Experimental fragmentation of pipe bombs

with varying case thickness

ARTICLE INFO
ABSTRACT
Article history:
Received
Among all the improvised explosive devices (IEDs) known, pipe bombs
Received in revised form
are one of the most popular devices used by terrorists. They are simple
Accepted
to use, easy to construct and materials are readily available. For this
Available online
IED, fragmentation is the primary injury mechanism, which makes them
a desirable weapon for terrorists aiming to inflict maximum human
Keywords:
casualties. Although the investigation of fragmentation pattern is not
Pipe Bombs
novel, there is limited data available on pipe bombs performance in the
forensic science
open literature. Therefore, this research is looking at validating results in
fragmentation
current literature, which showed limited repetition and weak
experimental design so far; by trial with six pipe bombs with two different
thickness (3 of each). The pipe bombs consisted of mild steel casing
and aluminised ammonium nitrate as the explosive filler. Fragments
were collected, with an average recovery of 72%, and measured
regarding mass and velocity. The experiment results show a correlation
between the pipe thickness and both the size and velocity of fragments.

(3871ºC) during an explosion and depending on how close

1 Introduction
victims and/or materials were to the explosion, these effects
may occur in the form of severe burning or charred and
Typically pipe bombs are improvised explosive devices
deformed materials [2]. The fourth outcome, the
(IEDs) which consist of a container, either plastic or metallic,
fragmentation effect, is the main injury mechanism of pipe
usually filled with low explosive and threaded with two end
bombs [5] and can be a result of either the rupture of the
caps to provide enough confinement for a detonation [1–5].
casing (primary fragments) or the materials accelerated by
Low explosives are known for having a velocity of explosion
the blast waves (secondary fragments) [5,7]. For the purpose
lower than the speed of sound, subsonic reactions and not
of this study, secondary fragments are not considered.
exhibiting a detonation wave. However, when confined within
an adequate container (i.e. a pipe bomb) and initiated by a
1.1 Post-blast investigation
detonator (e.g. a blasting cap), they can display a physical
behaviour similar to high explosives (HE) [2]. This
In post-blast scenes, from the fragmentation pattern
mechanism is commonly referred to as deflagration to
found from a detonated explosive device, a forensic
detonation transition [1].
investigator, with the help of a forensic laboratory, have to be
able to formulate a hypothesis regarding the explosive filler
Following a chemical explosion, there are four physical
and the initiation system used in the bombing. This analysis
effects: blast, ancillary, thermal, and fragmentation. The blast
is part of an investigative process that begins with an initial
effect is a result of the rapid expansion of gaseous detonation
response when various first responders (emergency services)
products, which creates a blast wave that travels at very high
arrive at the scene. After the first responder’s initial
speed, usually starting with a Mach number of about three
assessment, a forensic team is called upon to conduct a
and then decaying monotonically with to ambient speed of
more in-depth evaluation of the scene with aims to determine
sound (340 m/s) over large distances, and that increases the
whether it was an accidental or criminal event and document
ambient pressure to peak incident pressures [6]. The ancillary
its conclusion. After recording the scene through various
effects, also known as secondary blast pressure effects,
forms, evidence is identified, collected and sent to the lab.
relate to the interaction of the blast wave with water, soil and
After conducting a final survey to make sure all the
other objects. This leads to reflection and refraction of the
investigative steps have been done, the team release the
blast wave which in turn can cause blast focusing and higher
scene to a competent authority and the field investigation
than expected damage [2]. The thermal effect is a direct
starts. The field investigation is where questions such as
consequence of the heat produced by a chemical reaction.
what the explosive filler and the bomb material were, from
As the temperatures can reach or even exceed 7,000 ºF
where it came and to whom the bomb belonged to, are made
by the forensic team. To find those answers, an extensive velocities. The gases and the shockwave drive and compress
laboratory examination followed of further investigation is the shell wall, causing elastic strain. Once the stress
conducted [2]. actuating at the inner surface of the casing overcomes the
yield strength, the material undergoes through a plastic
According to the Global Terrorism Database (GTD), deformation expanding the area under stress towards the
created by the University of Maryland, between 1970 to 2016, outer face of the wall. The shell remains increasing its
there have been 2,456 reported incidents of pipe bomb usage diameter until the hoop stress reaches a level for which the
worldwide [8]. While this data is important to note, much of casing fails and break [20,21].
the data is extracted from worldwide news outlets, thus the
data quality may be limited. An alternative database is the The fracture process happens through crack formation
Explosives Incident Report (EIR) published annually by the and propagation and can undergo through two modes: ductile
United States Bomb Data Centre (USBDC), which is the or brittle fracture. The ductile fracture is represented by an
result of several forensic investigations on terrorist and extensive plastic deformation in the vicinity of an advancing
domestic attacks involving explosives in American soil [9–12]. crack, which means that the crack resists any further
extension unless more stress is applied. Whereas, the brittle
fracture is recognized when cracks can spread very quickly,
Vehicle 2 with very little plastic deformation, and once the crack
