CRANFIELD UNIVERSITY

DEPARTMENT of ENVIRONMENTAL and ORDNANCE

SYSTEMS

FOR PRESENTATION as Ph. D. THESIS

SUBMITTED BY

JOHN ALLUM

A STUDY OF THE BEHAVIOUR OF EMULSION

EXPLOSIVES

Supervisor: Dr. Michael Cartwright

March 2002

This thesis is submitted in partial fulfilment of the requirements for the Degree of Ph.D.

o Cranfield UniverSity. All tights reserved. No part of this publication may

be reproduced without the written permission of the copyright holder

 CRANFIELD UNIVERSITY

DEPARTMENT of ENVIRONMENTAL and ORDNANCE

SYSTEMS

FOR PRESENTATION as Ph. D. THESIS

SUBMITTED BY

JOHN ALLUM

A STUDY OF THE BEHAVIOUR OF EMULSION

EXPLOSIVES

Supervisor: Dr. Michael Cartwright

March 2002
is
This thesis submitted in partial fulfilment of the requirements for the Degree of Ph.D.

Cranfield University. All rights reserved.No part of this publication may

be reproduced without the written permission of the copyright holder

 Abstract

This study investigated the formulation and characterisation of emulsion

explosives. This included the manufacture of more than 120kg of emulsion
explosive of which around 105kg was used on the explosive ordnance range in
over 350 individual firings. For each emulsion composition, an average of eight
firings was undertaken with which to substantiate the explosive performance data.
The formulation was varied to determine the effects of water content upon the
physical characteristics of the emulsion. These physical effects included thermal
conductivity, particle size, viscosity and the explosive performance of the
emulsion. In respect of explosive performance, microballoons were added to
sensitise the emulsion and the proportions of microballoons added were altered to
look at their effect on velocity of detonation, sensitivity and the brisance of the
emulsions.

Emulsion explosives are commonly referred, in literature, as Type 11non-ideal

explosives. This is due to their non-linear behaviour with respect to the variation
of velocity of detonation with density. Traditionally, when an emulsion explosive
was commercially manufactured, the water content has been kept at a minimum
(12-17%). This was accepted as the way to achieve the best explosive
performance, based upon the belief that an emulsion with the highest
concentration of active ingredients, ammonium nitrate and oil, would give the best
explosive performance.

This study examined a wider range of emulsion explosive water contents than has
been previously studied, from 12% to 35% water. It was found, during this study,
that higher water content emulsions, specifically 25% water, had a marked
increase in explosive performance. The highest velocity of detonation recorded
was in a 39mm diameter tube, at 25% water content with 3% microballoons, was
5558ms-1. This was some 15% higher than any other VOD recorded in this study.
The high velocity of detonation, at 25% water content, was one of a number of
physical characteristics in which this water content varied from the other emulsion
water contents.

II
This study endeavored to show that emulsion explosives could exhibit two differing
types of explosive reaction, thermal explosion and grain burning. This was based
on the velocity of detonation and plate dent data, both of which indicated that there
was a change in reaction with water content. Emulsion explosives, with a high
water and high microballoon content, exhibited a thermal explosion type reaction.
They exhibited Type I ideal explosive behaviour, with increasing velocity of
detonation with density. Lower water content emulsion explosives, displayed the
more commonly expected Type 11non-ideal behaviour and reacted in a grain
burning type detonation.

III
 Acknowledgments

I would like to acknowledge the help that I have received throughout my research
from a great many people. Dr Michael Cartwright for his supervision, help and
support throughout the period of my research and a good while beyond! Professor
Alan Bailey for his pragmatic advice, which has saved me many hours of fruitless
work. The personnel from the field chemistry laboratory, Arthur Burns, Jim
Clements Mike Warden and Adrian Rothan for all their advice and help throughout
some very foul weather!

I would also like to thank all my work colleagues for their help and advice and for
many borrowed and purloined pieces of equipment, specifically Dr Anthony
Bellamy, Richard Hall, Lindsay Dunn, Dr Simon Ward and Dr Michael Williams.

IV
 Contents
Glossary XII
...............................................................................................................
1 Introduction 1
..................................................................................................
1.1 Explosivesbackground
...................................................................................
1.1.1 Chemical explosives 2
.........................................................................
1.2 Deflagration 6
.....................................................................................................
1.3 Detonation 7
.......................................................................................................
1.3.1 Simple detonation theory 9
..................................................................
1.4 Initiation Theory 13
.............................................................................................
1.4.1 Hot Spots 13
........................................................................................
1.4.2 Priming 17
............................................................................................
3.1.1 Calculation of detonation pressure 19
..................................................
1.5 Explosive behaviour classification 20
.................................................................
Type 1 Ideal explosives 20
...................................................................
1.5.2 Type 11non-ideal explosives 21
............................................................
1.5.3 Methods for VOD determination 23
......................................................
1.6 Emulsions 25
......................................................................................................
1.6.1 Introduction 25
.....................................................................................
1.6.2 Emulsifiers 28
......................................................................................
1.6.3 Break down of emulsions 30
................................................................
1.7 Emulsion Rheology 32
.......................................................................................
1.7.1 Viscosity 32
..........................................................................................
1.7.2 Particle size analysis 34
.......................................................................
1.8 Backgroundto Emulsions Explosives 35
............................................................
1.8.1 Developmentand use of Emulsion explosives 36
................................
1.9 Ammonium Nitrate (AN) 39
................................................................................
1.100bjective of the research project 41
...................................................................
2 Experimental 42
..............................................................................................
2.1 Material Sources 42
...........................................................................................
2.2 Emulsion Preparation 42
....................................................................................
2.3 Physical Characterisation 45
..............................................................................
2.3.1 Emulsion densities 45
..........................................................................

V
 2.3.2 Particle size analysis 45
.......................................................................
2.4 Thermal Analysis 46
...........................................................................................
2.4.1 Differential scanning calorimetry (DSC) 46
..........................................
2.4.2 Thermal Conductivity 48
......................................................................
2.4.3 Viscosity 52
..........................................................................................
2.4.4 Emulsion breakdown 54
.......................................................................
2.4.5 Velocity of detonation measurements 54
.............................................
2.4.6 Plate dent measurement 56
.................................................................
2.4.7 Double Pipe Test 57
............................................................................
3 Results and Discussion 58
.............................................................................
3.1 Emulsion densities 58
........................................................................................
3.2 Emulsion breakdown 61
.....................................................................................
3.3 Rheology 67
.......................................................................................................
3.4 Particle size analysis 70
.....................................................................................
3.5 Thermal Measurements 73
................................................................................
3.6 Differential scanning calorimetry 74
...................................................................
3.6.1 Heat capacity 74
..................................................................................
3.6.2 Thermal analysis data 76
.....................................................................
3.7 Explosive performance results 95
......................................................................
3.7.1 Introduction 95
.....................................................................................
3.7.2 Hot spot temperatures in the emulsion 98
...........................................
3.7.3 Velocity of detonation data 100
............................................................
3.7.4 VOID data plotted as functions of water content 103
............................
3.7.5 Plate dent data 115
..............................................................................
3.7.6 The effect of mixer type on VOID 122
...................................................
3.7.7 Effects of mixing time on VOID 123
. .....................................................
3.7.8 Viscosity versus VOID relationship 125
................................................
Further Discussion 127
...................................................................................
128
4.1 Initiation.......................................................................................................
4.2 Grain burning and thermal explosion 130
...........................................................
4.3 Low water content emulsions - Grain burning reaction 136
...............................
4.4 High microballoon and high water content emulsions- Thermal explosion.. 139

vi
4.5 Discussion 141
...................................................................................................
5 Conclusions 144
.............................................................................................
6 Further Work 146
............................................................................................
7 References 148
..............................................................................................

Vil
 Figures
Figure 1 Structure of a shock wave q
........................................................................
Figure 2 Schematic of the detonation process in a condensed explosive 10
. ...........
Figure 3 Variation of pressure, density and temperature across a
detonation wave 12
. ....................................................................................
Figure 4 Effect of priming 18
. ....................................................................................
Figure 5 Variation of VOD with density for a type I explosives VOD (ms") 21
..........
Figure 6 Type 11non-ideal explosive. Decrease in detonation velocity with
increasing density of ammonium perchlorate. (According to
Price,1967 )(45) 22
. .......................................................................................
Figure 7 Dautriche method of VOD measurement 23
. ..............................................
Figure 8 Electron micrographs of butter and a low fat spread. (Both at the
same magnification)...............................................................................27
Figure 9 Action of emulsifying agents in a water-in-oil emulsion 29
. .........................
Figure 10 Sorbitan monooleate molecule 30
.............................................................
Figure 11 Processes of emulsion breakdown 31
.......................................................
Figure 12 Mixing apparatus 43
..................................................................................
Figure 13 DSC Schematic 47
....................................................................................
Figure 14 Lee's apparatus for thermal conductivity measurement 49
. ......................
Figure 15 Cooling Experiment 50
..............................................................................
Figure 16 Schematic of Viscolog MRV-8 53
..............................................................
Figure 17 Measurement of VOD with ionisation probes 55
. ......................................
Figure 18 Plate dent measurement 56
. .....................................................................
Figure 19 Double Pipe Test 57
..................................................................................
Figure 20 The effect of water and microballoon content on density 58
. ....................
Figure 21 % TIVIDversus % microballoon content 60
................................................
Figure 22 Picture of an emulsion from an optical microscope x1550
(expanded for report) 61
............................................................................
Figure 23 Shows the onset of coalescence occurring in an aged sample
(xl 550) (expanded for report) 62
...............................................................
Figure 24 Coalescence now more advanced in sample (x1550) (expanded
for report) 63
..............................................................................................
Figure 25 Coalescence, now catastrophic and seen on a macroscopic
scale (x1 550) (expanded for report) 64
. ....................................................

Vill
Figure 26 Example of flocculation in emulsion sample (xl 550) (expanded
for report) 66
..............................................................................................
Figure 27 Variation of viscosity with spindle speed (viscosity/log plot) 68
.................
Figure 28 Viscosity's for emulsions 69
. .....................................................................
Figure 29 Effect of water content on particle size 70
.................................................
Figure 30 Effect of mixing time on particle size 71
. ...................................................
Figure 31 Effect of mixer type on particle size 72
......................................................
Figure 32 Thermal conductivity of emulsion explosives 74
........................................
Figure 33 Heat capacity as measured by DSC 75
.....................................................
Figure 34 DSC trace of AN prill 77
. ...........................................................................
Figure 35 DTA (AT) and heating (T) curves for ammonium nitrate, showing
three polymorphic transitions (at 32"C, 84'C and 1250C) 78
.....................
Figure 36 Ammonium nitrate phase changes from DSC 78
. .....................................
Figure 37 DSC trace of 35,30 and 25% water emulsions 80
.....................................
Figure 38 20% water emulsion 80
.............................................................................
Figure 39 15% water emulsion 81
.............................................................................
Figure 40 12% water emulsion 81
.............................................................................
Figure 41 Crystallised emulsion sample (12% water) heated and allowed
to cool then reheated 83
............................................................................
Figure 42 Crystallised sample, (12% water) heat cycled 3 times 84
. ........................
Figure 43 DSC trace crystallised sample (15.3mg) -20OC-1800C 85
.......................
Figure 44 Second run, after Figure 43 85
..................................................................
Figure 45 Final run, after Figure 44 86
. .....................................................................
Figure 46 Large sample size, 100.8mg 87
. ...............................................................
Figure 47 Second run, after Figure 46 87
..................................................................
Figure 48 Third run, after Figure 47 88
......................................................................
Figure 49 Final run, after Figure 48 88
. .....................................................................
Figure 50 Method used for gas analysis of crucibles 89
............................................
Figure 51 GC-MS traces m/z 44 91
...........................................................................
Figure 52 Mass analysis of peak at RT 8.2min 92
.....................................................
Figure 53 Mass analysis Of C02 peak, RT 7.6min 93
................................................
Figure 54 Explosive ordnance research range, with a charge prepared for
firing 95
. ....................................................................................................

ix
Figure 55 Detonation of charge- note the white cloud 95
. .........................................
Figure 56 Emulsion initiating 97
. ...............................................................................
Figure 57 Velocity of detonation versus %TMD 101
..................................................
Figure 58 Velocity of detonation versus microballoon and water content 102
...........
Figure 59 12% water content emulsion explosive 104
. .............................................
Figure 60 15% water content emulsion explosive 105
. .............................................
Figure 61 20% water content emulsion explosive 106
. .............................................
Figure 62 25% water content emulsion explosive 107
. .............................................
Figure 63 Showing the tube set up for a 25% water, 2% microballoon,

emulsion ............................................................................................. 108

Figure 64 27.5,30 and 35% water content emulsion explosives 110
........................
Figure 65 Average VOID for emulsion water contents 111
........................................
Figure 66 The linear section of density versus VOID 112
. .........................................
Figure 67 Plate dent volume versus TMID 115
. .........................................................
Figure 68 Plate dent versus TMID for 12% water content 116
. ..................................
Figure 69 Plate dent versus %TMD for 15% water content 117
...............................
Figure 70 Plate dent versus %TMD for 20% water content 118
................................
Figure 71 Plate dent versus %TMD 25% water content 119
.....................................
Figure 72 VOID versus plate dent for 12% and 15% water content

emulsion ............................................................................................. 120

Figure 73 VOID versus plate dent for 20% and 25% water content

emulsion ............................................................................................. 121

Figure 74 Effect of mixing time on VOID for 20% water emulsion/3%
microballoon ....................................................................................... 124
Figure 75 Effect of median particle size versus VOID 20% water 3%
microballoons . .................................................................................... 125
Figure 76 Viscosity, as calculated from emulsions with no microballoons
present, versus VOID for emulsions with 3% microballoons . .............. 126
Figure 77 All explosive firing data plotted on a VOD versus density graph 127
. .......
(96)
Figure 78 Plot of reaction rate versus normalised time 133
. .................................
Figure 79 Density versus VOD calculated by Cheetah using BKW equation 134
.....
Figure 80 Grain burning detonation VOD versus density 137
...................................
Figure 81 Type 11non-ideal explosive behaviour plate dent volume versus
density 138
................................................................................................

x
Figure 82 Thermal explosion VOD versus Density 139
.............................................
Figure 83 Plate dent versus density thermal explosion 140
......................................
Figure 84 Comparison of the initial slope in VOD versus %TMD 142
.......................
Figure 85 The variation in the slope of VOD versus water content 143
.....................

xi
Glossary
2, -Thermal conductivity.

oc- Burning Rate Index- The linear burning rate is the rate at which the chemical
combustion reaction is propagated by both thermal conduction and radiation.
p- Coefficient of burning rate (dependent on the material).

p- Density in kgd M-3.

8- Shape factor 0.88 for an infinite slab, 2.00 for an infinite cylinder, and 3.32 for a
sphere.
%-Thermal conductivity in J/cm/s/K.

y- Ratio of specific heats (Cp/Cv).

AH - Enthalpy of reaction.
AH -The heat of decomposition reaction in J/mol.
AN- Ammonium Nitrate.
ANFO- Ammonium Nitrate Fuel Oil mixture. Oil is mixed with ammonium nitrate
prills to give a cheap explosive, widely used in quarrying.
C -Heat capacity.
Chapman-Jouguet Point- The point at which the Hugoniot curve and Rayleigh
line meet. At this point, the velocity of detonation is a minimum subject to the
constraints of the theory.
Continuous phase- In a water-in-oil emulsion the phase that surrounds the water
droplets.
Critical Diameter- The minimum diameter of explosive required for detonation to
be sustained.
DDT- Deflagration-Detonation-Transfer.The process of going from a burning
(deflagration) reaction to a detonation.
Disperse Phase- The phase surrounding droplets in solution; see also continuous
phase.
DSC -Differential Scanning Calorimetry, a method of thermal analysis.
Ea-The activation energy in J/mol.
Emulsion -A colloid in which small particles of one liquid are dispersed in another
liquid.
FK- Frank-Kamenetskiiequation for heat transfer.

xii
Glass Microballoons- Small glass microballoons which can vary in both size and
collapse pressure. In this study 3M B15/250 Silica glass balloons with an average
diameter of 15ýtrnand a collapse pressure of 250MPa were used.
Inviscid- A hypotheticalfluid without any viscosity effects.
lonisation Probes- Polyurethane coated wires inserted into the explosive column
in order to determine velocity of detonation.
Lead Azide- Pb(N3)2a primary explosive mainly used in detonators.
P- Pressure.
PIBSA- Polyisobutylene Succinic Anhydride Ethanolamine- a dark brown viscous
liquid with a density 0.91 - 0.93 kg/dM-3.
Prills- Spheres of ammonium nitrate used as fertiliser.
Q- Rate of heat evolved per unit volume.
R- Universal gas constant.
r- Linear burning rate.
Solid Solution- A crystalline material that is a mixture of two or more components,
with ions, atoms, or molecules of one component replacing some of the ions,
atoms, or molecules of the other component in its normal crystal lattice.
Sorbitan Mono-Oleate- Viscous, amber liquid, insoluble in water. Density 1 kgdM-3,
Viscosity 100 MPa (250C).
T- Temperature.
Tc - The critical temperature in K.
TNT Trinitrotoluene,a yellow highly CH3C6H2(NO2)3.
explosive, crystalline solid,
VOD- Velocity of denotation.
Z- The pre-exponentialfactor in s".
A- Laplacian operator (x, y and z co-ordinates).

xill
1 Introduction

1.1 Explosives background

An explosion is defined as a violent expansion of gas that usually generates heat,

sound and a pressure wave. They can be conveniently separated into three
distinct categories.

9 Physical explosions, these are the most common in nature. They are pressure
bursts that do not release heat, but release large amounts of gas. A balloon
bursting would be a common example. Physical explosions can be much more
violent than this, a volcano being an example of a huge pressure burst.
Nuclear explosions can occur naturally, with the sun being the prime example
of nuclear fusion. This results in the liberation of immense quantities of heat,
but no gas. A nuclear explosion is by far the most energetic type of explosion.

9 Chemical explosions are the result of very rapid chemical reactions that
release large quantities of heat and gas. Chemical explosives are discussed in
more detail below and they are the most important type to the explosive
industry.

An explosive is material, with fuel and oxidiser elements, that is capable of

undergoing an exceptionally fast reaction without the participation of external
reactants. In general, explosives have three basic characteristics:

1) They are chemical compounds or mixtures ignited by heat, shock, impact,

friction or a combination of these conditions

2) Upon ignition, they decompose very rapidly (detonation and deflagration).

3) Upon detonation, there is a rapid release of heat and gas, which expand
rapidly, usually with sufficient force to overcome confining forces.

I
A chemical reaction involves the conversion of reactants to products. For this to
occur, without a significant energy input, the reactants must have a higher internal
energy than products. Energy is required to produce activated molecules and to
initially break some of the bonds in the molecule, in an endothermic process.
Once this has occurred products are then formed. The formation of products
results in liberation of energy, which, overall, yields an exothermic reaction. After
initiation, the reactions involved in the burning and detonation process are always
exothermic, and therefore always liberate energy.

1.1.1 Chemical explosives

Chemical explosives contain both fuel and oxidiser components. There are,
however, two major types of chemical explosives:

Composite explosives- these have two, separate fuel and oxidiser components,
which react to form the products. In order for a rapid reaction to proceed, the fuel
and oxidiser must be intimately mixed. The smaller the particle size, for solids, the
faster the reaction can occur. An example of a composite explosive is ammonium
nitrate fuel oil (ANFO). This comprises solid prills of ammonium nitrate, the
oxidiser, surrounded by oil, the fuel. The smaller the ammonium nitrate prill size is
the faster the reaction can occur. This is an effect, which is due to surface area,
the smaller the particle size the larger the surface area. Composite explosives are
commonly used in the commercial blasting industry, with only limited use in
military blasting charges, usually amatol/TNT fillings. Cost and availability are the
main reasons for their widespread use in commercial blasting, with poor
performance the reason for their limited military use. Safety is an important
parameter with any explosive and composite explosives can often be stored as
separate non-explosive ingredients.

Single molecule explosives- the oxidiser and fuel elements are contained within
the same molecule. This can be considered an intimate mix on the atomic scale
with a separation of nanometres between the fuel and oxidiser. A single molecule
explosive commonly reacts much faster than two-phase explosives, leading to a

2
higher velocity of detonation. Single molecule explosives are commonly used in
military ordnance due to their superior performance characteristics.

Single molecule explosives can be further subdivided into three distinct categories
based on the ease of ignition for each type:

Primary explosives are the most sensitive and are therefore easily initiated.
Examples of primary explosives are substances such as lead azide, silver azide,
and mercury fulminate. Primary explosives have a limited use due to their high

sensitivity and poor performance. The most common use for primary explosives is

as an initiator in detonators. Some primary explosives are able to transit from

surface burning to detonation within a very small distance. A 0.2mm thick grain of
lead azide when ignited will transit from burning to deflagration to detonation
(DDT) within the grain thickness. The lead azide molecule decomposes in a two-
step reaction producing high molecular weight reaction products. The reaction
products are generated at the surface faster than they can expand away, which
in build up of pressure at the burning surface leading to detonation (1).
results

Secondary explosives, substances such as trinitrotoluene (TNT) and

cyclotrimethylenetrinitramine (RDX) can also burn to detonation. However, a
larger quantity of explosive and some form of confinement is often required. With
small quantities, they will often burn with an example of this being a stick of TNT,
which, if lit, will burn slowly without undergoing DDT. This is because there is little
pressure build up from the reaction products, so the reaction does not speed up
and transfer to detonation does not often occur. When the quantity is increased or
the explosive is confined a pressure build usually occurs and transfer to detonation
often results.

Tertiary explosives, which include ammonium nitrate, are extremely difficult to

initiate and usually require a sensitising agent. Notwithstanding this, they are,
under certain conditions, capable of undergoing a burning to detonation transition.
Some of the largest accidental explosions have involved ammonium nitrate,
without any obvious sensitising agent.

3
Most secondary explosive compounds contain nitrogen and oxygen. When
oxygen combines with nitrogen, little energy is released. However, when oxygen
combines with carbon or hydrogen a relatively large amount of energy is released.
As the explosive detonates, the nitrogen-oxygen bond is broken and the oxygen
combines with carbon or hydrogen to form bonds. This results in the formation of
compounds such as carbon monoxide, carbon dioxide, water vapour and free
nitrogen. This is formed in varying proportions, which depend on the relative
proportions of fuel and oxidiser. These reactions liberate considerable quantities
of heat and gas, with the amount of energy actually released dependent upon the
explosive compound involved. An example reaction of a common secondary
explosive, RDX, is shown below.

