Explosive Loading of Engineering Structures - P S Bulson, 1997
Explosive Loading of Engineering Structures - P S Bulson, 1997
Explosive Loading of Engineering Structures - P S Bulson, 1997
ENGINEERING STRUCTURES
Earthquake Engineering
Y-X.Hu, S-C.Liu and W.Dong
For more information about these and other titles please contact:
The Marketing Department, E & FN Spon, 2–6 Boundary Row, London,
SE1 8HN. Tel: 0171–865 0066
P.S.Bulson
Consultant to the Mott MacDonald Group
Visiting Professor of Civil Engineering, University of Southampton
E & FN SPON
An Imprint of Chapman & Hall
London · Weinheim · New York · Tokyo · Melbourne · Madras
Chapman & Hall USA, 115 Fifth Avenue, New York, NY 10003, USA
Chapman & Hall Japan, ITP-Japan, Kyowa Building, 3F, 2–2–1 Hirakawacho,
Chiyoda-ku, Tokyo 102, Japan
Chapman & Hall Australia, 102 Dodds Street, South Melbourne, Victoria 3205,
Australia
Chapman & Hall India, R.Seshadri, 32 Second Main Road, CIT East, Madras
600 035, India
Apart from any fair dealing for the purposes of research or private study, or
criticism or review, as permitted under the UK Copyright Designs and Patents
Act, 1988, this publication may not be reproduced, stored, or transmitted, in any
form or by any means, without the prior permission in writing of the
publishers, or in the case of reprographic reproduction only in accordance with
the terms of the licences issued by the Copyright Licensing Agency in the UK,
or in accordance with the terms of licences issued by the appropriate
Reproduction Rights Organization outside the UK. Enquiries concerning
reproduction outside the terms stated here should be sent to the publishers at
the London address printed on this page.
The publisher makes no representation, express or implied, with regard to the
accuracy of the information contained in this book and cannot accept any legal
responsibility or liability for any errors or omissions that may be made.
A catalogue record for this book is available from the British Library
Preface
Acknowledgements
Notation
Introduction
A long time ago I walked over a snow-covered marsh to place explosives near
unsafe grenades; later cutting charges were set on steel joists under a desert sun,
and in another part of the world camouflets were bored in the rock of monsoon-
lashed mountains. Tables were used in military handbooks to calculate charge
weights, and instructions were read from the same books on the safety of handling
and storage. We never asked who calculated the figures or drew up the precautions,
or where the history of the subject could be found. We were told to concentrate,
keep cool and make sure that we avoided blowing up the explosive store, the
headquarters bunker and above all, the Colonel. The source or level of accuracy
of our technical information was not questioned by us.
The world of explosive technology and the science of explosive loading has
come a long way since those days, but it was always a secret world. The
development of military weapons, the response of civil and military structures,
and the legal overtones of great disasters involving dust, vapour and gas explosions
all require circumspect behaviour by those in the know. Scientists and engineers
conduct target response tests in nether regions or behind high fences. Analysis
and simulation are hidden in a parallel world of security passes and personal
vetting. Much research is subject to restrictions on open publication, and the
history of research and development is not always available to newcomers in
the field. Nevertheless, whether we are concerned with gas, vapour, dust, plastic
explosive, Semtex or nuclear warheads, the need to be aware of history and to
use a sense of history in judging the way forward is still desirable.
This book is an attempt to review developments over the years in methods
of calculation, measurement and prediction of the dynamic loading on structures
from explosions. It endeavours to trace the history of the subject and to
summarize some of the latest published thinking, and considers finally a range
of structures from buildings and bridges to ships and aircraft. It is not concerned
with the design of protective structures or the hardening of existing structures,
but there are brief passages on the analysis of response, residual strength and
safety. The work and achievements of major contributors to the subject over
the past century are discussed, because it is important to realize that the subject
has been taken forward by a collection of mathematicians, physicists, chemists,
military researchers and manufacturing experts as well as by engineering
scientists specializing in the interaction of loads and structures.
The fundamental science of explosive loading has been relatively slow to
develop because the level of independent academic investigation of the subject
in universities and similar institutions has been less than some would have
wished. Apart from the cost of facilities and the difficulty of ensuring safety,
many authorities supporting research with a military flavour have been the
subject of attention from pacifist groups. These groups, of course, never have
to grasp the nettle of responsibility for the protection of our population, our
heritage and our way of life from destruction by explosion or threat by terrorism.
It is therefore hoped that this survey will help to increase interest in the subject
by the academic workers in structural behaviour.
In the preparation of this book, I have received much administrative help
from the Mott MacDonald Group, and benefited from continuing contacts
with the Defence Research Agency and one or two universities. I am indebted
to two word-processor experts, Miss Rosemary Mardles and Mrs Sue Starks,
who have given legibility to the script, and to the editorial staff of Chapman
& Hall, who waited patiently for the final draft.
