp-Tl*J lTlo-TlospJ
According to the pressure continuity assumption, the air pressure distribution in
bin-standpipe structure will be subject to the requirement
eqn (4.22) + eqn (10.5) + eqn (10.7) + eqn (10.9) = 0 (10.10)
which simply states that the air pressure across the combined bin - standpipe
configuration is zero.
256
E q n (10.10) and eqn (10.3) provide a model to determine the unknown variable
K ^ . The flowrates are predicted by the flowrate model, eqn (5.17), as derived in
Chapter 5.
10.4 Theoretical Predictions and Discussion
To verify this theoretical model, the flowrates of alumina flowing from the test bin
with 0.0445 m outlet have been predicted and compared with the observed
flowrates. A comparison of the theoretical results with the experimental results, for
a sequence of fixed values of the coefficient X, is depicted in Figure 10.7. The
value of X = 0 implies no effect of the standpipe on the m i n i m u m pressure
position.
1.2 i
•
i
•
1
•
1
•
i
1
r
L / D 0 Ratio
Figure 10.7 Comparison of Theoretical with Experimental Results
The comparison indicates that the predicted flowrate increases with increase in
standpipe length, although for the standpipe with a lower L / D 0 ratiothe theoretical
results are over-predicted. The over-prediction of the flowrate at lower L / D 0 ratio
m a y be caused by a different flow m o d e in the standpipe. In particular the
theoretical results are based on the assumption of the standpipe being completely
filled with solids. It is noticed that in experiments the belt feeder has only been
257
used for a long standpipe (L/D 0 = 5). It m a y be possible that the conditions for the
lower L / D 0 ratio standpipes used in the experiments did not match the assumption.
Hence a lower suction effect m a y have occurred in the experimental standpipe.
F r o m Figure 10.7, it is considered that the case X = 0.75 yields the most confident
prediction.
The theoretical model also predicts that the standpipe provides a greater increase in
flowrate for fine material than for coarse material. Defining an enhancement factor
F c „ as the ratio of the flowrate with standpipe to that without standpipe, Figure
10.8 shows the variations of enhancement factor F s p with L / D 0 ratio for alumina,
Sand M D 2 and Sand M l , which have the permeability constants 398.384,
1054.413 and 6517.522 * 10" 9 ( M 4 N" 1 Sec"1) respectively. The results indicate
that a standpipe can create a higher flowrate forfinerbulk solids than for coarse
materials. Specifically, the flowrate enhancement factors produced by a standpipe
with L / D 0 = 7 are 4, 2 and 1.06 for alumina, Sand M D 2 and Sand M l
respectively. These results coincide with other researchers' results, as illustrated in
Table 10.1, indicating that the use of a standpipe increases the flowrate more
effectively for fine material than for coarse material.
0
L_—I_J
0
•
1
2
i
•
1
3
•
•
4
'
•
5
'
•
6
•
'
7
L / D Q Ratio
Figure 10.8 Flowrate Enhancement Factor for Different Materials ( X = 0.75)
Table 10.1 A Summary of the Results in Using Standpipes
Material Used Diameter of Dimensionless Enhancement
Standpipe
Length of
Factor
Researcher (Particle Size
or
Standpipe
D0(m)
Permeability
*sp
IVD 0
Constant)
Fine Sand
Chen et al.
8
in 1984[85] d 50 = 154 p:m
0.0254
130
(Experimental Coarse Sand
2.5
Results)
d 50 = 556 p.m
Ginestra et al. Unnamed
Material
in 1980[84]
0.030
100
7-8
8 = 30
(Predicted
V t = 0.1 m/sec
Results)
Sand
Knowlton et al 53 -177 \sm
6.4
Sand
in 1986t861
0.038
185
5.4
(Measured 177 - 420 *im
Sand
420 - 840 pjn
Glass Beads
dso = 127 p:m
Yuasa et al. Glass Beads
in 1972 [122] dso = 254 Jim
& Predicted)
Glass Beads
Results) d 50 = 505 urnGlass Beads
d50=1015 urn
Alumina
Current
Experimental d 50 = 100 p.m
(C0=398.38)
Results
Alumina
d 50 = 100 p:m
Predicted
(C 0 = 398.38)
Results
SandMD2
by
d 50 = 200 (im
(C 0 = 1054.4)
Current
Sand M l
Model
d 50 = 310p:m
(C 0 = 6517.5)
3.6
5.9
0.0091
165
(Measured
3.25
2.11
1.5
0.0445
5
3.14
4
0.0445
7
2
1.06
259
The increase in flowrate using the standpipe is caused by the vacuum suction at the
hopper outlet and a reduction of the negative air pressure gradient. This is evident
from Figure 10.9 and Figure 10.10 which show the variations of predicted air
pressure and air pressure gradients at the hopper outlet for fine material alumina
and coarse particles Sand M l , respectively. Corresponding to these results,
Figures 10.11 and 10.12 show the air pressure distributions in the bin and
standpipe generated by alumina and Sand M l respectively. From Figure 10.11 and
Figure 10.12, it can be seen that the predicted air pressure distribution in the
standpipe is a linear function of the depth for short standpipes, while it is a
nonlinear variation for long standpipes. In contrast, Yuasa et al. (1972)^
^ and
Chen et al. (1984)™ ^ observed an almost linear relationship between negative air
pressure and the depth in the standpipe; Knowlton et al. (1986)^ ^ obtained similar
observations in most of their experiments. It is possible that the nonlinear relation
produced by the current model is caused by the constraint of the m i n i m u m air
pressure position within the hopper section. Indeed, for a very long standpipe, the
experimental results obtained by Yuasa et al., Chen et al. and Knowlton et al. (for
standpipes of UDQ>100)
show the occurrence of the m i n i m u m air pressure below
the hopper outlet In this case, a higher flowrate can, theoretically, be produced by
the standpipe since a positive air pressure gradient developed at the hopper outlet
accelerates the particle flow. Within the time constraints imposed on this thesis,
only limited experimental results been obtained to verify and/or to provide a basis
to develop the current standpipe model quantitatively. Hence, further work in this
area is needed
260
0
1
2
3
4
5
6
7
L/D 0 Ratio
Figure 10.9 Predicted Air Pressure at the Hopper Outlet (X = 0.75)
0
1
2
3
4
5
6
7
L/D n Ratio
Figure 10.10 Predicted Air Pressure Gradient at the Hopper Outlet (X = 0.75)
261
-o-
L/Do=0
L/Do=l
L/Do=2
I7Do=3
L/Do=4
L/Do=5
I7Do=6
L/Do=7
Transition
level
Hopper outlet
-400 -300 -200 -100
-l.o Q
Air Pressure Distribution (Pa)
Figure 10.11 Predicted Air Pressure Distribution in the Bin and the Standpipe
for Alumina (X = 0.75)
Air Pressure Distribution
Figure 10.12 Predicted Air Pressure Distribution in the Bin and the Standpipe
for Sand M l (X = 0.75)
Chapter
11
Conclusions
Based on the continuum mechanics theory, a theoretical model for predicting the
flowrate of bulk solids from mass flow bins, eqn (5.17), was developed. T o
ensure this flowrate model predicts adequately the flowrate required the selection of
relevant flow properties of the bulk solids (especially the bulk density and
permeability) and an estimation of the air pressure gradient at the hopper outlet.
Considerable work was carried out to simplify the original flowrate model, eqn
(5.17), and to examine the strategies for increasing the limiting flowrate of fine
bulk solids. A series of experiments were performed to verify the theoretical
models developed in this thesis. F r o m above work, several conclusions can be
drawn relating to the particle flow from mass flow bins.
a) From the comparisons of consolidation-related bulk density models and
permeability models with the experiments, the best bulk density models for
practical applications are models 9, 14 and 16, as presented in Table 3.2; the
best permeability expression is Jenike and Johanson1 s model, as shown in eqn
(3.15).
b) 'Permeability' is a useful parameter for use in models to describe the flowrate of
coarse and fine as well as size-distributed particle mixtures in terms of the effect
of air pressure gradient on particle flowrate from mass flow bins. The theoretical
and experimental results indicate that the effect of the air pressure gradient on the
flowrate is significant for lower permeability material (e.g.,fineparticles and a
particle mixture with a size range which includesfineparticles) and insignificant
for higher permeability material (e.g., coarser material).
c) G o o d agreement between the theoretical and experimental results for predicting
the air pressure distribution and the particle flowrate were obtained. This
suggests that the pressure distribution model (as presented in Sections 4.5 and
4.6) and the flowrate model, eqn (5.17), be used for practical applications with
the following distinguishing features:
• they provide adequate predictions (the flowrate model provides the most
accurate prediction relative to the other researchers' models examined in
this thesis);
• they can be used to study the effect of the material surcharge level on
particle flow;
• since the effect of consolidation on the particle flow was considered by
using consolidation-related bulk density and permeability equations, both
models can be used for compressible or incompressible materials;
•
since the permeability w a s used to relate the particle motion to the air
resistance, the models can be used for size-distributed bulk solids.
d) Both theoretical and experimental results indicate that the flowrate of bulk solids
flowing steadily from a mass flow bin increases rapidly at first and then more
gradually as the particle permeability constant increases. A b o v e a critical
permeability value C c r i the flow of the particles can be considered to be
unhindered by interstitial air effects and the flowrate is only determined by the
bulk density and such bin geometry parameters as the hopper outlet diameter and
the hopper half angle (for D 0 » dp cases), as expressed in eqn (5.18).
e) The criterion C^ for distinguishing fine material from coarse material must be
related to the hopper geometry. In addition, as the bulk solids handled often
have a range of particle sizes, one particle size (say 500 Jim which is widely
recognised nowadays^ 28 ' 50 ' 51 ' 55 ' 61 ' 62 ]), cannot describe the criterion very
adequately. A hopper outlet-related criterion in terms of the permeability C ^ , as
presented in eqn (7.7), provides a more reasonable criterion for classifying fine
and coarse bulk solids in terms of the effect of air pressure gradients on the
particle flowrate.
f) Theoretical sensitivity analysis on the original flowrate model indicated that the
bulk density gradient and stress gradient at the hopper outlet can be ignored. The
analysis also verified the air pressure gradient at the hopper outlet is one of the
most important factors in determining the magnitude of the particle flowrate.
This conclusion is valid for unaided (gravity) flow or aided (e.g., air injected)
flow and also valid for any low cohesion material.
g) The theoretical and exrjerimental studies on the effect of material surcharge leve
on the particle flow suggested that particular attention should be paid to the
effect of material level on the flowrate when a fine material being considered. In
this study, three regions of particle permeability, as indicated in Figure 8.12,
have been proposed to generalize the effect of surcharge level.
Region A:
bulk material:
very fine or with very low permeability
particle flow:
unsteady with flooding from the bins likely
effect of surcharge level: significant; the flowrate increases with
increasing material surcharge level.