Equation 1 Decompositionof RDX

C3H6N606->
3N2+ 3H20 + 3CO

This is the initial decompositionsequence of RDX but some of the products remain
reactive after the initial reaction is complete. These products expand and undergo
further reactions. As the products expand, they may mix with oxygen in the air,
thereby oxidising the carbon monoxide and producing a secondary fireball. This
occurs as the enthalpy of formation for carbon dioxide is such that 283 kJ/mol of
energy is released by this reaction (AH f kJ/moIC02-393.5, AH f kJ/mol CO -110.5).

In this reaction, RDX decomposes to give three main product gases. The fact that
explosives are mostly self sufficient in oxygen gives rise to a distinction between
explosives and other flammable substances. Whilst the combustion of explosives
can take place equally well in the presence or absence of air, the combustion of a
flammable substance requires oxygen. As explosives supply their own source of
oxygen, less energy is available compared to a flammable substance such as
petrol. Burning an explosive in air will generate about eight times more heat than
detonation does. The reason explosives are so destructive, in comparison to
flammable substances, is the speed at which an explosive releases its energy.

4
An explosion can be differentiated from combustion by the speed and type of

reaction. The speed of the reaction allows an explosive to do useful work and, on

an energy basis, explosives do not have a high internal energy in comparison to

fuels such as petrol. As petrol contains no oxidiser, it burns with oxygen from air,

which is a rate-determining step in its combustion. As an explosive has its own

oxidiser and fuel, it is not therefore limited by this step. A good solid explosive,

such as RDX, converts energy, at the detonation front, at a rate of approximately

1010 WCM-2 As a comparison, the total solar energy intercepted by earth is about
.
4x 1016W Therefore, a 20m square detonation wave gives the same power
.
that the from the (2)
output as earth receives sun

An explosive is a material capable of a violent exothermic chemical reaction(3) but

,
the normal resting state is at a metastable chemical equilibrium. An explosive is
stable under 'normal' conditions because the reaction rate is so slow. A local
stimulus can initiate a reaction wave, which spreads through the material. This
reaction wave can be one of two types: a deflagration or a detonation wave.
Deflagration is a surface-burning phenomenon and a detonation is characterised
by the presenceof a shock wave. These are described in the following section.

5
1.2 Deflagration

A deflagration (flame) is a slow (subsonic) wave where transport processes such

as viscosity, heat conduction and matter diffusion are dominant. Changes in
momentum and kinetic energy are small and the pressure change through the
wave, to an approximation,can be neglected.

Burning reactions are surface phenomena and as such, the flame will spread over
the surface of the material quicker than through the main body. In 1893, Paul
Vieille determined a mathematical relationship between burning speed and
pressure. He showed that linear burning rate, the speed of burning, r, is
determined by the pressure P according to the equation below, Vieille's law(4)
.

Equation 2 Vieille's burning rate law

PPOC

r Linear burning rate

p Coefficient of burning rate (dependent on the material)
P Pressure

oc Burning rate index (dependent on the material)

The rate of burning index, cc,can vary from 0.3 to greater than 1.0. With values of
less than one, a flame front is generated and the reaction proceeds rapidly, but
only as a surface burning phenomena. An example of this is black powder, which
when unconfined will burn at a rapid rate (-O. lkms-1), which is below that of
detonation. If the explosive has a burning rate index of greater than 1.0, then
pulses are generated. These pulses can accelerate the linear burning rate to
sonic velocity, thereby leading to detonation. When an explosive is ignited, they
often burn with no oxygen being required for this process to occur. If the explosive
was confined, such that the reaction product gases cannot escape, then the

6
pressure, at the reaction zone, builds up. Burning reactions are functions of
pressure and temperature; therefore, the reaction rate increases as the pressure
increases. The high pressure forces the hot gases into the surrounding material
thereby accelerating the process. Pressure waves are generated in the
deflagrating region, these compact and compress the explosive material, causing
greater confinement and hence even greater pressure build up. The
compressional waves cause shock pile up, which, given sufficient time and
distance, form the conditions necessary for detonation to occur. A shock wave is
then formed, and the transition to detonation occurs. This process is commonly
referred to as deflagration-to-detonation transition, DDT. DDT transition is an
important mechanism for the initiation of explosives and Mallard(5) in 1881, was
,
the first person to show that this could occur.

There have been many examples of DDT occurring accidentally, such as the
Texas container ship explosions (6) container ship carrying ammonium nitrate
-A
had a minor fire in the main hold, which was not considered a risk. After burning
for many hours, the ammonium nitrate eventually exploded. The ammonium
nitrate underwent DDT due to the large mass of explosive present (-3000 tons),
which then destroyed everything within a 2-mile vicinity.

1.3 Detonation

A detonation runs at supersonic speeds, 2000 to 8000 ms-1in solids and liquids,
with the velocity limited to the local speed of sound in the explosive. The transport
properties, which were important for combustion, become relatively unimportant in
a detonation. Changes in momentum and kinetic energy are dominant, with
compressibility and inertia being important properties. The leading element of a
detonation wave is a shock front, and the high temperature and pressure produced
by this shock front is maintained by inertial confinement. That is the outer layers
of the material are not appreciably displaced before the reaction in the centre is
over.

The process of an explosive going from an undisturbed state to an explosion is

described as the detonation process. The detonation process can be explained by

7
Detonation theory, which is based on the study of shock propagation through inert
materials. Both Berthelot(7) and Vieille(8)determined the existence of detonation
waves in 1881. Using a Le Boulenge chronograph, they measured the
propagation velocity of detonation waves in approximately 50 different mixtures of
fuels and oxidisers. Their conclusions were that the propagation velocity of an
explosive is a constant, which depends upon the mixture and not on external
conditions, such as tube diameter or tube material. This discovery, within twenty-
five years, allowed the growth of detonation theory, in thermodynamic and
hydrodynamicterms, to be developed.

This development began, in the 19th century, with theoretical studies of the
detonation process. Much of the understanding of detonation phenomena was
based on research into gas explosions. This began with the early work of
Berthelot(9) (1882) and Dixon(lo) (1893), who attributed the high detonation
velocities to the properties of explosion product molecules. Schuster (1893) was
the first to point out similarities between detonations and unreactive shock waves.
He suggested, the now accepted view, that detonation is a shock wave supported
by chemical reaction.

A shock wave (11,12,13,14,15)

is a region of very high pressure that passes through a

material at the local speed of sound for that material. An important feature is that
the local speed of sound is the speed of sound at the shock front, where the
material is under high pressure. A representation of a shock wave in an explosive
material is shown in Figure 1. This shows the shock impinging on the material and
immediately exerting a maximum pressure, referred to as the Von Neumann spike.
After the shock front has passed, the pressure reduces in an exponential manner.

In an explosion, the shock wave and reaction zone are called the detonation front.
The shock wave compresses the explosive, which breaks apart the chemical
bonds, causing a chemical reaction. This chemical reaction continuously adds
energy to the shock front, which as well as driving the shock front forwards, is also
lost laterally. This is because the detonation wave expands sideways as well as
forwards through the column. If more energy is added to the shock front, by the
chemical reaction, than is lost laterally, then the velocity at the front increases.

8
This process of acceleration continues until the energy added, to the detonation
front, equals the lateral losses due to expansion and heat flow. Once this has
occurred, and the losses from the system equal the energy being added to it, then
a steady state velocity of detonation is achieved.

Figure 1 Structure of a shock wave

4-Von Neumann
spike

C-J plane

AL

Reaction
zone Shockfront

Shockwavetravelling in this direction

1.3.1 Simple detonation theory

The first attempts at a thermodynamiC(16)analysis of a detonation were

independently reported by Chapman (17) (1899), Jouguet("') (1903), and
(1891). This is commonly referred to as the Chapman-Jougettheory.
Michelson(19)
In Chapman-Jouguet (CJ) theory, the detonation process is treated as a one-
dimensional wave of infinitesimal thickness, without mass, momentum or energy

9
 loss, in an inviscid (no viscosity) fluid-, this is done for mathematical simplicity. The
(20)
detonation process is shown schematically in Figure 2

Figure 2 Schematic of the detonation process in a condensed explosive.

Fxpanding
OýIscs Chapinan-
.
Jouguict region

Shock front in the

z explosive

FXPI, IVF
pp,
Direction of
detonation

Primary reaction
zone

Stable by-
products mainly
gases
I Jndisturbcd
Shock wave in the
explosive
surroundim, media

It can be seen, from Figure 2, that the CJ region, described by the equations of

conservation, is at the end plane of the reaction zone. This is a point after the
shock wave has passed through the material and the initial reactions are
complete, although the gases produced have not expanded and reactions with
oxygen in the air have yet to take place.
Detonation theory has significantly advanced since the 1900's, but the Chapman-
Jouguet theory is still used as the basis for modern interpretation. In the 1940s
Zeldovich (21),
Von Neumann (22)and Doering(23)independently developed a model
to describe the detonation process. This is referred to as the simple theory or the
ZND model. This model makes some assumptions in order to simplify the

mathematics of the system.

1) Flow is one-dimensional (no lateral losses).

2) The front of the detonation is a jump discontinuity (happens instantly).

3) The reaction-productgases leaving the detonation front are in chemical and

thermodynamic equilibrium and the reaction is complete (the entire
explosive is consumed and all the energy is released).

4) The chemical reaction-zone length is zero A typical high explosive has a

reaction zone of about 1mm thickness. For less sensitive explosives such
as amatol (ammonium nitrate TNT mix) this can be 10 times this length.

5) The VOD is constant; this is a steady state process; the products leaving
the detonation remain the same independent of time

These assumptions are known to be incorrect but this allows the mathematics to
become tractable and the results given by this method give reasonable agreement
to experiment.

The basic theory allows profiles for temperature, pressure and density to be
determined through the reaction zone and Figure 3 shows this variation across the
reactionzone.

11
Figure 3 Variation of pressure, density and temperature across a detonation wave.

Shock wave direction

Pressure variation
Through reaction zone

Unreacted
exvlosive
Reacted
explosive

Density variation
through the reaction zone

Shock wave.. End of reaction zone

T
Temperature variation through the
reaction zone

This shows that the pressure and density reach an instant maximum at the shock
front, which then drops off through the reaction zone. Whereas the temperature
continues to rise throughout the reaction zone, as secondary reactions occur.

12
1.4 Initiation Theory

An explosivecan be initiated(24)by various different forms of energy delivery.

These stimuli could be friction, shock, electrical or thermal. Using the Frank-
Kamenetskii equation (see later), with Arrhenius kinetics, bulk heating and thermal
initiation can treated. Bowden and Yoffe(25)
came up with a treatise in 1952 on the
mechanism of initiation in explosives.

Whatever form of energy is supplied to an explosive to cause initiation, once the

activation energy for that mechanism is exceeded, the detonation process will
begin. An explosive can be initiated by applying a shock of greater energy than its
critical activation energy. Below this critical energy, the reaction will not generate
to
enough energy support itself. This does not mean that reaction does not occur,
but that the reaction 'fades'. If the explosive contains sensitising agents, such as
grit or bubbles then the minimum shock pressure required for initiation is
(25)
decreased by approximately one order of magnitude . Voids in the explosive
cause irregularities in the mass flow when shocked and the explosive is initiated
by local hot spots formed at these irregularities. The hot spot mechanism is
important in the propagation and failure of the detonation wave.

1.4.1 Hot Spots

(26,27,28,29,30,31,32,33,34,35,36,37,38,39,40)
Hot spots have been well characterised and are
small areas where heat is generated when an explosive is subjected a variety of
stimuli. Although they are large, compared to molecular dimensions they are
to 10-3
small, within the range of
_10-5 CM3, in comparison to the physical
dimensions of the charge. Stimuli, which can produce hotspots in explosives, are
described below.

13
e Adiabatic compression of small entrapped bubbles of gas

9 Frictional hot spots 1) 1nter-crystalline friction

2) Grit particles

3) On confining surfaces

-m Viscous heating of rapidly flowing explosive

* Shear of crystals

For emulsion explosives, the main area of interest is the adiabatic compression of
gas bubbles. When an adiabatic reaction occurs there is no transfer of heat in or
out of the system, all the work supplied being converted to heat at the hot spot
site. An emulsion explosive usually has micro-balloons or gas bubbles added to
the emulsion to provide a source of hotspots. Micro-balloons (3M B15/250) are
glass balloons with an average diameter of 60gm; they are filled to 0.2 bar with
sulphur dioxide and molecular oxygen.

The initiation of an explosion, when small gas bubbles are trapped in liquid, is due
to the adiabatic compression of the bubble. For an ideal gas the temperature
reached inside the bubble (T2 degrees absolute) depends on the compression
ratio given by;

Equation 3 Compression ratio for hot spot temperature determination.

(7-1)
P2 y
T2 ý-Tl
P,

P, = Initial pressure in gas bubble.

P2= Final Pressure in gas bubble.

y= Ratio of specific heats (Cp/C, ) (y has an accepted value of 1.4 for air and 1.29
for sulphur dioxide (41)).

T, = Initial temperature. T2= Final temperature.

14
The initial pressure would usually be atmospheric pressure but micro-balloons are
filled at a reduced pressure of 0.2 bar. The higher the initial pressure of the
bubble the lower the temperature obtained by the hot spot. By having, the micro-
balloons filled to only 0.2 bar pressure the pressure differential is greater and
hence temperature obtained will be higher. This gives a means of determining the
hotspot temperature obtained from an applied pressure.

There are other processes, beyond adiabatic compression, which occur when a
bubble is subject to a shock wave. These include jetting, where the top surface of
the bubble forms a jet, which impinges on the bottom surface generating a high-
pressure, high temperature zone, and vaporisation where a small amount of the
explosive is vaporised in the cavity as it collapses causing it to violently react.
Although both these processes are important, they are not the dominant process
in hotspot initiation.

1.4.1.1 Fran k-Kamenetskii Equation

Frank-Kamenetskii(42) proposed the classical three-dimensional heat transfer

equation, which relates the rate of heat production to the rate of temperature rise
in a reacting material. The equation shows that if heat is evolved by a reaction
faster than it can be transferred away, then the temperature of the reacting
material must increase. Increasing the temperature increases reaction rate, as
shown by the Equation 4.

Equation 4 Rate of heat evolved to reaction temperature.

-Ea /RT
Q=pAHZe
Rate of heat evolved per unit volume.
Density.

AH - Enthalpy of reaction.

15
Ea -Reaction activation energy.
R Universal gas constant.
T Temperature.
Z Pre-exponentialfactor

The heat produced by a reaction is transferred to the surrounding material; this

can be shown using the Frank-Kamenetskii equation. This equation gives the heat
transfer rate, a figure that is dependent upon temperature, as well as thermal
conductivity, heat capacity and density.

Equation 5 Frank-Kamenetskii(FK) equation for heat transfer.

Where;

7, iV2
dT ) -Ea/RT
_ T+pC = pAH Ze
dt
X -Thermal conductivity
C -Heat capacity
V- Laplacian operator (x, y and z co-ordinates)

The FK equation predicts that for each explosive a critical temperature can be
obtained, which, if exceeded, will result in explosion. This critical temperature is
the point at which heat loss exactly equals heat gain, any increase in heat gain
results in a runaway reaction and thereby an explosion. The heat transfer rate
given by the equation depends upon sample size and geometry. With the same
geometry, the larger the mass of the explosive, the slower the rate of internal heat
conduction. This leads to the conclusion that the critical temperature is dependent
on the size and geometry of the explosive, as well as its composition. When the
FK equation is solved, to obtain the critical temperature, the following expression
(43)
is obtained

16
Equation 6 Frank-Kamenetskiiequation for critical temperature.

! ý'- r2p AH Ma
=R In
Tc Tc2UR

Where;

r The radius of a sphere, cylinder, or half slab thickness in cm

P Density in kg/d M-3

AH The heat of decomposition reaction in J/mol

z The pre-exponentialfactor in s"
Ea The activation energy in J/mol

Tc The critical temperature in K

R The universal gas constant

X Thermal conductivity in J/cm/s/K

8 Shape factor 0.88 for an infinite slab 2.00 for an infinite cylinder and 3.32 for

a sphere

The shape factors are used to correct for sample geometry, as different shapes
will have a different surface areas and therefore lose heat at different rates. The
equation represents a condition where the rate of heat generation exactly matches
that of heat lost.

1.4.2 Priming

Priming is the energy required by the explosive to successfully propagate to allow

detonation to occur. In order that adequate priming is achieved, the minimum
detonation pressure of a primer must be at least equal to that of the explosive,
preferably it should be higher. If the detonation pressure applied to the explosive
is higher than required, then the explosive will be overdriven. This means that the
explosive will initially detonate faster than its theoretical maximum velocity of
detonation. This is an effect, often used in the commercial quarrying industry,
where a more powerful explosion is often required at the toe of a quarry face. This

17
is only a short-lived effect, which rapidly dies away with distance. The distance
the over-driven state exists for is dependent on the diameter of the explosive and
the source of initiation. With a primer of equal diameter to that of the explosive
charge, the overdriven state can exist for up to three charge diameters.

This is an important effect and must be accounted for when measuring the VOIDof
an explosive initiated with a primer. Conversely, if the pressure of detonation and
VOID applied by the primer is lower than the explosive's own VOID then the
explosive will be under-driven, reacting at a rate lower than its theoretical VOID,
until the detonation wave either fades or picks up. This is often referred to as the
run up distance, and if the explosion is to propagate, the VOIDwill pick up within
three charge diameters. Figure 4 illustrates the effect of under-driving and over-
driving the explosive.

Figure 4 Effect of priming.

Overdriven

VOD
SteadyStateVOD

ýd 3d 4d 5d '
ýýDistance frompnmer(d=dianwter
of primedexplosive)

Underdriven

18
3.1.1 Calculation of detonation pressure

Cook(44)developed an empirical method for determining the detonation pressure, if

the density and the velocity of detonation are known. It can be shown that the
Chapman-Jouguetpressure can be determined by:

Equation 7

Pcj

Where Pcj is the CJ or detonation pressure in gigapascals (GPa)

p is density in kgld M-3

y is the ratio of specifics heats of the detonation product gases

D is the velocity of detonation in km/s

This allows the detonation pressure to be predicted to within an error of less than
5%. An alternative method, which allows the calculation of detonation pressure to
a similar degree of accuracy, without requiring the ratio of specific heats for the
detonation gases, the following equation can be applied.

Equation 8

* 10-7 * VOD2 *p
Dp = 2.325

As Equation 8 shows, a high density and a high VOD lead to high detonation
pressures. An emulsion explosive has both a low VOD and a low density; it
therefore follows that the detonation pressure will be low. This signifies that to
successfully initiate an emulsion explosive, a primer with a high detonation
pressure of detonation is not necessarily required.

19
1.5 Explosive behaviour classification

Chemical explosives, as described by Price(45) can be classified as one of two

,
distinct categories. The first, and most common category, is type I ideal
explosives and the second, less common, is type 11non-ideal explosives. This
classification is based upon the variation in velocity of detonation as a function of
density.

1.5.1 Type I Ideal explosives

Type 1 explosives are the standard military high explosives such as, RDX, HMX
and TNT. The density of an explosive affects its performance in terms of the
velocity of detonation, heat and pressure. The term of most interest is velocity of
detonation (VOD) and for a Type 1 explosive, increasing density leads to higher
velocity and pressure of detonation. For type I explosives, the relationship
between VOIDand the density of the unreacted explosive is close to linear. This
leads to the relationship shown in Equation 9, which is shown graphically as
Figure 5.

Equation 9 Relationship between VOIDand density.

D=a+bp

D- Velocity of detonation.
a, b- Empirical constants specific for each particular explosive and identified from
experimental studies.
p- Density.

The density of an explosive is an important factor in determining the properties of

an explosive although direct comparison of performance cannot be made from the
density alone. Comparing lead azide with RDX shows this, lead azide is four
times more dense than RDX, but its velocity of detonation is half that of RDX. For
the same type 1 explosive the higher the density of the explosive the higher the

20
higher the VOID will be. As a result of this, type I ideal explosives, are often
loading density to give maximum performance. (46)
packed at maximum

Figure 5 Variation of VOD with density for a type I explosives VOD (ms-1)

km/sec
RDX
PETN
60RDX/40TNT
Picric acid

Di I-

6 +
+
.1-11,
13

4 #*+ ýed
.010Iol*
e-.

0-5- 1 1-5 g/cm3

R
0

1.5.2 Type 11non-ideal explosives

The behaviour of type 11explosives is more complex, although density is still a

parameter of interest. Examples of type 11explosives are ANFO, emulsions and
slurries. These are not single molecule explosives and increasing the density

often causes differing reaction mechanisms to be observed. At low densities, the

velocity of detonation increases as a linear function of density, but as the density is

increased, the velocity of detonation drops off rapidly until failure occurs. A graph
of this behaviour, as first reported by Price, is shown in Figure 6.

21
Figure 6 Type 11non-ideal explosive. Decrease in detonation velocity with
increasing density of ammonium perchlorate. (According to Price,1967 )(45)
.
/
.0
D
km/sec
d --p- oD
5
d= 76.2 mm
-4!

50-8
4

349.9

Failure

0 0-2 0-4 0-6 0.8 1-0

PO/PM.
U

There are many explanations for this process, with reaction zone length being of
prime importance, but at present, the process has not been fully elucidated.
Emulsion explosives, as used in this study, have been categorized as type 11non-
ideal explosives, although there has been little published data on the
characterisation of them.

22
1.5.3 Methods for VOD determination

VOID can be measured by a variety of different methods, varying from point

methods such as ionisation probes and fibre optic lines to continuous systems
such as resistance wire and streak camera's.

One of the original methods, and a method that is still sometimes utilised, was the
Dautriche method for measuring velocity of detonation. It was developed (1906)

as a method of measuring the velocity of detonation in short columns of explosive.

The practice before this had been developed was to use columns of explosive tens
of meters long.

Figure 7 Dautriche method of VOD measurement.

23
The(47)test sample is inserted into a cylindrical column and blasting caps are
inserted into the sample at a set distance apart (L) (see Figure 7). A loop of
detonating cord, of known VOID,is connected to the blasting caps and placed on a
lead sheet (P). On the lead sheet, the centre of the cord is marked out (E). The
explosive column is then initiated at one end. The cord is ignited as the detonation
wave passes over the blasting caps and sets them off. The meeting point of the
two detonation waves in the cord makes a notch in the lead witness plate (F). The
distance between the meeting point and the centre of the cord (m) is a measure of
the reciprocal detonation rate to be determined as shown by Equation 10.

Equation 10 Determinationof VOD using the Dautriche method.

DxL
2m

Dx- Detonation velocity of sample.

D- Detonation velocity of cord.
L- Length of distance to be measured.
M- The distance between the notch and the centre of the cord length.