P.S.Bulson
Winchester
1997
Note: Because this is a work with many references to history, readers are
asked to accept a mixture of fps and SI units in the text. A conversion table is
provided at the end of the notation section on p. xv.
All symbols are defined in the text where they first appear. The symbols listed
below are those that appear repeatedly, or are of greatest interest.
m mass of projectile
me equivalent mass
n attenuation coefficient circumferential modal index
p pressure, overpressure, stress in soil, vertical penetration
pa atmospheric pressure
pd drag pressure
pe duct entrance static pressure
pi initial pressure, peak stress in soil
pm peak pressure
p0 peak overpressure, static yield pressure
pr reflected pressure
ps pressure outside duct
p1 instantaneous pressure, peak instantaneous pressure pressure at
tunnel entrance
pmax maximum pressure
px peak pressure at a distance x along a tunnel
pw internal overpressure in a shock tube
q dynamic pressure due to blast winds
qL equivalent uniformly distributed load
qD dead load per metre
r pipe radius radial distance
s distance charge spacing
t time slab thickness; thickness of casing time measured from
moment of arrival of blast wave
ta time of arrival of shock
tA intervals
tc time of reflection effects
td duration of blast pressure
tp minimum slab thickness
tm time of maximum response
t0 time for shock front to reach given point
tr duration; rise time
u dilational seismic velocity
ū shockwave velocity
ua speed of sound in air
up permanent horizontal movement
ux outburst speed
CAPITAL LETTERS
GREEK LETTERS
a peak stress attenuation factor angle between a radial line and the
vertical time constant factor applied to distance to establish peak
pressure at tunnel entrances
ß time constant
γ ratio of specific heats
γ ratio of specific heats
γ0 gas constant for expansion chamber
γ3 gas constant for compression chamber
θ angle of orientation of a bomb time for pressure to fall to half
initial value
µ refractive index
Conversion factors
Explosions can threaten people’s lives. They can also threaten the integrity of
dwellings, industry and the security of communications, transport and services.
Explosions can be man-made or result from tragic accidents, and can range
from nuclear explosions to the firing of a shotgun; or from the detonation of
unconfined vapour clouds to a bursting tyre. Explosions can be used as weapons
of war as well as instruments of peace. The military sappers and miners of one
generation become the quarry blasters of another.
Almost every evening our television screens show the effects of explosions
on structures. A shattered hotel, a damaged police post, a domestic gas explosion,
the explosive failure of an aircraft pressure bulkhead or of a jet engine. In spite
of this the engineering profession in general is not well versed in the design of
static or moving structures to withstand explosions, partly because in the past
specifications have rarely included explosive loading as a factor in design, and
partly because the various dynamic effects of explosions on structures have
only been examined as research subjects in a small number of research laboratories.
The cost of providing a safe environment for research and testing is high, and
experimental work in most countries has been left to the armed services, government
research establishments, or to large industrial explosive manufacturers. Very often
the results are not openly reported because of security restrictions. The author,
however, senses a change, and a growing need for information. The risk of chemical
plant explosions is a demanding design problem. A level of protection against
nuclear and conventional weapons of attack is often specified in new civil works,
and there is a terrorist danger, which threatens marine and ships’ structures as
well as buildings, bridges and aircraft. The Gulf War in 1991 emphasized the
structural damage that can result from modern weapons of great accuracy.
Where can the designer turn for advice? There are several good texts on the
physics of explosions, the science of detonics and the design of protective structures.
A fund of knowledge lies in the archives of government departments and the
services on the testing of devices of war. The collection of wartime home security
reports in the British Public Records Office is worth travelling a long way to read
and a recent paper has summarized this work. The output of laboratories and
agencies in the USA since the Second World War has been of great value and
many important technical memoranda have been published over the years by the
American military services. There have been many conferences recently on related
subjects, but their proceedings have not always been widely distributed. Military
and Naval research has been aimed at the effects of explosive attack weapons,
and at the improvement of defences against them, but the results are closely guarded.
Much of the work has been ad hoc testing in support of development, with no
specific examination of the fundamentals, and without the discovery of parameters
and coefficients that could influence future design and research.
This book has been written to try to collect the historical philosophy of the
subject together in a way that will be a useful introduction to scientifically
minded newcomers and a quick reminder of the fundamentals to experts in
the field. Where possible the principles have been drawn out to reveal the
underlying science. Where the principles are clouded, an attempt has been
made to clarify them by examining other methods of approach. It seemed
logical to begin with a review of the nature of explosions, so we deal first with
the basic science, starting with a brief history. The beginnings of gunpowder,
or black powder as it was called, go back a very long way, and volcanoes and
natural gases have been exploding since the morning of time; but most modern
work stems from the discovery of nitroglycerine by the Italian scientist Sobrero
less than 150 years ago. His name is less well known than that of the two
Nobels, father and son, who invented the simple manufacturing process and
discovered the importance of shock, rather than heating, as an initiator.