Unknown or N/A 29 formation process started, it will continue without requiring
Tubing 103 the increase of stress applied [22].
Sports Item 3
Pipe 161 Due to shear and radial fractures, the cracks begin
Other 76 outside of the shell and move toward the interior. On the
Grenade 3 other hand, the inherent materials flaws, such as voids,
Gas Cylinder 26 inclusions and weak grain boundaries, also generates cracks
Envelope 3 that moves from the interior towards the exterior at a 45º
End Cap/Plugs 126 angle. When both kinds of cracks meet, fragments are
CO2 Container 24 produced and the gases from detonation keep accelerating
Can 21 those fragments though the opens until they reach their
Box 23
maximum values [20]. Depending on the toughness, density
Bottle/Jug 288
and grain size of the casing material, the warhead breaks into
Bag 7
fragments with different shapes and sizes.
0 100 200 300 400
Over the years, several theories investigating the
fragmentation process were conceived, however, two notable
Figure 1 – Number of explosives incidents reported for each mathematical models are the Gurney [23] and Mott [19]
container type in EIR between 2014 and 2017 by USBDC. models. In 1943, Gurney developed a mathematical model
Unknown or N/A is applied when either a container was unknown
used to predict the rule of dispersion of fragments driven by
at the time of entry or the device found had no container.
HE in a variety of casing geometries [23]. In his approach,
According to EIR (Fig. 1), between 2014 and 2017, when the warhead is on-axis-initiated, the fragments velocity
among all the containers used in attacks in America, pipe of a cylindrical bomb with no end caps, behave as follow:
bombs took the second place with 161 events out of 895 √2
=
incidents. As these devices are easy to use, simple to build 1 (1)
+
and their materials are easy to obtain, they have become 2
quite popular among terrorist groups over the years. This
ease of acquisition is represented by a percentage increase where v – fragment velocity, √2 – Gurney constant
in the use of pipe bombs in 2014 (9% of 137 events), 2015 (depends only on the explosive), M – casing mass and C –
(20% of 254 events) and 2016 (25% of 240 events) [9–11]. explosive mass.
Even though those numbers declined in 2017 (14% of 264
events) [12], in each of these events, fragmentation was the However, in the case of pipe bomb events, HE are not
primary cause of injury and thus remaining an important area commonly used, thus the suitability of this model to predict
of study [5]. To date, fragmentation and its effects have been fragment velocities from these events may be limited.
extensively studied.
In 1947, Mott [19] developed a theory for predicting the
length of the average fragment of cylindrical “ring-bombs”
1.2 Previous Work caused by radial fracture. In this theory, two assumptions
were made: the casing wall consists of a series of
Fragmentation is a process that happens when a material
axisymmetric circular rings, with the same inner and outer
undergoes multiple fractures when subjected to stress. The
diameter, stacked one above the other with a thin steel wall
stress can be due to static forces, such as grinding and
and once the fracture begins, it happens so fast that it can be
crushing, or dynamic loadings, such as the impact of a
considered instantaneous relative to the rate of strain of the
projectile into a target or the expansion of gaseous products
metal. As an outcome of their work, they derived an equation
from a detonation [13]. Extensive work has been done on the
determining that the mean fragment size depends on the
formation of fragmentation of warheads, which are usually
radius and velocity of the ring at the moment of rupture and
build from a suitable steel casing and filled with HE [13–19].
other mechanical properties of the casing.
When a detonation occurs, the expansion of gaseous
products creates a shockwave that travels at very high
 fillers, initiation system, pipe material and size, device
1+
∝ (2) orientation, as well as, the degree of volume filled with
explosives. Among all the results, two notable findings
deserve highlighting. First, they found that the fragmentation
Where x – average fragment length [m], F – true fracture behaviour could be reproducible among pipe bombs set up
stress [MPa], F – fracture strain based on reduction of area, under the same explosive conditions and thus, it would be
 – density of casing material [kg.m-3], 2 – coefficient of possible to identify the energetic filler used based on the pipe
stress strain curve for large strains [MPa], r – radius of ring at fragmentation found [25]. Secondly, as a way to quantify
break-up [m] and v – outward velocity of ring at break-up IEDs effects, they produced a numerical evaluator (Equation
[m/s]. 7), called Fragment Weight Distribution Mapping (FWDM),
which provides a method to assess this distribution by being
Elek and Jaramaz noticed that the mechanical properties relatively insensitive to the percent of recovery of fragments.
of real materials, such as fracture stress and strain, depend
on heterogeneities (i.e., micro flaws) present in the material (7)
log =
[13]. Therefore, modelling the fragmentation of an axis-
symmetric casing, which is the case of a pipe bomb, requires
Where Mx – mass of a particular fragment plus all fragments
the imposition of some microscopic flaws which is where
heavier than it, Mr – mass of all the recovery fragments, m –
failures and cracking initiate. By randomizing the failure
slope and mx – mass of a particular fragment.
stress or strain of the material, each element will have a
different failure stress or strain, thus creating weaker spots in
Several studies have since been undertaken to generate
the material. Based on this approach, known as stochastic
additional post-blast signatures. To better evaluate
model, Mott derived a distribution to determine the variance
fragmentation behaviour of pipe bombs, studies have been
in failure stress or strain [24].
conducted focusing on mass and velocity distributions, blast
. overpressure, fragment thrown distance and how different