24
1.6 Emulsions

1.6.1 Introduction

An emulsion comes under the broad heading of a colloid, which Thomas Graham
originally defined, in 1861, as substances, such as starch or gelatine, which will
not diffuse through a membrane. He distinguished colloids from crystalloids (i.e.
inorganic salts), which would pass through membranes. Later it was recognized
that colloids could be distinguished from true solutions by the presence of particles
that were too small to be observed with a normal microscope, yet were much
larger than normal molecules. Colloids are now regarded, as systems in which
there are two or more phases with one, the dispersed phase, distributed in the
other, the continuous phase. Moreover, at least one of the phases has small
dimensions (in the range 10-10-6 m). Colloids can be classified by various
subdivisions.

Sols are dispersions of small solid particles in a liquid. The particles may
be macromolecules or clusters of small molecules. Lyophobic sols are
those in which there is no affinity between the dispersed phase and the
liquid. An example is silver chloride dispersed in water. In such colloids,
the solid particles have a surface charge, which tends to stop them coming
together. Lyophobic sols are inherently unstable and, in time, the particles
aggregate and form a precipitate. Lyophilic sols, on the other hand, are
more like true solutions in which the solute molecules are large and have an
affinity for the solvent. Starch in water is an example of such a system.
Association colloids are systems in which the dispersed phase consists of
clusters of molecules that have Iyophobic and Iyophilic parts. Soap in water
is an association colloid.

Gels are colloids in which both dispersed and continuous phases have a
three-dimensional network throughout the material, so that it forms a jelly-
like mass. Gelatine is a common example. One component may
sometimes be removed (e.g. by heating) to leave a rigid gel (e.g. silica gel).

25
 Other types of colloid include aerosols (dispersions of liquid or solid

particles in a gas, as in a mist or smoke) and foams (dispersions of gases in

liquids or solids).

Emulsions are colloidal systems in which the dispersed and continuous

phases are both liquids, e.g. oil-in-water or water-in-oil. Such systems

require an emulsifying agent to stabilize the dispersed particles.

Emulsions, formed by the process of emulsification (48,49,50,51,52,53)

are
heterogeneous systems consisting of at least one immiscible liquid intimately
dispersed in another in the form of droplets. The preparation of an emulsion

requires the formation of a large interfacial area between two immiscible liquids.
They are described as thermodynamically unstable mixtures that are stabilised by
the presence of a third phase, an emulsifying agent. The diameter of the
dispersed phase droplets is generally in the range of about 0.1 to 1OPm, though
droplets as small as 0.01[trn and as large as 100prn are often observed. Simple

geometric calculations indicate that if a sample of 10ml of oil is emulsified in water

to give an average droplet diameter of 0.2pm, the resulting interfacial area will
have been increased by a factor of approximately 106.

Milk is the most common food emulsion with cow's milk comprising about 87%

water, 3.6% lipids 3.3% protein, 4.7% lactose, and, in much smaller amounts,
vitamins and minerals. The stability of raw milk(54)is accomplished by a complex
interfacial lipid-protein membrane, formed during the secretion of milk. To avoid

separation of the fat phase during storage, milk is homogenised. This process
reduces the fat globule size to below 1 micron in diameter, and thus keeps the fat
in suspension by Brownian motion. During the homogenisation process, the
surface area of the fat phase is greatly increased due to the formation of many
new fat particles. These new fat particles are covered with absorbed milk proteins,
stabilising them against coalescence. If the particle size of the fat droplets is
lowered too much (below 0.5gm), the surface load of absorbed proteins may
cause the oil droplets to precipitate due to the increased density.

26
 Some common food emulsions are shown in Table 1, they are distinguished by
which is liquid is the disperse phase.

Table 1 Emulsion types

Oill-in-water Particlesize Water-lin-olill Particlesize

Milk >1ýtm Margarine 5-20ýtm
Mayonnaise 11-5ýtm Butter 1ýtm

There is a large variation in droplet size between differing foods and this has
significant effects on their properties and behaviour. Other common emulsions
include oil products such as bitumen and a whole range of medical and cosmetic
products. Figure 8 shows two electron micrographs of butter and a low fat spread,
these show the variation in droplet size between them. (55)

Figure 8 Electron micrographs of butter and a low fat spread. (Both at the same
magnification).

Ilk"
.:

w. ;*;
-
".

ý0i 4k
91
it

Butter Low fat spread

Confocal laser scanning microscopy Confocal laser scanning microscopy
image of butter. Dark water droplets in image of a low fat spread. Dark water
the continuous fat phase. A water in oil droplets in the continuous fat phase,
emulsion. Image by Grethe Hyldig inside the globular water droplets. An
oil in water in oil emulsion
Image by Grethe Hy1dig.

27
1.6.2 Emulsifiers

In order for an emulsion to successfully disperse droplets in a continuous phase, it

is often necessary to use an emulsifier. The addition of an emulsifier allows a
thermodynamically unstable system to achieve a degree of stability, by separating
the two phases. The emulsifying agent minimises the tendency of the globules to
coalesce, or join to form larger globules, which would eventually lead to the
separation of the two liquids. The stability of an emulsion is dependent upon the
properties of the emulsifier and the film it forms at the interface between the two
phases. The film at the interface must be tough and elastic and should be rapidly
formed during the preparation.

Emulsifying agents can be divided into three categories:

1. Surface-active agents work by congregating at the oil-water interface

reducing the interfacial tension.
2. Hydrophilic colloids work by forming multi-molecular films about the
dispersed particles.
3. Finely divide solids work by being absorbed at the interface between the
two liquid phases and form a film of particles around the dispersed
globules.

The surface-active emulsifying agents represent the principle type used in

industry(56) on a weight basis at least. Common emulsifiers are substances, such
,
as detergents, that have both hydrophobic and hydrophilic components. An
illustration of an emulsifier separating the two phases is shown in Figure 9. The
emulsifier that surrounds the droplet will often be more than one molecule thick,
frequently several layers separates the two systems.

28
 Figure 9 Action of emulsifying agents in a water-in-oil emulsion.

Surrounding oil phase separated

from water by the emulsifier Water

Al a
loý

1*
or

Hydrophilic end of
Hydrophobic component of emulsifier
emulsifier

The choice of emulsifier depends on the type of emulsion required, oil-in-water or

water-in oil, and on the solubility of the emulsifier in each phase. As a rough rule,
the phase in which the surfactant is more soluble forms the continuous phase.
The exact mechanism of emulsion formulation and emulsifier choice is as much an

art as science, with trial and error along with previous experience often deciding
the choice of emulsifier.

The most common emulsifier for emulsion explosives, and one of the emulsifiers

used in this study, is sorbitan monooleate. Sorbitan monooleate has low water

solubility and acts as a good solubiliser of water-in-oils. The structure of this is

shown in Figure 10

29
Figure 10 Sorbitan monooleate molecule.

OH OH
\\
11
C-C-O-C CH2)7 (CH2)
-CH3
017
-
OH
cýC

HH

Emulsification of oils and fats in an aqueous solution of proteins, carbohydrates

and salts is normally achieved by using high-pressure homogenisation. Although
emulsifiers reduce the interfacial tension between oil and water, this only has a
minor effect on the final particle size. The particle size of the emulsion is mainly
due to the energy density (57)
applied . This means that the more energy supplied to
the emulsion on formation, the smaller the droplet size will usually be. In a water-
in-oil emulsion, the emulsifiers used will help provide a finer particle size. This is
partially due to the reduction in interfacial tension between the oil and water and
due to a reduction in the rate of re-coalescence of water droplets before the fat
phase crystallises.

1.6.3 Break down of emulsions

Emulsions are thermodynamically unstable, and their relative stability is affected

by factors such as flocculation and aggregation. Lowering the interfacial tension,
with emulsifiers, allows the emulsion to form, if the interfacial tension then
increases, the emulsion will begin to breakdown. An emulsion can breakdown in
one of four different modes of failure, as shown in Figure 1

30
Figure 11 Processes of emulsion breakdown.

A C

069
040

B D

A) Coalescence B) Breaking C) Flocculation D) Creaming

A) Coalescence
This is the joining of two (or more) drops to form a single drop of greater volume,
but smaller interfacial area. Such a process is energetically favourable in all cases
in which there exists a positive (even if small) interfacial tension. Although
coalescence results in significant microscopic changes in the condition of the
dispersed phase (size and distribution) it may not immediately result in
macroscopic change.

B) Breaking
This refers to the process of gross separation of the two phases. The process is
the macroscopically apparent consequence of the microscopic process of
coalescence.

31
C) Flocculation
This is the process where individual droplets in the emulsion congregate to form
flocs or loose assemblies of particles in which the identity of each droplet is
maintained. This condition is clearly different from coalescence, as the droplets

remain separate. Flocculation, in many cases, can be reversed by the input of

energy, such as stirring. Much less energy is required for this than was originally
required to form the emulsion.

D) Creaming
This is relatedto flocculationin that it occurswithoutthe loss of individualdroplet
identity. This processwill occurover time in almostall emulsionsin whichthere is
a difference in the densities of the two phases. The rate of creaming will be
dependenton the physicalcharacteristicsof the system,especiallythe viscosityof
the continuousphase. It can easily be reversed. If the dispersephaseis denser
than the continuousphase,then the processis calledsedimentation.

1.7 Emulsion Rheology

1.7.1 Viscosity

viscosity is a measurement of the flow properties of a product. It is the ratio of

shear force applied and the amount of resulting deformation. The deformation of
the fluid is expressed as the rate of shear. Therefore, viscosity is the relationship
between shearing stress and the rate of shear. The concept of viscosity came
about with Newton's postulate (1690's) that; Shear stress a is related to the
velocity gradient or shear rate ý through the equation;

Equation 11 Shear stress.

Where il is the shear viscosity.

32
In, the simplest cases, like water or aqueous solutions, the shearing stress is
directly proportional to the rate of shear. The proportionality constant is called the

viscosity coefficient. Fluids where the proportion is direct are called Newtonian
fluids. For Newtonian fluids il is independent of ý and therefore does not change
with respect to ý. For Newtonian fluidsTj is usually called the viscosity.

Table 2 Example of viscosity values.

Liquid Approximate Viscosity

(P)
o4U
Glass 1

Bitumen loll

Liquid Honey 10'

10-2
Bicycle Oil
Water i 0-ý,
Air 10-5

There are many types of non-Newtonian fluids, each having distinct properties.
Measurement of these viscosities is more involved, requiring the additional
function of time. The viscosity behaviour, when the stress is applied over time,
determines the type of a non-Newtonian fluid. With emulsions it is usual for the
apparent viscosity to lower with increasing shear.

33
1.7.2 Particle size analysis

An important parameter for emulsions is the droplet size, with this possibly having
the greatest effect on the rheological and physical properties of the emulsion.
There are a number of ways to determine particle size, although particle sizing of
emulsions is not an exact science, errors compound every process but these can
often be minimised by careful selection of methods and the use of good practice to
minimise error.

The most common method, of particle size analysis involves the use of

microscopy. Samples of emulsion are placed on a microscope slide and a digital

picture taken. This is converted by computer manipulation to a black and white

image, which allows the size of each droplet to be interpolated. A histogram report
of droplet size and number of drops is then obtained from this. The larger the
sample group size, the greater the accuracy of this method.

An alternative technique for particle size characterisation is hydrodynamic

chromatography. Hydrodynamic chromatography has emerged, in the last 20

years, as a method of choice for particle size analysis. Samples are isolated in a
solvent and by passing them through a column packed with tiny, porous beads

separation occurs. The carrier liquid used is inert to the particles and the beads
are packed in long columns, usually with no immobilised coating. Separation
depends on the hydrodynamic flow of the fluid flowing in the Poiseuille regime.
This means that large particles move quickly with the inert flow and smaller
particles move at a slower speed. The turbulence generated around the beads
causes the smaller particles to take a longer path and therefore pass more slowly
through the column. Mixed particle sizes introduced at the head of the column are
separated according to size. The large particles emerge first, followed in order by
smaller and smaller particles.

34
1.8 Background to Emulsions Explosives

Emulsion explosives consist of three basic ingredients, an aqueous phase, an oil

phase and a surfactant. The aqueous phase, which makes up 85% or more of the
formulation, usually, consists of a supersaturated aqueous solution of ammonium
nitrate (AN). Other oxidisers such as sodium and calcium nitrate are sometimes
added as an alternative to AN. The salts are dissolved in hot water (-900C), as
the solubility of these salts greatly increases with temperature. These solutions
often have a crystallisation point around 70'C and have to be mixed above this
temperature. This solution is added to a heated oil phase and mixed. Once
properly formulated, the emulsion can be stable down to temperatures as low as
minus 20'C, although at low temperatures crystallisation is encouraged, so
prolonged storage at low temperature leads to emulsion failure.

The oil phase plays a critical role in the stability of the emulsion system, as well as
providing the fuel for the ammonium nitrate oxidiser. Oils such as refined paraffin
or microcrystalline waxes are often used. A variety of chemicals has been
investigated, but paraffin is by far the most commonly used oil.

Surfactants help to form and stabilise, the high internal phase water-in-oil
emulsion, at processing temperatures. Once the emulsion has formed, it is
allowed to cool to ambient temperature. This means that the water solution
becomes supersaturated,so the emulsifier has to be capable of limiting nucleation
of the nitrate salts. The surfactants surround each of the water droplets with the
hydrophilic layer limiting the nucleation of the nitrate salt. A basic emulsion
explosive composition, as used in commercial applications, is shown in Table 3.
This contains about 13 times as much ammonium nitrate by mass, or about 10
times as much AN solution by volume, as oil.

35
Table 3 Emulsion formulation.

Component % by weight
Water 15
Oil 6
Nitrate salts 78
Surfactant 1

The reaction between ammonium nitrate and fuel oil is relatively simple, with
water, carbon dioxide and nitrogen gas formed as reaction products.

Equation 12 Stoichiometryof the reaction between AN and oil.

N03 Cn C02+
3NH 4 +IH': 2n
* 7H 20+ 3N 2
n

The reaction is an idealized one involving ammonium nitrate and oil, which
assumes the complete reaction of the ammonium nitrate with the hydrocarbon.

1.8.1 Development and use of Emulsion explosives

Emulsions explosives were first formulated in the early sixties. Egly disclosed, in a
US patent in 1964 (58),
a water-in-oil emulsion that was explosive. This was based
on a concentrated solution of a metal nitrate in paraffin oil. Atlas powder applied
for a whole series of patents covering the use of nitric acid in an oil phase (59)
Their emulsion was based upon a solution of 30 to 80% aqueous nitric acid mixed
with ammonium nitrate and a carbonaceous fuel, such as diesel oil. This formed a
dispersion that was stable for at least eight months. These early emulsions gave
VOID figures, using the Dautriche method, of 1700ms-I to 4500ms-1. The patent
stated that these figures were obtained from firings with a 3" (76mm) diameter and
10" (254mm) long cylinder. The emulsion was primed with %lb (340g) of
B. Using the Dautriche(60)method of VOID determination, with the
composition

36
size of priming used, there is some ambiguity over the velocity of detonation data
obtained.

In the results reported by Egly, the first blasting cap (for the Dautriche VOID
measurement) was well within three charge diameters of the initiation point. The
detonation wave from the composition B primer would not have had time to fade,
thereby producing erroneously high VOID values. This could be a reason for the
lack of further work on nitric acid emulsions, notwithstanding the extremely
corrosive nature of the emulsion, which made storage and use of the explosive
difficult. There was also the toxicity and safety aspect of using bulk quantities of
an explosive containing up to 80% nitric acid.

Bluhm's patent(61)led to the development of the first commercially viable emulsion

explosive. This was based upon a water-in-oil emulsion, made with a mixture of
hot nitrate solution and hot oil. Micro-balloons (gas bubbles) were used to
sensitise the emulsion and to lower the density. Other studies have since
produced refinementsto this initial system. The problem with the initial formulation
was that it was not detonator sensitive and required extensive priming.

The next major advance was in 1977 when Wade(62)

of Atlas powder patented an
emulsion explosive system using a closed void material as a sensitising material.
The preferred material was glass micro-balloons, which gave the emulsion greater
sensitivity to initiation. Using the glass micro-balloons, the emulsion could be
initiated with a standard detonator. The closed void material sensitises the
emulsion by providing a source of hot spots. When the emulsion is shocked, the
glass micro-balloons collapse and heat the immediate area around them via
adiabatic compression.

Emulsion explosives are water-in-oil emulsions with the water forming the disperse
phase. Emulsions offered an intrinsic advantage over more basic ammonium
nitrate fuel oil (ANFO) explosives in that they are water resistant. The emulsion
does not break down in water because of the impervious oil layer surrounding the
disperse phase. ANFO is composed of ammonium nitrate prills and fuel oil and is
extremely sensitive to water. A thin oil layer surrounds the prills but does not

37
completely cover them. This layer is easily washed away by water allowing the
ammonium nitrate to be dissolved. This desensitises the explosive leading to
detonation failure or possibly, more significantly, partial detonation. Emulsions

with water resistance do not suffer from this problem. This is probably the single
(63)
most important fact that has lead to the success of emulsion explosives .

Emulsions are widely used in tunnelling where a most important feature is the
explosive's fume characteristics. In addition to the production of gases such as
C02 and H20, some explosives produce nitrogen oxides and carbon monoxide.
These gases are poisonous with low exposure limits and compositions that
produce high concentrations of them are inappropriate for tunnelling. A high
oxygen balance leads to the production of nitric oxide and nitrogen dioxide in the
explosion fumes, which are highly toxic. When the explosive is not oxygen
balanced, it develops a lower heat of explosion. This is due to an incomplete
reaction occurring and products being formed that result in a lower energy release,
such as CO rather thanC02-

Equation 13 Energy of CO and C02 formation.

C+ 02 C02 Energy of Formation -393.5 U/mol

->
C+ Y202

-+ CO Energy of Formation W/mol

-110.5

For composite explosives, the intimacy of mixing of the ingredients influences the
overall fume characteristic of the explosive. Oxygen balanced explosives with
coarse ingredients produce higher quantities of noxious gases than products that
are more intimately mixed. This is due to kinetics and homogeneity of the
explosive. As the reacting material rapidly expands not all the reactions go to
completion and explosives that are not resistant to water will, often, form large
quantities of noxious gases when fired wet. Re-entry times to the tunnel face are
critical to the overall rate of advance of a tunnel and emulsion products have
shown up to a tenfold reduction in the quantity of nitrogen dioxide and a two fold
in
reduction the quantity of carbon monoxide produced when they have replaced
(64)
nitroglycerine based explosives .

38
1.9 Ammonium Nitrate (AN)

In 1659 Glauber first synthesised AN by reacting nitric acid with ammonium

carbonate. In the eighteenth century, it was considered to have medicinal value
and by 1867, a patent was issued in Sweden to cover its use as an explosive
additive. FollowingWorld War 1, when large stocks of AN were abundant, AN was
first used as a fertiliser(65) The main difference between fertiliser and explosive
.
grade AN is that explosive grade AN has smaller sized prills and have larger pore
spacing. This is because the major use of explosive grade AN is for ANFO
explosives and the smaller prill gives a larger surface area, whilst the larger pore
spacing allows more oil to enter the prill. The increased porosity of explosive
grade AN is obtained by adding extra water to the AN liquor before the prilling
process. The prill is dried out and as the water escapes away, void spaces are
created. For this study, fertiliser grade prills were used, but since it was dissolved
in water, there is no discernible difference between the two grades.

AN is used in massive quantities, with world usage of AN in 1985 being about 44

million metric tons(66) About 75% of this is used for fertiliser production, with the
.
majority of the remainder being used for explosives.

AN (67)is stable in different allotropic forms below its melting point of 442K. These
phases are shown in Table 4, as reported by Ahtee (68)
and Fecloroff(69) The data
.
shown is that obtained from moist AN, with a residual amount of water present in
the AN before analysis.

Table 4 Ammonium nitrate phases.

Phase Symmetry Density ( kg/dm-ý') Temperature Range (*C)

v Tetragonal 1.710 at -250C -170.0 to -16.0
IV Orthorhombic 1.725 at 25'C -16.0 to 32.1
III Orthorhombic 1.661 at 400C 32.1 to 84.2
Tetragonal 1.666 at 930C 84.2 to 125.2
Cubic 1.594 at 1300C 125.2 to 169.6
Melt 169.6

39
Dry AN has a different thermal profile to that of moist AN and this is shown in the
(70)
Table 5, which gives the temperatures at which each phase change occurs

Table 5 Phase change temperatures for dry AN compared to moist AN.

Phase Moist transition temperature Dry transition temperature

V-1v -18 -18
IV-111 32 -
111-11 84 -
IV-11 - 50
11-1 125 124

The important phase changes, in the commercial use of AN, is the phase IV to III
transition. This occurs at approximately 320C in moist AN, whilst in dry AN the
320C phase change does not occur. Instead there is a transition straight to phase
11,which occurs at about 500C, with the exact temperature dependent upon the

water content. The phase change at 32"C is important, as this increases the

volume of the AN by about 3.6%(71),which can have major implications in the use

of AN.

40
1.10 Objective of the research project

The objectives of this research project were to characterise the behaviour and

physical characteristics of emulsion explosives. Emulsion explosives, as a

system, are used by the quarrying and mining industry, but as such, little core
research, beyond that required for commercial use, has been carried out into their
characteristics. This project set out to determine how the behaviour of the

explosive varies with formulation. The parameters for this were closely controlled,
with the water content and microballoon content of the emulsion being changed,
whilst keeping a constant ratio between the oil and AN. This ratio was kept

constant throughout all of the experimentation undertaken in this project. Other

studies (84,85)
have looked at varying the ratio between the oil phase and the AN,

usually replacing a proportion of AN with sodium nitrate or aluminium.

The overall objective was to determine if increased explosive performance could

be achieved by varying the water and microballoon content. If this varied the
performance, then how this affected the physical properties of the emulsion, such
as the thermal and viscosity properties, was to be investigated. If a positive link
could be established between the physical properties and the explosive properties
of the emulsion, then the mechanism for the explosive reaction could be further
understood. The objective was, the development of emulsion explosives on a
scientific basis, rather than on an empirical basis, as much of the data is at
present.

41
 Experimental

2.1 Material Sources

Ammonium nitrate was manufactured by ICI and was fertiliser grade 'Nitram',
which had a 34.5% Nitrogen content and was supplied in 50kg bags. This was
prilled AN, which had 0.1% of an anticaking agent, which limited water absorption.

Sorbitan mono-oleatewas supplied by Sigma Aldrich (catalogue number 38,891-2)

and came in 1L glass bottles.

and came in a 25kg sealed

Glass micro-balloons were supplied by 3M (13151250)
box.

Oils as used to manufacture the emulsion, polyisobutylene succinic anhydride

ethanolamine adduct in xylene and mineral oil, Eldex 13, were supplied by ICI
Nobel Explosives.