The physics of the detonation process is of interest, from initiation to the
formation of the shock front and blast wave. This has been well documented
in works by Kinney and Graham, Baker, and Henrych. It must be remembered,
too, that explosions can take place underwater and underground, as well as
in air, and the influence of the surrounding medium can be considerable. Many
manmade explosions, particularly in weapons of war, are accompanied by
the fragmentation of a disintegrating casing. Nuclear explosions are followed
by hurricane-like winds of great magnitude. Structures in the explosive field
can be damaged by the instantaneous rise of air-blast pressure, by fragments,
and by blast winds. In nuclear explosions humans can be killed by radiation
effects, although this aspect is not a subject for the book.
It was not until 1919 that the scaling laws for simple explosions were
expressed succinctly. A presentation by Hopkinson to the British Ordnance
Board was not conveyed mathematically, but was of classic significance. He
pointed out that if two structures were made to the same drawings and of
similar materials, but on different scales, and if charges were detonated to
produce similar structural effects, then the weights of the charges needed to
be proportional to the cubes of the linear dimensions. This law was apparently
discovered independently in 1926 by Cranz.
This breadth of experience led to the discovery of the explosive scaling law
which was mentioned earlier and which is still such a fundamental part of our
analysis of the effects of explosions on structures. It underlines the need for
cross linking in our training of engineers and scientists. The second reason is
to emphasize the quality of the British contribution to the science of explosions
in the early part of this century.
The fundamentals also owe much to three other outstanding British scientists.
The first of these was Horace Lamb, a mathematician born at Stockport in
1849 who was professor of mathematics in Manchester University from 1885
to 1920. He was the recognized authority on hydrodynamics and wave
propagation, among many accomplishments, and it was he who set down in
his famous book Hydrodynamics the physics of plane waves diffracted by
striking discs, cylinders or circular apertures in plane screens. He also focused
attention on the work of Riemman on the formation of shock waves, and to
the earlier work of Earnshaw on the mathematical theory of sound.
The second scientist was William John Macquorn Rankine, born in
Edinburgh in 1820, who was trained as an engineer. He became professor of
civil engineering at Glasgow University in 1855, and died at the relatively
early age of 52. He was the author of the first formal treatise on thermodynamics,
as well as a remarkable manual of civil engineering. He discovered the changes
in pressure, density and velocity of a gas passing through a shock wave, and
published this work in 1870. This analysis was of great significance in the
study of the behaviour of explosions, and was also discovered by Hugoniot in
Paris in 1889. Rankine, like Hopkinson, was outstanding in a number of
areas of physics and engineering at the same time.
The third scientist was Geoffrey Ingram Taylor whose work on the dynamics
of blast waves from explosive charges was of great value to the defence research
effort in Britain in the period between 1936 and 1950. His earlier papers dealt
with the propagation and decay of blast waves from conventional weapons,
but his later work was devoted to the behaviour of blast waves from the first
atomic explosion in New Mexico in 1945, and from underwater atomic
explosions. Readers are recommended to examine the collection of the scientific
papers of Sir Geoffrey Ingram Taylor, edited by G.K.Batchelor and published
by the Cambridge University Press in 1963. Volume 3 of this work deals with
the aerodynamics and mechanics of projectiles and explosions.
We are concerned in this book with the shock and dynamic loading acting
on structures from various types of explosion. The accuracy of the figures
depends on the quality of the instrumentation used to measure instantaneous
pressure, the variation of pressure with time, the duration of the pulse, and
the velocity of associated impulsive effects such as blast winds. In certain
instances the response of the structure influences the nature and level of the
applied dynamic loading.
Accurate measurement of pressure and duration first became historically
important during the rapid development of the science of military ordnance
could be inserted. When the instantaneous pressure acted on the end of the
tool, it drove the other end (which was shaped as an indenting point) into a
copper disc and the amount of penetration was an indication of the pressure.
Apparently this method was originally proposed by a Major Wade and it
enabled the effect of increasing the weight of the charge and the weight of the
projectile to be assessed accurately for the first time. As might be expected,
the peak pressure due to the explosion was found to diminish rapidly as the
distance along the gun barrel from the source of the explosion increased.