( )= exp − ; = ; = (3)
2 2 temperatures may affect the results [4,26,27]. In these
studies, the explosive filler selected was either black powder,
where is: N(m) – the cumulative number of fragments or total double base smokeless powder (Alliant Bullseye, Alliant Red
number of fragments with the mass greater than m, M0 – the Dot, Alliant Herco, Alliant 2400, Alliant Reloader 22, WC 870,
total mass of fragments, N0 – the total fragment number and Winchester Action Pistol or Hodgdon Pyrodex), single-base
– the average fragment mass. The elemental premise in smokeless powder (IMR-PB or IMR-4227), or a high
Mott’s work is that the probability of an unfractured specimen explosive (e.g., C4, TNT or nitromethane) to be used as
of unit length fracturing when the strain increases from ε to reference [4,25,26].
ε+dε has the following form:
Ammonium nitrate (AN) is an important
(4) ingredient in the manufacture of fertilizers and commercial
explosives. The aspects that make this product so desirable
in the industry are: low cost, easy access and its role as
where parameters C and γ represent the characteristics of
oxidant. In 2016, according to a Grand View Research report,
material and they can be calculated by the following
the global ammonium nitrate market size was estimated at
equations:
USD 4.67 billion [33]. For many years, AN alone was not
( . ) considered an explosive. However, since the beginning of its
≈ (
;
)
≈ (5)
mass production, several accidents involving AN have been
reported (Oppau, 1921; Texas City, 1947; Toulouse, 2001;
where  – stochastic variance, p – strain-hardening, σF – Texas, 2013).
fracture stress, F – fracture strain and C is a material
parameter. According to the previous model, following Mott’s Due to the large number of accidents involving AN and
distribution law, the fracture probability is determined for each the fact that this product has become the weapon of choice
element and takes the following form: for many terrorist groups, Kirk Yeager, an FBI forensic expert,
carried out tests to study the detonability of fertilizer grade
(6) ammonium nitrate (FGAN) with a booster [2]. In his report,
=1− − [exp( ) − 1]
the test results showed that FGAN prills alone (without a fuel)
could be detonable and once initiated, could endure a stable
Even though the investigation of fragmentation patterns
detonation without requiring a high external confinement [2].
and velocities is not new, there is a need to study how these
However, it should be noted that this result did not imply that
fragments behave in improvised explosive devices, such as
any amount of AN was detonatable. A critical diameter, which
pipe bombs. During the past decade, researchers have been
is the minimum diameter of the explosive in which a steady
trying to characterize IEDs from measuring the mass and
detonation may still occur, is necessary so that the detonation
velocity distributions of container fragments or microstructure
of the booster can be transmitted to the product. Although
deformation and hardness, as a way to correlate explosive
this information was not given by Yeager in his report [2],
properties with material response [3,4,25–27].
Zygmunt [34] undertook some studies on detonation
parameters of mixture containing ammonium nitrate and
To study fragmentation pattern of pipe bombs, it is
aluminium, known as ammonals and hydro-ammonals, and
necessary to carry out a series of controlled experiments to
noticed that some additives, such as aluminium, TNT, nitro
evaluate how certain parameters can affect the fragmentation
glycerine, and others, could improve the AN detonation ability
effect. In 2001, Oxley et al. [25] analysed the fragments from
by reducing the critical diameter and increasing the
56 pipes bombs. The parameters assessed were: energetic
detonation velocity. It was concluded that the addition of 1%
of aluminium to AN could give a critical diameter below 20 as the explosive filler as literature indicates that aluminium is
mm, while the same value could only be seen with the a better sensitizer to AN, in comparison with TNT and nitro
addition of approximately 20% of TNT. glycerine, and explosive properties, such as critical diameter
and detonation velocity of ammonals depend on the
Overall, pipe bombs remain a significant modern threat aluminium content of those mixtures [34].
and a weapon of choice for both domestic and international
terrorism. While studies have indicated that their 2.1 Experimental Protocol
effectiveness is affected by parameters such as material,
explosive, initiator, etc. work has yet to be undertaken to Six pipe bombs were manufactured from seamless mild
investigate the effects of pipe wall thickness. steel S235J tube supplied by Precision Profiles, UK. This
same metal was also used for end caps (top and bottom).
1.3 Aims and Objectives Pipe-bomb dimensions were 38 mm by 229 mm (3), and 34
mm by 229 mm (3). Each device had one of its caps drilled
The aim of this study is to explore the influence of wall (7.5 mm hole) to allow the insertion of an electric blasting cap
thickness on fragmentation behaviour experimentally, in order #8 (1g of PETN). As shown in Fig. 2, every pipe had a
to assess the suitability of such models in the aid of forensic constant inner diameter of 28 mm whilst the outer diameter
investigation of post blast events. was either 34 mm or 38 mm (threads excluded). The thread
region in every device was filled with Styrofoam to assure
To achieve this, in-house manufactured pipe bombs were that the explosive filler (110 g of AAN) remained only in the
designed and tested with two different wall thickness. non-thread region. The styrofoam in the top-end cap had a
2 Materials and Methods groove to accommodate a 5g of booster (64% of PETN and
36% of nitrocellulose). To help differentiation among
This project assessed fragmentation pattern of metallic fragments within the building, all devices were painted in
pipe bombs. For the experiments, the pipes were filled with different colours (Fig. 3).
aluminized ammonium nitrate (AAN) – mass ratio of 9:1 AN
to aluminium – at a density of 1.03 g.cm-3. AAN was selected