All other materials, such as plastic tubing, wires and metal plates were locally
sourced.

2.2 Emulsion Preparation

Aspects of emulsion explosives have been characterised )

(72,73,74,75,76.
with limited
features having been investigated. However, only limited research has been

undertaken on the effect that varying the water content has on explosive
performance. Research has been undertaken to investigate the effect of adding
different microballoon types and ratios to the (77)
emulsion , but only at a single fixed
water content. By altering the water content the sensitivity, and the performance,
of the emulsion explosive can be modified. The aim of this work was to

characterise emulsion explosives systems and these were prepared in an adapted

commercial food mixer, as shown in Figure 12.

42
The study initially investigated a single type of emulsion explosive. The

microballoon and water content were varied, whilst keeping a constant fuel to
oxidiser ratio. The water and microballoon content were altered, as these were
considered the two most pertinent factors affecting performance. The fuel/oxidiser
ratio was not altered, as this ratio (an approximate fuel/oxidiser ratio of 6.7/100)
gives an oxygen-balanced explosive.

Figure 12 Mixing apparatus.

Food Mixer 0

Sealedheating
jacket

The compositions of the emulsions, as used in this study, are shown in Table 6.

Table 6 Composition of emulsion matrix.

Water 12% 15% 20% 25% 30% 35%

H20 H20 H20 H20 H20 H20

Ammonium nitrate 82.4% 79.6% 74.9% 70.3% 65.6% 60.9%

Oil 4.0% 3.8% 3.6% 3.4% 3.2% 2.9%
Polyisobutylene 1.3% 1.2% 1.1% 1.1% 1.0% 0.9%

succinic anhydride
ethanolamine
(PIBSA)
Sorbitan mono-oleate 0.30% 0.30% 0.29% 0.27% 0.25% 0.23%

43
PolylsoButylene Succinic Anhydride ethanolamine adduct in xylene (PIBSA). A
mainly amide derivative of succinic anhydride, which is a dark brown viscous liquid
with a density of 0.91-0.93 kg/dM-3. PlBSA is a surfactant/emulsifier and was used
to stabilise the emulsion system. This was a proprietary mixture supplied by ICI
Explosives.

Mineral oil. Eldex 13, a light paraffin oil, which was the foundation fuel for the
explosive.

Sorbitan mono-oleate. A viscous amber liquid, insoluble in water with a density of 1

kg/d M-3 and a Viscosity 100 MP (25'C). This was a surfactant/emulsifier, which

encouragedemulsion formation.

Glass micro-balloons. 3M B15/250 silica glass balloons with an average diameter

of 15[tm and a collapse pressure of 250MPa. These provide a source of hot spots
in order to lower the initiation sensitivity of the explosive.

Example preparation of aI kg batch of 12% water emulsion

Ammonium nitrate (824g) was mixed with distilled water (120g) and heated to
900C to dissolve the ammonium nitrate. The oil phase (40g of mineral oil, 13g of
PIBSA and 3g of sorbitan mono-oleate) was placed in an adapted Kenwood Chef
Major food mixer, which was heated by water jacket to 900C. The aqueous

ammonium nitrate solution was slowly added to the hot oil phase, with slow
stirring, until an emulsion just started to form. Once the emulsion had formed, the

remaining aqueous ammonium nitrate was then quickly added and the speed of
the mixer increased to maximum. This was maintained for five minutes to ensure
a homogeneous mixture. The hot emulsion was then transferred to a plastic bag
(high temperature cooking bag), allowed to cool naturally to ambient and stored

until required.

Micro-balloons (3M B15/250) were added to the emulsion at room temperature

within 24h of using the explosive. They were combined with the emulsion by

slowly adding the required weight fraction and mixing at a programmed mixing

44
speed for five minutes. The emulsion was then split into five 200g portions, with
one portion being retained for later analysis.

2.3 Physical Characterisation

2.3.1 Emulsion densities

The densities of the emulsions were determined by weight measurements. Initially

small containers were filled with water and weighed to determine their volume.
Emulsions samples were then placed in the container and weighed to determine
their densities. This was repeated, at least ten times, for each emulsion to
eliminate the chance that any trapped air or an inhomogeneity would affect the
result. This was repeated in three different sized containers. The containers used
were a small glass container -75pl, a standard aluminium DSC container -50pl
and a large DSC container -1 00pl

2.3.2 Particle size analysis

Emulsion droplet sizes were determined using a Kontron IBAS 2000 image
analyser, a dedicated particle size analyser. The emulsion sample was placed on
a microscope slide covered and placed under a light microscope. The microscope
was then manually focused and the resulting image was converted into a digital
image. A computer then measured the number and size of each of the droplets.
The instrument was capable of measuring droplet sizes from 1pm to 40pm and a
printout was produced from this giving a histogram analysis of the sample, which
showed droplet size as a percentage of the total number of droplets.

45
2.4 Thermal Analysis

Thermal analysis includes a group of techniques in which the specific physical

properties of a material are measured as a function of temperature. In differential
thermal analysis, IDTA,a sample is heated, usually in an inert atmosphere, and a
plot of temperature of sample versus reference material is made. In differential
scanning calorimetry, IDSC,heat is electrically added to or removed from a sample
as the temperature is increased. This allows enthalpy changes, due to phase
changes and thermal decomposition, to be studied. The thermal properties of a
substance are important in determining its behavioural characteristics. IDSCwas
used in order to determine the level of crystallinity in emulsion samples. The re-
dissolution of ammonium nitrate crystals into the emulsion matrix should produce
an endotherm. Thermal conductivity has been determined as an important
parameter(78)for explosive materials as evidenced by the Frank-Kamenetskii
equation, where the rate of heat loss is dependent on thermal conductivity.
Thermal conductivity is not easily measured using DSC or IDTA,so a method for
examining relative thermal conductivities was developed.

2.4.1 Differential scanning calorimetry (DSC)

DSC and DTA measure the rate and degree of heat change as a function of time
(79)
and temperature In addition to direct energy measurements, the precise
.
temperature of the sample material, at any point during the experiment, is also
monitored. As DSC can measure both the temperature and heat of transition or
reaction, it has replaced DTA as the main thermal analysis technique, except in
certain high-temperatureapplications.

The DSC instrument, a Mettler TA4000 control unit fitted to a Mettler DSC 30 head
with a CDL50 liquid nitrogen attachment, was capable of cooling samples to -
1700C and heating to 6000C. Sample size was usually 5 to 10mg but the head
could also accommodate larger containers, which can take up to 100mg of

46
 sample. Samples were contained in sealed aluminium cups. After the cups have
been sealed, they are gas tight. Initially the containers were pierced in order to

stop a build up of gas pressure. However, since this could cause an event that

would damage the sensor, it eventually became apparent that a sealed system
could be used. The advantage of using a sealed system is that any gas created
by the heating of the sample could be later analysed. The essential components
of this are shown in Figure 13. Heat is supplied to the sample and the reference,
allowing the enthalpy changes due to phase changes to be studied.

Figure 13 DSC Schematic.

Sample + reference pans

Gas
Thermo
escape
couples tube

I Processor I
I Heater I
IIII nffge gas
connector
Fumace

Nitrogen I
heater

A DSC trace highlights phase changes, which take place within a sample, these
increase or decrease the temperature of the sample in comparison to a reference
and this difference is measured. It is a quantitative technique with the area
enclosed by the curve of signal against time being directly proportional to the

energy change.

47
In power compensated DSC(80), the sample and a reference material are

maintained at the same temperature (Ts= Tr, -T, = 0) throughout the controlled
temperature program. Any energy differences, in the independent energy supplies
to the sample and reference, are then recorded against the program temperatures

2.4.2 Thermal Conductivity

Thermal conductivity is a measure of the ability of a substance to conduct heat.

This was measured to determine the rate of heat transfer through the emulsion
with respect to time. This allows a determination of the maximum temperature
attained when a microballoon collapses adiabatically in the emulsion. Thermal
conductivity is considered important because the higher the rate of heat
conduction, the quicker the heat will spread from the hotspot

For a block of material of cross section A, the energy transferred per unit time
(Q/t), between faces a distance, L, apart is given by;

Equation 14 Lees equation for thermal conductivity

Q/t = kA(T2- Tl)/L

Where
k is the conductivity,
T2, and T, are the temPeraturesof the faces.

This equation assumes that the opposite faces are parallel and that there is no
heat loss through the sides of the block. The Sl unit for thermal conductivity is
Wm-1K-1
-

Thermal conductivitywas measured using a self-designed Lees apparatuS(81).The

sample was sandwiched between two aluminium blocks as shown Figure 14. This
widely used apparatus was designed by Lees for measuring the thermal

48
conductivity of bad conductors that can be made in the form of a slab. To make
the heat flow measurable, the area A is made large and the temperature gradient
is made high.

The aluminium block was heated from above by a metal beaker of boiling water.
This provided a constant heat conduction source that limited the affect of
convection and radiation, allowing them to be neglected. Samples were placed in
the centre of the blocks with an insulator surrounding it; this was used to stop the
emulsion escaping. The temperature at the top and bottom of the two aluminium
tubes, along with room temperature, was continuously recorded until a steady
state was achieved.

Figure 14 Lee's apparatus for thermal conductivity measurement.

A
AlumMIUM block

T1

Th
/ Insulator

T2

ion
, sample

If, for example, T2= 67'C the heat per second (Q/t) flowing through the sample in
the steady state is equal to the heat per second lost by B due to radiation and
convection at 67'C.

49
A second experiment was necessary to find the rate of loss of heat by B, see
Figure 15. The top block and heater were removed and with the sample still in
place, B was heated to a few degrees above its steady state temperature. The
heating was removed and the sample allowed to cool, and readings of its
temperature fall with respect to time were taken. From the temperature time

graph, the tangent at its steady temperature was taken to give the slope of the
line. This allowed the thermal conductivity of the sample to be determined using
Equation 15.

Figure 15 Cooling Experiment

nulsion

ilator
T
T2

Equation 15

Q/t = heat lost by cooling at 670C =mc a/b

Where m is the mass of the bottom aluminium block and c is the heat capacity of
the sample, which was obtained from the DSC. Recalling Equation 14 from the
first experiment:

Equation 14
Q/t'-: kA M- Tl)/L

Where A is the area and L is the thickness of the sample. This allows k, the
thermal conductivity to be calculated.

50
This method gave the thermal conductivity values for emulsions on a comparative
basis. The method was validated using a thermal paste of known conductivity.
The value obtained from this paste was used as a reference value. It is known
that the method is not rigorous, but it did give comparative thermal conductivity
data for the emulsions. This allowed the different emulsions to be ranked

according to their relative thermal conductivity.

There are several methods, which could be used to obtain more rigorous thermal
conductivity values but these all require high value instrumentation. The reasons
for not applying them in the present study were both fiscal and time limitations. It
was also felt that rigorous results for thermal conductivity were not required to a
high degree of accuracy.

51
2.4.3 Viscosity

Viscosity is an important physical parameter for characterising emulsion

behaviour. It was measured using a Viscometers UK Ltd Viscolog MRV-8
instrument. The Viscolog operates on the principle of a rotating cylinder or disc
immersed in the material being tested, and measures the torque necessary to
overcome the viscous resistance to rotation. The rotating disc (spindle) is coupled
via a spring to a drive shaft, which is turned at a known speed. The angle through
which the spindle is deflected is measured electronically, giving a measurement of
torque. Calculations performed within the Viscolog from the torque measurement,
the spindle speed and the spindle characteristics give a direct readout of apparent

viscosity.

Various spindle sizes are available and the Viscolog has a range of spindle
speeds, giving it a wide range of viscosity measurement capability. For a fluid of
given viscosity, the drag will be greater as the rotational speed of the spindle, or its
size, is increased. The Viscolog was set up to take account of speed setting and
spindle size to give readings in P. Measurements can be made using the same
spindle at different speeds, to determine the rheological properties of a material at
different shear rates. Figure 16 Schematic of Viscolog MRV-8 shows the Viscolog.
Samples, for analysis, were placed in 500ml plastic, wide-mouth bottles with a

minimum of 300g of emulsion placed in each bottle. The spindle was inserted up
to the immersion point, and the speed varied from 0.5 to 1OOrpm.

Table 7 shows the full-scale readings for the spindle, an R7, used. The maximum
sensitivity of the instrument is 0.1% of full-scale deflection, which means that at 1
rpm the maximum sensitivity was 4P.

Table 7 Spindle speed to maximum viscosity reading for an R7 spindle

Spindle 100 50 20 10 5 2.5 1.0 0.5

speed/rpm
Viscosity/P 40 80 1200 400 800 1,600 4,000 8,000

52
Figure 16 Schematic of Viscolog MRV-8.

I INPUT/OUTPU
2 IMMERSION P
3 SPINDLE
4 SPINDLE NUT
5 SHAFT
6 SPEED CONTRi
7 MEASURING HI
8 BUBBLE LEVEI
9 SUPPORT
10 BEAD CLAMP 10
11 POWER SWITCE
12 FUNCTION SWI
13 SPINDLE SWIl
14 SC CONTROL
15 LIQUID CRYSI
16 LEVELLING FE

11

L4 11

53
Figure 16 Schematic of Viscolog MRV-8.

I INPUT/OUTPU
2 IMMERSION P
3 SPINDLE
4 SPINDLE NUT
5 SHAFT
6 SPEED CONTRi
7 MEASURING H!,
8 BUBBLE LEVEI
9 SVPPORT
10 HEAD CLAMP 10
11 POWER SWITC1
12 FUNCTION SWI
13 SPINDLE SWIl
14 SG CONTROL
15 LIQUID CRYSI
16 LEVELLING FE

IzII

53
2.4.4 Emulsion breakdown

Emulsion breakdown was determined by examining the emulsion with a light

microscope. A standard microscope with a camera attachment was used for this
purpose. This was connected to a Polaroid printer, which produced photographic
images from the eyepiece perspective. Emulsions were placed on a glass slide
and examined directly, without the use of a cover slip. No cover slip was used as
it was thought that the pressure of the cover slip on the emulsion, could lead to a
distorted picture of the emulsion structure. Pictures were taken at a magnification
of 1550. Breakdown of the emulsion structure, at this magnification, could be
observed easily through the microscope.

2.4.5 Velocity of detonation measurements

The only used method for VOID determination, in this study, was the use of
ionisation probes. This involved inserting twisted pairs of insulated wire into the
explosive column. The wires were not joined, but when the explosive detonates,
an ionisation cloud is formed in the reaction zone, and before the wire is destroyed
this shorts the circuit, allowing timings to be recorded. Using a minimum of two
probes, at a set distance apart, and a timer, a record of the VOID at various points
in the explosive was created.

A further refinement of the ionisation probes method would be to use resistance

wire to measure the VOID continuously. In this system, a current is passed down a
high resistance wire, which is connected to a fast recording oscilloscope. The
drop in resistance is recorded as the wire is consumed by the detonation wave.
As there was no access to a fast recording oscilloscope, it was not possible to use
this method in this study

Firings were carried out on the Explosive Ordnance Range. Cylindrical

polypropylene pipes (39mm diameter) were cut to the required length and small
holes, for the ionisation probes, were cut into the pipe exactly 100mm apart. Each

54
cylinder was sealed, with bathroom sealant, onto to steel witness plate and filled
with 200g of emulsion. The density of the emulsion varied from 0.8 to 1.2 kgIdM-3.
Due to the variations in density, the length of the pipe varied from 180 to 280mm.
The explosive was initiated by detonator, usually an 8*, and if priming was
required Demex 200 sheet explosive was used. On one set of firings number 8
detonators were used due to a misunderstanding (8 contains 0.69g of PETN base
charge whilst 8* contains 0.85g of PETN). lonisation probes (polyurethane coated
wires) were inserted through the pre-cut holes in the wall of the cylinder to
determine the velocity of detonation of the emulsion (Figure 17)

Figure 17 Measurementof VOD with ionisation probes.

D=39mm

lonisation
probes
loomm

Mild steel
plate

12mm

100mm by I 00mm

lonisation probes were connected by wire to an electronic interface unit, which

operated an electronic timer (Gould Advance Timer Counter TC314). The shock
wave on reaching the wires causes a current to pass starting the timer. When the
shock wave reaches the second probe the timer stops. The velocity of detonation
is determined by the time taken for the shock front to travel between the probes.
Velocity of detonation data was recorded for each emulsion type. Each emulsion

55
type was tested at least 4 times, with many undergoing 8 firings, to allow for
statistical variation and eliminate anomalous results. Results of particular interest,

such as 25% water content emulsion at 3% microballoon content, had up to 32

individual firings.

2.4.6 Plate dent measurement

The plate dent test was used as a technique to determine the brisance of the
(82)
explosive . The charge was placed on a mild steel plate (100mm x 100mm x
12mm) and the dent size, after the event, was measured. The charge and plate
were placed on a steel anvil, -1 50mm by 150mm by 500mm, to provide an infinite

slab below the plate. Although the slab is in no way infinite, it did give a consistent
base for the metal plates. Dent size was measured by volume rather than by
depth as this gave a more realistic figure for the brisance effect. The dents
produced by the explosive did not produce sharply defined edges so a small glass
cylinder, of known volume and larger internal diameter, was used (Figure 18). The

cylinder was placed on the metal plate, with a small amount of grease to form a
seal, and then filled with water. The volume of the water less that contained by the
cylinder determined the dent volume.

Figure 18 Plate dent measurement.

Glassring placed over

dent and greasedto
Mild steel allow measurement
plate%
Plaitedent

56
2.4.7 Double Pipe Test

The double pipe test was used as a method to determine the brisance of the

explosive and the degree of reaction, as the detonation proceeds through the
explosive column (83)
as shown in Figure 19.

Figure 19 Double Pipe Test.

Mild steel anvil Tubcs taped togolier

The emulsion explosive charge was confined in a plastic or steel tube, which was
taped to a hollow mild steel pipe of known dimensions and composition. The

assembly was then placed on a steel anvil and the charge detonated. The test

can be used to determine the sensitivity, the output and a figure for the brisance of
the explosive. Pivotal to the procedure is that it places an emphasis on monitoring
the performance of the emulsion explosive under conditions which match, as
closely as possible, those under which the explosive is used for rock blasting. The
tests allow parameters to be altered, such as pipe size and wall thickness, to
simulate differing conditions in boreholes.

57
3 Results and Discussion

3.1 Emulsion densities

The effect of water and microballoon content on the density and the explosive
properties, of the emulsion, was examined. The error in the density measurement
was minimised by repeating the experiment, until at least eight consistent results
were obtained. The density of the emulsion matrix is controlled, in part, by the
water and microballoon content. Changing either of these variables affects the
density of the emulsion. In Figure 20, the effect of water and microballoon
content on density is shown. It is apparent, as would be expected, that
microballoon content had a more pronounced effect on the density of the emulsion
than water content.

Figure 20 The effect of water and microballoon content on density.

Density versus microballoon content

Water(%) 12% 15% 20% 25% 30% 35%

0
U
C
0
0

0.9 1.1 1.2 1.3 1.4 1.5 1.6

Density (kg/dm-)

58
The average density of the microballoons used in this study was 0.15 kg/d M-3and,
therefore, even a small percentage increase in microballoon content significantly
alters the density of the emulsion. Water, having a much higher density (-1
kg/dM-3) has a less pronounced effect on density. Variation of the microballoon
,
content allows the direct comparison of density for differing water contents. The
microballoons are made of silica glass, with a sulphur dioxide and molecular
oxygen filling (0.2 bar). Microballoons provide hotspots in the emulsion, when
subjected to shock, but they provide no energy contribution to the explosive.
Therefore, as the microballoon content is increased, the total explosive energy of
the emulsion decreases.

Figure 20 shows that emulsion explosives have a narrow density range (0.92-1.55
kg/dM-3) The observed densities compare well with those of similar studies by
.
Lee and Hattori. Lee (84)
et al, prepared similar emulsions and reported a density
range of 0.80-1-31 kg/dM-3. (85)
Hattori et al noted a variation in density of 1.00-
1.30 kg/dM-3 although neither study varied the microballoon content to the same
,
degree as in this study.

Table 8 Detonationdata for high explosives(86)

.

Explosive D/ms" p/ kgIdM-3 P/kbar ca1c.

HMX 9110 1.89 392
RDX 8440 1.70 300
DATB 7520 1.79 253
TATB 7658 1.80 260
TNT 6950 1.57 190
Tetryl 7160 1.50 192
Emulsion (Highest density) 3394 1.42 38
Emulsion (Highest velocity) 5558 1.13 81

D -Velocity of detonation (ms-1) p- Density used for VOIDand pressure data

P- Pressure of detonation (kbar)

59
In comparison to standard high explosives, emulsions have a low density. Table 8
compares the data for highest density emulsion to successfully initiate and the
highest VOID emulsion in this study, to that of standard military explosives. The
density, VOID and detonation pressure of the military explosives are all appreciably
higher than the emulsion.

A more useful measurement, for explosives, than density is the percentage of

theoretical maximum density (TIVID). Percentage TIVIDis used in place of density
in analysis of explosive performance as this allows direct comparisons of different
explosives to be undertaken. TIVID is taken, for solid explosives, as the actual
density divided by the maximum theoretically calculated density, and is quoted as

a percentage. Emulsions as used in this study are thick liquids and as such are at
their maximum density before the microballoons are added. Figure 21 shows a

graph of microballoon content versus %TMD.

Figure 21 % TMD versus % microballoon content

Microballoon content versus %TMD

Water 12% 15% 20% 25% 30% 35%

7 -

6-

iS 5

NA- -
4- -

3 -

0
65 70 75 80 85 90 95 100
% of TMD

60
3.2 Emulsion breakdown

Emulsions degrade slowly over time and one of the most effective ways to monitor
this is using microscopy. Emulsions were examined under a light microscope for
early signs of failure, which eventually, would lead to more severe macroscopic
failure. Pictures of the emulsion samples were taken over time, and comparisons
taken between new and aged samples.

Figure 22 shows a recently manufactured emulsion, less than a week old, with a
typical particle size and distribution. The two phases can be seen with the
disperse phase (water droplets), containing dissolved ammonium nitrate,
surrounded by the continuous oil phase.

Figure 22 Picture of an emulsion from an optical microscope x1550 (expanded for

report).

Water (disperse) Oil (Continuous) Glass microballoon

phase Phase

61
It can be seen from Figure 22 that there was a significant, and large, variation in
droplet size, but the emulsion was stable and most of the droplets were spherical.
This, Figure 22, was a typical emulsion explosive, which has been well mixed and
is stable. The droplets correspond to a closely packed sphere model, with little

wasted space between the droplets.