During the first part of the twentieth century, particularly during the years
leading up to both world wars, there was a large increase in the amount of
thought given to methods of measurement. Other mechanical gauges that
were developed by organizations such as the US Naval Ordnance Laboratory
or the UK research department at Woolwich Arsenal included the ball crusher
gauge, in which a spherical copper ball was compressed between a sliding
piston, activated by the pressure, and a fixed anvil. The permanent deformation
of the ball gave an indication of pressure. A number of varieties of spring-
piston gauge were also developed, and several types of foil and cylinder gauge
were used in practical experiments. The foil gauges consisted of thin diaphragms
of copper or aluminium foil, or even paper. These were clamped over the open
ends of cylinders connected face-on or side-on to the blast, and the deformation
of the discs gave a measurement of pressure. Other gauges used aluminium
strips or wires clamped as cantilevers to steel posts, to form cantilever ‘flags’.
The permanent tip deflection of the cantilevers varied with charge distance
for various charge sizes. Simple gauges of this type were often used after the
Second World War when there was great research activity into the nature of
nuclear explosions. The ultimate in simplicity was the deployment by the
British scientist, William Penney (later Lord Penney), of petrol tins at the
atomic weapon test on the Bikini Atoll. The degree of deformation of the tins
by the blast wave, in terms of the change in internal volume, enabled a good
approximation to be made of the magnitude and distribution of the peak
pressures. These test measurements often succeeded when more sophisticated
gauges failed. Later this technique was used in experimental research on
conventional explosions.
The greatest step forward in pressure measurement, however, was the
development of the pressure transducer, using piezoelectrically active crystalline
materials that produce electric charges when strained. Tourmaline, quartz and
ammonium dihydrogen phosphate are substances that have been used in successful
gauge design, which in addition to the crystals require amplifiers, calibration
circuits and electrical recording systems. Condenser-microphone gauges have
also been developed, using changes in capacity under pressure, and resistor
gauges which use changes in resistance under stress as the measurement. All of
these gauges are particularly useful for specified characteristics such as low-
frequency response, pressure range, duration and temperature. Many defence
laboratories and research institutes in the UK and USA have produced their
own versions of the pressure transducer, ranging in size from miniatures with
diameters of a few millimetres to large-scale components with diameters of 10
centimetres and more. These and others have been reviewed in detail by Wilfred
Baker in his book Explosions in Air, published by the University of Texas Press
in 1973 in conjunction with the Southwest Research Institute of the USA. In a
most useful survey he also discussed the design of gauges to measure dynamic
pressures due to blast winds and to record the time of arrival or time of start of
release of blast energy.
The accurate measurement of the characteristics of the dynamic loading
due to explosive shock has, as we have seen, been of great interest to structural
engineers. The instrumentation for a major nuclear or high explosive test can
be astronomically expensive, but unless the pressure/duration characteristics
of a blast wave at various distances from the source and at various parts of a
structure can be measured the calculation of structural response in air, water
or underground cannot be carried out with scientific confidence. Care must
also be taken to distinguish between detonation and deflagration, the absence
of sudden shock, and changes in the shape of the pressure/duration curve.
These factors are pursued later.
It is important in the analysis of loading and response not to overlook the
lack of uniformity in explosive actions. The pressure distribution depends on
the shape of the charge, but even with a perfectly spherical charge under ideal
conditions it would be wrong to assume that the peak pressure at a given
distance from the centre of the sphere was uniform and regular. There are
peaks and troughs in the distribution pattern that make it difficult to treat the
loading analysis as an exact science. Charges of nominally the same weight
and geometry do not necessarily yield similar pressures, durations or impulsive
characteristics. The physical conditions are so variable that great care must
be taken when drawing analytical conclusions, because experimental scatter
in laboratory or field tests is high.
Some of the earliest methods of calculation of structural response were
associated with the assessment of the strength and safety of gun barrels. Cast
guns often burst on discharge because of minute flaws, and in the fifteenth
century in England accidental bursts of military ordnance were considered
important enough to alert the privy council. The idea of proving the strength
of ordnance by ‘proof testing developed in successive centuries, and today we
have ‘proof and experimental establishments’ like that at Shoeburyness as
historic links with earlier days. The structural strength and safety of shotgun
barrels are still tested today by proof testing, where a charge well in excess of
that normally associated with the explosion of a cartridge is fired in the chamber
of the gun.
As the design of guns progressed, early versions of ‘thick cylinder’ theory
began to be used to predict strength, and this was probably the first use of
structural analysis to judge response to shock loading. The earliest guns fired
simple stone or cast-iron cannonballs, or canisters containing grape shot or
flint pebbles. The destructive power of these weapons was mainly due to
collision impact between the balls and the outer fabric of structures, whether
stone structures such as castles or walls, or the structure of the human body.
However, a major and far-reaching development occurred in the fifteenth and
sixteenth centuries when the explosive shell was invented by the French. The
solid cannonballs were replaced by hollow spheres of iron containing a fuse,
a bursting charge, and smaller fragments. This meant that explosions now
took place when the projectile made contact with, or penetrated, the target.