Figure 2 - Diagrammatic representation of Pipe Bomb construction for all test

samples with an outer diameter of X being either 34 or 38 mm.
Figure 3 - Pipe bomb samples (#1, #2, #3, #4, #5 and #6) from left to right (left) and an illustration of pipe #1 in the
arena.

A total of six devices were fired. The experiment was

carried out within the containment building at Explosives
Ranges on the Explosives Research and Demonstration
Area, located in the Defence Academy, in Shrivenham, UK.

Due to financial limitations, the detonation area was

prepared (Fig. 4) by placing only two witness plates
perpendicularly to each other to capture pipe fragments.
Each witness plate was equipped with recovery packs,
consisting of 20 cardboard sheets (dimensions of 1015 mm
height x 760 mm width x 3.8 mm depth per sheet) supplied by
Limehouse Board Mills, UK. Each pack was labelled and its
position recorded. For each detonation a pipe bomb was
placed in between the witness plates (1000 mm apart) with Figure 4 – Experimental set up for capturing fragments
the bottom-end cap facing the first witness plate. The pipe and record their velocities.
bomb was suspended at a height of 500 mm (Fig. 4) and
initiated from the “top” end cap. After each firing, the fragments were collected from the
environment (ground, walls and ceilings) as well as extracted
The fragments velocity was measured with the aid of four from recovery packs with the help of pliers and magnets. Post
copper foils acting as electrical conductive surfaces: the first blast fragment masses were determined using a Mettler PE
and second were positioned in front of the recovery packs, 16 analytical balance with a 0.1 g precision. Fragments with
while the third was 10 mm away from the bottom-end cap and mass smaller than 0.1g were counted, but the mass was
the fourth was attached to the “top” end cap. The third and assumed as zero.
the fourth positions were used to trigger the software-
3 Results and Discussion
controlled logic analyser, whereas the first and second
positions were used to measure the arrival time of fragments
3.1 Experimental Results
from the pipe walls and the endcaps.
Fig. 7 shows the fragments profile produced by three
Every copper foil is an open circuit, so whenever
samples with 3 mm of pipe wall thickness (top) and three
fragments hit these foils, they close the circuit and are
samples with 5 mm of pipe wall thickness (bottom). Even
recorded digitally. The comparison between the arrival time of
though the explosive filler was the same for all devices, the
fragments and the initial time recorded by the closest foils,
differences can be noticed by fragments size, velocity and
combined with the distance between the foils, is supposed to
mass distribution (Fig. 7). For 38 mm diameter x 229 mm
allow the determination of fragments velocities when they hit
length mild steel seamless pipes, the number of fragments
the panels. However, a limitation of this approach is: if by any
collected were 61, 75 and 68, whereas for 34 mm diameter x
chance, a fragment hit the circuit and remains in contact with
229 mm length pipe bombs, the number of fragments
it, the copper foil apparatus will only be able to record the
gathered were 70, 56 and 66. Every fragment was weighed
speed of the fastest fragment.
individually, however, to preserve the pattern of
fragmentation, pipe thread pieces that became attached to
the end caps after the detonation, were kept and weighed
together. For the six samples, the percent recovery ranged
from 65 to 76%, averaging at 72.04 ± 4.23%. Table 3 shows
the pipe and fragment weight, the percent recovery and the
fastest fragment velocity for each pipe bomb.
 Figure 5 - Fragments of seamless steel pipes filled with Aluminized Ammonium Nitrate
categorized as bottom end cap (top line), pipe body (middle) and top end cap (bottom line),
where the detonation took place. Three samples (pipe #1, #2 and #3) of 38 mm diameter x 229
mm length (top) and three samples (pipe #4, #5 and #6) of 34 mm diameter x 229 mm length
(bottom).