Region B:
bulk material:
fine or with low permeability
particle flow:
steady; flowrate increases with increasing
permeability of the bulk solids.
effect of surcharge level: significant; the flowrate decreases as the
material surcharge level increases; the lower the
permeability, the more significant the effect of
surcharge level on flowrate becomes.
Region C:
bulk solids:
coarse or with high permeability
particle flow:
steady; the flowrate depends on the bin
geometry and bulk density of the particulate
material and independent of the permeability of
the bulk solid.
effect of surcharge level:
insignificant.
The criterion for judging a significant effect of surcharge level (the boundary
between Region B and Region C ) depends on the bin geometry. F r o m the
experimental results this criterion is considered to be about 2 0 ^ .
h) For fine powders the median particle diameter cannot alone be used to predict
accurately the flowrate. Account must be taken of the particle size distribution.
i) A low bulk density occurs at both ends of a bin and the maximum bulk density
occurs at the transition level for a hopper with surcharge or in the converging
section for a hopper without surcharge.
j) Theoretically, the air pressure distribution in a bin has a pressure minima near
the hopper outlet and a pressure maxima near the top surface of the material. The
absolute value of the pressure maxima is very small compared with the absolute
value of the pressure minima. Therefore, the pressure m a x i m a is visible in
practice only w h e n the hopper outlet is large and/or the permeability of the bulk
solid is low.
k) Under steady flow conditions, the air pressure gradient at the hopper outlet
increases with increasing the outlet size, material surcharge level and/or
decreasing particle permeability, while the particle flowrate increases with
increasing the outlet size, permeability of bulk solid and/or decreasing surcharge
level.
1) Experiments indicated that using a double bin apparatus enables a steady air
pressure distribution and a steady particle flowrate to be observed for a
significant period oftime.The experimental results confirmed that the negative
air pressure generated by flowing particles increased with an increase in material
surcharge level. The measured m i n i m u m pressure position agreed with the
assumption, eqn (4.29).
m) An unsteady flow caused by a high surcharge level has two flowrate values,
Q l o w and Q^gh, where the Q h i g h is the flooding flowrate and the Q l o w is the
flowrate without being affected by fluidisation. The value of Q l o w is also the
m a x i m u m attainable steadyflowratewhen a feeder is installed below the hopper
outlet to diminish the fluidisation effect
n) The dynamic deaeration coefficient Kdea mainly depends on bulk density
constants (p 0 , b l f b 2 ), surcharge level H/D, internal friction angle 5 and the
hopper outlet diameter D 0 . Three simplified models for the K d e a (which resulted
from a combined method of M o n t e Carlo simulation and an optimization
technique) provide a simple but accurate method to approximate the original
Kdea m °del.
o) Based on the simplified K^ model, a simplified flowrate model, as presented
in Section 9.3, w a s obtained having very good agreement with the original
flowrate model, eqn (5.17) but without requiring the numerical analysis used to
solve the integrations in the original model. The simplified flowrate model has
all the distinguishing features of the original model and therefore has more
practical implications.
p) Strategies for increasing the limiting flowrate include:
• low surcharge level with double hopper arrangement;
• addition of coarse material into fine particles;
• air injection;
• use of a standpipe.
The use of standpipes to increase the limiting flowrate by gravity was examined
both theoretically and experimentally. The results indicate that the use of
standpipes can increase the flowrate significantly for fine particles (lower
permeability materials) and insignificantly for coarse particles (higher
permeability materials). The longer the standpipe, the more significant the
suction effect induced by the standpipe and the higher the flowrate that can be
obtained. However, a long standpipe is only efficient when it is kept full of bulk
solid. This suggests that some precaution needs being taken to ensure that the
standpipe remains full, for instance, by using a feeder under the outlet of the
standpipe to control the flowrate.
Chapter 12 Suggestions for Further W o r k
The work completed has been concerned with axisymmetric bins. For general bulk
solids flow, the work needs to be extended to the flow of bulk solids from the plane
flow bins.
More experimental work should be carried out to measure the air pressure
distribution in the mass flow bins, to give further evidence for the m i n i m u m
position expression eqn (4.29) and to provide a wider range of experimental data to
validate and/or improve the theoretical model for predicting the air pressure gradient
at the hopper outlet
The critical permeability value C^, eqn (7.7), as a criterion of distinguishing the
air retarded flow from non-air retarded flow is deduced from the experimental
results for the bulk solids flowing out of the hopper outlet with the diameter less
than 0.045 m . It seems that the eqn (7.7) is acceptable for the hopper outlet up to
about 0.1 m in diameter. M o r e research needs to be carried out for larger outlet
diameter to examine this critical permeability expression.
The critical permeability value to decide the surcharge level-affected flow is
suggested as 20^. M o r e research needs to be done for large hopper outlet cases to
provide a greater degree of generality.
The theoretical model for predicting the flowrate in this dissertation has not includ
the effect of the fluidization on the flow behaviour. M o r e work should be carried
out to better describe the flow of very fine materials. This work should relate to the
studies on the flooding phenomenon of very low permeable materials.
In this work, the dynamic deaeration coefficient Kdea is used to predict
successfully the air pressure distribution in a mass flow bin and the particle flowrate
from the bin. Both original and simplified models are presented to evaluate this
coefficient. Further work is suggested to study the concept of dynamic deaeration,
defined as the deaeration of a bulk solid in a mass flow bin under flowing
condition, and the prediction of K ^ .
The theoretical model presented for the standpipe predicts a reduction of the
negative air pressure gradient at the hopper outlet. For longer standpipes, a positive
air pressure gradient could be created by the standpipe, which is excluded in current
model. Therefore, the further work should be done to improve the theoretical model
to predict the performance.
The main flow properties for predicting the flowrate of bulk solids from a mass
flow bin include essentially the bulk density (p 0 , b l s b 2 ), permeability (C 0 , a) and
internal friction angle (8). All of these parameters are related to the physical particle
characteristics, say particle size, particle size distribution, particle density, as shown
in Figure 3.1. F r o m the point of view of rheology and the packing of the particles,
these variables could be predictable from the particles size and particle size
distribution. Therefore, a theoretical interest becomes apparent in predicting
flowrate directly from the physical characteristics of the particles, while this is an
outcome which will be difficult to achieve, it should be the alternate aim of m u c h of
the research relating to powder technology.
References
Jenike, A. W., Gravity Flow of Bulk Solids, Bulletin 108, Utah
Engineering Experiment Station, Univ. of Utah, 1961.
Jenike, A. W., Storage and Flow of Solids, Bulletin 123, Utah Engineering
Experiment Station, Uni. of Utah, 1964.
Arnold, P. C , McLean, A.G. and Roberts, A.W., Bulk Solids: Storage,
Flow and Handling, T U N R A Ltd, 2nd Edition, 1980.
Nedderman, R. M., Tiiziin, U., Savage, S. B. and Houlsby, G. T., The
Flow of Granular Materials -1: Discharge Rates from Hoppers, Chem. Eng.
Sci., Vol. 37, No. 11, 1982, pp 1597-1609.
Nedderman, R. M., The Flow of Granular Materials through Orifices.
Powtech'85: Particle Technology - I. Chem. E. Symp. Series, No. 91,
1985, pp 281-294.
Schofield, C , Recent Progress in Solids Processing: Recent Research on
the Storage of Particulate Materials in Hoppers, Chem. Eng. Res. Des.,
Vol. 64, No. 2, 1986, pp 89-90.
Tiizun, U., Critique of Recent Theoretical Advances in Particulate Solids
Flow. Int. Symp. ~ Workshop on Particulate and Multi-Phase Processes
and 16th Annual Meeting of the Fine Particle Society, Florida, 22-26
April, 1985.
Tiizun, U., Houlsby, G. T., Nedderman, R.M. and Savage, S.B., The
Flow of Granular Materials - II: Velocity Distributions in Slow Flow,
Chem. Eng. Sci., Vol. 37, No. 12, 1982, pp 1691-1709.
Deming, W . E. and Mehring, A. L., The Gravitational Flow of Fertilizers
and Other Comminuted Solids, Ind. Eng. Chem., Vol. 21, No. 7, 1929,
pp 661-665.
Newton, R. H., Dunham, G. S. and Simpson, T. P., The T C C catalytic
cracking process for motor gasoline production, Trans. Amer. Instn. Chem.
Engrs., Vol. 41, 1945, pp 215-232.
11.
Franklin, F. C. and Johanson, L. N., Flow of Granular Material through a
Circular Orifice, Chem. Eng. Sci., Vol. 4, 1955, pp 119-129.
12.
Fowler, R. T. and Glastonbury, J. R., The Flow of Granular Solids
Through Orifices, Chem. Eng. Sci., Vol. 10, 1959, pp 150-156.
13.
Rose, H. E. and Tanaka, T., Rate of Discharge of Granular Materials from
Bins and Hoppers, The Engineer, Vol. 208, 1959, pp 465-469.
14.
Beverloo, W . A., Leniger, H. A. and Van de Velde, J., The Flow of Granular
Solids through Orifices, Chem. Eng. Sci., Vol. 15, 1961, pp 260-269.
15.
Bosley, J., Schofield, C. and Shook, C. A., A n Experimental Study of
Granule Discharge from Model Hoppers, Trans. Instn. Chem. Engrs., Vol.
47, 1969, pp T147-T153.
16.
Van Zuilichem, D. J., Van Egmond, N. D. and de Swart, J. G., Density
Behaviour of Flowing Granular Material, Powder Technol., Vol. 10, 1974,
pp 161-169.
17.
Brown, R. L. and Richards, J. C , Exploratory Study of the Flow of
Granules Through Apertures, Trans. Instn. Chem. Engrs., Vol. 37, 1959,
pp 108-119.
18.
Brown, R. L., Minimum Energy Theorem for Flow of Dry Granules
through Apertures, Nature, Vol. 191, No. 4787, 1961, pp 458-461.
19.
Brown, R. L. and Richards, J. C , Principles of Powder Mechanics,
Pergamon Press, London, 1970.
20.
Jones, D. R. M . and Davidson, J. F., The Flow of Particles from a
Fluidised Bed through an Orifice, Rheo. Acta, Band 4, Heft 3, 1965,
pp 180-192.
McDougall, I. R. and Evans, A. C , A Theoretical Equation for the Flow
of Granular Solids from a Hopper, Rheo. Acta, Band 4, Heft 3, 1965,
pp 218-225.
22.
Johanson, J. R., Method of Calculating Rate of Discharge from Hoppers
and Bins, Tran. Soc. Min. Engrs., March, 1965, pp 69-80.
23.
Savage, S. B., The Mass Flow of Granular Materials Derived from Coupled
Velocity - Stress Fields, Brit. J. Appl. Phys., Vol. 16,1965, pp 1885-1888.