After time the emulsion begins to break down and the two phases separate by

either coalescence or flocculation. Figure 23 shows an emulsion sample at the

onset of coalescence. This was a two-year-old emulsion explosive sample with a

25% water and 1% microballoon content. The overall droplet size has started to
increase with more medium to large droplets being observed.

Figure 23 Shows the onset of coalescence occurring in an aged sample (xI 550)
(expanded for report).

'vow

%-4, ,

The gap between droplets has also increased and the disperse phase can be
clearly seen. At this early stage, almost no difference in the sample can be seen

62
on the macroscopic scale. The sample can become slightly more opaque and less
free flowing, but this change was minimal.

Figure 24 shows an aged sample in a more advanced stage of coalescence. It

can be seen that the droplets were far less numerous and were no longer
spherical. There were still a number of small droplets, but droplets can be seen
amalgamating with others, their shape matching that of the joined droplets. At this
stage, macroscopic changes were observed in samples. At the edge of large
samples, there was a small amount of crystallised ammonium nitrate. Samples, at
this stage, were very opaque and noticeably less free flowing, although the sample
could still be easily converted back to a stable emulsion. The action of stirring
caused the droplets to reform as a stable emulsion. This occurs when little or no
crystallisation in the bulk of the sample has occurred. Even if the coalescence
was slightly more advanced, the emulsion could still be recovered by the
combined action of heat and stirring. The colour and flow properties of the
emulsion were both affected at this stage, but macroscopically it still behaved as
an emulsion.

Figure 24 Coalescence now more advanced in sample (xl 550) (expanded for
report).

Ax..
.
I-I'll

jr

Joý

I
,, "Oor

63
Figure 25 show the most advanced stage of coalescence with the effects very
apparent on the macroscopic scale. The sample was completely opaque,

extremely viscous and did not flow, but behaved more like a solid. At this stage,
the effects of coalescence were irreversible and the droplets were very large, with
virtually no small droplets. The shape of the droplets was haphazard, with no real
structure to them. This sample was a 12% water emulsion with 5% microballoons,
aged for 21/2years.

Figure 25 Coalescence, now catastrophic and seen on a macroscopic scale

(xl 550) (expanded for report).
9Vý J'Aw
't 0
It ,, 10 Ic
00
t,
101

"4.

'0

Crystallisation of samples sometimes occurred, but usually only on a minor scale.

This was noted from the texture of the emulsion, which felt 'gritty' to touch. This,

and the opaque colour, showed that a degree of crystallisation had occurred in the
emulsion. Crystallisation was easily reversed, by heating the emulsion above
600C and mixing, forcing the AN back into solution.

64
Initially 10% water content emulsions were manufactured, but these continually

suffered catastrophic failures, due to crystallisation. This occurred spontaneously,

either on cooling of the emulsion, or on the addition of microballoons. A 10%

water emulsion was never manufactured in a stable enough state for explosive
testing.

The lower water content emulsions, 12,15 and 20%, showed a much greater
propensity to spontaneously crystallise. On three separate occasions, 12% water
emulsions began to crystallise on addition of microballoons. This was corrected
by heating and stirring, which restored the emulsion matrix. Above 20% water
content, crystallisation was rarely seen and this never occurred spontaneously on
addition of microballoons.

An alternative breakdown process for emulsions is that of flocculation. Figure 26

shows a 20% water emulsion with no microballoons. This had been kept in a
sealed container for 6 months and before analysis. Figure 26 shows that the
droplets, although having grouped together, have still kept their individual identity.
The size of this large 'droplet' is substantial in comparison to the normal droplet
size for the emulsion. Emulsion explosives are not generally thought to undergo
(87)
flocculation and this was only noted in a few samples. This process was easily
reversed with a minimum of stirring of the sample.

Coalescence and crystallisation were the more common processes of emulsion

failure, and subsequent breakdown. By adding microballoons to the emulsion
within 24h of explosive testing, this alleviated the problem of failure of the
emulsion, by mixing the emulsion and thereby reforming a stable system before
explosive testing.

65
Figure 26 Example of flocculation in emulsion sample (x1550) (expanded for
report).

00ý
% "A' "',

41
19

lw

41

66
3.3 Rheology
Rheological measurements were made on the emulsions, specifically viscosity,
using a Brookfield viscometer. Emulsions exhibit thixotropic behaviour, which is a
time-dependant analogue of shear thinning and plastic behaviour. When
subjected to shear at a constant rate, the apparent viscosity decreases with time
until a balance between the structural breakdown and structure re-formation is
reached. If the system is allowed to stand, it regains its original structure and an
initial high viscosity is noted, before this again decreases. For emulsions, the
smallest spindle size available was used, which allowed the highest viscosities, for
the instrument, to be measured. Limitations of the instrument restricted the
emulsions, which could be examined, and it was not possible to measure the
viscosity of an emulsion with a higher than 1% microballoon content.

The speed of the spindle through the medium has an effect on the viscosity range
that can be studied. By using a slower spindle speed, mediums that are more
viscous can be studied. With the emulsion explosives, varying the spindle speed
has a dramatic effect on the apparent viscosity; this was due to their thixotropic
nature.

Figure 27 shows the effect of spindle speed on viscosity. As the spindle speed

was increased, the apparent viscosity dropped; this effect is called shear thinning.
This is a well-known property of emulsions, which has been utilised by paint
manufacturers, for emulsion paints. The solution viscosity decreases with applied
force, being initially thick and viscous at low force and becoming free flowing as
the force is increased. The drop in apparent viscosity can be seen as the spindle
speed increases. When the apparent viscosity becomes stable, the emulsion

matrix turns with the spindle, at low speed. However, as the spindle speed
increases the apparent viscosity dramatically drops. This was shown for every

emulsion composition, with the apparent viscosity dependent upon the water
content.

67
Figure 27 Variation of viscosity with spindle speed (viscosity/log plot).

The effect of shear on apparent viscosity

Water: 35% 30% 25% 20% 15% 12%
10000

1000

100-

10
0 20 40 60 80 100

Spindle speed (rpm)

Figure 28 shows the data obtained at Irpm. At this speed the most stable

viscometer readings were obtained. With no microballoons in the emulsion matrix,

it can be seen that as the water content dropped from 35%, the viscosity began to
drop. At lower than 25% water content, where the lowest viscosity was recorded,
the viscosity increased with decreasing water content. With the addition of
microballoons, the overall viscosity of the emulsion increased for all water contents
and again, at 25% water content, the lowest viscosity is recorded.

68
Figure 28 Viscosity's for emulsions.

Viscosity of emulsions for 0 and 1% microballoons

0%

3600
3200
2800
2400
4,2000
1600
W
1200
800
400
0
10 15 20 25 30 35
Emulsion water content

The reason for the low viscosity at 25% water content is thought to be a balance
between the water, ammonium nitrate and the disperse phase droplet size.
Increasing the percentage of water, in comparison to ammonium nitrate content,
initially has the effect of lowering the viscosity. This is believed to occur because
the density of the ammonium nitrate solution, the disperse phase, decreases.
With further increase in water content, the continuous phase becomes, in
comparison, sparser, leading to a greater number of disperse phase droplets. It is
thought that this has the effect of increasing the viscosity, as noted when the water
content increases from 25% to 35%. At 25% water, these two interactions, the
droplet size and droplet density, are thought to be at a mutual low point. This
would account for the lowest viscosity readings being recorded at 25% water
content.

69
3.4 Particle size analysis

Water content has an effect on the particle size of the emulsion and this, as
previously described, was investigated. Three different emulsion water contents
were examined, 12,20 and 30%, and the data from this is presented in Figure 29.

Figure 29 Effect of water content on particle size.

The effect of water content on droplet size

X 12% water X 20% water 30% water
60

50

40

30 -

20

10

0 1F r-T/f-M T I/
........

123456789

10 11 12 13 14 15 16
Droplet size (pm)

The chart shows droplet size as a percentage of the total, however, due to
limitations of the instrument the maximum droplet size that could be recorded was
50pm. From Figure 29 it can be seen that 30% water emulsion had the lowest
average droplet size. This was notwithstanding the fact there could have been a
greater proportion of large, but not recorded, droplets in the emulsion. At 20%
water content, the highest average particle size, 4gm, was recorded, although
there was only a limited difference in droplet size between the various emulsions.
The difference in droplet size could be linked to the viscosity of the emulsions, but

70
no clear trend was observed in the data. Overall, there was only a limited
difference in the particle sizes between the differing water contents.

The effect of time of mixing on particle size was investigated using a 20% water
content emulsion. The emulsion was prepared, following the standard method,
and samples taken from this at set times. The samples were stored for later
analysis with the results of this shown in Figure 30.

Figure 30 Effect of mixing time on particle size.

Effect of mixing time on particle size (20% water)

4 nin nýx )v 10 min nix 22 min mix 30

90

80

70

60

50

40

30

20

10

0
1 10
Droplet size (pm)

The lowest average droplet size was recorded in a newly formed emulsion and in
this sample over 80% of comprised 1pm droplets. Notwithstanding this, it should
be stressed that as the instrument could only record a maximum droplet size of
50pm, the true state of the sample cannot be ascertained. By increasing mixing
time the average droplet size of the emulsion increased.

71
Further mixing of the emulsion after formation can cause catastrophic failure of the
emulsion. After the emulsion is formed, and is then subjected to further vigorous
mixing, flocculation and coagulation can occur as failure of the emulsion matrix
begins. A degree of mixing after the initial emulsion formation is necessary but
this was limited to stop these effects. On this basis, emulsions were mixed for two
minutes after formation, before being used in explosive testing.

Figure 31 Effect of mixer type on particle size.

Comparsion of mixer types

50
Homogeniser Food miýxýer
45

40

35

30

25

20

15

10

0 T=t-7,01-1
02468 10 12 14 16 18 20

Particlesize(pm)

As an alternative to the standard mixing apparatus, a homogeniser was tested in

an attempt to lower the droplet size, to gain an increase in explosive performance.
To this end, a 25% water content emulsion was mixed with the standard mixer and
then compared to a 25% water content emulsion mixed with a homogeniser. The
results of this are shown in Figure 31.

72
The homogeniser produced a refined emulsion much more quickly than the

standard mixer. The average droplet size of the emulsion produced by the
homogeniser was lower, the majority of the droplets being at or below 1gm.
However, when the homogeniser was used to mix the emulsion, water was prone
to separate out from the emulsion matrix. The mixing head provided a much more
severe environment than a standard food style mixer, which led to phase
separation. Emulsions formed with the homogeniser were prone to catastrophic
failure, the ammonium nitrate crystallising out of solution in the emulsion. It is
thought that this process occurs because of the small size of the disperse phase
droplets, which provide more sites for recrystallisation to occur. The homogeniser

was not used for the addition of microballoons to the emulsion. This was
undertaken on safety grounds due the severity of the mixing provided by the
homogeniser.

3.5 Thermal Measurements

The thermal properties of emulsions, without microballoons, were examined using

a DSC instrument and a Lee's disk apparatus. A Lee's disk apparatus, which as
previously described, gave comparative thermal conductivity measurements.
Figure 32 shows the data presented for emulsion explosives along with the
thermal conductivities of kerosene and water, which are given as a guideline for
thermal conductivities.

As the water content of the emulsion increased, the thermal conductivity

decreased. This implied that as the ammonium nitrate concentration increased,

with respect to water content, so the thermal conductivity increases. This would
be expected, as an emulsion with a more concentrated AN solution (low water
content) would conduct heat faster than emulsions with less concentrated AN
solutions (higher water contents). The rate of heat dispersal is important for
explosives in the context of initiation, with shock and heat being two of the major
mechanisms for this. This is discussed later, in more detail, in the context of
explosive performance.

73
Figure 32 Thermal conductivity of emulsion explosives.

Thermal conductivity of Emulsion Explosive

Water Kerosene
3-

2.5 -

2-

Se
1.5 -
E

0.5 -

0- 11111
10 15 20 25 30 35
water content (%)

3.6 Differential scanning calorimetry

3.6.1 Heat capacity

The heat capacity, for emulsion explosives, was measured using differential
scanning calorimetry. The heat capacity is the energy required to heat 1kg of
sample by 10C. The energy required to do this varies with the temperature of the
sample, with heat being conducted at different rates depending on the
temperature. The data presented in Figure 33 was measured for the emulsions at
40'C. This temperature was used since temperatures higher than this are unlikely
to be encountered in the normal operational use of emulsion explosives and below
this temperature it could be affected by phase changes.

74
Figure 33 Heat capacity as measured by DSC.

Heat capacity as measured by DSC

2.65

2.6

2.55

2.5

2.45

2.4

2.35

2.3

2.2 5

2.2
10 15 20 25 30 35
Water content (%)

This data shows that as the water content increases, so the heat capacity
increases. The data observed at 15% water content seems to be higher than
would be expected. This analysis was repeated four times for each sample,
although this used the same bulk sample, with respect to water content, for each
emulsion. The high result recorded at 15% water content could be due to a
degree of crystallinity in the sample. Energy would then be utilised redissolving

some of the AN, before an increase in temperature was recorded. Overall, as

water has a higher heat capacity, in comparison to oil, these results show an
expected trend of increasing heat capacity with increasing water content.

75
3.6.2 Thermal analysis data

Ammonium nitrate (AN) is an enantiotropic material, which is a crystalline

substance capable of a reversible transformation, from one allotrope to another, at
a definite fixed temperature. In an emulsion, the AN is dissolved in water droplets,
which are surrounded by the continuous oil phase. Inside the water droplets, a
degree of crystallisation can occur, which would not be readily identifiable on the
macroscopic scale. DSC was used as a technique with which to identify the
degree of sample crystallinity. Initially AN prills were examined by placing a prill in
an open crucible and running the instrument. The results of the first run are shown
(Figure 34) and this was compared to a standard DTA trace for AN (Figure 35)(88).

Figure 34 shows the AN prill underwent three polymorphic transitions, which

occurred at 52'C, 92'C and 134'C, respectively. The literature DTA trace showed
three transitions at 32"C, 84'C and 125C, which correspond to, as shown in Table
and 11-1
5 below, phase IV-111,111-11 transitions respectively.

Table 9 Phase change temperatures for dry AN compared to moist AN.

Phase Moist transition temperature Dry transition temperature

V-1v -18 -18
IV-111 32 -
111-11 84 -
IV-1I - 50
11-1 125 124

The phase transitions, in the AN prill, occurred at higher temperatures than the
reported literature values for powdered AN. The initial phase change, at 52"C in
the AN prill, could be the IV-11phase transition or a late IV-111
transition. Whilst it
would be expected that the centre of the AN prill would be free from moisture,
allowing the core of the prill to undergo a direct IV-11
phase transition, this would
not account for the fact that the moisture rich exterior of the prill did not undergo

76
the IV-111phase transition at 320C. However, as the exterior of the prills were
sprayed, at manufacture, with 0.5% of an anti-caking agent and this could lower
the IV-111phase transition temperature, at the moisture rich exterior of the prill.
The implication of this being that the IV-111phase transition would have already
occurred, at room temperature, before the DSC thermal trace began.

It could also be possible that because of the surface coating, moisture was not in
transition did not occur as there
direct contact with the AN, and therefore the IV-111
was no moist AN. If this were the case, then there would be difficulty explaining
the phase transition at 920C, which is likely to be a late 111-11
phase transition, as
seen in moist AN. In my opinion, it is most likely that the surface coating of anti-
caking agent does in fact lower the initial IV-111 transition temperature and this is
the reason why it is not seen on the thermogram. There was no direct evidence to
prove that this occurred, nevertheless, some of the AN in the prill must have been
orthorhombic state, as evidenced by the 111-11
in the phase 111, phase transition at
920C.

Figure 34 DSC trace of AN prill.

Ammonium Nitrate File: 00002.001 DSC METTLER I I-Feb-97

10.800 Mg Rate: 10.0 *C/min Ident: 2.0 R. M. C. S.
-M

C;

1340c

20.40.60. SO. too. 120.140. C

77
Figure 35 DTA (AT) and heating (T) curves for ammonium nitrate, showing three

polymorphic transitions (at 320C, 84'C and 125'C).

Time, min

Figure 36 Ammonium nitrate phase changes from DSC.

IV-III

Decomposition

40*
910
AT

174*

130*

Melting

0 100 200 30
Temperature in OC

78
In the literature there is some ambiguity over the exact temperatures at which the
phase changes occur in moist AN. Figure 36(89),a IDSC trace, shows the phase
transition temperatures in moist AN. The temperature that the phase changes

occurred in Figure 36, in comparison to the present study and the data from Figure
35, is shown below in Table 10.

Table 10 Comparison of AN phase transition temperatures.

Phase Transition Figure 35 DTA Figure 36 DSC Present Study

Trace Trace DSC Trace
IV-111 32'C 40'C -
IV-1I - - 52 OC
111-11 84 OC 91 11C 92 'C
11-1 125 OC 130'C 134 OC

The phase transitions temperatures in the present study were higher than those
recorded in the literature. One plausible explanation is that because the AN was
in a prill form, there could be instrumental temperature lag due to uneven heating.
This would be expected as the prill is spherical and has, in comparison with
powdered samples, a low contact area with the surface of the DSC pan.

Figure 37 to 40 show DSC traces for emulsions with water percentages 35,30 25,
20,15 and 12% respectively. The traces for the higher water contents (35,30 and
25%, Figure 37) show a smooth trace with, any phase transitions masked by the
broad water evaporation band. The water evaporation band occurred because the

sample containers were pierced. There is evidence of an endotherm on the 25%

emulsion, but this was hidden in the water evaporation band and therefore difficult
to determine quantitatively.

79
Figure 37 DSC trace of 35,30 and 25% water emulsions.

25,30 and3 5% emulsion

-9 Possible
A endothermat this
0, 04 point (-55'C)
0- x . 35%

cu
r-
T 50

Figure 38 20% water emulsion.

80
Figure 39 15% water emulsion.

Itpr
15% mmteremulsion
"d-
E-
A
-0

50 100

Figure 40 12% water emulsion.

12% livater emulsion

-0

0ý

50 100

81
As the water content decreased from 25% to 20%, Figure 38, an endotherm can
be seen at approximately 55C. This is could be to be due to a IV-11phase
transition, although this would only be expected to occur in a dry sample. A more

plausible explanation for the endotherm would be that a degree of crystallisation

had occurred in the sample and no phase change occurred, but the AN

redissolved into the water. AN dissolving in water is a well-characterised

endothermic process and this would seem an entirely plausible process to occur in
the sample. The same endotherm is apparent but reduced for the 15% water

emulsion, Figure 39, signifying that perhaps less crystallisation of this sample had

occurred.

Figure 40 showed an unexpected result, with two endotherms in the 40-60'C

region of the trace and another two endotherms at -90'C and -1500C
respectively. These extra endotherms could be accounted for if crystallisation had

occurred in parts of the emulsion to differing degrees. If the mixture was

inhomogeneous, then the phase change could occur at different temperatures,
due to differing conditions. An alternative explanation is that some of the AN had
entirely crystallised, and come out of solution, and as heating of the sample
occurred, the crystallised AN dissolved. The endotherms at -500C could then be
separated with one endotherm being the redissolution of AN in one part of the
sample, and the other endotherm from another part of the sample being a IV-11
phase transition. This explanation is not entirely satisfactory and a more detailed
examination of this process combining TGA and DSC could be utilised to further
elucidate the process.

To further look at the effect of sample crystallinity, a 12% water emulsion sample,
which had crystallised into a single solid piece, was analysed. Figure 41 shows an
initial trace from this sample. This was placed in a sealed crucible, heated from
200C to 1700Cat 1OK/min,and then allowed to cool to room temperature. Without
removing the sample from the DSC, the cycle was repeated.

The first run shows an endotherm at approximately -60'C, which could be the
redissolution of part of the sample. There was a second endotherm at

82
 approximately -130"C, which, given the temperature this occurred at, is likely to
be a 11-1phase transition. The second run shows an initial endotherm at -45'C
rather than 60'C. This is likely to be a phase change, rather than the AN
redissolving, although it was not possible to determine if this was a IV-11transition
transition. There was another endotherm at -900C, which was not seen
or a IV-111
in the initial trace.This endotherm was seen in the dry AN prills and was
determined to be the 111-11
phase transition. The process occurring in the sample,
which causes the phase transitions to become apparent after initial heating was
not apparent. Notwithstanding this, it is possible that the endotherms, at 600C in
the first trace, were the combination of the redissolution of AN and the 111-11
phase
transition, which combined to form a single endotherm. Again, this is not an
entirely credible explanation and more samples were run in an attempt to
determine a pattern.

Figure 41 Crystallised emulsion sample (12% water) heated and allowed to cool
then reheated.

Crystallised emulsion (12% water)

C;

IL,
1

50. i0o. 150.

83
This was undertaken using the pieces of the same crystallised sample whilst
keeping the conditions constant. The sample was cycled from 200C to 1800C
three times and this is shown in Figure 42.

Figure 42 Crystallised sample, (12% water) heat cycled 3 times.

:LeA owater crysta. Liona

15.000 mv fttm-. io. o rimin

iou. 150.

A similar pattern to Figure 42 was seen, with only two endotherms in the first run,
although, in this trace, only two endotherms were seen on the second run. it was
not until the third run that all three endotherms were seen. The polymorphic
transitions that occur on heating of ammonium nitrate are changes in the crystal
structure and as such should only occur at definite temperatures. The formation of
solid solutions can cause the effects, due to polymorphic transitions to be
displaced, or even disappear, owing to the retarding effect of isomorphous
substitution on the process of phase transition. A solid solution being a crystalline
material, which is a mixture of two or more components, with ions, atoms, or
molecules of one component replacing some of the ions, atoms, or molecules of
the other component in its normal crystal lattice. Compounds can form solid
solutions if they are isomorphous. However, the crystallised emulsion was not a
solid solution, so this would not seem to be a conceivable explanation of the
process involved.

84
To prove that the previous results were not spurious they were repeated and some
the results of this are shown on the following pages.

Figure 43 DSC trace crystallised sample (15.3mg) -20'C-180'C.

A
0
x
cu

C;
ru
I a

14114194148aa114-411 -1--T-T--T-

50.100.150. *

Figure 44 Second run, after Figure 43.

.-

LO
I

0.1

0.100.

85
 Figure 45 Final run, after Figure 44.

0
cu
I

001

I- -IIIIIIIIIIIf

100
04

-86
 Figure 46 Large sample size, 100.8mg.

, -0
C)- X
ý a)

0
0

ioo. 6c

Figure 47 Second run, after Figure 46.

o-
-

C;
LO
I

100. l' c

87
Figure 48 Third run, after Figure 47.

0-
1

Cý
LO
I

ioo. 0
Figure 49 Final run, after Figure 48.

:r.