An attempt was made to organize the fragments collected It was also noticed that the pipe thread attached to the
according to the initial pipe bomb parts: top end cap (where top end cap of pipe #4 failed and opened completely. It was
the devices were initiated), pipe body and bottom end cap. In suspected that this behaviour was due to intrinsic material
five samples, it was noticed that the bottom end cap defects in the metal casing.
remained attached to the pipe thread. This showed that there
is a certain degree of reproducibility of fragmentation among
these samples due to the same explosion condition.

Table 1 – Pipe bomb parameters and results from the experiments.

Fragment weight

Fastest Fragment Velocity

% Recovery
Pipe weight

Fragments
Number of
Pipe size
Pipe #

End Cap
Pipe

1 1093.1 g 755.2 g 69% 61 N/D* N/D*

38 mm x
229 mm

2 1087.8 g 831.5 g 76% 75 967 m/s 517 m/s

3 1084.3 g 824.8 g 76% 68 N/D* N/D*

mm x

mm
229

4 1392.2 g 899.8 g 65% 70 854 m/s 605 m/s

34
 5 1391.6 g 1040.3 g 75% 56 805 m/s 630 m/s

6 1393.1 g 992.5 g 71% 66 807 m/s 614 m/s

*N/D – non-detectable.
thin pipes showed that 89% of the number of fragments have
As expected, thick pipes produced fragments with bigger their masses ranging from 0 to 15 g; whereas for the thick
sizes, thus heavier, than the thin ones. Hence, as displayed devices, the same 89% of the number of fragments was
by Figs. 8 and 9, histograms with cumulative relative achieved with masses ranging from 0 to 20 g. Therefore, we
frequencies distribution were used to represent fragments may assume that thickness may affect the fragmentation
weight distribution and to allow comparison between the thin behaviour. It should be noted that this result is in agreement
and the thick pipes. with Mott's equation for fragment size, under the assumptions
of using the same casing gives the same mechanical
For the thin pipes, the pipe wall fragments ranged from 0 properties, thus only the radius and the velocities at break-up
to 15 g, while the end-cap fragments ranged from 5 to 300 g. are varying. Based on this, thin walls give smaller fragments,
For the thick devices, the pipe wall fragments ranged from 0 while thick walls give larger ones [19].
to 35 g, while the end-cap fragments ranged from 5 g to 300
g. The cumulative relative frequency distribution for the three

35 100%

Cumulative Relative Frequency

90%
30
80%
25 70%
Frequency

20 60%
50%
15 40%
10 30%
20%
5
10%
0 0%

Fragments Weight (g)

Pipe #1 Pipe #2 Pipe #3 Cumulative Relative Frequency Distribution
Figure 6 - Histogram of fragment weight distribution of mild steel devices filled with AAN with 3 mm of
pipe wall thickness.

30 100%
Cumulative Relative Frequency

90%
25 80%
20 70%
Frequency

60%
15 50%
40%
10 30%
5 20%
10%
0 0%

Fragments Weight (g)

Pipe #4 Pipe #5 Pipe #6 Cumulative Relative Frequency Distribution

Figure 7 - Histogram of fragment weight distribution of mild steel devices filled with AAN with 5 mm of
pipe wall thickness.

This happened because, after pipe bombs (initiated on-axis)