24.
Savage, S. B., Gravity Flow of a Cohesionless Bulk Solid in a Converging
Conical Channel, Int. J. Mech. Sci., Vol. 9, 1967, pp 651-659.
25.
Mullins, W . W., Stochastic Theory of Particle Flow under Gravity, J.
Appl. Phys., Vol. 43, No. 2, February, 1972, pp 665-678.
26.
Mullins, W . W., Nonsteady - State Particle Flow under Gravity - A n
Extension of the Stochastic Theory, J. Appl. Mech. - Trans. A S M E ,
December, 1974, pp 867-872.
27.
Davidson, J. F. and Nedderman, R. M., The Hour-Glass Theory of
Hopper Flow, Trans. Instn. Chem. Engrs., Vol. 51, 1973, pp 29-35.
28.
Williams, J. C , The Rate of Discharge of Coarse Granular Materials from
Conical Mass Flow Hoppers, Chem. Eng. Sci., Vol. 32, 1977, pp 247-255.
29.
Nedderman, R. M . and Tiizun, U., A Kinematic Model for the Flow of
Granular Materials, Powder Technol., Vol. 22,1979, pp 243-253.
30.
Drescher, A., Cousens, T. W . and Bransby, P. L., Kinematics of the Mass
Flow of Granular Material through a Plane Hopper, Geotechnique, Vol. 28,
No. 1, 1978, pp 27-42.
31.
Brennen, C. and Pearce, J. C , Granular Material Flow in TwoDimensional Hoppers, J. Appl. Mech., Vol. 45, March, 1978, pp 43-50.
32.
Nguyen, T. V., Brennen, C. E. and Sabersky, R. H., Gravity Flow of
Granular Materials in Conical Hoppers, J. Appl. Mech. - Trans, of the
A S M E , Vol. 46, 1979, pp 529-535.
33.
Kaza, K. R. and Jackson, R., The Rate of Discharge of Coarse Granular
Material from a Wedge - Shaped Mass Flow Hopper, Powder Technol.,
Vol. 33,1982, pp 223-237.
34.
Michalowski, R. L., Flow of Granular Material through a Plane Hopper,
Powder Technol., Vol. 39, 1984, pp 29-40.
35.
Hiroshi Takahashi, Eiji Obata and Takao Takeuchi, Velocity Discontinuity
of Particles Flowing in a Mass-Flow Hopper and the Analysis of Its
Characteristics, K O N A , No. 6, 1988, pp 47-56.
36.
Pitman, E. B., Stress and Velocity Fields in T w o - and Three - Dimensional
Hoppers, Powder Technol., Vol. 47, 1986, pp 219-281.
37.
Pitman, E. B., The Stability of Granular R o w in Converging Hoppers,
S I A M J. Appl. Math., Vol. 48, No. 5, October 1988, pp 1033-1053.
38.
Polderman, H. G., Boom, J., de Hilster, E. and Scott, A. M., Solids Flow
Velocity Profiles in Mass Flow Hoppers, Chem. Eng. Sci., Vol. 42, No. 4,
1987, pp 737-744.
39.
Ravi Prakash, J. and Kesava Rao, K., Steady Compressible Flow of
Granular Materials through a Wedge - Shaped Hopper: the Smooth
Wall, Radial Gravity Problem, Chem. Eng. Sci., Vol. 43, No. 3, 1988,
pp 479-494.
40.
Haussler, U. and Eibl, J., Numerical Investigations on Discharging Silos,
J. Eng. Mech., A S C E , Vol. 110, No. 6, 1984, pp 957-971.
41.
Runesson, K. and Nilsson, L., Finite Element Modelling of the
Gravitational R o w of a Granular Material, Bulk Solids Handling, Vol. 6,
No. 5, October, 1986, pp 877-884.
42.
Arteaga, P. and Tiizun, U., Flow of Binary Mixtures of Equal-Density
Granules in Hoppers — Size Segregation, Rowing Density and Discharge
Rates. Chem. Eng. Sci., Vol. 45, No. 1, 1990, pp. 205-223.
43.
Bulsara, P. U., Zenz, F. A. and Eckert, R. A., Pressure and Additive
Effects on R o w of Bulk Solids, I & E C Process Design and Development,
Vol. 3, No. 4, October, 1964, pp 348-355.
44.
Miles, J. E. P., Schofield, C. and Valentin, F. H. H., The Rate of
Discharge of Powders from Hoppers, I. Chem. E. Symp. Series, No. 29,
1968, pp 25-33.
45.
Miles, J. E. P., Some Aspects of the R o w of Particulate Materials from
Hoppers, J. Soc. Cosmetic Chemists, Vol. 21, 1970, pp 53-72.
46.
Holland, J., Miles, J. E. P., Schofield, C. and Shook, C. A., Ruid Drag
Effects in the Discharge of Granules from Hoppers, Trans. Instn. Chem.
Engrs., Vol. 47, 1969, T154-T159.
47.
Shook, C. A., Carleton, A. J. and Rain, R. J., Effect of Ruid Drag on the
Row-rate of Granular Solids from Hoppers - Shorter Communications,
Trans. Instn. Chem. Engrs., Vol. 48, 1970, pp T173-T175.
48.
McDougall, I. R. and Knowles, G. H., R o w of Particles through Orifices,
Trans. Instn. Chem. Engrs., Vol. 47,1969, pp T73-T78.
49.
Carleton, A. J., The Effect of Ruid - Drag Forces on the Discharge of Free Rowing Solids from Hoppers, Powder Technol., Vol. 6, 1972, pp 91 - 96.
50.
Crewdson, B. J., Ormond, A. L. and Nedderman, R. M., Air-Impeded
Discharge of Fine Particles from a Hopper, Powder Technol., Vol. 16,
1977, pp 197-207.
51.
Spink, C. D. and Nedderman, R. M., Gravity Discharge Rate of Fine
Particles from Hoppers, Powder Technol., Vol. 21,1978, pp 245-261.
52.
Arnold, P. C , The R o w Rate of Bulk Solids from Mass-Flow Bins,
Unpublished Lecture Notes, 1986.
53.
Smith, J. G., Feeders for mass flow hoppers, Thesis (B.E.) Univ. of
Wollongong, 1978.
54.
Willis, G. J., Interstitial Gas Pressures during R o w of Simple Bulk Solids
from a Plane R o w Bin, Thesis (B.E.), Univ. of Wollongong, 1978.
275
55.
Head, J. M., The Discharge Rate of Granular Solids from Mass Flow
Hoppers, Thesis (PhD), University of Bradford, 1979.
56.
Nedderman, R. M., Tiiziin, U. and Thorpe, R. B., The Effect of Interstitial
Air Pressure Gradients on the Discharge from Bins, Powder Technol.,
Vol.35, 1983, pp 69-81.
57.
Ducker, J. R. Ducker, M . E. and Nedderman, R. M., The Discharge of
Granular MaterialsfromUnventilated Hoppers, Powder Technol., Vol. 42,
1985, pp 3-14.
58.
McLean, A. G., R o w Rates of Simple Bulk Solids from Mass-flow Bins,
Thesis (PhD), Univ. of Wollongong, 1978.
59.
McLean, A. G. and Arnold, P. C , T w o Phase R o w in Converging Channels,
Proc. Euro. Symp. Particle Technol., Amsterdam, N L , June 3-5, 1980.
60.
McLean, A. G., Simplified Prediction of the R o w Parameters during the R o w
of Fine Bulk SolidsfromMass R o w Bins, Private Communication, 1984.
61.
Shamlou, P. A., Handling of Bulk Solids - Theory and Practice,
Butterworths, London, 1988.
62.
Ford, L. J. and Davies, R., International Fine Particle Research Institute The First 10 Years and Beyond, Proc. Second World Congress P A R T I C L E
T E C H N O L O G Y , Kyoto, Japan, September 19 -22,1990, pp 25-53 (Part II).
63.
Bird, D. A., Cammack, P. and Sproson, J. C , Coal Handleability and the
Design of Storage Bunkers for Coking Smalls and Power Station Blends,
Proc. 7th Inter. Coal Preparation Congress (Partridge, A. C. ed.),
Sydney, 1976.
64.
Bo, M . K., Freshwater, D. C. and Scarlett, B., The Effect of Particle Size
Distribution on the Permeability of Filter Cakes, Trans. Instn. Chem.
Engrs., Vol. 43, 1965, pp T228-T232.
65.
Lloyd, P. J. and Webb, P. J., The Rooding of a Powder - The Importance
of Particle Size Distribution, Powder Technol., Vol. 51,1987, pp 125-133.
276
66.
Marecek, J., Krpata, M . and Mondry, F., The Effect of Particle Size
Distribution on the Removal of the Liquid Phase from a Centrifuged Filter
Cake, I. Chem. Eng., Vol. 12, No. 3, 1972, pp 384-388.
67.
M e m o n , M . A. and Foster, P. J., R o w of Mixtures of Large and Small
Particles from a Hopper, Powtech'85: Particle Technology -1. Chem. E.
Symp. Series No. 91, 1985, pp 187-196.
68.
Molerus, O. and Nywlt, M., The Influence of the Fine Particle Content on
the R o w Behaviour of Bulk Materials, Powder Technol., Vol. 37, 1984,
pp 145-154.
69.
Noda, K., Uchida, S., Makino, T. and Kamo, H., Minimum Ruidization
Velocity of Binary Mixture of Particles with Large Size Ratio, Powder
Technol., Vol. 46, 1986, pp 149-154.
70.
Van Lier, W . C , The Influence of Particle Shape and Particle Size
Distribution on Ruidisation Behaviour of Granular Activated Carbon in the
Aqueous Phase, Part. Charact Vol. 1, 1984, pp 143-152.
71.
Thomson, F. M., Granular R o w : Industrial Problems Needing Solutions,
Advancements in Aerodynamics, Ruid Mechanics, and Hydraulics, (ed. by
Arndt R.E.A., Stefan, H.G., Farell, C. and Peterson, S.M.), Amer. Soc.
of Civil Engrs, N e w York, 1986, pp 481-486.
72.
Rathbone, T., Nedderman, R. M . and Davison, J. F., Aeration,
Deaeration, and Rooding of Fine Particles, Chem. Eng. Sci., Vol. 42,
No. 4, 1987, pp 725-736.
73.
Langmaid, R. N. and Rose, H. E., Arch Formation in a Non-cohesive
Granular Material, J. Institute of Fuel, April 1957, pp 166-172.
74.
Jenike, A. W . and Leser, T., A Row-No-Flow Criterion in the Gravity
R o w of Powders in Converging Channels, Proc. 4th Int. Congr. on
Rheology, Part 3,1963, pp 125-141.
75.
Jenike, A. W., Steady Gravity Flow of Frictional - Cohesive Solids in
Converging Channels, J. Appl. Mech. - Tran. A S M E , March, 1964, pp 5-11.