For the first time it became necessary to predict the response of heavy stone or
rock fortifications to contact explosions. Later the projectiles were elongated
and formed into the now recognizable shapes of bombs, mortars, armour
penetrating and air-bursting shells and rockets. Much of the inventive British
development in this field occurred at the beginning of the nineteenth century,
due to illustrious men like Sir Henry Shrapnel and Sir William Congreve. The
explosive shell or bomb became the major attacking weapon against the
structures of land fortifications, naval vessels and military targets. The designers
of fortifications had to analyse much more closely the response of masonry to
localized explosions. Repeated hits on small areas of masonry could lead to
breaching, and explosion after penetration could cause fragmentation, cracking
and collapse.
Although there was much development in the science of demolition and in
the effect of shock waves on the structure of stone and masonry, the analysis
of structural response as we now know it began with the invention of reinforced
concrete and the rapid growth of steel- and concrete-framed building structures
in the early years of the twentieth century. This growth coincided with two of
the most dreaded inventions of man, the military aircraft and the aerial bomb.
Thus in the First World War the main structural damage on land was of
traditional building structures due to the high-explosive artillery shell, but in
the Second World War the structural problem had shifted to the effect of
bombs on framed structures. In Britain many of the outstanding engineers
and scientists of the day came into the field in support of the Ministry of
Home Security. Professor John Fleetwood Baker (later Lord Baker), Professor
William Norman Thomas and Professor Dermot Christopherson (later Sir
Dermot Christopherson) wrote papers and reports of great significance at the
end of the Second World War on structural response. It was realized that
design methods involving elastic stress distribution for static structures were
not suitable for analysing the effects of blast loading, and that the critical
factor was the capacity of a structure to absorb energy. The ability to absorb
energy in the plastic range was seen to be very significant in comparison with
the capacity in the elastic range for most types of ductile structure. The carrying
capacity of simple beams made of steel when bending beyond the limit of the
elastic distribution of stress had been examined earlier by Ewing in Britain
and Maier-Leibnitz in continental Europe. Their work was extended and the
so-called ‘plastic method’ was used in the design of basements, surface shelters,
indoor shelters and factory walls to withstand the effects of blast energy from
closely exploding bombs, and to give protection against bomb fragments,
crater debris, and other objects projected violently by the explosion.
Much full-scale test work was undertaken during the wars of the twentieth
century by most of the major participating nations. The need to calculate the
charges required to demolish structures, particularly reinforced and unreinforced
concrete structures, became important, and most countries produced simple
formulae linking the dimensions of concrete protective structures with the
weight of the explosive charge and the internal volume of air space. In steel
and wrought-iron building frames the possibility of progressive collapse was
investigated, and the need for alternative load paths became an important
consideration. Problems due to weakness in the connections of framed structures
were evident, and designers were reminded that the more nearly that connections
could be made to approach the continuity and ductility of the main members
they join, the less the damage that would result from a given bomb explosion.
The framework needed to be capable of resisting collapse if one main member
was suddenly removed, and in no circumstances should it be liable to progressive
failure.
The modelling of structures as simple mass-spring systems and the
comparison of the natural periods of oscillation of such systems with the
duration of the pressure pulse of an explosion were important in the
determination of the level of structural response before the main energy
absorption activity began. The advent of computer-based software in the mid-
twentieth century has enabled quite sophisticated modelling to be undertaken
in an effort fully to understand the physics of blast-structure interaction. More
complex, multi-degree of freedom systems can be investigated quickly by the
designer, and the influence on behaviour of structural damping assessed. This
has led to a great increase in the computer-based analysis of structures subjected
to shock loading. It has also become much easier to investigate theoretically
the diffusion of shock waves in structural materials, and to use this to predict
material behaviour resulting from explosive shock characteristics. The spalling
and scattering of steel and concrete, and localized fragmentation at the point
of application of the impulsive loading, can be investigated.
It was clear from the earliest time that structural damage would be
considerably increased if the explosion of a charge took place within the material
of the structure, so the penetration of attacking missiles into structures followed
by subsequent explosion became an important aspect of design. It is not possible
to discuss structural loading fully without discussing penetration, whether it
is penetration of concrete by free-falling bombs or of the armour plating of
military vehicles, naval vessels or military aircraft by armour-piercing shells,
bombs, rockets or missiles. The penetration of high-velocity objects into soils,
stone, metals and concrete has historically been a subject for military engineers,
and much of the work in this field originated in the military establishments of
Europe in earlier centuries.
The most famous of the early contributors to this field was Jean Victor
Poncelet, an illustrious product of the French Ecole Polytechnique system,
who was born at Metz in 1788. Part of his career was spent as a military
engineer in Napoleon’s Army, and part as an expert in engineering mechanics
in the arsenal in Metz. He made outstanding contributions to the fields of
geometry, structural dynamics, and engineering statics. He is said to have
introduced the effect of shearing force into the calculation of beam deflection.