From Table 3, we can observe that the velocities of the had overcome the metal case strength, their fragments
pipe wall fragments are higher than the end-cap fragments. dispersed uniformly around the axis and since threaded parts
of the end caps were heavier and thicker than pipe walls, lead to lower fragment velocities. Therefore, due to the fact
they travelled slower. Moreover, the velocity data for pipe #2 that thick pipes are heavier, the pipe wall fragments velocities
showed that pipe wall fragments from thin pipes travelled at will be lower.
higher speeds relative to the fragments generated by the
thick pipes, while the opposite behaviour was observed for An attempt was made to verify if the Gurney equation
end-cap fragments. As discussed, the copper foil apparatus would be applicable to the experiments. By looking at the
only measured the speed of the fastest fragment. After the original Gurney equation (equation 1), it would be reasonable
first fragment hit the copper foil, the circuit was closed and it to assume that the initial velocities are independent of the
didn’t record any more hit. Due to human error prediction, for geometry size. However, Weinland [40] observed that real
pipe #1 and #3, the time-base for the data capture was set to experiments were affected by it. In his work, he studied the
be faster than the actual phenomena, therefore the software effect of scaling by assessing the ratio of length over
stopped recording before the fragments had arrived at the diameter (L/D) for cylinders and he found out that if L/D ≥ 6,
witness plates and their velocities were not registered. the Gurney constant is not affected by scaling. Since the
outer diameter of the pipe bombs used in this project were
As a way to explain the fragment velocities behaviour, the either 34 or 38 mm, the ratio L/D was equal or above 6, thus
Gurney model for cylindrical bombs was discussed [23]. In it was assumed that the Gurney constant shouldn’t be
this study, the explosive and its mass were kept the same for affected by anything other than the casing mass. The values
all devices, thus the only parameter changing was the casing found were compared to the Gurney constant obtained from
mass. From this equation, we may assume heavier casings the literature for AAN mixtures with 10% of aluminium [41].

Table 2 – Calculated Gurney constants.

Pipe # Casing Mass (M) Explosive Mass (C) Fragment Velocity Gurney Constant (√2 ) Gurney Constant [41]