0
0
cu
I

100. l' c

88
Figure 43 to Figure 49, along with numerous repeated experiments, show that the
results were not abnormalities. In a crystallised emulsion sample repeated heating
cycles were required before the endotherms become apparent. Although
repeating the experiments confirmed that this was an actual process, it did not
help to clarify the mechanism by which it occurred.

To help elucidate the reaction process occurring in the crystallised emulsion

sample, analysis of the gaseous reaction products was undertaken. Large
(100mg) samples, of crystallised 12% water emulsion were sealed in gas tight
aluminium crucibles. The crucibles were then heated, in the DSC, in a repeated
heating cycle, in the same manner as the previous samples. After the heating
cycle, the crucibles were placed in a glass sample vial Figure 50, which was
sealed with a self-sealing rubber septum. The crucibles were then pierced and the
gaseous products were analysed by headspace GC-MS.

Figure 50 Method used for gas analysis of crucibles.

Septum, self-sealing Syringe with

preventing gas loss sealed needle
when syringe is used to pierce
removed; crucibles
Purged with N2
before sealing

Gas-tight glass
samplevial

Aluminium J[]
crucibles -4QQ

89
It would be expected that ammonium nitrate would begin to decompose at the

upper temperature limit of the experiment (1700C). If this happened then N20

would be formed, along with water, as shown in the reaction below.

Equation 16 Decompositionof AN.

N20
NH4NO3 + 2H20

Analysing the gas from the crucibles, for the presence of N20, would show

whether decomposition had occurred. N20 is a neutral oxide and does not form
hyponitrous acid (H2N202) with water or hyponitrites with alkali. Therefore, water

should remain unreacted (if formed) in the container. If decomposition were

occurring, then water could be the catalyst for the polymorphic phase changes in

the ammonium nitrate.

The samples were analysed and compared to two blanks. One blank consisted of
an empty sample container purged with N2 before analysis. The other was a
sample blank, which consisted of a glass container with sealed crucibles of
unheated emulsion.

90
Figure 51 GC-MS traces m/z 44.

Heatedsample 1
Scan
EI+
7.664 44,
1.444
,
RT
C02

peak

Probable N20 peak

8,268
9.355 10-467 1081611.097
6ý399 696A

0 El+
can
JARH AA

91
The retention time for C02 and N20 was 7.6min and 8.2min respectively. It can be
seen that there is a small peak at 8.2min in the sample trace, which is absent in

the blank traces. The two peaks, from the sample trace, were analysed by mass
spectrometry to determine their fragmentation pattern and identities. These show
the expected fragmentation for N20 andC02 respectively.

Figure 52 Mass analysis of peak at RT 8.2min.

N2
too-

NO

N20
%
44

02

N, 32
14

30
0

Is
L111
I

219
20 34 40
17 1 1 3,8 1
+ --, -==-7= ý1 1 111111T =r-= "ft
O ,, .1111
W 11 12 13 14 15 1i 19 20 21 22 23 24 25 26 27 ý8 29 31 31 32 33 34 35 36 37 36 39 40 41 42 43 44 45 46 6 48 49 50
_I

92
Analysis of the atomic masses, in Figure 52 shows that the peak at RT 8.2min is
almost certainly N20. There was no peak at m/z 12 whilst there are peaks at mass

weight 14,30 and 32, which were not present in the C02 spectrum (Figure 53).

C02
Figure 53 Mass analysis Of peak, RT 7.6min.

93
If N20 is formed in the heating of the sample, then water will be produced as well.

N20
NH4NO3 io + 2H20

Water, specifically H30+, is suggested to be the catalyst for the phase

transitions(90)in AN. As water is being generated through, the decomposition AN,
this could be providing a source of H30+ and thereby catalysing the reaction.
Notwithstanding this, the crystallised emulsion sample should have water present
and thereby the H30+ion should be present.

A reason for the water in the emulsion not catalysing the reaction could be that oil
and emulsifiers are present in the sample. If, when the sample completely comes
out of solution and crystallises, the water is expelled from around the crystal
structure and oil then surrounds this, it could be argued that there is no water
available to catalyse the reaction. This argument is purely based on supposition,
with, as yet, no evidence and much further research beyond the scope of the
present study required, but in my opinion this is plausible

94
3.7 Explosive performance results
3.7.1 Introduction
The explosive ordnance range (ERDA) is a purpose built facility where explosive
firings can take place. Up to 400g of explosive material may be fired outside and
up to lkg inside a purpose built containment building. All firings below 400g, for
this study, were conducted outside so that they could be observed. Figure 54
shows a prepared charge as seen from the firing point.

Figure 54 Explosive ordnance research range, with a charge prepared for firing.

Containment
building

Preparedexplosive
charge

WL -

Figure 55 Detonation of charge- note the white cloud.

95
Figure 55 shows a charge initiating; it was noted that there was a bright yellow
flash upon detonation and a large plume of white smoke. All firings that
propagated showed this, with the intensity of the flash and amount of smoke
produced being variable.

Charges were prepared, mixed and initiated within 24h on the ERDA range. This
was done to minimise any effects of ageing on the emulsion matrix. Prepared
emulsions were placed into plastic tubing (39mm polypropylene) as these provided
a minimum degree of confinement whilst still providing the necessary degree of
support. The tubing was attached to steel witness plates (10mm thick mild steel)
using an elastomer. Once on the explosive research range, priming, if required,
was added in the form of cut to fit sheets of Demex 200 (sheet explosive).
Charges were placed outside in the firing area and the range-conducting officer
(RCO) attached detonators directly to the top of the tubing. After checking the
circuits, the RCO would initiate the electric detonator.

Figure 56 shows the initiation process as recorded on a standard VHS video

camera. Standard video can only record at 15 frames per second, so only a brief
snapshot of the process can be seen. Figure 56 shows this process as a bright
flame flash as detonation occurred followed by the production of a large amount of
gas. The smoke cloud did not disperse rapidly, but slowly dissipated into the
surrounding area in a process that took minutes rather than seconds. In Figure 56
frame 2, the charge can be seen at the beginning of the initiation process. The
detonator, and primer have initiated and the emulsion is just starting to propagate.
The escape of the hot gases and flame can be clearly seen to be coming

predominately upwards, from the top of the charge.

96
Figure 56 Emulsion initiating.

10

4,

'VA

14
14

5i

97
3.7.2 Hot spot temperatures in the emulsion
By calculatingthe detonationpressureapplied by the primingexplosive,used in
the trials, Demex200 sheet explosive,it was possibleto determinethe hot spot
temperatureof the microballoons.The detonationpressurefor Demex200 was
first calculated:

Equation 17 Calculation of detonation pressure for Demex 200

* 10 -6 * VOD 2*p
Dp = 2.5

VOD = 7,200 ms -1
p=1.63

Dp = 2.5 * 10 7,200 2*1.63

= 211 kbar

Using the calculated detonation pressure and knowing the initial pressure of the
microballoons (0.2bar), the hot spot temperature can be determined using Bowden

and Yoffe's equation.

Equation 18
(Y-1)
T2 =T
J(P2
A

y=1.29, p, = 0.2bar

P2 = 207.5kbar
T, z 293K

(1.29-1)
( 207500 1-29

T 293
0.2
T2 6596K

98
The average hot spot temperature would be in the region of 6000'C. Eyring(91)in
his treatise stated that the calculated critical hot spot temperature for initiation is
dependent on the radius of the hotspot. The larger the radius the lower the critical
temperature required before initiation occurs.

Table 11 Critical hot-spot temperatures for hot-spot radii as calculated by Eyring.

10-4M 1 O"m I 0-6m

Explosive r= 10'3m
PETN 31 O'C 385 *C 495 OC 640 OC
RDX 380 OC 485 OC 615 OC 815 OC
HMX 4100C 5100C 645 OC 825 OC
Tetryl 425'C 570'C 815 OC 1250 OC I
Ammonium 590'C 825'C 1230 OC 2180 *C
nitrate

The average radius for the microballoons, before being crushed by the shock
wave, is in the region of 3 X10-5 m. The hot-spot temperatures achieved are
therefore higher than that required to initiate ammonium nitrate, without the
presence of oil. The critical temperature for emulsions is not known but, given the
high temperature achieved, microballoons would be expected to provide an
adequate initiation source for the shocked explosive.

99
3.7.3 Velocity of detonation data

Emulsions were formulated, as described in the experimental section, and velocity

of detonation measurements taken. For each variant of the emulsion, change in
water and microballoon content, at least four explosive firings (and an average of
eight) were undertaken, to acquire representative data. The data from this was
then tabulated, and is shown in Figure 57, as a plot of percentage theoretical
maximum density (TIVID)versus VOID. (Percentage theoretical maximum density
is taken as a percentage of the density compared to the base emulsion with no
added microballoons). The error bars on Figure 57, show the standard deviation
calculated for each emulsion based on the variation in measured VOID. As plotted
in this form the data is hard to interpret, although it is possible to see a slight
parabola. This data is also presented, in a different form, in Figure 58.

From Figure 58 it is possible to see that the data shows a hill type profile. At low
water contents, there is a high ridge between 2.5 and 3.5% microballoon content.
As the water increases to 25%, a peak is seen where the highest VOID is
observed. Beyond 25% water content, there is a rapid drop as the observed VOID
fails. Figure 57 also shows that at low water contents the VOID is more stable
across the microballoon range.

100
 Velocity of detoantion (ms-1) -n
(a
c
-A, -Cý (JI CA 0) (D
0 (n C) (n 0 01 0
0 0 C) C) C) 0 CD
0 0 00 C) 0 C)
(3) III I
(-TI -
0* (D
0
:) 0) 0
0
(D

0
4%
CL
CD
0
=3
-4
C)

=3

CD
W

-4

0
<
CD

U)
CD
9 C) i <
-T
0
LTI
IR
0
OD ,4
Cn
CD

w 3
Cý LVI

(D
Ul -
-
 co
CD
al
OD

0
0
CD
0
C> 4ýk 00 CN Cý a
C) 0 Cý 0 0 0 C> CD
a 0 CD C) CD 0 0 I-
0
JI 11 !A
týj týj t!A
01 00 ON 0 00
0 0 C) C> C) 0 0 rA
C) Cý CD 0

C:
0.- 0
C)
t-j 3
0
0I
cr
w
0 8
D
0
91)
=3
a

(D
0

(D
::3

t.

ON
t-j

Water
3.7.4 VOD data plotted as functions of water content

To analyse the data the results were split into those which required boosting for
initiation and those, which were detonator sensitive. Low water content emulsions
required less boosting than the higher water content emulsions. As the water
content increased, the ease of initiation decreased; this showed that the water had
a retarding effect on the initiation of the emulsion.

The level of priming required for initiation is shown below in Table 12. This is
shown by how many sheets of Demex 200 explosive were required before the
emulsion propagated and an explosive reaction occurred. It can be seen from the
Table 12, that at low water content, no priming was required, but as the water
content increased priming became a requirement. At 25% water content, priming
was required for every formulation. At higher than 25% water content the
emulsion often failed to propagate and no reaction occurred. Three layers of
sheet explosive was the maximum priming used. If the charge did not propagate
at this level then it was considered insensitive to shock initiation and consequently
not listed on the table.

Table 12 Priming required for initiation (sheets of Demex 200 plastic explosive).
Water Microballoon content (%)
Content 0122.5 3 3.5 4 4.5 56
N
12 f 0 0 0 0 0 0 0 0 0
15 f 1 0 0 0 0 0 0 0 0
20 f 3 2 2 0 0 0 0 0 0
22.5 n f n n 0 1 n n n n
25 n f 1 1 1 1 1 1 1 0
27.5 n f f f f n 2 n n n
30 n f f f 2 n 1 n 1 n
35 n f f f f f f f 1 n

f- failed to propagate n- not tested

103
Table 12 shows that 12% water content emulsion required no priming, but with a
microballoon content of less than 1%, it failed to propagate, even with four sheets

of Demex 200. When the water content was increased to 20%, at low

microballoon content, priming was required for initiation. More priming was

required at 20% water content than for 25% water content for equivalent low
microballoon contents. For this to have occurred there must be a change in the
initiation mechanism, which increased the sensitivity of the 25% water emulsion.

To analyse the VOD data for the emulsions it was determined that splitting the
data into separate water contents, and plotting these against percentage TIVID,

would aide interpretation of the data. The error bars, as shown in Figure 59, are
the standard deviation as calculated from the data. The fitting parabolas shown
are that of a line of best fit for a second order polynomial, as plotted by the

spreadsheet software.

Figure 59 12% water content emulsion explosive.

12% water content

5500

5000

f- 4500
,
cn
E
0
0
> 4000

3500

3000
60 65 70 75 80 85 90 95
% TMD

104
In Figure 59, the VOID showed an initial increase with increasing TIVID followed by
a plateau region (70-83% TIVID) where the VOID remained reasonably constant at
around 4750ms-'. As the TIVID increased beyond 83%, there was a rapid drop in
VOID down to 3400ms-1, followed by failure to initiate. This was a classical Type 11

non-ideal explosive behaviour, which mirrors the type of results recorded by Price.

Figure 60 15% water content emulsion explosive.

15% water content

5000

4800

4600

4400

4200

4000

3800

3600

3400

3200

3000
65 70 75 80 85 90 95
% TMD

Figure 60 shows a similar trend to that of 12% water content emulsion except the
plateau region was less pronounced. The VOD increased with TIVIDuntil at 82%
TIVIDthe VOIDbegan to drop off rapidly until failure occurred, at above 93% TIVID.
In comparison to the 12% emulsion, the average VOD was lower across the whole

105
TIVID range but this was not a significant drop in performance. Again, this showed

a classical type 11non-ideal explosive behaviour.

Figure 61 20% water content emulsion explosive.

20% water content

5000

4800

4600

4400

4200
cn

4000
0
0
3800

3600

3400

3200

3000
65 70 75 80 85 90 95
% TMD

Figure 61 shows the data for a 20% water content emulsion and it can be seen
that there was approximately 30% difference in VOID between the highest,
4918ms-1, and lowest, 3472ms-1, VOID. This was not considered excessive and

would be considered typical for VOID versus TIMIDrelationships. The data shows
that the 20% water emulsion underwent a rapid rise in VOID with TIVID before
peaking at around 80% TIVID. At higher than 80% TIVIDthere was a slow drop off
in VOID, although not as rapid a drop as observed with the lower water content

106
emulsions. This slow drop off could be accounted for as at high TIVID the 20%

water emulsion required priming for initiation, and once the reaction started it was
sustained.

Figure 62 25% water content emulsion explosive.

25% water content

5800

5600

5400

5200

5000
E
0
0 4800

4600

4400

4200

4000
65 70 75 80 85 90
% TMD

Figure 62 showed a similar pattern to the lower water content emulsions, except
for the initial drop in VOID at 67 to 70% TIVID, there was an upward trend in VOID
followed by a rapid drop. However, this occurred at a much higher VOID than

exhibited by the lower water content emulsions. The highest VOID recorded, for a
25% water emulsion with 3% microballoons, was 5558ms-1. This VOID was

observed at approximately 80% TIVID, which was a similar TIVID at which the
maximum VOD for the lower water content emulsions was observed.

107
At 25% water, content priming was required for the entire density range except the
lowest TIVID. Although this showed that, the 25% water emulsion was less

sensitive to shock initiation than the lower water content emulsions, once the
explosive initiated the reaction occurred at a much faster rate. As the trend was

similar to that of the lower water content, emulsions it must be considered

plausible that the high detonation velocities recorded at 25% water content could
be due to overdriving the explosive. As the plastic sheet explosive, which was

used to prime the emulsion, has a VOD of -8000 ms-1 this was a reasonable
proposition. In this regard at 2% microballoon content, the highest TIVID to initiate
(and therefore at 25% water content occupies the least amount of space in the
tube), the length of plastic tubing used was 170mm. Given that the diameter of the
tube was 39mm, it was clearly possible that this could be an effect of overdriving
the explosive, as the gap between ionisation probes was 100mm with a 10mm
offset at the bottom of the tube, see Figure 63, leaving a 60mm gap to the top of

the explosive column.

Figure 63 Showing the tube set up for a 25% water, 2% microballoon, emulsion.

39111111
10

Figure 63 shows that, in this case, the separation of the first lonisation probe from
the top of the explosive column was less than three times the diameter of the
column. As was discussed previously (see Figure 4), this would have the effect of

108
At 25% water, content priming was required for the entire density range except the
lowest TIVID. Although this showed that, the 25% water emulsion was less

sensitive to shock initiation than the lower water content emulsions, once the
explosive initiated the reaction occurred at a much faster rate. As the trend was
similar to that of the lower water content, emulsions it must be considered
plausible that the high detonation velocities recorded at 25% water content could
be due to overdriving the explosive. As the plastic sheet explosive, which was

used to prime the emulsion, has a VOID of -8000 ms"' this was a reasonable
proposition. In this regard at 2% microballoon content, the highest TIVID to initiate
(and therefore at 25% water content occupies the least amount of space in the
tube), the length of plastic tubing used was 170mm. Given that the diameter of the
tube was 39mm, it was clearly possible that this could be an effect of overdriving
the explosive, as the gap between ionisation probes was 100mm with a 10mm

offset at the bottom of the tube, see Figure 63, leaving a 60mm gap to the top of
the explosive column.

Figure 63 Showing the tube set up for a 25% water, 2% microballoon, emulsion.

39mm
10

Figure 63 shows that, in this case, the separation of the first ionisation probe from
the top of the explosive column was less than three times the diameter of the
column. As was discussed previously (see Figure 4), this would have the effect of

108
overdriving the explosive causing an artificially high VOID, over the measured
section, to be observed. In this regard it should be noted that the increase in VOID
due to the explosive being overdriven drops off exponentially with distance and in
this case the distance, was just over 1.5 times the tube diameter. The explosive
could still be overdriven at the point of the first ionisation probe but not excessively
SO.

Two differing approaches were adopted to show that the observed VOID results
were not anomalous results due to overdriving of the explosive. The first approach
was to increase separation between the first probe and the top of the explosive.
This was undertaken by decreasing the separation between the ionisation probes
to 5cm, which gave a distance of 11cm from the top of the explosive column. This
distance was still inside three times the diameter of the column, which was
11.7cm. The results from this were entirely consistent with the data already
obtained from the previous firings and showed no decrease in VOID, just the
normal statistical spread of results. A second approach was taken, to remove any
uncertainty with the validity of the data. The amount of emulsion in a tube was
increased to 1kg, which increased the column length to a minimum of 77cm, for
the highest TIVID. For these trials, conducted inside the containment building, the
gap between the ionisation probes was increased to 50cm, whilst the separation
from the top of the explosive column was increased to a minimum of 25cm. Again,
data from these results were consistent with previous firings, indicating that any
effect due to overdriving the explosive was a short-lived effect and did not
noticeably affect the results.

This work was considered important, in that it showed that although overdriving an
explosive is a noted effect, when a single sheet of priming was used it had no
perceptible effect on the observed VOID data. Therefore although many of the
firings were undertaken where the first ionisation probe was just within 3 charge
diameters of the top of the explosive column, when a single sheet of priming was
used these are considered valid results.

109
Figure 64 27.5,30 and 35% water content emulsion explosives.

High water content emulsion explosives

Water content -27.5% 35%j

4000 -

3950 -

3900 -

3850 -

3800 -

E
3750 -
0
0
3700 -

3650- M

3600-

3550 -

3500
72 73 74 75 76 77 78
% TIVID

At higher than 25% water content, the emulsions became extremely insensitive to
initiation.The %TMD range across which initiation could occur was narrow with

the explosive often failing to successfully propagate to detonation, even inside

these narrow %TMD limits. This failure would sometimes occur after the explosive
had begun to initiate, which had the effect of throwing the unreacted emulsion
across the firing area.

The VOID results for these high water content emulsions were consistently below
that of the lower water content emulsions. It was difficult to fully characterise the
high water content emulsions, as failure to detonate occurred on such a regular
basis, that firing became a problem, due to the clean up operations required.

110
Figure 65 Average VOID for emulsion water contents

Average VOD for each emulsion water content

5000

4750

4500

4250

4000

3750

3500
10 12 14 16 18 20 22 24 26 28 30

Water content (%)

Figure 65 shows the average VOID for each emulsion water content, as calculated
from every result for each emulsion, giving the average VOID for the emulsion

across the entire %TMD range. This shows an expected result of decreasing VOD

with increasing water content until, at 25% water content, this trend stops, and a
huge rise in VOID occurs. This cannot be considered anomalous as the average
was based on over sixty individual firings for the 25% water emulsion, with a
similar number for the other, low water content, emulsions.

The low water content emulsions showed an initial rise in VOID with %TMD and if
this initial behaviour is examined more closely, a trend can be seen in the data as
Figure 66 shows.
Figure 66 The linear section of density versus VOID.

12% water emulsion 15% water emulsion

6000 6000

5500 5500

C, 5000 5000 4
.2
r- r-
2
.
45001 -0 4500

0
gýI 1y= 19.757x + 3262.8 2ý.
:
0 i R2 =0 4354
Z .
> 4000 - 4000
y= 47.92x + 921.45
R2 = 0.8498

3500 4
3500
65 70 75 80 85 65 70 75 80 85
TMD
TW(%)

20% water emulsion 25% water emulsion

60001
6000

5500 5500
E

0 500011
5000

4500 4500
0
26x - 3245.3 A-_ y= 116.89x - 3840.7
0 0.883 0
R2=0.9896
4000 0 4UOO
>

3500 3500
65 70 75 80 85 65 70 75 80 85

TMD(%) TMD (%)

Figure 66 shows that as the water content increases, so the rate of increase in
VOID with respect to %TMD increases. At 12% water content, the VOID across the

entire density range is almost invariant. No priming was required and the
showed a comparable VOID across the %TMD range.
emulsion

112
 Figure 66 The linear section of density versus VOID.

12% water emulsion 15% water emulsion

6000 6000

5500 5500
V)
-
E U)
E
C:
0 5000 5000
,;--
cu
C:
a)
_0 -0
4500 4500
y 19,757x + 3262,8 21
ü
02= 0
a) R 0.4354
> Q)
4000 > 4000
y= 47 92x + 921 45
R2=0,8498

3500 3500
65 70 75 80 85 65 70 75 80 85
TMD (%)
TMD (%)

20% water emulsion 25% water emulsion

6000 6000

5500 5500
(n
E
c C:
0 2 5000
5000
C:
ýo
ý! 4500 4500
0
y 101.26x - 3245.3
y= 116 89x - 3840 7
0 R2=0.883 0
Z 4000 4000 W=0 9896

3500 3500
65 70 75 80 85 65 70 75 80 85

TMD (%) TMD (%)

Figure 66 shows that as the water content increases, so the rate of increase in
VOID with respect to %TMD increases. At 12% water content, the VOID across the

entire density range is almost invariant. No priming was required and the
emulsion showed a comparable VOID across the %TMD range.