277
76.
Jenike, A. W., Gravity Flow of Frictional - Cohesive Solids Convergence to Radial Stress Fields, J. Appl. Mech. - Tran. A S M E ,
March, 1965, pp 205-207.
77.
Jenike, A. W., A Theory of R o w of Particulate Solids in Converging and
Diverging Channels Based on a Conical Yield Function, Powder Technol.,
Vol. 50, 1987, pp 229-236.
78.
Fried, I., Carson, J. and Park, Y., Nonlinear Finite Element Analysis of Gas
R o w in Contact Bed Reactor, 2nd Int Symposium on Finite Element Methods
in R o w Problems, Santa Margherita Ligure (Italy), 1976, pp 695-700.
79.
W u , Y. H., Static and Dynamic Analyses of the R o w of Bulk Materials
through Silos, Thesis (PhD), Univ. of Wollongong, 1990.
80.
Spink, C. D., Gravity Discharge Rate of Fine Particles from Hoppers,
Thesis (PhD), Univ. of Cambridge, 1976.
81.
Papazoglou, C. S. and Pyle, D. L., Air - Assisted R o w from a Bed of
Particles, Powder Technol., Vol. 4,1970/1971, pp 9-18.
82.
Altiner, H. K., R o w of Solids from Aerated Hoppers: Effect of Aeration
Methods, AIChE Symposium Series 222, Vol. 79,1983, pp 55-59.
83.
de Jong, J. A. H. and Hoelen, Q. E. J. J. M., Cocurrent Gas and Particle
R o w During Pneumatic Discharge from a Bunker through an Orifice,
Powder Technol., Vol. 12, 1975, pp 201-208.
84.
Ginestra, J. C , Rangachari, S. and Jackson, R., A One-Dimensional
Theory of R o w in a Vertical Standpipe, Powder Technol., Vol. 27, 1980,
pp 69-84.
85.
Chen, Y. M., Rangachari, S. and Jackson, R., Theoretical and
Experimental Investigation of Fluid and Particle Flow in a Vertical
Standpipe, Ind. Eng. Chem. Fundam., Vol. 23, 1984, pp 354-370.
86.
Knowlton, T. M., Mountziaris, T.J. and Jackson, R., The Effect of Pipe
Length on the Gravity R o w of Granular Materials in Vertical Standpipes,
Powder Technol., Vol. 47, 1986, pp 115-128.
87.
Allen, T., Particle Size Measurement (3rd ed), Chapman and Hall Ltd,
London, 1981.
88.
Ramanujan, T. K. and Venkateswarlu, D., Size, Shape and Distribution of
Particles, Particle Technology - Proc. Seminar, (ed. by Venkateswarlu, D.
and Prabhakara Rao, A.), Indian Ins. of Technol., Madras, 1971, pp 91-105.
89.
Jelinek, Z. K., Particle Size Analysis, (English Edition), John Wiley &
Sons Inc., N.Y., 1974.
90.
Molerus, O., Effect of Interparticle Cohesive Forces on the R o w Behaviour
of Powders, Powder Technol., 20, 1978, pp 161-175.
91.
Molerus, O., Invited Review: Flow Behaviour of Cohesive Materials,
Chem. Eng. Commun. Vol. 15, 1982, pp 257-289.
92.
Schwedes, J., Evolution of Bulk Solids Technology since 1974, Bulk
Solids Handling, Vol. 3, No. 1, March, 1983, pp 143-147.
93.
Schwedes, J. and Schulze, D., Measurement of Flow Properties of Bulk
Solids, Proc. of Second World Congress P A R T I C L E T E C H N O L O G Y ,
Kyoto, Japan, September 19 -22,1990, pp 61-70 (Part II).
94.
Schwedes, J. and Schulze, D., Measurement of Flow Properties of Bulk
Solids, Powder Technol., Vol. 61, 1990, pp 59-68.
95.
Svarovsky, L, Powder Testing Guide : Methods of Measuring the Physical
Properties of Bulk Powders. Elsevier Applied Science, London, 1987.
96.
Kim, K., Hot Compaction Equations for Metal Powders and Porous
Preforms, Int. J. Powder Metallurgy, Vol. 24, 1988, pp 31-37.
97.
Kawakita, K. and Liidde, K. H., Some Considerations on Powder
Compression Equations, Powder Technol., Vol.4, 1970, pp 61-68.
98.
Johanson, J. R. and Cox, B. D., Practical Solutions to Fine Powder
Handling, Powder Handling & Processing, Vol. 1, No. 1, 1989, pp 61-65.
99.
Schofield, A. N. and Wroth, C. P., Critical State Soil Mechanics, McGraw
- Hill, N e w York, 1968.
. German, R. M., Particle Packing Characteristics, Metal Powder Industries
Federation Princeton, 1989.
101. Marquardt, D. W., A n Algorithm for Least-Squares Estimation of Nonlinear
Parameters, J. Soc. Indust Appl. Math., Vol. 11, No.2, 1963, pp 431-441.
102. Yu, A. B. and Gu, Z. H., Private Communication, 1990.
103. Lambe, T. W . and Whitman, R. V., Soil Mechanics, John Wiley & Sons,
Inc., N e w York , 1969.
104. Mahinda Samarasinghe, A., Huang, Y. H. and Drnevich, V. P.,
Permeability and Consolidation of Normally Consolidated Soils, A S C E J.
Geotech. Engng. D M . , Vol. 108, (GT6), 1982, pp 835-850.
105. Johanson, J. R. and Jenike, A. W., The Effect of the Gaseous Phase on
Pressures in a Cylindrical Silo, Powder Technology, Vol. 5, 1971 / 1972,
pp 133-145.
106. Johanson, J. R. and Jenike, A. W., Settlement of Powders in Vertical
Channels Caused by Gas Escape, J. Appl. Mech., Dec, 1972, pp 863-868.
107. Ergun, S. and Orning, A. A., Ruid Flow through Randomly Packed
Columns and Ruidized Beds, Ind. Eng. Chem., Vol. 41, No. 6, 1949,
pp 1179-1184.
108. Carman, P. C , Ruid R o w Through Granular Beds, Trans. Instn. Chem.
Engrs., Vol. 15, 1937, pp 150-166.
109. Levenspiel, O., Engineering R o w and Heat Exchange, Plenum Press,
N.Y., 1984.
110. Cornell, D. and Katz, D. L., R o w of Gases through Consolidated Porous
Media, Ind. Eng. Chem., Vol. 45, No. 10, 1953, pp 2145-2152.
111. Walters, J. K., A Theoretical Analysis of Stresses in Silos with Vertical
Walls, Chem. Eng. Sci., Vol. 28, 1973, pp 13-21.
112. Blight, G. E., Pressure Exerted by Materials Stored in Silos: Part II, Fine
Powders, Geotechnique, Vol. 36, No. 1, 1986, pp 47-56.
. Mullins, W . W., Experimental Evidence for the Stochastic Theory of
Particle R o w under Gravity, Powder Technol, Vol. 9, 1974, pp 29-37.
114. W e n , C. Y. and Yu, Y. H., Mechanics of Ruidization, AIChE Symposium
Series, Vol. 62, Vol. 62, 1966, pp 100-111.
115. McLean, A. G., Private Communication, 1990.
116. de Jong, J. A. H., Vertical Air-Controlled Particle R o w from a Bunker
through Circular Orifices, Powder Technol., Vol. 3,1969/1970, pp 279-286.
117. Sutton, H. M . and Richmond, R. A., Improving the Storage Conditions of
Fine Powders by Aeration, Trans. Instn. Chem. Engrs., Vol. 51, 1973,
pp 97-104.
118. Sutton, H. M . and Richmond, R. A., H o w to Improve Powder Storage and
Discharge in Hoppers by Aeration, Process Engineering, September 1973,
pp 82-85.
119. de Lazzari, R. E., The Performance of an Air Permeated Bin, Thesis (B. E.)
Univ. of Wollongong, 1981.
120. Johanson, J. R., Retrofitting Bins and Hoppers to Prevent Fine Powder
Rushing. Binside Scoop, Vol. 4, No. 1, Winter 1990.
121. Wes, W . J., Stemerding, S. and van Zuilichem, D. J., Control of R o w of
Cohesive Powders by Means of Simultaneous Aeration and Vibration,
Powder Technol., Vol. 61, 1990, pp 39-49.
122. Yuasa, Y. and Kuno, H., Effects of an Efflux Tube on the Rate of R o w of
Glass Beads from a Hopper, Powder Technol., Vol. 6, 1972, pp 97-102.
123. McDougall, I. R. and Pullen, R. J. F., The Effect on Solids Mass Rowrate
of an Expansion Chamber between a Hopper Outlet and a Vertical
Standpipe, Powder Technol., Vol. 8, 1973, pp 231-242.
124. de Jong, J. A. H., Aerated Solids R o w through a Vertical Standpipe Below
a Pneumatically Discharged Bunker, Powder Technology, Vol. 12, 1975,
pp 197-200.
Leung, L. S. and Wilson, L. A., Downflow of Solids in Standpipes,
Powder Technol., Vol. 7, 1973, pp 343-349.
126. Leung, L. S. and Jones, P. J., A n Analysis of Moving-Bed Flow of
Solids D o w n Standpipes and Slida Valves, Powder Technol., Vol. 19,
1978, pp 7-15.
127. Leung, L. S. and Jones, P. J., R o w of Gas-Solid Mixtures in Standpipes.
A Review, Powder Technol., Vol. 20, 1978, pp 145-160.
128. Bird, R. B., Stewart, W . E. and Lightfoot, E. N., Transpot Phenomena,
John Wiley & Sons, Inc., N e w York, 1960.
129. Wieghard, K., Some Experiments on Flow in Sand, Ingenieur. - Archiv.
Vol.20, 1952, pp 109-115.
130. McCabe, R. P., R o w Patterns in Granular Material in Circular Silos.
Geotechnique. N o 1, 1974, pp 45-62.
131. Iiu, D. G., Fei, J. G., Yu, Y. J. and Li, G. Y., Collected Algorithms (in
Chinese), Vol. 2, National Defence Industry Press House, Beijing, 1983.
132. Reinsch, C. H., Smoothing by Spline Functions, Numer. Meth., Vol. 10,
1967, pp 177-183.
133. Arnold, P. C. and Gu, Z. H., The Effect of the Material Level on the
Rowrate of Bulk Solids from Mass R o w Bins. Proc. Third Inter. Conf.
on Bulk Materials, Storage, Handling and Transportation. Newcastle,
1989, pp 196-199.
134. Cumberland, D. J. and Crawford, R. J., The Packing of Particles.
Handbook of Powder Technology, Vol. 6. Elsevier, N e w York , 1987.