1 604.0 g 110 g N/D* -

2 598.8 g 110 g 967 m/s 2.36 km/s

3 595.4 g 110 g N/D* -

1.78 – 1.98 km/s
4 903.3 g 110 g 854 m/s 2.52 km/s

5 902.4 g 110 g 805 m/s 2.37 km/s

6 903.8 g 110 g 807 m/s 2.38 km/s

Using the fragment velocities displayed at Table 3, 4 Conclusion

Gurney velocities were calculated and shown in Table 4.
Since Gurney Overall, the experiments showed that the thicker the pipe,
the larger the fragments produced and that the fragments
model was developed for open cylinders, only the pipe weight velocities obtained from the detonations followed the Gurney
was used for calculating the ratio M/C. Out of four pipes, model for initial fragment velocities. The impact of endcaps
three (pipes #2, #5 and #6) were proven to best fit the model, on the pipe bomb and resultant variation in fragment velocity,
showing calculated values ranging from 2.36 to 2.38 km/s. and the potential for Al to cause higher velocities were
Thus, we may assume that, under the experimental interesting findings that require more experimentation to
conditions set up in this project, the Gurney model could explain.
predict fragment velocities. However, also note that all the
5 Further Work
values calculated were higher than the published range of
1.78 and 1.98 km/s. It was assumed that this happened
Regarding the experiments, due to financial limitations, it
because this study used an AAN with a higher density (1.03
was not possible to set up the experiment with a 360º
g.cm-3) than the density reported (0.953 g.cm-3) by the
approach and many fragments were lost within the
literature, which increased the velocity of detonation, thus
containment building. Hence, this approach could not only
making the conversion of mechanical energy from the
increase the percentage of recovery but also be an
detonation into fragment velocities more efficient.
alternative way of measuring fragment velocities by
assessing the penetration on the recovery packs. Since pipe
Gurney approach for cylindrical bombs didn’t include end
bombs are very common due to materials being readily
caps, thus it could not explain the end-cap fragments
available, we recommend that further studies using regularly
behaviour. However, considering that thick pipes take longer
available metal tubes be conducted. Moreover, a
time to expand than thin pipes, a more intense build-up of
metallography study should be done pre and post experiment
pressure inside the pipe is expected to happen before
to evaluate the effects of flaws in the metal casing.
fracture. With high pressures acting on the inner area of the
cylinder, there will be a higher initial acceleration, which gives
higher initial velocities to the end-cap fragments. Besides,
since part of the kinetic energy was lost in pulling the pipe
apart, thicker pipes posed more resistance to fragmentation,
resulting in lower pipe wall velocities.
References [17] I.G. Cullis, P. Dunsmore, A. Harrison, I. Lewtas, R.
Townsley, Numerical simulation of the natural
[1] B. Mohanty, Forensic Investigation of Explosions, fragmentation of explosively loaded thick walled
2nd ed., CRC Press, 1998. cylinders, Def. Technol. 10 (2014) 198–210.
doi:10.1017/CBO9781107415324.004. doi:10.1016/j.dt.2014.06.003.
[2] J.T. Thurman, Practical Bomb Scene Investigation, [18] H. Grisaro, A.N. Dancygier, Numerical study of
3rd ed., CRC Press, 2017. velocity distribution of fragments caused by
[3] O. Gregory, J. Oxley, J. Smith, M. Platek, H. explosion of a cylindrical cased charge, Int. J.
Ghonem, E. Bernier, M. Downey, C. Cumminskey, Impact Eng. 86 (2015) 1–12.
Microstructural characterization of pipe bomb doi:10.1016/j.ijimpeng.2015.06.024.
fragments, Mater. Charact. 61 (2010) 347–354. [19] N.F. Mott, Fragmentation of shell cases, Proc. R.
doi:10.1016/j.matchar.2009.12.017. Soc. A Math. Phys. Eng. Sci. 189 (1947) 300–308.
[4] D. Bors, J. Cummins, J. Goodpaster, The Anatomy doi:10.1098/rspa.1947.0042.
of a Pipe Bomb Explosion: Measuring the Mass and [20] A. Rotariu, E. Trana, Modelling and simulation for
Velocity Distributions of Container Fragments, J. ballistic protection, in: Proc. Greener Safer Energ.
Forensic Sci. 59 (2014) 42–51. doi:10.1111/1556- Ballist. Syst. Conf., Bucharest, Romania, 2016: p.
4029.12294. 27.
[5] A.J. Gibbons, J.N. Farrier, S.J. Key, The pipe bomb: [21] E. Trană, A. Rotariu, F. Bucur, Numerical Simulation
a modern terrorist weapon., J. R. Army Med. Corps. Study on the Ring, MTA Rev. 2 (2015) 179–188.
149 (2003) 23–26. doi:10.1136/jramc-149-01-04. [22] H.A. Richard, M. Sander, Fatigue Crack Growth:
[6] J.M. Dewey, Experimental Methods of Shock Wave Detect-Assess-Avoid, 2016.
Research, in: Exp. Methods Shock Wave Res., [23] R. W. Gurney, The Initial Velocities of Fragments
Springer, Cham, 2016: pp. 53–86. doi:10.1007/978- from Bombs, Shell, Grenades, Aberdeen, Maryland,
3-319-23745-9. 1943.
[7] S. Lichorobiec, V. Kavicky, L. Figuli, Comprehensive [24] N.F. Mott, E.H. Linfoot, D. Grady, A Theory of
Assessment of Potential,Threats to All Kinds of Fragmentation, in: Fragm. Rings Shells - Leg. N.F.
Events Arising from the Explosion of Pipe Bomb, Mott, 1st ed., Springer-Verlag Berlin Heidelberg,
Key Eng. Mater. 755 (2017) 219–228. 2006: pp. 207–225. doi:10.1007/978-3-540-27145-
doi:10.4028/www.scientific.net/KEM.755.219. 1_9.
[8] Global Terrorism Database, GTD Search Results for [25] J.C. Oxley, D. Ph, J.L. Smith, E. Resende, E.
terrorist incidents in the US, 1970-2016, Natl. Rogers, R.O. Jc, S. Jl, E. Resende, E. Rogers,
Consort. Study Terror. Response to Terror. (2017). Improvised Explosive Devices : Pipe Bombs *, J.
[9] United States Bomb Data Center, United States Forensic Sci. 46 (2001) 510–534.
bomb data center (USBDC) Explosives Incident doi:10.1520/JFS15000J.
report (EIR) 2015, (2015). https://www.atf.gov/rules- [26] J.C. Oxley, J.L. Smith, E.T. Bernier, F. Sandstrom,
and- G.G. Weiss, G.W. Recht, D. Schatzer,
regulations/docs/report/2015usbdcexplosiveincidentr Characterizing the Performance of Pipe Bombs, J.
eportpdf/download. Forensic Sci. 63 (2018) 86–101. doi:10.1111/1556-
[10] United States Bomb Data Center, United States 4029.13524.
bomb data center (USBDC) Explosives Incident [27] D. Bors, J. Cummins, J. Goodpaster, The anatomy
report (EIR) 2014, (2014). of a pipe bomb explosion: The effect of explosive
https://www.atf.gov/explosives/docs/report/2014- filler, container material and ambient temperature on
usbdc-explosive-incident-report/download. device fragmentation, Forensic Sci. Int. 234 (2014)
[11] United States Bomb Data Center, United States 95–102. doi:10.1016/j.forsciint.2013.10.017.
bomb data center (USBDC) Explosives Incident [28] A. Gullerud, J. Hollenshead, Coupled Euler-
report (EIR) 2016, (2016). Lagrange simulations of metal fragmentation in pipe
https://www.atf.gov/explosives/docs/report/2016- bomb configurations, Procedia Eng. 204 (2017)
explosives-incident-report/download. 202–207. doi:10.1016/j.proeng.2017.09.774.
[12] United States Bomb Data Center, United States [29] X. Quan, B. Gerber, M. Cowler, N. Birnbaum,
bomb data center (USBDC) Explosives Incident Simulation of Blast and Fragment Loading Using a
report (EIR) 2017, (2017). Coupled Multi-Solver Approach, Trans. Tianjin Univ.
https://www.atf.gov/resource- 12 (2006) 182–187.
center/docs/report/2017-explosives-incident-report- [30] G.E. Fairlie, The Numerical Simulation of High
eir/download. Explosives using AUTODYN-2D & 3D, Inst. Explos.
[13] P. Elek, S. Jaramaz, Modeling of Fragmentation of Eng. 4th Biannu. Symp. (1998) 13.
Rapidly Expanding Cylinders, Theor. Appl. Mech. 32 [31] Y. Li, Y. hua Li, Y. quan Wen, Radial distribution of
(2005) 113–130. fragment velocity of asymmetrically initiated
http://www.doiserbia.nb.rs/Article.aspx?ID=1450- warhead, Int. J. Impact Eng. 99 (2017) 39–47.
55840502113E%5Cnpapers3://publication/doi/10.22 doi:10.1016/j.ijimpeng.2016.09.007.
98/TAM0502113E. [32] W. Arnold, E. Rottenkolber, Fragment mass
[14] G.Y. Huang, W. Li, S.S. Feng, Axial distribution of distribution of metal cased explosive charges, Int. J.
Fragment Velocities from cylindrical casing under Impact Eng. 35 (2008) 1393–1398.
explosive loading, Int. J. Impact Eng. 76 (2015) 20– doi:10.1016/j.ijimpeng.2008.07.049.
27. doi:10.1016/j.ijimpeng.2014.08.007. [33] Grand View Research, Ammonium Nitrate Market
[15] M. Ugrčić, M. Ivanišević, Characterization of the Analysis And Segment Forecasts To 2025, USA,
Natural Fragmentation of Explosive Ordnance Using 2017. https://www.grandviewresearch.com/industry-
the Numerical Techniques Based on the FEM, Sci. analysis/ammonium-nitrate-market.
Tech. Rev. 65 (2015) 14–27. [34] B. Zygmunt, Detonation parameters of mixtures
doi:10.5937/STR1504016U. containing ammonium nitrate and aluminium, Cent.
[16] M. Ugrčić, Numerical Simulation of the Eur. J. Energ. Mater. 6 (2009) 57–66.
Fragmentation Process of High Explosive [35] ANSYS Inc., ANSYS AUTODYN User’s Manual,
Projectiles, Sci. Tech. Rev. 63 (2013) 47–57. 15317 (2013) 1–502.
http://www.vti.mod.gov.rs/ntp/rad2013/2-13/3/3.pdf. [36] B. M Dobratz, P. C Crawford, LLNL Explosives