112
As the water content was increased to 15%, still, no priming was required,
however there was a noticeable relationship between increasing density and
increasing VOID. The slope was at a shallow angle, and varying the formulation
did not have a significant effect on the VOID.

At 20% water content the slope was such that increasing the %TMD gave a
significant increase in VOD. There was no overall improvement in the VOD in

comparison to the lower water content explosives, but there was a stronger
correlation between density and VOD.

At 25% water content there was a significant, and strong, link between VOD and
density. The link between %TMD and VOID was important as an increase in
%TMD markedly increased the VOID. There was also a noteworthy improvement
in VOID in comparison to lower the water contents emulsions. However, priming
was required to initiate the emulsion, although it was a minimum-priming
requirement and overall there was a significant improvement in explosive
performance.

To elucidate the processes occurring in the emulsion at detonation, the process of

initiation must be understood. When an explosive initiates, the phenomena can be
into three (93)
stages ;
separated

1. The trigger mechanism or primary activation process.

2. The spreading or fading out of deflagration from the primary action.
3. The transition from cleflagratingexplosion into detonation.

When the initiating hot spots are small, as with microballoons, it is possible to
separate the three processes, which in such a case succeed one another at some
interval in both space and time.

When the area of initiation is large, the build up to detonation can occur much
more rapidly, without the clearly defined stages of build up of detonation from
hotspots.

113
As the 25% water emulsion required priming, it could be argued that, the
increased VOIDwas an effect of a large initiation area, which caused the explosive
to initiate without undergoingthree separate stages for hot spot initiation.

To prove that this was not the case in this study, selected, although limited, firings
at 12,15 and 20% were initiated using single sheets Demex 200. The data from
these firings were similar and within the expected range for emulsions initiated
using only detonators.

Whilst this goes some way to showing that, the size of the initiator was not the
cause of the high VOID'srecorded at 25% water content, it has been subsequently
noted that for clarity all the firings should have been conducted using the same
level of priming.

114
3.7.5 Plate dent data

Plate dent measurementswere taken, as previously described, and plotted against

%TMD for each emulsion and a graph of this is shown in Figure 67.

Figure 67 Plate dent volume versus TIVID.

Plate dent versus %TMD all data

Figure 67 shows a similar spread of data to Figure 57 with no pattern easily

discernable, except possibly a trend of decreasing dent volume at higher TIVID,
this is not in any way clear. Again, by splitting the data up into individual
although
water contents, the data can be more easily interpreted.

115
3.7.5 Plate dent data

Plate dent measurements were taken, as previously described, and plotted against
%TMD for each emulsion and a graph of this is shown in Figure 67.

Figure 67 Plate dent volume versus TIVID.

Plate dent versus %TMD all data

.
3.5
""
" ""
££ ""
3

E 2.5

E
0 2

_0
15
.
.

1
.

0.5

0[
.-ý-, "- -- --. - ý-

60 70 80 90 100
TMD (%)

Figure 67 shows a similar spread of data to Figure 57 with no pattern easily

discernable, except possibly a trend of decreasing dent volume at higher TIVID,

although this is not in any way clear. Again, by splitting the data up into individual

water contents, the data can be more easily interpreted.

115
The data for 12% water content emulsion can be seen below in Figure 68, which
shows, that apart from a low dent volume at -65%TMD, the dent volume is

reasonably consistent. The average volume, barring the low result, is 2 .9CM3with
little variation away from this.

Figure 68 Plate dent versus TIVIDfor 12% water content.

12% water content emulsion

3.5

3,3

3.1

2.9

E 2.7

E
2.5

cu 2.3

21

1,9

1.7

1.5
60 70 80 90 100

TMD

This shows a similar trend to the VOID data where there was an initial low result,
followed by consistent data across the TIMIDrange.

Figure 69, shows initially the plate dent volume is stable until around -72%TMD

when there is a large increase in dent volume. The dent volume then drops at
the dent volume drops before again increasing this time to a maximum
-79%TMD
at -82%TMD.

116
Figure 69 Plate dent versus %TMD for 15% water content

15% water content emulsion

3.5

clý-

2.5

1.5
60 70 80 90 100

TMD (%)

The exact reason for the up and down nature of the plate dent results cannot be

adequately explained, except to comment that whilst the plate dent test gives a
permanent record it is subject to a certain degree of error. These errors can
permeate themselves before the explosive is initiated, one such error being that if
there was an air gap at the bottom of the cylinder. If this occurs then the plate
dent achieved is not an accurate refection of the brisance of the explosive. Whilst
it was not certain that this occurred, measures were taken to overcome this effect.
These included drilling small holes at the base of the cylinder and then pushing the

emulsion down the tube with a purpose made plunger until emulsion came out of
the holes. Plate dent measurements were also subject to variations due to the
quality of the witness plate, and the state of the backing plate below.

117
Notwithstanding this, they did provide a guide that ties in with the VOID data to give
a general trend.

Figure 70 Plate dent versus %TMD for 20% water content.

20% water content emulsion

3.5

3.3

3.1

2.9

E 2.7
a)
E
:32.5
0
2.3

2.1

1.9

1.7

1.5
60 70 80 90 100

TMD

Figure 70 shows a similar profile to that of the 15% water content, except that
there was a more pronounced initial rise in the plate dent depth with %TMD.
overall, the plate dent depth at 20% water content was more consistent than that

observed at 15% water content emulsion.

Figure 71 shows the data for 25% water content emulsion. There was a drop off in
dent volume at -83% TIVID but this was followed by a subsequent increase at

-87% TIVID.

118
Figure 71 Plate dent versus %TMD 25% water content

25% water content emulsion

3.5

3.3

3.1

2.9

IE
2.7
(V
E
:32.5
0
2.3

2.1

1.9

1.7

1.5
65 75 85 95

TMD

The plate dent data does not show the same correlation as seen with the VOD.
This would indicate that either the test was to an extent flawed where the data
produced was unusable or it indicates that VOID is not the only parameter to affect
the brisance of the explosive. The size of the reaction zone itself could be such a
parameter, with a detonation with a large reaction zone releasing energy over a
longer time period, and therefore showing a lower brisance value to that of a
detonation with the same VOID but a smaller reaction zone. This is discussed later
in the context of a grain burning and thermal detonation.

In order to determine if there was a direct relationship between VOID and plate
dent a graph of VOID versus plate dent was plotted and this is shown in Figure 72

and Figure 73.

I 19
1
Figure 72 VOIDversus plate dent for 12% and 15% water content emulsion.

VOD versus plate dent for 12% and 15% water content emulsion
12% water 0 15% water
5600-

5200 -

4800

lfý
,Cn a
E
4400
0
0
>
4000

3600

3200
1.8 2.2 2.6 3 3.4 3.8
Plate dent (cm)

Figure 72 and Figure 73 shows that whilst it can be argued that there is an
underlying trend of increasing dent volume with VOIDthe data is ambiguous.

120
Figure 73 VOID versus plate dent for 20% and 25% water content emulsion.

VOID versus plate dent for 20% and 25% water content emulsion

20% water 25% water

5600 -

5200 -

4800 -

(n 0
E 4400-

4000-

3600

3200
1.8 2.2 2.6 3 3.4 3.8

Plate dent (CM)

Plate dent tests are not the best method with which to determine the brisance
effect of an explosive. The advantage they did offer, and a reason for their

widespread use, is that the test was easily combined with VOID observations and
was relatively inexpensive, in both materials and equipment. It was hoped that the
double pipe test would produce data that would correlate directly to the plate dent
data, but, unfortunately, this test was unsuccessful. Each firing undertaken using
the double pipe test resulted in a steel cylinder being entirely flattened along the
bottom surface. This was not the expected result as, according to the literature(M
this test has been used quantitatively for emulsion explosives.

For the plate dent tests itself, there are many improvements, which could have
been adopted, to obtain results that are more consistent. These were not
undertaken for fiscal reasons. Improvements would have included using a much

121
larger backing plate, replacing this as and when necessary, and using steel tubing

welded directly to the witness plates, to eliminate the possibility of the tube falling

during the firing process. Notwithstandingthis, the plate dent data does provide a
reasonable, if limited, insight into the brisance effect of emulsion explosives.

3.7.6 The effect of mixer type on VOD.

The effects of mixer type were investigated by comparing the standard food mixer,
with a homogeniser, and looking at the variation in VOID from the resulting
emulsions. A 25% water emulsion was prepared using the homogeniser and then
microballoons were added to this using the standard food mixer. This was

compared to a 25% water emulsion mixed using the food mixer alone.

Using standard 39mm plastic tubing, the homogenisedemulsion failed to initiate in

any of the sixteen firings undertaken. Larger tubing (65mm diameter) was used as
an alternative and this did initiate. The results of these later tests are shown in
Table 13.

Table 13 VOD of a homogenised 25% water emulsion versus a standard mixed

25% water emulsion.

Standard emulsion 65mm diameter Homogenised emulsion 65mm diameter

tubing tubing
(25% water/3% microballoons) (25% water/3% microballoons)

5208ms-1 3876ms-1

5208ms-1 3676ms-1

5502ms-1 3759ms-1

5324ms-1 3817ms-1

5265ms-1 3782ms-1

The homogenised emulsion did not perform as well as a standard emulsion,

exhibiting only around 70% of the VOD of a standard mixed emulsion. This was
thought to have been due to the mixing action of the homogeniser, which had a

122
higher rate of energy deposition than the standard mixer. This had the effect of

causing phase separation to occur on mixing. It was occasionally noted that, after
homogenising, water would separate from the emulsion itself.
As the results from the homogeniser produced low VOID's it was not considered

useful to examine the homogeniser further as a technique for emulsion

manufacture.

3.7.7 Effects of mixing time on VOD.

The effect of mixing time on the VOD of an emulsion was measured. This was
achieved by mixing a 5kg batch of 20% water emulsion and removing 1kg at a
time from the mixer at various time intervals (2,10,20 and 30min). Before
microballoons were added to the emulsion, particle size was measured for each
batch (as shown in Figure 30). Microballoons were then added to make a
standard mixture of 20% water emulsion with 3% microballoons. The explosive
performance properties of these were then measured.

Figure 74 shows that the VOIDof the emulsion steadily increased with mixing time
until about 20min of mixing time. After 20min of mixing, the mixing process begins
to have a detrimental effect on the emulsion VOID. This effect could be compared
to the use of the homogeniser, as the emulsion is mixed it reaches an optimum
after which breakdown of the emulsion matrix begins to occur. The homogeniser
applied energy to the mixing process very rapidly, not giving the emulsion time to
recover, whereas the food mixer applied energy much more evenly through the
emulsion. Notwithstanding this, there was a point at which the addition of more
energy, by mixing, has a detrimental effect on the VOD of the emulsion.

123
Figure 74 Effect of mixing time on VOD for 20% water emulsion/3% microballoon.

Effect of mixing time on VOD for 20% water emulsion/3% microballoons

4000

3800

f-
ý.
cn
E

3600 -

3400 -
0 5 10 15 20 25 30 35'
Mixing time (min)

Figure 75 shows the same VOID data as in Figure 74, but this time plotted as a
function of emulsion droplet size versus VOID. As discussed earlier, the more
intimately mixed an explosive composition is, then the faster the reaction would be
to occur. It would be expected that decreasing the average droplet size
expected
would lead to increase in VOID.

124
Figure 75 Effect of median particle size versus VOID 20% water 3% microballoons.

Effect of median particle size versus VOID for

20% water/3% microballoons

4000

3800

E
a
0

3600

3400
4 4.5 5 5.5 6 6.5 7
Median particle size (ýjm)

Figure 75 shows that for emulsion explosives this trend of increasing reaction rate
with decreasing droplet size did not occur. From Figure 75 it can be seen that the

converse is almost true, as the droplet size increases, so the VOID increases.
Thus, whilst droplet size is important to emulsion stability, there was not a strict

relationship between disperse phase droplet size and emulsion explosive VOID. It
was also apparent that the change in average droplet size was not as important an
effect on VOID, as mixing time itself, which showed a clear trend of increasing
VOD with mixing time.

3.7.8 Viscosity versus VOD relationship

it was expected that emulsion explosives would display a relationship between

and VOID. The viscosity of the emulsion determines the rate of
viscosity
Movement of microballoons through the emulsion, as, in time, the low-density

microballoons would begin to separate out. The behaviour of the emulsion under
shock loading, to a certain degree, is also determined by the viscosity of the

125
emulsion. Figure 76 shows a graph of viscosity versus VOID, for various different

water content emulsions.

Figure 76 Viscosity, as calculated from emulsions with no microballoons present,

versus VOID for emulsions with 3% microballoons.

VOD versus Viscosity with 3% microballoons

6000

^IAZ -- --

5500

5000

0
>
4500 iter

4000

3500 -ý--
300 350 400 450 500 550 600 650 700,
Viscosity (P)

The graph shows that there was only a limited correlation between viscosity and
VOID. The low viscosity at 25% displays the highest VOID but the expected trend
decreasing VOID with viscosity did not occur and no real trend can be
of
determined from the data.

126
4 Further Discussion

The preceding results have shown that the explosive behaviour of emulsion
explosives cannot easily be defined by their physical characteristics. Knowing the
viscosity, density or the thermal conductivity of the emulsion does not allow the
prediction of the VOID. This section attempts to account for the differing behaviour
of the emulsions and show how emulsion explosives, whilst displaying type 11
non-
ideal behaviour, could be further classified by the type of reaction that occurs upon
detonation.

Figure 77 All explosive firing data plotted on a VOD versus density graph.

All data plotted on a VOD versus density graph showing data separation

6000

5500 -

5000 -

17
U)
E
4500 -
0

4000 -

3500 -

3000 -
0.9 1 1.1 1.2 1.3 1.4 1.5
Density (g CM)

Figure 77 shows the data for all the explosive firings plotted as density versus
VOID. Looking at this it was possible to see two distinct trends in the data. On this
it
basis, was subjectively split into one side of the graph, which showed a linear
trend of increasing VOID with density, and one side, which showed a parabola.

127
This split was entirely based on the data as seen in Figure 77. Inspection of the
data showed that the split in the data was not random. All the data, which formed
the linear trend, was either high water content or high microballoon content. All
the data in the parabola was low water content and microballoon content. This

process of splitting the data was repeated for the plate dent data and again the
same trend was noted.

Discussions with colleagues(92)suggested that it was possible that emulsion

explosives could display two differing reaction regimes, that of grain burning and a
thermal explosion. How these two reaction regimes exist and how emulsions
could display these regimes is discussed below.

4.1 Initiation

Bowden and Yoffe(93)

have suggested that;

'A knowledge of the effect of heat on explosives is of importance for almost

every aspect of their behavidur. For example, a rise of temperature is
involved either directly or indirectly in most explosion hazards. Again,
fundamental information on the process of thermal decomposition of
explosives is required for elucidating various mechanisms of sensitiveness.
Finally, in the build up of the detonation wave in an explosive, the
mechanisms of energy release involve thermal decomposition, though under
rather different conditions of pressure and temperature from those which
apply under normal sensitiveness.

Once an energetic material has been exposed to external energy and hot spots
have been generated, the surface of the material will soon reach temperatures
high enough to volatilise it. This process, referred to as the induction lag or
ignition delay time, occurs in a very short time period (measured in microseconds).
There is no standard lag time for explosive material since induction lag is
dependent on the composition of the explosive mixture.

128
They further commented that the standard hydrodynamic equations of detonation
do not refer to the physico-chernicalmechanisms where energy is released in the
detonation wave. In the case of mixed explosives some of the energy may be
released a considerable way behind the detonation front. In fact, the usual
theoretical equations refer to the behaviour of an infinite plane wave of detonation
travelling through the explosive. However, experimental studies have shown(94)
that detonation velocities depend, to some extent, on charge diameter, when this
is small. This can be explained in terms of energy release. If the time of
expansion of the charge is comparable with the time required to release the
chemical energy of the explosive, the maximum pressure of the detonation wave
will be less than that calculated for an infinite plane wave, and the detonation
velocity will be correspondingly lower. Qualitative measurements of times of
energy release have been (95)
computed and these are summarised below.

1. TNT and mixtures of TNT and tetryl have energy release times of the order
of 10-7s. The true figure depends, to some extent, on crystal size, as the
size of the crystal increases the time increases.
2. In amatols (TNT/AN), considerably longer times, for the energy release, are
observed. This is attributed to AN, which reacts only at a finite rate with the
detonation products of the TNT. The rate of reaction increases as the size
of the ammonium nitrate crystal decreases.

This can be used to explain the nature of failure for propagation to occur. When
the detonation wave arrives, from the detonator or primer, the rise in pressure
leads to lateral expansion at a rate, which is determined by the charge diameter
and the nature of the peripheral confinement. If the energy release is rapid,
compared with the rate of lateral expansion, detonation will be stable. If the
energy release is slow, the pressure and temperature may drop to a level, at the
periphery of the charge, that the explosive in this region never releases its energy
effectively. When this occurs detonation fails at the edges. The failure occurs in
successive layers from the edge working inwards progressively, which explains
both the fading inwards from the edges and the decrease in velocity as the
diameter of what is left of the detonation front decreases. Failure of the explosive

129
to propagate can therefore be attributed to the slow rate of energy release in the

explosive.

In single molecule explosives, and in mixtures of these, the critical diameter

decreases with decreasing grain size and increasing initial density up to a density
close to a critical density. At or very near the crystal density, the critical diameter
is often found to increase, growing from a few millimetres to tens of millimetres.
One explanation for this behaviour is the hot spot theory. A powdered or pressed
explosive will contain a huge number of 'hot spots', or initiation centres and the
reaction can be assumed to be dominated by the spread of reaction from the hot
spots. It is assumed that the rate of reaction at each hot spot will be greater with a
greater shock pressure therefore implying that there will be a greater concentration
of energy, initially, at each hot spot. There will be an induction lag before the
energy from the hot spots dissipates and the bulk of the explosive material reacts.
The length of this lag depends upon the way the explosive reacts. This reaction
can either be a grain burning type reaction or a thermal explosion.

4.2 Grain burning and thermal explosion

A grain burning type reaction can be thought of as a combustion type process,

where the reaction proceeds stepwise through the explosive. The reaction is
constantly fading and picking up as hot spots are formed. There is little time lag
between the formation of the hot spot, and the subsequent reaction of the
explosive.

in a thermal explosion, the hot spots are created by the initial shock wave and for
a period there is no discernable reaction from the explosive, until the bulk of the
explosive reaches a temperature where it all reacts in a short time period.

There is no increase in energy output or gas production in one process over the
other, but there is an increase in the speed of the reaction rate in a thermal
explosion.

130
This was demonstrated by Leiper et al(96)when they investigated the link between
the physical form and ingredients of an explosive and its detonation performance.
They were looking at low and high order detonations, as shown by explosives

such as nitro-glycerine. They found that in a high velocity regime, thermal

explosion was the process of heat release, but in low order detonation grain
burning kinetics predominate. They went on to establish a link between charge
diameter effects and the nature of heat release. They used data from previous

studies and placed this in a slightly divergent flow detonation model, CPEX, to

produce constant pressure to heat release rates for the formulations. Thermo-
hydrodynamic calculations for the ideal Chapman-Jouguet state for a variety of

explosives were carried out using the JCZ3 equation of state.

Equation 19

(p
h - a,t+p a
=(1-ß) SP
T
r T

where
Mass fraction reacted
Time differentiation
p pressure
ai shape functions depending only on initial formulations
,rj time constants
p,,critical pressure
Subscripts h,l,s refer to various phases in the explosive, nominally hot spot, liquid
and solid.
The four fitting parameters were ri, ThITs, and p,

They used an automatic fitting algorithm, based on a simple line search followed
by a constrained minimisation routine, to generate a least squares fit to the
experimental data with a failure point that lay at a smaller diameter and lower

131
velocity than experimental data. The explosives used, and their compositions, are
shown below in Table 14.

Table 14 Key to graph and composition of explosives examined by Lieper et al(96)

Explosive Line Composition

label
NG98 (high velocity) A 98% Nitro-glycerineblasting gelatine
NG98 (low velocity) B 98% Nitro-glycerineblasting gelatine
NG29 (high velocity) c 29% Nitro-glycerineplaster gelatine
NG29 (low velocity) D 29% Nitro-glycerine plaster gelatine
NG15 E 15% Ni ro-glycerine powder
NG8 F 8% Nitro-glycerine powder

NG98 and NG29 both had a high velocity regime, a low velocity regime, and an
unstable transition regime.

Their graph, Figure 78, shows the differing reaction regimes. Line A (NG98 high
velocity) shows an induction time where no or very little reaction occurs. After this,
the reaction is exceptionally fast, rising to completion very quickly. Line C (NG29)

shows a similar trend, but the induction period is shorter and the rise time slower.
Both these reactions show typical thermal explosion type characteristics as would
be expected by an ideal military high explosive. Thermal explosions produce high
VOID's with the explosive reacting rapidly.

The remaining lines (D, B, E and F) show the characteristics of a grain burning
reaction. In this, the explosive reacts almost immediately with the shock wave but
in a cellular type process. This means that the reaction takes a relatively long time
to go to completion. Line B shows that the reaction rate is at a maximum as the
shock wave passes and slowly tails off in an exponential manner. This leads to
lower velocities of detonation and lower detonation pressures.

132
 (96)
Figure 78 Plot of reaction rate versus normalised time
.

1.2

C A
I

0.8

/'\ i
E

0.6
I- . "\ .1
:
\I
':\
(D
16
E \».
.. \.
Ir 0.4 "..i
/ N:. 'N

0.2 - '.

0
0 0.2 0.4 0.6 0.8
Time / Time to Complete Reaction -

To demonstrate that emulsion explosives do not display standard type I explosive

behaviour, their explosive performance was modelled, using a computer-modelling
program. This was accomplished using a standard military computer modelling
code, CHEETAH, which is based on ideal explosive behaviour, utilising a BKW
equation. Emulsions explosives, as used in this study, were modelled with this
software.

This is a standard Lawrence Livermore computer code that is used for modelling
military high explosives. It allows no divergent flow and gives a linear response in
respect of density versus VOID. Figure 79 shows the results obtained from using
this code, based upon the density and formulation of the emulsion explosives.

133
Figure 79 Density versus VOD calculated by Cheetah using BKW equation

Density versus VOD calculated by Cheetah

6600 -
Water content 0 25 -A- 20 15 --*- 12

6200

5800
E
0
0
ý! 5400
'a

5000-

4600 -

4200-
60 65 70 75 80 85 90 95
TIVID (%)

As the graph shows the lower water content emulsions, having a higher energy
content, are predicated to have a higher VOID. This is the expected response for a
Type I military explosive and at no point should the higher water content
explosives give a higher VOID than a lower water content. The actual velocities
predicated by the code are, principally, significantly higher than seen in this study,
although the code makes no account for charge diameter effects. The code
shows that the highest expected VOID for a 25% water content emulsion was
around 5500ms at 93%TMD, which was similar to that found in this study,

although at a different %TMD.