135. Leitzelement, M . , Lo, Cho Seng and Dodds, J., Porosity and
Permeability of Ternary Mixtures of Particles. Powder Technol. Vol. 41,
1985, pp 159-164.
136. Standish, N. and Borger, D. E., The Porosity of Particulate Mixtures.
Powder Technol. Vol. 22, 1979, pp 121-125.
Standish, N. and Collins, D. N., The Permeability of Ternary Particulate
Mixtures for Laminar R o w . Powder Technol. Vol. 36,1983, pp 55-60.
138. Rubinstein, R. Y., Simulation and the Monte Carlo Method. John Wiley &
Sons. N.Y. 1981.
139. Spendley, W., Hext, G. R. and Himsworth, F. R., Sequential Application
of Simplex Designs in Optimisation and Evolutionary Operation,
Technometrics, Vol. 4,1962, pp 441-461.
140. Nelder, J. A. and Mead, R., A Simplex Method for Function Minimization,
Computer J., Vol. 7, 1965, pp 308-313.
141. Avriel, M., Nonlinear Programming: Analysis and Methods, Prentice-Hall,
Inc., N e w Jersey, 1976.
142. Subrahmanyam, M . B., A n Extension of the Simplex Method to
Constrained Nonlinear Optimization, J. Optimization Theory and
Applications, Vol. 62, No. 2, August 1989, pp 311-319.
143. Himelblau, D. M., Applied Nonlinear Programming, McGraw - Hill Book
Company, N.Y., 1972.
144. Prabhakara Rao, A. and Venkateswarlu, D., Particle Technology : its Scope
and Developments, Particle Technology - Proc. Seminar, (ed. by
Venkateswarlu, D. and Prabhakara Rao, A.), Indian Ins. of Technol.,
Madras, 1971, pp 1-10.
145. Matsen, J. M., R o w of Ruidized Solids and Bubbles in Standpipes and
Risers, Powder Technol., Vol. 7, 1973, pp 93-96.
Appendix I :
Theoretical
Analysis
Section
Appendix I -1
The Young's Modulus Coefficient K y for Bulk Solids under Uniaxial Test
Since the Young's modulus coefficient K y for a bulk solid varies with the voidage
of the bulk material. The attempt in this section is to discover this variation under
uniaxial condition.
From equation (3.4), the relation between strain { E} and principal stress { (J)
for bulk solids is in the form
{ e } = K,
1 -0) -1)
-V 1 -1)
-1) -V 1
(3.4)
{o- }
For a three dimensional compacted bulk material
bo
(A-l.l)
P= Po v
The relation between bulk density and voidage is
(A-1.2)
p = ps(l-e)
For a element which is compacted by three stresses (a ,wa
, a ) , Fig.A-I-1.1,
x ' y ' vz z
Figure A-I-l.l A n Element of Bulk Solids Compacted by Three Stresses
v
bO=(1+ex)dx(l+Ey)dy(l+ez)dz
Vjj =dxdydz
from eqn (A-l.l)
p= p0(l+ £*) (l+ey)(l+ez)
or on taking a first order approximation
P = P o ( 1 + ex + ey
+e
(A-1.3)
z)
For the Compressibility Tester, using cylindrical coordinates (r, 9, z), Figure A-I1.2, er = ee = 0.
<*Z
V^yH111 llll IIIIII III 111IITTTTTTT
~r*
y
y
IWMv^^WWx.
Figure A-I-1.2
Compressibility Tester
+r
which, from eqn ( 3.4), suggests that
CTr= e =
° T VC*
and
E
2\) 2
z= K y ^ ' l-i)
TT"^
(A_L4)
In neglecting the second term in the bracket of eqn (A-1.4), it follows that
e, = Ky cz (A-1.5)
where Ky is a function of strain
From eqns (A-1.3) and (A-1.5), the measured bulk density ( since e^= e = 0 )
p= p0(l+ Kyoz)
On comparison with model 16, the factor Kj can be evaluated as follows. Firstl
noting
p = b1azb2 + b3
Let b3~p0,then
p = b1azb2+p0 (A-1.6)
The relation between voidage and strain,fromeqns (A-1.2) and (A-1.3), is
e = e0-(l-e0)e2 (A-1.7)
It follows on substituting eqns (A-1.6) and (A-1.7) into (A-1.2), that
D
l bo
(A-1.8)
h = — °z
Po
Comparing eqn (A-1.8) with eqn (A-1.5), yields
K
= J- e (M/b2)
(A-1.9)
^y K s - z
where the constant K s = N —
l/b2
or
Ky =
fe
fe)
( A
"U0)
Inserting eqn (A-1.9) into eqn (A-1.5), yields
J_ e d-l/b2) CT
z K s Ez
°z
e =
E
Hence, finally this suggests the relation between stress and strain is
a^K.E,1^
or
(Po V *
(A-l.ll)
Appendix 1-2
Equivalent Major Principal Stresses in Honner Region of a Mass Row Bin
A s presented in Section 3.5.4, it is assumed that the bulk density can be evaluated
by the mean stress acting. Alternatively the equivalent major principal stress ale f
different situations can be used in place of the mean stress om, as applies in the
compressibility tester. From Appendix I -1, for the material in the Compressibility
Tester (Figure A-I-1.2), Ej. = eQ = 0, then
°r = a9 =~T~~
1- "0 °z
The mean stress in comressibility tester becomes
°mt- 3
i.e.,
CT
l+l)
mt =
mt
°7
3 ( 1 -v
)
z
where the major principal stress az is the equivalent major principal stress ale in
this case, i.e.,
omt = ale (A-2.1)
mt
i)
3 ( 1 -v
)
le
Tn a plane flow hopper
The motion of bulk solids in a plane flow hopper (Figure A-I-2.1) is a two
dimensional problem since ez = 0.
288
Figure A-I-2.1 The Cylindrical Coordinates for the Plane R o w Hopper
or+CTfl
As the computed mean stress o ~ m w = — h y — a , from eqn (3.4), then
oz=\)( ar+CTe )
Therefore, the mean stress acting on the bulk solids in the plane flow hoppe
becomes
q
a
m =
z+ g r + q e
i.e.,
CT
1 +\) .
.
m = — 3 - ( °r+ CTe )
(A-2.2)
According to the assumption, a m = o m t . Combining eqn (A-2.1) into eqn (A-2.2),
the equivalent major principal stress in the plane flow hopper is
CT1P=
'le
2(1-D) a
mw
(A-2.3)
289
ii)
In conical hoppers
Here the spherical coordinates (r, 6 , 0 ) are used (Figure A-I-2.2).
Rgure A-I-2.2 The Spherical Coordinates for the Conical Hopper
The computed m e a n stress is a
m c
= — — f
* . The equivalent major
principal stress for use in the density equation, therefore, is
a le
3(1-1) )
'mc
1 + v
Hence with knowledge of the computed mean stress c
(A-2.4)
m w
and a m c , use of eqns
(3.6), (A-2.3) and (A-2.4) and the density models (9), (14) or (16) provide a
means to evaluate theoretically the appropriate bulk density in the hoppers.
For example, inserting eqn (A-2.3) and eqn (A-2.4) together with eqn (3.6) into the
density model 16 and refitting the bulk density equation, the bulk density equations
can be expressed as follows:
in plane flow hopper
p = p0 + b l w ( c r m w ) b 2 w
(A-2.5)
in conical hopper
p = p0 + blc ( o ~ m c ) b 2 c
(A-2.6)
Once the experimental data are obtained on the Compressibility Tester, the bulk
densities in both plane flow hopper eqn (A-2.5) and conical hopper eqn (A-2.6) can
be obtained together with that in vertical section of the bin eqn (3.7) during the
processing of the uniaxial measured data. For instance, the bulk density equations
of Shirley Phosphate are obtained as follows:
in the vertical section
0 09544
p = 1246.61 + 231.16 ( o z )
in plane flow hopper p = 1246.61 + 244.84 (omw)
in conical hopper
n no^^i
p = 1246.61 + 252.48 ( a m c )
Appendix I - 3
Difference Caused bv Considering Variable Bulk Density in Walters' Equation
for the Vertical Stress in Cylindrical Section of the Bins
The differential equation describing the stress distribution in vertical section of the
bin presented by Walters^1u ^ is in the form
do z
4KW
(A-3.1)
-fr "D"°«=pg
+
where o*_ is mean vertical stress (Pa)
Kw =
tan <}> cos 8
(1 + sin 8) - 2 y sin 8
(under dynamic conditions)
y = j7(i-(i-c)»l
V tan8 )
From eqn (3.7), the bulk density equation is
>2v
p = p 0 + b lv
(3.8)
W000 )
Inserting eqn (3.7) into eqn (A-3.1) yields
daz
4K„
'2v i
p 0 + b lv
^ 1000 )
g
(A-3.2)
or
daz
4KW
( az
IT + ~rTa'-MToooJ
^2v
g=
P° g
This equation is difficult to solve analytically.
Considering the Walters' solution for constant bulk density pb, the mean vertical
stress becomes
CT
4K.
De z
PbgD
* = Tir(1-
>
(A 33)
-
Using variable bulk density p instead of constant bulk density pb in eqn (A-3.3),
the mean vertical stress becomes
Following work attempts to judge if it is acceptable to use eqn (A-3.4) as the
solution of eqn (A-3.1).
Differentiating eqn (A-3.4) with respect to z, gives
4K„
^ - D e e " "
dz ~ p g e
Z
+
fziL^
p daz dz
(A-3.5)
293
Combining eqn (A-3.5) with eqn (A-3.4) provides
do
z ( , 4^1^^ daz
"dr=pgl1-^P-F J +G"dT (A-3-6)
or
.
da z
(1_G)
~d7
+
4K,,
z=pg
"D"°
(A"3-7)
where
Gz
n
G
=
d
P
7 d^
<A-3-8)
from eqn (3.7), eqn (A-3.8) can be rewritten as
blv b2V f CTZ >|b2v b2
G=
"p—ITOOOJ = —(P"PO)
(A"3-9)
or
G =b
2 v
(l-y)
Since b 2 v « 1 and p > p n , G « 1.
(A-3.10)
Table A-I-3.1 shows the G values at the assumed stress levels (1 and 50 kPa) for
the materials used in experiments described in Chapter 6. For the powders under
dynamic conditions, 50 kPa in vertical stress may never be reached. This table
corifirms that the G values are very small. Therefore, eqn (A-3.4) can be used with
confidence as the approximate solution of eqn (A-3.1) in terms of variable bulk
density.