9
 Handbook Properties of Chemical Explosives and
Explosive Simulants, 1985.
[37] P. Sherkar, A.S. Whittaker, A.J. Aref, Modeling the
Effects of Detonations of High Explosives to Inform
Blast-Resistant Design, MCEER Tech. Reports. 10
(2010) 188. doi:10.13140/RG.2.1.3978.4565.
[38] M.-H. Yu, J.-C. Li, 3D Simulation of Normal and
Oblique Penetration and Perforation, in: Comput.
Plast. With Emphas. Appl. Unified Strength Theory,
Springer Berlin Heidelberg, Berlin, Heidelberg, 2012:
pp. 321–331. doi:10.1007/978-3-642-24590-9_13.
[39] G.R. Johnson, W.H. Cook, Fracture characteristics
of three metals subjected to various strains, strain
rates, temperatures and pressures, Eng. Fract.
Mech. 21 (1985) 31–48. doi:10.1016/0013-
7944(85)90052-9.
[40] C.E. Weinland, A Scaling Law for Fragmenting
Cylindrical Warheads, (1969).
[41] D.L. Robbins, E.K. Anderson, M.U. Anderson, S.I.
Jackson, Cylinder Test Characterization of an
Ammonium Nitrate and Aluminum Powder
Explosive, 15th Int. Symp. Detonation. (2014).
http://public.lanl.gov/sjackson/papers/2015-
AndersonAmmonal.pdf.
[42] E. Galuta, W. Regig, Numerical Simulations of RC
Panels Subjected to High Speed Projectile - Erosion
Selection in AUTODYN-3D code, 4 (2017) 25–30.
http://ijiset.com/vol4/v4s8/IJISET_V4_I08_03.pdf.
[43] J.M. McGlaun, S.L. Thompson, M.G. Elrick, CTH: A
three-dimensional shock wave physics code, Int. J.
Impact Eng. 10 (1990) 351–360. doi:10.1016/0734-
743X(90)90071-3.
[44] V.M. Gold, Fragmentation model for large L/D
(Length over Diameter) explosive fragmentation
warheads, Def. Technol. 13 (2017) 300–309.
doi:10.1016/j.dt.2017.05.007.

10
Appendix
Pipe bomb design

Figure A-1 – Pipe and thread design.

A-1