This variation in performance from that predicated can be accounted for by the
slightly divergent flow expected from an emulsion and from the highly curved
detonation front. Lateral losses, as mentioned are also not accounted for by the
code, which, again, would lead to lower observed velocities than expected. None

134
of this accounts for why the 25% water emulsion achieves the highest VOD, as all
the parameters for the code are based upon giving the maximum energy release
for each formulation.

If emulsions as mentioned exhibited thermal explosion behaviour as well as the

more standard grain burning detonation regime then some of the anomalies could
be accounted for. With a thermal explosion, there is a time lag after the explosive
is shocked in which localised heating occurs before the whole explosive reacts
very rapidly. This rapid reaction allows the release of energy from the explosive to
occur much more rapidly, ultimately giving a higher VOD.

With higher water content explosives, the emulsion requires a greater initial energy
input from priming before the explosive goes to detonation. With high
microballoon content emulsions there is much more localised heating of the
explosive due to the adiabatic compression of the glass balloons. Both these
explosive compositions require or have a greater initial heat input than lower
microballoon and water content emulsions.

135
4.3 Low water content emulsions - Grain burning reaction
These emulsion formulations exhibited the classical type 11non-ideal explosive
behaviour. This data was taken from Figure 77, and comprised points on the
graph described by the parabola. Although, as previously discussed, this data
was separated subjectively (initially without recourse to the identity of the data), all
these data points, except two, were low water content emulsions. Table 15 shows
the emulsion formulations that made up this group.

Table 15 Emulsions exhibiting grain burning type detonation

Water Microballoon
content content

35
30 3
25 2 1 1
20 1 2 2.5 3 3.5 4 4.5 5
15 1 2 2.5 3 3.5 4 4.5 5
12 I ll 2 2.5 3 3.5 4 4.5 5

All these formulations are low water content emulsions, which were relatively

easily initiated and when plotted individually as separate water content displayed
type 11non ideal explosive behaviour.

The only exceptions to the low water content emulsion were the 25 and 30% water

content emulsions. Although there was only one data point from each, and that
was the highest density of each emulsion and the most difficult to initiate. These
two results can be accounted for as they were the highest density of each
formulation to successfully initiate and thereby the detonation front in these would
probably have been unstable. This means that the detonation was close to failure
and parts of the explosive would not have been entirely consumed in the reaction.
On this basis, the fact that results from higher water content emulsions are
included in the grain burning type reaction is not considered significant.

Figure 80 shows this data plotted as density versus VOD and this shows the
classical type 11non-ideal profile.

136
Figure 80 Grain burning detonation VOD versus density

Type 11 non-ideal explosive behaviour (Grain burning detonation)

3.5

2.5
0
>
2 i
12
"0 1
y -25.993'x 2 + 64.092x - 36.334
R 0.7105-
1.5 /*

0.5
0.9 1 1.1 1.2 1.3 1.4 1.5
Density (kg/dM-3 )

Figure 81 shows the same emulsion formulations plotted as density versus plate
dent. The same trend, of initially increasing plate dent volume with density
followed by a region of stability before rapidly leading to failure, is noted.

137
Figure 80 Grain burning detonation VOD versus density

Ty I) ell non-ide aI explosive I) ch aN iour (Crain I) urning cleton at ion)

.
3.5 . .

.
""";. ":. .

.
2+ 64.092x 36.334
y --- -25.99')x -
R2 -- 0.7105
1
.5 .

0.9 1 1.1 1.2 1.4 1.5

Density (kg/dni-

Figure 81 shows the same emulsion formulations plotted as density versus plate
dent. The same trend, of initially increasing plate dent volume with density
followed by a region of stability before rapidly leading to failure, is noted.

137
Figure 81 Type 11non-ideal explosive behaviour plate dent volume versus density.

Type 11non-ideal explosive behaviour (Grain burning detonation)

3.5
. .

n
P

4
2.5

-25.993x' + 64.092x - 36.334

R 0.7105

1"
m- 1
.5

0.5
0.9 1.2 1.3 1.4 1.5
Density (kg/drn-

When the plate dent data was previously plotted no significant correlation with the
VOID versus density data could be seen, but when the data was separated out in
to two reaction regimes, a relationship can be seen between density and plate
dent and VOID and density.

138
 4.4 High microballoon and high water content emulsions-
Thermal explosion
Figure 82 and Figure 83 are graphs of the linear section removed from Figure 77.
These, as shown Table 16, comprise of the high microballoon content emulsions

and the high water content emulsions. The data follows a linear trend of
increasing VOD with density.

Table 16 Emulsions exhibiting thermal detonation behaviour.

ý-Contc
Wal C %licr oballoon
I Content

35
_no
25 4.4 3.5 3 2.
20
15 0
12 0

Figure 82 Thermal explosion VOID versus density

VOD Nersus densit-, ideal cxl)14)si%

e beha%iour

6000

00

1000

ý00 .

IMOO

1.05

I)(-Ilsit.N (kg/dIll

131)
The trend shows that VOID increases directly with density, as would be expected

with a type I ideal explosive. Only a small density range was represented by this

graph, in comparison to Figure 80, but there was a clear linear trend.

Figure 83 shows the plate dent data and this shows a linear trend of increasing

plate dent volume with density.

Figure 83 Plate dent versus density thermal explosion

Plate dcnt vcrsus density Ideal explosive bchaviour

(thermal explosion)
3.5

ci

1.5

0.5
0.95 1 1.05 1.1 1.15
I
Dcnsity (kg/dm--')

If only the VOD versus density graph showed this linear trend then it could be
argued that this was an anomaly of the data. However, the plate dent data also
shows the same trend. When the plate dent data was plotted for individual water
content emulsions no clear trend could be seen. It was previously discussed as to
whether the plate dent test was actually showing the brisance of the explosive.

140
With the data now separated, even though somewhat arbitrarily, the dent volumes

show a relationship with the VOD of the explosives.

4.5 Discussion

Splitting the data into two sets fits with the concept of two differing detonation

regimes, but it must be disputed as to whether the results do support the premise
Given that the plate dent data for the linear trend, thermal
of two reaction regimes.
burning detonation, supports the VOID data,
explosion, and the parabola, grain
there must be merit in an argument for two differing reaction regimes.

There are points that suggest that the differing profiles are not actually differing
but affects from priming of the explosive. This does not
explosion regimes,
for the high microballoon content emulsions, which did not require
account
priming, but did fit in with the observed data. Notwithstanding this, priming was
required for the low microballoon content emulsion with 20% water and this did not
fit in with the linear data. Given this, in my opinion, the split in the data leading to
differing reaction regimes can be justified. Unpublished work by Dr Graham
Cooper has shown that ion pairing can occur with high water content emulsions,
this might have a strong effect on the properties of the emulsion. This could be

one effect on the emulsion, which leads to the change in reaction regime.

In my view, the change in regime does not occur suddenly, but is a gradual
process, with the degree of thermal explosion behaviour increasing as the water
content increased. Figure 84 shows the linear sections of VOD versus density for

each emulsion (as shown separately in Figure 66). This shows that as the water
content of the emulsion increases so the rate of increase in VOIDwith respect to
density increases. This indicates that there is an increase in thermal explosion
type reaction with the increasing water content. At low water content, there is little
difference in VOD with density because the rate of energy release is slow, even
though there is more energy available to be released. At higher water content,
the energy is released more rapidly, which indicates thermal explosion type
behaviour.

Figure 84 Comparison of the initial slope in VOD versus %TMD

Comparision of intial slope In VOD versus O/6TMD

relationship
I 20% water *25% water
*12%waterml5%water

6000

5500

5000

4500
'a
I-0
91
.
4000

3500

3000 i
68 72 76 80
%TMD

The 25% water content emulsion was the only emulsion to achieve the VOID
predicated by CHEETAH. This program was based on ideal explosives, where all
the energy is released rapidly, in a thermal explosion reaction. For the emulsion to
have obtained this VOID, the rate of energy release must be equal to that of a type
I military high explosive, with thermal explosion kinetics. As the lower water
contents failed to reach the predicated VOID's then the rate of energy release must

142
be lower than that of a thermal explosion type reaction, which indicates a grain
burning reaction.

There was further evidence that indicated the slow change from grain burning to
thermal explosion kinetics. This is shown in Figure 85, which shows the angle
from the line of best fit for the linear section of each of the emulsions. This graph
shows that the rate of increase in the slope is at a maximum at about 24% water
content. This would indicate that perhaps increased performance could be

obtained using a 24% water content emulsion in comparison to a 25% water

content emulsion.

Figure 85 The variation in the slope of VOIDversus water content.

The variation in the angle of the initial VOIDversus water content

140

120

100

80

60

40

20

0
10 12 14 16 18 20 22 24 26 28 30 32
Water content (%)

in my opinion, the evidence suggests that emulsion explosives do show a change

in reaction regime with increasing water content.

143
5 Conclusions

This study set out to characterise emulsion explosives with the aim of gaining a
better scientific understanding of their behaviour. Whilst some of these objectives
have been achieved, such as showing that low water content is not a prerequisite
to high VOD performance, others objectives have not been achieved and the
results obtained require further work to elucidate their significance.

It was shown that microballoons had a greater affect on the density of the

explosive, than the water content, and emulsions can be compared directly by
altering the microballoon content.

In investigating the possible breakdown mechanisms for emulsion explosives, it

was discovered that flocculation, whilst rare, was in fact a possible mode of failure.

The thermal properties of the emulsions were observed and it was ascertained
that the highest thermal conductivity was exhibited by the lowest water content
emulsion (12%), whilst the lowest heat capacity was also exhibited by the 12%

water contentemulsion.

The affect of mixing time on the explosive properties of the emulsion and the
droplet size was investigated. It was ascertained from this that the optimum
mixing time, for droplet size, was 2 minutes of mixing, whilst the optimum time for
explosive performance was 20 minutes of mixing.

Viscosities of the emulsion were investigated and it was found that the lowest
viscosities were observed at 25% water content with the highest viscosities
recorded at 35% water content. It was noted that emulsion explosives displayed a
thixotropic nature and as such viscosities varied with time.

The study endeavoured to link the physical properties of the emulsion directly to
the explosive performance, but no physical property was found which allowed the
accurate predication of the emulsion explosive performance.

144
The study showed that emulsion explosives exhibit two distinct explosion regimes,

namely that of a thermal explosion and a grain burning regime. At low water

contents the explosive follows a grain burning regime, with the reaction occurring
rapidly with respect to the shock front and the explosive exhibiting a classical type
11explosive behaviour. As the water content increased, then the regime began to

change to that of a thermal detonation. At 25% water content, the emulsion

displayed this thermal explosion behaviour, and as such increased VOID and
brisance was seen for this formulation.

The study showed that at 25% water content with 3% microballoons the VOID
achieved (5558ms-1) was similar to that predicated by computer modelling
software (CHEETAH). This indicated that the emulsion was undergoing a thermal
explosion reaction, which is type I ideal explosive behaviour.

The study has also shown that there is commercial work that could be undertaken
in this area as the increased performance in the emulsion explosive occurs at
higher water contents. This makes for a more stable emulsion as the increased
water helps to inhibit breakdown of the emulsion matrix. The increase in water
content also leads to an increase in safety of the explosive, as more initial energy
is required for initiation. Notwithstanding this, increasing the water content, whilst
decreasing the oil and ammonium nitrate content, makes commercial sense on a
cost basis.

145
6 Further Work

There is a great deal of further work that could be undertaken in this area, some of
it having commercial implications.

Commercially emulsion explosive formulations could be further developed with

investigations into the increased performance noted at 25% water content. Initially
this would require looking at different diameter tubes to determine if the high VOID
noted is an affect of the diameter of the charge. Current formulations use low
water contents, and this is an area where significant cost benefits, along with
increased performance, could be gained.

The thermal results obtained in this study were mostly comparative. With modern
thermal instrumentation, it is possible to determine thermal conductivity whilst
running TGA and DSC. Undertaking this would help to elucidate the phase
changes that occur in the emulsion. If this was coupled with X-ray crystallography
then the processes involved, which cause the phase changes to occur in AN could
be more fully understood.

Whilst particle size was measured in the study, this was undertaken using an old
instrument at the limit of its design capability. Using a modern technique, which
would allow the whole emulsion including large droplet sizes to be analysed, would
better determine if droplet size has a direct effect on explosive performance.

It is believed that emulsion explosives undergo two distinct reaction regimes and
there are experimental techniques, which could help elucidate this. The
measurement of VOD with resistance probes to allow continuous measurements
would determine run up distances and reveal more information on the way the
explosive undergoes transition to detonation. This could be coupled with fibre
optic probes connected to a fast video recording device, and if placed on the edge
of the charge, would allow limited measurements of the reaction zone to be made.
The use of high-speed video itself would allow some determination of the reaction
processes occurring in the explosive. To further determine what is happening

146
inside the explosive reaction zone then the use of flash x-ray would yield much
information.

The main problem with all the above techniques is that they usually require large
testing facilities and expensive equipment to usefully undertake the analysis.
Fiscal limitations severely hampered the present study, but without this limitation, it
would be possible to fully characterise emulsion explosives and link their explosive
behaviour to their physical properties.

147
7 References

I 'Explosives Engineering', P.Cooper; Wiley-VCH, 1997.

2 'Detonation' W. Fickett and W. Davies; University of California Press, 1979.
3 'The Mathematics of Combustion', Editor J. Buckmaster, SIAM, 1985.
4 P.Vieille Comptes Rendes Academie des Sciences, 1893,130.
5 E.Mallard, Le Chatelier, Comptes Rendes Academie des Sciences, 1881,93.
6 'Explosives in the Service of Man, Editors J.Dolan and S.Langer: The Royal
Society of Chemistry 1997; W. Mather 120-138.
7 M.Bertholet, Comptes Rendes Academie des Sciences, 1881,93.
8 M.Bertholet, P.Vieille, Annales de Chimie at de Physique, 1883,5 th Seiers,
28,289.
9 M.Bertholet, Comptes Rendes Academie des Sciences, 1882,94.
10 H.Dixon, Philosophical TransitionsA, 1893,184.93-98.
11 'introduction to the Technology of Explosives', P.Cooper, S.Kurowski; VCH
1996.
12 'Numerical Modelling of Detonations', C.Mader; University Of California
Press, 1979.
13 'Detonation of Condensed Explosives', R.Cheret; Springer-Verlag, 1992.
14 'Detonation in Condensed Explosives', J.Taylor; Oxford University Press,
1952.
15 'Theory of Detonation', Z. Kompaneets;Academic Press, 1960.
16 'Combustion: a study in theory, fact and application', J.Chomiak; Abacus,
1990.
17 D.Chapman, Philosophical Magazine, 1889,47.
18 E.Jouguet, Comptes Rendes Academie des Sciences, 1904,140.
19 'On the Normal Ignition Velocity of Explosive Gaseous Mixtures',
V. Michelson; Moscow University Printing Service, 1893,10.
20 'Explosives and Rock Blasting,' R.Morhard; Atlas Powder, 1987.
21 Y.Zeldovich, Zh. Exp. / Teor. Fiz., 1940,10.
22 J.Neumann, 'Theory of Stationary DetonationWaves' Selected Works,
1942,6.

148
23 W. Doering, Annalen der Physik 1942 43.
24 'Explosive Effects and Applications', Editors: A.Zukasj, W.Walters; Springer-
Verlag, 1998.
25 'Initiation and Growth of Explosions in Liquids and Solids', F.Bowden,
Y.Yoffe; Cambridge University Press 1952.
26 J. Dear, J. Fluid Mech, 1988,190,409-425.
27, N.Bourne, J.Field, J. Fluid Mech, 1992,244,225-240.
28 N.Bourne, J.Field, Proc. R. Soc. Lond. A, 1991,435,423-435.
29 J. Dear, J. Field, A.Walton, Nature, 1988,332,505-508.
30 K.Feng, W.Chung, J.Yu, Symposium on Detonation, Maryland, 1981,343-
351.
31 J.Austing, A.Tullis, D.Brzycki, Symposium on Detonation, Maryland, 1981,
47-56.
32 J.Johnson, P.Tang, C.Forest, Journ. AppL Phys, 1985,. 57,4323-4334.
33 J. Field, G.Swallowe, S.Heavens, Proc. R. Soc. Lond. A, 1982,382,231-
244
34 J. Field, N.Bourne, S.Palmer, S.Walley, Phil. Trans. R. Soc. Lond.A, 1992,
339,269-283.
35 S.Walley, J. Field, S.Palmer, Proc.R.Soclond. A, 1992,.438,571-583.
36 N.Bourne, Proc. Ninth Symp. on Detonation, 1989,869-878.
,
37 J.Field Acc. Chem. Res, 1992,25,489-496.
38 P.Tang, J.Johnson, C.Forest Proc. Ninth Symp. on Detonation 1989,375-
384.
39 C.Mader, J.Kershner Proc. Ninth Symp. on Detonation 1989,366-374.
40 P.Taylor, Proc. Ninth Symp. on Detonation,1989,358-365.
41 'American Instiute of Physics Handbook', 2ndEdition.
42 Frank-Kamenetskii,Acta Physiochimica, 1939,10,365.
43 'Explosives Engineering' P.Cooper; Wiley-VCH, 1997.
44 M.Cook, 'The Science of High Explosives', Chapman & Hall London, 1958
(Reprint 1971).
45 D.Price, Proc.1Ith Symp (International) on Combustion, 1967,693-702.

149
46 'Detonics of High Explosives' C.Johansson, P. Persson, Academic Press,
1970.
47 'Explosives', R. Meyer; VCH, 1987.
48 'Emulsions and Foams', Berkman & Egloff; Reinhold, 1941.
49 'Introduction to Colloid & Surface Chemistry', D.Shaw; Butterworth-

Heinemann, 4th Edition, 1994.

50 'Emulsifiers Functionality and Applications', Editors: K.Berger & R.Hamilton;
Society of Chemical Industry, 1998.
51 'An Introductionto Rheology', H.Barnes,J.Hufton & K.Walters; Elsevier, 1989.
52 'Surfaces, Interfaces, and Colloids: Principles and Applications', M.Drew;
VCH, 1990.
53 'Basic Principles of Colloid Science', D.Everett; RSC Paperbacks,1988.
54 'Food Emulsions -What Are They?' Emulsifiers Functionality and
Applications', N.Krog; SCI, 1998.
55 Grete Hyldig: PhD Thesis 'Influence of Technological Parameters on the
Enzymatic Reaction and Gel Formation in Milk and LIF-Retentates, 1993,
KVL.
56 'Emulsions: Theory and Practice', Becher; Reinhold, 1957.
57 'Encyclopaedia of Emulsion Technology', Editor: P. Becher; Marcel-
Deker, 1983.
58 US Patent 3,161,551, R. Egly, Dec 1964.
59 US Patent 3,164,503, N. E.Gehrig, Jan 1965.
60 H. Dautriche, Comptes Rendes Academie des Science, 1906,143.
61 US Patent 4,110,134, H. F. Bluhm, June 1969.
62 US Patent, 4,110,134, Wade, 1978.
63 'Encyclopaedia of Emulsion Technology, H. Bampfield, J. Cooper; Marcel
Deckker, 1988.
64 'Explosives in the Service of Man' Editor: J. Dolan, S. Langer, Royal Society of
Chemistry, 1997.
65 P.Worsy, B.Weeks, J. Explos. Eng Des, 1997 18-21.
66 'Encyclopedia of Chemical Technology, Kirk-othmer;Wiley, 1991,2.
67 B.Luca, M.Ahtee, A. Hewat, Acta Cryst, 1979, B35,1038-1041.

150
 68 M.Ahtee, K.Smolander, B.Luca, A. Hewat Acta Cryst, 1983, C39,651-655.
69 'Encyclopaedia of Explosives' Picatinny Arsenal',B.Fedoroff, 1960 11.
70 'Differential Thermal Analysis', R.Mackenzie; Academic Press, 1973.
71 D.Tramoundanis, 'Phase Stabilisation Of Ammonium Nitrate' Msc Thesis
Nov.1988.
1 72 U.Teipel, Propellants, Explosives, Pyrotechnics, 1995,20,20-265.
73 O.Listh, Propellants and Explosives, 1978,3,36-41.
74 Y.Alymova, V.Annikova, N.Vlasova, B.Kondrikov, Combustion, Explosion and
Shock Waves, 1994,30,340-345.
75 L.Wang, J. of ThermalAnalysis, 1995,45,261-268.
76 A. Martin, D.Bastos-Netto, Propellants, Explosives, Pyrotechnics, 1999,24,
308-313.
77 V. Mohan, T.Tang, Propellants, Explosives, Pyrotechnics, 1984,9,30-39.
78 E.Lee, C.Tarver, Phys. Fluids, 1980,23,2362-2372.
79 'Analytical Chemistry Handbook', J. Dean, Mcgraw-Hill, 1995.
80 'Introduction to Thermal Analysis Techniques and Applications', M. Brown;
Chapman and Hall, 1988.
81 'Advanced Physics', M. Nelkon, M. Detheridge; Pan Books, 1982.
82 G. Shankar, A. Kumar, V. Mohan, Propellants, Explosives, Pyrotechnics, 1996,
21,70-73.
83 M. Kenndy, D. Kerr, Proc. Ninth Symp. on Detonation, 1989,1352-1359
84 J. Lee, P. Persson, Propellants, Explosives, Pyrotechnics, 1996,15 208-216.
85 K.Haftori,Y. Fukatsu, Journal of the Industrial Explosives Society Japan, 1982,
43,295-301.
86 'Explosives, Propellants and Pyrotechnics' A. Bailey, S.Murray; Brasseys, Vol
2,1989.
87 Private communicationswith Dr J.Cooper and G.Lieper.
88 'Differential Thermal Analysis', R.Mackenzie; Academic Press, 1970.
89 AN Trace Provided by Dr.Cartwright.
90 Private Communicationswith Dr Cartwright.
91 H.Eyring, R-E-Powell, G.H.Duffy, R.B.Parlin, Chemical Reviews, 1949,45,
69-172.

151
92 Private communicationswith Dr J.Cooper and G.Lieper.
93 Science of High Explosives' Editor:C.Bawn and G.Rotter; Her Majesty's
Stationary Office, 1956.
94 H.Jones, Proc. R. Soc. Lond. 1947,189,415-422.

95 J.Copp, A. R.Ubbelohde, Trans. Faraday Soc. 1948,44,646-658.

96 G.A. Leiper and J.Cooper, 28th ICT, Karisrhue, 1997.

152