Table A-I-3.1 The G Values for the Materials Used in Experiments
Bulk Material
a z = 50kPa
alumina
0.010311
0.007655
P V C powder
0.011574
0.007904
sugar
0.011220
0.007325
Sand M l
0.005282
0.004406
SandM2
0.005539
0.004610
SandM3
0.006675
0.005574
SandM4
0.007885
0.006572
SandM5
0.006476
0.005381
SandM6
0.017146
0.013800
SandM7
0.012393
0.010130
SandMDl
0.009267
0.006894
SandMD2
0.008861
0.007394
SandMD3
0.008235
0.006949
SandMD4
0.008380
0.006882
CTZ
= 1 kPa
Appendix
II :
Experimental Measurement
Section
Appendix I -1
The Results of Particle Size Analysis bv Laser Particle Sizer
Following particle size analysis results are obtained by Malven Laser Particle Sizer
(MLPS). The tables show the output of M L P S for alumina, P V C powder, Sand M l
to Sand M 6 and Sand M D 1 . In these tables, the data of the cumulative particle size
distribution are detailed; D (v, 0.5), D (v, 0.9) and D (v, 0.1) represent the volume
diameters at 5 0 % , 9 0 % and 1 0 % point of the cumulative particle size distribution
respectively.
A-II-1.1
For Alumina
Malvern Irifctrurnerits
Size
rn i ci-oriw
188.0
87. a
53. 5
37.6
28. 1
£1. 5
16.7
13. 0
10. 1
7. 9
s. £
4. s
3.
3.
2.
1.
8
0
4
9
MttSTER Particle- S u e t - M3. O
1
'A wncier 1
lOO. 0
1
38. a
4.8
4.7
2. 5
2. 3
2.3
1. O
0. 7
0. 7
0. 6
0.2
O. 1
0. 1
O. 1
0. 1 1
S i ;: eb.ir.ci
micv •Oi-l-r
188.0
87.2
53. 5
37.6
28. 1
21. 5
16. 7
13. 0
10. 1
7.9
6.2
4.8
3.8
3. 0
2. 4
S a m p l e detai 1 s :-al urnina 22/7/88
**Averaged
Result**
•/.
07. 2
5^« *-»
37.6
23. 1
21. 5
1.=,. 7
13. 0
10. 1
7.9
6.2
4.8
3.8
3. 0
a. 4
1. 9
61. 2
34.0
0. 0
2. 2
0. 2
0. 0
0. 5
1. 2
0. 0
0.0
0. 4
0. 1
0. 0
0.0
0. 0
i
1
-1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
-•<.- .1.3. i bunrc^^Ssuip
&
i"-i:oi-d Nio.
=
0
Fucal length = 100 IOil'1.
Experiment t ype pi 1
Volume
dist ri but ion
Beam length =
14. 3 mm.
Obscuration =0.£343
Volume Cone. = 0. 0475 "/= 4. 91
Log. Diff.
Model indp
D(v,0.5)
100. 2
149. 5
D(v, 0.9)
D(v,0.1)
62. 4
D(4,3)
101. 0
D<3,2)
59.6
Span
= 0. 9
Spec. surf. area
0.05 so. n. /cc.
urn
>.im
urn
urn
pm
A-n-1.2
Malvern
For P V C Powder
Instruments
Size
m1crons
MPSTtR Particle Sizer M3.0
"/. under
188. 0
87.2
53. 5
37. 6
£8. 1
£1.5
16. 7
13. 0
10. 1
7.9
6. 2
4. 3
3. 8
3.0
2. 4
1.9
iOO. 0
5. S
2. O
1. 4
1. 3
1. 3
1. 3
1. 2
O. O
0. 0
0. 0
0. o
0. o
0. o
0. 0
0. o
Si: 'O band
microns
188. n
87. iz!
53. 5
37. 6
28. 1
21. 5
16. 7
13. 0
10. 1
7. 9
6. £
4. 8
3. 8
3. 0
2. 4
87.
53.
37.
28.
21.
16.
13.
2
5
8
1
5
7
0
io. 1
7. 9
6. 2
4. 3
3. 8
3. 0
i_. 4
1. 9
X
94. 2
3. 8
0. 6
0. O
O.O
0. 0
0. 1
1. 2
0.0
0. 0
0. O
0.0
0.0
0. 0
O.O
Result source-Sample
Record No.
0
Focal length = 100 mm.
Experiment t ype pi)
Volume
di^t r i but i.• n
Beam length
=
14. 3 mm.
Obscurat ion
= 0 . 1 4 ? :4
V o l u m e Cone. = 0. 04 1 3 •/
Log. Diff.
= 4. 74
Model indp
127. 2
D(v,0.5)
161. 5
D(v, 0.9)
D(v, O. 1)
97.5
126. 5
D(4,3) .
D(3,£>
113.7
Span
= 0. 5
Spec. surf. Bt^ea
0.05 sq. m. /cc
urn
Urn
Urn
urn
urn
S a m p l e detai Is :-corvi r- pvc powder 22/7/88
**flveraged R e s u l t * *
A-n-1.3
For Sand M l
Malvern I n s t r u m e n t s
Size
microns
564. 0
261. 7
160. 4
112. a
84.
64. 6
SO. 2
39. 0
30. 3
£3. 7
18.
14. 5
11. 4
9. 0
7. £
5. 8
MASTER Particle Sizer M3.0
Size band
%
under
100. 0
12. 1
2. o
1. 1
0. 7
0. 6
0. 6
0. 6 0. 6
0. 5
0. 2
0. £
0. £
0. 2
0. £
0. 1
rn i c r o n s
564. 0
£61. 7
160. 4
112. 8
84. 3
64. 6
50. £
39. 0
30. 3
£3. 7
18. 5
14. 5
11. 4
9. O
7. £
Sample d e t a i I s : - 3 0 0 - 4 2 5 u m river sand
**0veraged
Result**
£61. 7
160. 4
112. 8
84. 3
64. 6
50. £
39. 0
30. 3
23. 7
18. 5
14. 5
11. 4
9. 0
7. £
5. 8
54
87.9
10. 1
0.9
0.5
0. 1
0. 0
0.0
0.0
0. 1
0. 3
0.0
0. 0
0.0
0. 0
0.0
Result source=Sample
Record No.
0
Focal length = 300 mm.
Experiment t ype pil
Volume
dist ri but ion
Beam length
14. .i mm.
Obscurat ion
=0.2013
Volume Cone. = 0. 1557 •/
Log. Diff.
=4. 38
Model indp
D(v,0.5)
370. 8
D(v,0.9)
480. 0
D(v,0.1)
238. 3
D(4,3)
364. 5
D(3,£)
254. 9
Span
= 0. 7
Spec. surf. area
O.Ot: sq. m./cc.
urn
urn
urn
urn
urn
A-n-1.4
For Sand M 2
Malvern Instruments
Size
microns
564.0
£61.7
160. 4
112. 8
84.3
64.6
50.2
39.0
30. 3
23. 7
18.5
14. 5
11. 4
9. 0
7.2
5.8
%
MASTER Particle Sizer M3.0
under
lOO.O
35. 5
2. 1
2. 0
1. 1
O. 7
0.7
0.7
O. 7
0.7
0. 3
O. 3
O. 3
0. 3
O. 3
0. 2
Siz e band
microns
564. 0
261. 7
160. 4
112. 8
84. 3
64. 6
50. 2
39. 0
30. 3
£3. 7
18. 5
14. 5
11. 4
9. 0
7. 2
£61. 7
160. 4
11£. 8
84. 3
64. 6
50. 2
39. 0
30. 3
23. 7
18. 5
14. 5
11. 4
9. 0
7. 2
5. 8
X
64. 5
33. 3
0. 2
0.9
0. 4
0.0
0.0
0.0
0.0
0.3
0. 0
0.0
0. 0
0. 0
0. 1
Result source=Sarnpl e
Record No.
0
Focal length = 300 mm.
Experiment t ype pi'.
Volume
dist ri but ic• n
14. 3 mm.
Beam length
Obscurat i on =0.1893
Volume Cone. = 0. 1161 -A
Log. Diff.
=4. 93
Model indp
D (v, 0. 5)
= 310.2
453. 0
D(v,0. 9)
D(v,0. 1)
194. 6
212. 9
D(4,3)
D(3,2)
£31.4
Span
= O. 8
Spec. surf. BTea
0. 0 2 sq. m./cc.
pm
pm
pm
pm
pm
Sample detai Is: -£l£-350urn river sand
••Averaged
A-H-1.5
Result**
For Sand M 3
Malvern Instruments
Size
microns
564. 0
261.7
160. 4
112.8
84.3
64.6
50.2
39.0
30. 3
£3. 7
18. 5
14. 5
11.4
9.0
7. 2
5. 8
MASTER Particle Sizer M3. 0
•A under
100. o
93.8
£1.5
5.8
1. 8
1.3
1. 1 1. 1
1. 1
1. 1
0.7
0.7
0. 7
0.7
0. 7
0. 6
Size band
microns
564. 0
£61. 7
160. 4
11£. 8
84. 3
64. 6
50. 2
39. 0
30. 3
£3. 7
18. 5
14. 5
11. 4
9. 0
7. 2
S a m p l e detai Is :-180-£12urn river sand
••Averaged
Result**
261. 7
160. 4
112. 8
84. vf>
64. 6
50. £
39. 0
30. 3
23. 7
18. 5
14. 5
11. 4
9. 0
7. £
5. 3
54
6. 2
72. 3
15. 7
4. 0
0. 6
0. 2
0. O
0. 0
0. 0
0. 4
0. 0
0. 0
0. 0
0. 0
0. 1
Result source=Samp! e
Record No.
0
Focal length = 300
Experiment t ype pi!
Volume
dist ri but ion
Beam length
=
14. j. mm.
Obscurat ion =0.2075
Volume Cone. = 0.0754 %
Log. Diff.
= 4.31
Model indp
D(v,0. 5)
197. 6
246.9
D«v,0.9)
D(v, 0. 1 )
129. 0
D ( 4, 3 )
197. 8
D<3, 2>
152. 0
Span
= 0. 6
Spec. surf. area
U.U3 sq. m./ee.
pm
pm
pm
pm
pm
A-n-i.6
For Sand M 4
M a l v e r n Instruments
Size
microns
MASTER Particle Sizer M3.0
X under
564.0
£61.7
160.4
112. 8
84. 3
64. 6
50.2
39. 0
30. 3
23. 7
18. 5
14.5
11. 4
9. O
7.2
5.8
1
1
lOO.O
70.4
34.5
1
1
1
5. 1
2. £
1.6
1.2
1. 1
0.8
0.6
0.6
0. 6
0.6
0. 5
0. 4
1
1
1
1
1
1
1
1
1
1
1
1
i&. a
i
Si ze band
microns
564. 0
£61.7
160. 4
11£. 8
84. 3
64. 6
50. £
39. 0
30. 3
£3.7
18. 5
14.5
11. 4
9.0
7. 2
•/
£9. 6
35. 9
17. 6
11. B
2. 9
0. 6
0. 4
0. 1
0. 2
0. 2
0. 0
0., 0
0., 0
0., 1
0., 1
£61. 7
160. 4
112. 8
84. 3
64. 6
50. £
39. 0
30.3
23. 7
18. 5
14. 5
11. 4
9.0
7. £
5. 8
1 Result source=Sampl e
1 Record No.
0
= 300
1 Experiment t;ype pil
1 Volume
dist ri but icin
1 Beam length
14. 3 mm.