His influence on the subsequent development of structural analysis was
profound, and among his most far-reaching research was the study of
penetration. He saw that the kinetic energy of penetrators, proportional to
the square of the velocity, was a key ingredient of the equation, and that the
flow of soils or metals around the body of the penetrator could not be ignored.
There are two aspects of penetration mechanics relevant to our subject.
The first of these is the penetration of an explosive-carrying missile into the
target medium before the explosion occurs. The second is the penetration of
structures by high-velocity fragments of the metallic casings of bombs and
shells after the explosion has taken place at a short distance from the target.
The first aspect results in internal shock waves in the material of the target,
the second results in high-velocity impact damage.
High-velocity impact from the steel fragments of shell casings began to be
investigated after the advent of the artillery shell, and much work was in
progress in the nineteenth century. However, the penetration of warheads and
explosive-carrying devices did not become widely researched until the beginnings
of aerial bombing in the twentieth century. The influence of penetrator shape
on the depth of penetration, and the relevance of the strength of the penetrated
material was examined, and at the end of the Second World War there were
several empirical penetration formulae available. Most of these assumed that
the penetrator did not deform on contact with the medium. The problem of a
deforming plastic penetrator seems to have been considered firstly by G.I.Taylor,
but in recent times there has been an increase in research in this area resulting
from the development of three-dimensional finite element programs capable
of dealing with dynamic forces and massive plastic deformation.
Once the fundamentals of deformation, blast loading, structural response
and penetration have been examined the problems facing the structural designer
in many fields can be addressed. The first of these is the effect of local explosions,
where up to one or two tonnes of explosive is detonated on or near structures.
The targets might be aircraft structures, naval structures, underwater structures,
protective structures built on the surface of the ground, or protective structures
built below ground. We owe much to the research and testing aimed at the
evaluation of structural response to ‘conventional’ or ‘non-nuclear’ bombs
carried out in the USA, Great Britain and Germany during the past 30 years.
The leading American establishment in this work has been the US Army
Waterways Experiment Station at Vicksburg, Mississippi. Their test facilities
have been constructed at great expense, and have been used to examine
cities. Very detailed reports of bomb damage to reinforced concrete and steel
building frameworks were published in Britain, and from these and other
sources enough information was gathered to produce for the first time design
guidance for engineers who needed to calculate the response and strength of
structures that were specified to be resistant to accidental or man-made local
explosions.
Another military threat about which much has been written is the nuclear
explosion. Although the threat seems less immediate these days, the major
arsenals of the world still contain nuclear bombs, and the behaviour of structures
in the shock and blast wind phases of a nuclear explosion must still be assessed
by engineers. It is the duration of the dynamic pressure, or drag loading, that
is the main difference between nuclear and high-explosive detonations. The
duration of the positive phase of the dynamic pressure from a megaton nuclear
explosion can be several seconds, whereas the duration of the air blast from a
conventional high-explosive detonation may only be a few milliseconds.
Structures most likely to be damaged by the high instantaneous pressure
associated with shock front are dwelling houses. Structures likely to be damaged
by the dragforce of the blast winds are chimneys, poles, towers, truss bridges
and steel-framed buildings with light wall cladding. There are also the hazards
of fast-flying debris and fire. The threat is so great that nuclear resistant
structures are normally buried below the ground surface.
A great deal of information about the behaviour of structures of all types
was assembled after the Second World War from the controlled nuclear bomb
tests in the Pacific and at the US test site in Nevada during the 1950s. Publications
by Professor Nathan Newmark and others on the design of structures to
withstand nuclear effects, including the problem of radiation, were important
and progressive contributions to the structural mechanics of the problem.
The American Society of Civil Engineers was particularly active in this work.
The importance of underground structures led to a surge during the 1960s in
analytical and experimental research on soil/structure interaction in a dynamic
environment, and a number of simulation facilities were built in America and
Britain. As the political problems of detonation and fall-out from field nuclear
tests increased, nuclear bomb effects had to be approximately simulated by
exploding a great weight of TNT instead. Unfortunately much of the target
response information is hedged in by a high-security classification, and the
details are not freely available.
The effect of the heat flash associated with a nuclear explosion can also
damage certain types of structure, particularly when the structure is made of
aluminium. It is possible for aluminium military equipment, such as a rapidly
built bridge, to escape damage by pressure or wind, but be subjected to
temperatures that are high enough to reduce the strength of the alloy. The
other hazard, as mentioned above, is due to initial nuclear radiation. As a rule
structures designed to protect occupants against peak overpressures of 77
KPa and above should also be checked for radiation. Radiation can be
particularly damaging when peak overpressures exceed 315 KPa. The response
of structures and the humans in them to nuclear explosions is therefore a very
complex design field, and the engineering of the subject is often left to physicists,
with questionable results.