1 Obscuration =0.1898
1 Volume Core. = 0.0691 %
=4. 61
1 Log. Diff.
1 Model indp
1
202. 9 pm
1 D(v,0.5)
362. 9 pm
1 D(v,0.9>
1 D(v,0.1)
97. 3 pm
216. 0 pm
1 D(4,3>
1 D(3,£)
123. 4 pm
1 Span
1. 3
1 Spec. surf. area
1
0.02 sq. m. /cc.
Sninple detai Is:—0—350um river sand
"Averaged Result**
A-n-i.7
Malvern
For Sand M 5
Instrument;
Size 1
rn i c r o n s 1 under
564. O
487.0
420.0
36£.O
31E.O
£70.O
£33.0
201.0
173.0
14 9.O
129. O
111. O
95. 9
82. 7
71.4
61. 6
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
lOO. 0
99. 7
98. 9
97.8
36.6
95.7
92.3
8 2.3
65.6
42. 8
21.2
11.5
8.0
Urn 5
5. 0
4.4
MASTER Particle Sizer M3. 0
Size 1
1
•A
i n band 1 microns 1
0.3 1
0.8 1
1. 1 1
1. 2 1
1. O 1
3. 4 1
10.0 1
16.7 -1
22. 8 1
£1.6 1
9.7 1
3. 5 1
2. 5 1
0. 5 1
0.6 1
1.7 1
S a m p l e d e t a i I s : — r i v e r sand
••Averaged Result*^
•/.
1
under
in band 1
£.8
1. 5
1. 1
1. 1
1. 1
1. 0
1.0
1.0
0.9
0. 9
0.9
0. 9
1.3 1
0. 4 1
0. 0 1
0. 1 1
0. 0 1
0.0 1
0. 1 1
0. 0 1
0. 1 1
0. 0 1
0. 0 1
0. O 1
0.0 1
0.0 1
0. £ 1
0. 3 1
53. 1 1
45.8 1
33. 5 1
34. 1 1
£9. 4 1
25.4 1
£1.9 -1
18. 9 1
16.3 1
14. 1 1
1£. 1 1
10. 5 1
9. 0 1
7. 8 1
6. 7 1
5. 8 1
<98-154urn)
0. a
0.8
0.7
0.6
17-3-88
Result source=Samp le
Record No.
=
0
Focal length = 300 rn rn •
Experiment t ype pi:i
Volume
dist r i but i<:m
14.. 3 mm.
Beam length
Dbscurat ion
=0. £364
Volume Cone. = 0. 0 7 23 y.
Log. Diff.
=4. 34
Model indp
D(v,0.5)
= 155. 7
222. 8
D<v,0.9>
D<v,0. 1)
104. 1
164. 0
D(4,3)
106.0
D(3,£)
Span
= 0. 8
Spec. surf, area
0. 03 sq. rn. /cc.
pm
pm
pm
pm
pen
A-n-i.8
Malvern
For Sand M 6
Instruments
MASTER Particle Sizer M3.0
Size
microns
V. under
564.0
261. 7
160.4
112.8
84.3
64. 6
SO. 2
39.0
30. 3
£3. 7
18.5
14. 5
11.4
9.0
7.2
5. 8
lOO.O
99.6
95.9
BO. £
54.7
36. 4
£7. O
18. 3
11. 1
7.5
5.0
3.9
3. 3
2. 8
2. 3
1.4
S a m p l e d e t a i l s : - r i v e r sand
Si ze band
M icr ons
564.0
£61.7
160. 4
112.8
84. 3
64. 6
50. £
39.0
30. 3
£3. 7
18. 5
14.5
11. 4
9.0
7. 2
<0-98um>
£61. 7
160. 4
112. 8
84. 3
B4. 6
50. 2
39. 0
30. 3
£3. 7
18. 5
14. 5
11. 4
9. 0
7.£
5. 8
•/.
0.4
3. 7
15. 7
£5.5
18. 3
9. 4
8. 7
7. 1
3.7
£. 4
1. 1
0.6
0. 5
0. 5
0. 8
Result source-Sample
Record No.
=
0
Focal length = 300,rnm.
Experiment type pil
Volume
distribution
Beam length
= 14. 3 mm
Obscurat ion =0. 19£4
Volume Cone. = 0.0££6 X
Log. Diff.
=3.73
Model indp
D(v,0.5>
79. 6
131. 5
D(v,0. 9)
D(v,0. 1)
£8. 6
BI. 4
D<4,3)
D<3,£>
45. 8
Span
= 1. 3
Spec. surf. area.
0. 0 4 sq. m./cc.
prn
prn
pm
prn
prn
17-3-83
••Averaged Result**
A-n-i.9
For Sand M D 1
MASTER F'article Sizer M3. O
M a l v e r n Instruments
Size
microns
'A under
564.0
261. 7
160. 4
112.8
84. 3
64. 6
SO. 2
39. 0
30. 3
23. 7
18. 5
14.5
11.4
9. 0
7.2
5. 8
100. 0
97.8
13.2
2. 2
1. 3
1. 1
1. 1
1.1
0. a
0.2
0.2
0. 2
0.2
0. £
0.2
0. 2
S a m p l e detaiIs:-sand
••Averaged Result**
Size band
rn l crons
564.0
261.7
160. 4
112.8
84. 3
64. 6
50.2
39.0
30. 3
23. 7
16.5
14.5
11. 4
9.0
7. 2
261. 7
160. 4
112. 8
84. 3
64. 6
50. £
39.0
30. 3
23. 7
IS. 5
14. 5
11. 4
9. 0
7.2
5. 8
154 -223um 9/10/88
/.
2. £
84.5
11. 0
0.9
0. £
0.0
0. 0
0. 3
0. 5
0.0
0.0
0. 0
0. 0
0. 0
0.0
Result source=Sample
Record No.
=
0
Focal length = 3 0 0 mm.
Experiment type pil
distribution
Volume
Beam length
=
14. 3 mm
Obscurat ion =0.1637
Volume Cone. = 0. 070£ -A
=4. 19
Log. Diff.
Model indp
201.6
D(v,O.S>
241. 0
D<v,0.9)
147.3
D(v,0. 1)
139.5
D(4,3)
159. 1
D(3,£)
Span
= 0. 5
Spec. surf, area
O. 03 sq. rn. /cc
pm
pm
prn
prn
pm
Appendix II - 2
Instantaneous Yield Loci Measured for All Test Materials
A-II-2.1
For Alumina
B.
1-
2-
6.
7.
8.
NORMAL STRESS - kPa
9.
10.
11.
12.
INSTflNTRNEOUS YIELD LOCI
r
MATERIAL:alumina
MOISTURE CONTENT:as received
;GURE=
A-H-2.2
TESTED:14/7/88
TEMPERRTURE:RMBIENT
For P V C Powder
6- n-r
i . i i . i i M
M
I I I M
I I I I I I | I I I I | I I I I | I M
I | I I I I | 1 I I I | I I I I | I I I
in
•n
UJ
cc
co
cr
cr
LiJ
r
co
i i i I I i i i I I I I I I I I I U J I i i i
0.
1 .
2.
3.
4.
5.
6.
7.
8.
NORMRL STRESS - kPa
INSTflNTRNEOUS YIELD LOCI
FIGURE:
MRTERIRL:PVC Powder
MOISTURE CONTENT:as received
9-
10.
11.
TESTED:14/7/88
TEMPERRTURE:RMBIENT
12.
301
A-II-2.3
For Sugar
i i . | i i I I | I : i i | i i i i |
e-0.
i i i | i i I i | i I i I | i I i I i i I I i i I i i i | i i i i i i I i i
4.
5.
6.
7.
NORMAL STRESS - kPa
1 •
8.
INSTflNTRNEOUS Y I E L D
LOCI
for Sand M l
A-H-2.4
6 -
10.
TESTED:14/7/88
TEMPERATURE:AMBIENT
MATERIAL:sugar
MOISTURE CONTENT:as received
IGURE:
9-
I I
j i i | i i i i | i i i i | i i i i | i i i i | i i i i | i i i i | i i i i | i i i i | i i i i | i i i i | i i i i_
4. 01
CO
" 2
2.
3.
4.
NORMAL STRESS - kPa
5.
INSTANTANEOUS YIELD LOCI
FIGURE:
MATERIAL:sand-ml
MOISTURE CONTENT:as received
6.
7.
8
9.
10.
11.
TESTED:14/7/88
TEMPERATURE:AMBIENT
12.
A-n-2.5
For Sand M 2
i i i i i i i
i ' • • • i
z.
1 1 1
3.
1.
i
i""iIMIi
1 1 1 1 1 1 1 1 1 1 1 1 1 1
11 1 1 1 1 1 1
i ' ''
' i
111
i I i i i i 1 i i ii i I
10.
4.
5.
6.
7.
NORMAL STRESS - kPa
i
i
i
i
i
i
11.
i
i
12.
INSTANTANEOUS YIELD LOCI
TESTED:14/7/88
TEMPERATURE:AMBIENT
MATERIAL:sand-m2
MOISTURE CONTENT:as received
GURE:
A-H-2.6
6.
For Sand M 3
i i I i i i i i i i i i
I i i i i I i i i i j i i i i | i i i i | i i i i | i i i i | i i i i i|i
i i i
i i i
5.
4.
0.
1
3.
4.
5.
6-.
7.
I
NORMAL STRESS - kPa
I N S T A N T A N E O U S YIELD LOCI
IGURE:
MRTERIRL:sand-m3
MOISTURE CONTENT:as received
10.
11.
TESTED:14/7/88
TEMPERRTURE:RMBIENT
12.
A-H-2.7
For Sand M 4
:
1 i i I . i i i i i i i i . i i i || i< i i i | I I I I | 1 I I I | I I I I | l i | | | | | i i | ; i |
l""l
-• h
CD
en
cr
i—
CO
cr
cr
4.
NORMAL STRESS - kPa
5.
6.
7. * 8.
9-
10.
11-
12.
INSTANTANEOUS YIELD LOCI
FIGURE:
A-H-2.8
MATERIAL: sand-n.4
MOISTURE CONTENT:as received
TESTED: 1 4/7/88
TEMPERATURE:AMBIENT
For Sand M 5
cr
cz
3.
4.
NORMAL STRESS - kPa
5.
6.
7.
8
10.
11.
INSTANTANEOUS YIELD LOCI
IGURE:
MATERIAL:sand-m5
MOISTURE CONTENT:as received
TESTED:14/7/88
TEMPERATURE:AMBIENT
12.
A-n-2.9
For Sand M 6
4.
5.
6.
7.
NORMAL STRESS - kPa
INSTANTANEOUS YIELD LOCI
TESTED:14/7/88
TEMPERATURE:AMBIENT
MATERIAL:sand-m6
MOISTURE CONTENT:as received
FIGURE:
A-n-2.10
7 -
M
For Sand M 7
I I I | I I I I | I I I I | I I I I | I I I I |I I I I |I I I I | I I I I | I I I I | I I I I | I I I I |I ! I I |I I
CO
CO
DC
t—
CO
cr
cr
CO
4.
NORMAL STRESS - kPa
5-
6.
7.
INSTANTANEOUS YIELD LOCI
FIGURE:
MATERIAL:sand-m7
MOISTURE C0NTEN1:as received
8.
9-
10- 1
TESTED:23/9/89
TEMPERATURE:AMBIENT
A-II-2.11
For SandMDl
i l i ; i i |
M
i i | i i i i | i i
M
| :
M
i | i i I I | i i i i | I i i i | i i i i |
CO
CO
cr
i—
CO
cr
cz
DZ
CO
11.
4.
5.
6.
7.
NORMAL STRESS - kPa
12.
INSTANTANEOUS YIELD LOCI
FIGURE:
A-H-2.12
MATERIAL:sand-mdl
MOISTURE CONTENT:as received
TESTED:23/9/89
TEMPERATURE:AMBIENT
For Sand M D 2
CO
CO
UJ
cr
CO
cr
9.
10.
11.
NORMAL STRESS
INSTANTANEOUS YIELD LOCI
FIGURE:
MATERIAL:sand-md2
MOISTURE CONTENT:as received
TESTED:23/9/89
TEMPERATURE:AMBIENT
12.
A-II-2.13
For Sand M D 3
~i
i ; i i i r j I I i I | I"!
I ! I 1 1 ~1
1 T" T T 1 "I
T~
I I 1 I I I I I I I I I I I I I
I' ' ' ' I
11.
12.
NORMAL STRESS - kPa
INSTANTANEOUS YIELD LOCI
FIGURE:
A-H-2.14
MATERIAL:sand-md3
MOISTURE CONTENT:as
received
TESTED:23/9/89
TEMPERATURE:AMBIENT
For Sand M D 4
1'''' 1
I'''' I
I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
•j~>
NORMAL STRESS - kPa
INSTANTANEOUS YIELD LOCI
CURE:
MATERIAL:sand-md4
MOISTURE CONTENT:as
received
TESTE0:23/9/89
TEMPERATURE:AMBIENT
307
Appendix II - 3
Wall Yield Loci Measured for All Test Materials
For Alumina
A-n-3.1
1
r T,
1'''' i' ~r-"-r
r *r
.EC-END
C
Pers:ei
^
Gelvoniied
B.
1.
2.
3.
Steel
4.
5. 6. 7. 8. 9. 10. 11. 12- 13. 14. 15.
NORMAL STRESS - kPo
WALL YIELD LOCI
MflTERIRLrfllumino
MOISTURE C0NTENT:os
received
TESTE0:lB/7/88
TEMPERATURE: ombi snt
FIGURE:
For P V C Powder
A-H-3.2
1<!.
-""1 " " I 1 " 1 ! " " ! " " ! " " ! " 1
11.
1Z-
I •'" 1 " ' ' I " ' ' I " ' ' I ' ' • • I ' ' ' ' I ' ' ' ' I '
LEGEND
0
Perspex
Colvsniled
Steel
9-r
E. :7. r.
B. '-
-
{ ^ ^ S ^ t t t t l T i r i Z . 13. ,4. 15.
NORMRL STRESS - kPi
WALL YIELD LOCITESTED:16/7/88
ICU^C
M^TERIRLiPVC Powder
MOISTURE CONTENT:at
received
TEMPERATURE: ombi ent
308
A-II-3.3
For Sugar
• i n',_rn'
NORMAL
STRESS
- kPa
WALL YIELD LOCI
r
A-H-3.4
MqTERIflL:Sugar
MOISTURE CONTENT:os
received
TESTED:16/7/88
TEMPERATURE:»»bient
IGU=!
for Sand M l
1 1 1 1 1 1 1 H 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 | 1 1 1 1 1 1 1 " 1 1 ' " | " " | " '' | " " I " " I " " I " "•
LEGEND
O
Pcrspei
A
Golvenlied Steel
NORMAL STRESS - »Po
WALL YIELD LOCI
M3TERIAL:5.nd-»l
MOISTURE C0NTENT:ot
'iteinl
TESTED: 16/7/88
T E M P E R A T U R E : i.bient
309
For Sand M 2
A-II-3.5
12
1 1
\l
a
e
?
E
5
NORMAL STRESS - kPa
WALL YIELD LOCI
"ICL^E:
MHTERIAL:S«nd-(t,2
MOISTURE C0NTENT:«s
TESTED:16/7/88
TEMPERATURE: t»bient
received
For Sand M 3
A-H-3.6
| in
I| i I I i |i•rI|I i i i | l l l n i i l l | l i i l | i l M | i l l i | l
5.
6-
NORMAL
7.
8.
9- IB- H .
ni|i
12- 13. 14. IS-
STRESS - kPo
WALL YIELD LOCI
M=ITERinL:S.nd-»3
MOISTURE C0NTENT:oe
FIGURE:
received
TESTED:16/7/88,
TEMPERATURE:.«bii
310
For Sand M 4
A-II-3.7
12.
11.
ie.
"I""l""l""r"'l""l""l""l""l""l""l'
LEGEND
O
Perspai
&
Galvanised
> | " " I1
Steel
,,!.,, , l , , , , l , , . , l . . . i l n . . 1 . . t . l i i n l i i i i l i i i i l i i i i l i i n l
l
M l
M ,
,
, M
,1, n m m u i M
i ^
^
^"
^
^
i0. il. 12. 13-
illllll
14. 15
NORMAL STRESS - kPo
WALL YIELD LOCI
FIGURE:
MSTERIAL:S«nd-»4
MOISTURE CONTENT:os received
TESTED:16/7/88
TEMPERATURE: ombi ent
For Sand M 5
A-H-3.8
I M
i|l
I u
| • •
i ,. I I I | M
I I | I I I I | I I I I | I I i i | i . •' |
_l
,,,,,nil, M l , , .III. ll.M jin ^
NORMAL STRESS - kPo
^ ^ ''.'Jl ''.'i! ''.'ii ''.'it 15
WALL YIELD LOCI
riGURC:
M3TERIA t :Send--5
MOISTURE CONTEN":es
received
| • '
TESTED:16/7/88
TEMPERATURE:o*bi ent
311
A-II-3.9
For Sand M 6
i [ 11 i i ; ri-rr | i n i | i
1
1''" i",ri'
M""l'
rjTTTn
_EGENQ
C_ C
D Pers Pers =
A
Galvanised 5 I e e I
8-
1-
2-
3-
<•
5. B. 7. 8. S. 10. 11. 12. 13. 14. IS.
NORMAL STRESS - kPo
WALL YIELD LOCI
MATERIHL:Sond-m6
MOISTURE CONTENT:as received
iGURE:
A-H-3.10
TESTEO:16/7/88
TEMPERATURE:anbient
For Sand M 7
r" " I ' ' " I ' " ' I ' " ' I " ' ' I " '
"l""l""l""l'
F
f LEGEND
: O
Perspei
Galvanised
&
S.
!
2.
3.
4.
Steel
5. 6. 7. E. 9. 10- 11NORMAL STRESS - kPa
12- 13- 14.
WALL YIELD LOCI
"I CURE:
MaTERIfiL:Sond-m1
M3TERIfiL:Sond-m7
MOISTURE C0NTENT:a. received
TESTED:12/I0/8S
TEMPERATURE:embi
312
For SandMDl
A-II-3.11
12.
rp-rrry.
11.
ie.
"FT
LEGEND
O
PsrsDei
&
Galvanised
'l""l'
Steel
9.
4
-
5. 6. 7. 6. 9. 10. 11. 12. 13. 14.
NORMAL STRESS - kPo
15.
WALL YIELD LOCI
MBTERIRL:S«nd-mdl
MOISTURE CONTENT:as r e c e i v e c
FIGURE:
A-H-3.12
TESTED:12/10/89
TEMPERATURE:ambient
For Sand M D 2
12.
i 1.
!Z.
"'I""!""!""!""!""!'
1
1 " " I " " I " " I " " I " " I " " I " "J
LECENO
O
Perspex
^
Galvanised Steel
c,
E.
7- E-
'0.
1.
2.
3.
4.
h I i i I
I II i 111 , i i I i
5. 6. 7. 8. 9. 10. 11. 12. 13. 14. IS.
NORMAL STRESS - kPo
WALL YIELD LOCI
FIGURE:
MATERJBl:Sand-»d2
MOISTURE C0NTENT:a. received
TESTED:12/10/89
TEMPERATURE:ambient
313
For Sand M D 3
A-II-3.13
12.
i i i i i i i i i
' I • • ' ' I ' ' > •i Ii '
' • ' I
11 .
ie.
LEGENO
^ O
Perspei
£> Galvanised
|""|'"
i i i i i i
'I1111!""!""!""!1
Steel
c.
8.
7.
6.
5. 6. 7. 6. 9. 10. 11. 12- 13. 14.
NORMAL STRESS - kPa
WALL YIELD LOCI
FIGURE:
A-H-3.14
MATERIAL:Send-md3
MOISTURE CONTENT:os
received
TESTEO:12/10/89
TEMPERATURE:embi
For Sand M D 4
i i I I |
LEGEND
O
Perspe>
&
Galvanised
11 i i | i 1 1 i i i i
Steel
15.
5. B. 7. 8. 9. Jf
NORMAL STRESS - kPo
WALL YIELD LOCI
IGURE:
M3T ERIAL:Sand-md4
KOI STURE CONTENT:as
TESTED:12/10/89
TEMPERATURE:ombi
Appendix III Publications While P h D Candidate
Arnold, P. C. and Gu, Z. HM The Effect of the Material Level on the
Flowrate of Bulk Solids from Mass Flow Bins. Proc. Third Inter. Conf.
on Bulk Materials, Storage, Handling and Transportation. Newcastle,
1989, pp 196-199.
Arnold, P. C, Gu, Z. H. and McLean, A. G., On the Flowrate of Bulk
Solids from Mass-Flow Bins, Proc. Second World Congress PARTICLE
T E C H N O L O G Y , Kyoto, Japan, September 19-22,1990, pp 2-9 (Part II).
Arnold, P. C. and Gu, Z. H., The Effect of Permeability on the Flowr
of Bulk Solids from Mass-Flow Bins, Powder Handling & Processing,
Vol. 2, No. 3, September 1990, pp 229-233.