Another field of interest is what might be termed civil explosions. These
are non-military or non-terrorist explosions which are either the result of
accidents due to natural forces or deliberately produced to aid the requirements
of civilized society. The most frequent accidental explosions are probably due
to domestic gas leaks, and there have also been accidents due to the ignition
of methane gas in civil engineering works and mines. In Britain much structural
research has centred on the former Building Research Establishment. Many
reviews and reports have been written on the nature of serious gas explosions
and structural disasters are well documented. The action of domestic gas
explosions is nearly always from within the structure, and in structures that
are badly designed with respect to progressive collapse a relatively small
explosion can lead to spectacular results. The collapse of a high stack of corner
rooms at the Ronan Point high-rise apartment building in London following
a gas explosion led, in the 1960s, to a formal enquiry and adjustments to
future design requirements for all types of building. Internal gas explosions
are unfortunately still a regular occurrence, often resulting in the death of
house or apartment dwellers.
The possibility of internal explosions also occurs in buildings that house
test facilities for engines, and here it is good practice to limit the instantaneous
pressure rise of the explosion by using weak panels in the superstructure that
will collapse outwards. Other industrial manufacturers who require protective
construction in their facilities are the producers and storers of explosive
materials. The possibility of an accidental explosion causing detonation of
other explosive devices is a hazard that has to be carefully examined at the
building design stage. Hardened buildings that limit the propagation of mass
detonations are usually constructed of reinforced concrete with steel blast
doors. There are regular conferences, particularly in the USA, on protection
against the accidental explosion of hazardous mixtures, and safety control in
this field is becoming increasingly important in most countries.
Another civil hazard is the unconfined vapour cloud or dust cloud explosion,
which has often resulted in a major national disaster. In Britain the on-shore
explosion of the Flixborough chemical plant and the off-shore explosion at
the Piper Alpha oil rig have caused much damage and loss of life. Explosions
of this type cause secondary damage due to fragmentation, fire and loss of
structural stability. There is always much publicity and speculation when large
cloud explosions occur, and the political pressures to find the causes and
distribute the blame are usually as intense as the explosion itself. Smaller
explosions of this type have occurred from time to time in ships, under cargo
conditions that are not always foreseen as dangerous. The problem of design
against accidental explosions has been addressed by the US Department of
but the feeling in recent decades is that the prayer should be replaced by a
microchip that monitors the damage as it occurs and gives instructions to the
controlling computers.
It can be seen from these introductory remarks that there is a great deal of
data available and, as we said earlier, conferences on the subject are taking
place with increasing regularity. The aim is to review the history as concisely
as possible, beginning with the physics and ending with the engineering of the
subject. The future is not easily predicted. Nuclear explosions are less likely,
one hopes, but local explosions due to terrorism or revolution seem to be
growing in number throughout the world. The need for research and
development will no doubt remain, perhaps with a change of emphasis from
military to civil requirements. There is already a move to increase the amount
of university funding in this area, underwritten by the response of senior
committees to the need for fuller research in support of the ‘new’ subject of
Hazard Engineering, and this text may help to introduce the problems of
explosions and structures to a wider audience of academics, consultants and
advanced students.
propelling pressure; if packed too loosely most of the gases escaped around the
projectile and the subsequent speed of travel of the shot was much reduced.
This problem was overcome by mixing the powder with fluids to form a cake,
and then breaking the cake into uniformly sized crumbs or grains through a
sieve. Since two of the best fluids were alcohol and urine, the making of gunpowder
was not unlike a farming process, and as the original scientist in the field had
links with bacon and Somerset cider this was not surprising. Incidentally there
were still members of the Bacon family living in Yeovil, the author’s hometown,
about five miles from Ilchester, in the mid-nineteenth century, and the late-
twentieth-century telephone directory shows a number of Bacon entries in the
Yeovil, Crewkerne, Taunton and Bridgwater areas of Somerset.
The search for saltpetre was a serious activity over the centuries. In the
early days it was known that there were surface deposits in Spain and India,
but in most European countries no ready-made supply existed. Consequently
the business of ‘nitre’ beds prospered, in which layers of decaying animal and
vegetable matter, earth, sand and old mortar were moistened from time to
time with blood. Eventually potassium nitrate could be extracted. This method
was mainly overtaken when vast deposits of sodium nitrate, which could be
converted to saltpetre, were found in Chile. The military use of black powder,
with its powder kegs, powder horns and the commands to ‘keep your powder
dry’ was much reduced at the end of the nineteenth century, when ‘smokeless’
powder was produced in France. Apparently one of the last non-military uses
on a large scale was in the destruction of a huge rock in New York harbour
(the Pot Rock) in 1853. We are told that 200 000 lb of powder were used.
Gunpowder was also used in the mining of ore in Europe from the seventeenth
century, and for road widening in Switzerland at that time; and of course we
all know of the 36 barrels, weighing about one cwt each, concealed beneath
coal and faggots under the House of Lords in 1605.
The firing of black powder produced so much smoke that after a heavy
volley from land-based guns or a broadside from naval guns the firers and the
target became completely obscured. A new propellant, smokeless and without
residue, was needed, and this led to the development by the French scientist,
Paul Vieille, of a mixture of black powder and gelatin which was smokeless.
The action of all powders upon initiation is by burning, or deflagration. Each
grain of the powder burns at the same time as other grains and the internal
pressure throughout the mass of powder remains uniform and equal to the
external pressure. It is possible for the deflagration to change to a detonation in
which the chemical reaction spreads like a wave from the point of initiation, a
change that is completed in a few millionths of a second; but smokeless powders
are very difficult to bring to this stage. They are therefore ‘low’ explosives.
Most other conventional explosive substances have been developed from
nitric acid and nitrates, and the distillation of potassium nitrate, alum and
blue vitriol to form nitric acid is usually considered to have originated with
the Arabian chemist Greber (or Jabir) in the eighth century. He was the most
lower and the duration of the impulse much greater. Detonation, as opposed to
deflagration, is obtained by the use of a burster charge containing a mass of
conventional high explosive about one-hundredth of the FAE mass. The control
of FAE weapons was thought to be uneven, and research in the USA was reduced
in the 1970s, but more recently the Soviet air force used weapons of this type in
Afghanistan, and there are reports in the technical military press of the very
successful use of FAE bombs by the US forces in the Gulf War.
Let us now turn to the detonation process, which has been considerably
developed during the past century. There are now many methods for the safe
detonation of high explosives, and nearly all consist of the initiation of a
shock wave into a base charge which then detonates the main explosive charge.
Most recruits to the armed services in the first half of this century will have
handled the small aluminium tubes containing a priming charge of mercury
fulminate (HgC2N2O2), or lead azide (PbN6) and a secondary charge of PETN.
The primary charge could be set off by impact, as in the spring released system
of a hand grenade, or via a length of safety fuse capable of initiation by a
safety match.
The priming charge can also be ignited by electrical means, and electrical
shotfiring is now widely used. When an electrical charge is connected to the
detonator it heats up a fine wire which then ignites a flashing compound. This
‘fusehead’ fires the primer and base charges. Delaying elements can be
introduced between the fusehead and the primary charge, or the fusehead can
be set to explode so quickly that the entry of the firing current and the detonation
of the base charge is virtually instantaneous. The use of electrical detonation
must be carefully controlled in very sensitive environments where there is a
risk of premature firing by static electricity, or where there are dangerous
methane or dust clouds. Special detonators have been developed to overcome
these problems and to avoid accidental pre-detonation. These are described
in handbooks and in the marketing literature of manufacturers such as ICI
Nobel and Dupont.
The other main detonating medium is the ‘detonating cord’ and ‘Cordtex’
is a universally used version in blasting and quarrying operations, explosive
cutting, and underwater operations. A core of PETN is surrounded by tape
and wrapped with textile or synthetic yarn, and this cord is completely enclosed
by a white plastic tube. It has a high velocity of detonation which means that
it will initiate most commercial explosives, and it can be initiated itself by
most of the standard detonators. There is plenty of information on methods
of using Cordtex for various blasting operations, involving detonating relays,
jointing and branching methods, and initiating systems.
A major area of development has been the artillery fuse. These were originally
the powder train fuses, or rather inaccurate mechanical devices, but more
recently the need for high firing rates has led to the evolution of the electronic
time fuse and to the electronic fuse with a high-explosive boosting charge.
The fuse is the shaped end of a mortar or artillery projectile and consists of an
mass, and this must be produced in a very short space of time, either by firing
one sub-critical mass at another using explosive propellant, or by compressing
a sub-critical mass by the implosion forces of surrounding high explosive.
Information on the design of nuclear weapons was a matter of high security
for many years, but recently more information has become available on the
size, shape and action of the devices.
explosive gases escape and form an incandescent zone that expands so rapidly
that a shock wave or pressure pulse is formed.
Taylor wrote many valuable papers [1.7], [1.8] on the dynamics of shock
waves for the Civil Defence Research Committee in the early days of the
Second World War, and it is from these that much of the analysis has been
taken. His work was summarized lucidly by D.G.Christopherson in 1945
[1.9] and the latter’s summary has been consulted frequently by the author in
writing the chapters of this book. In the summer of 1945 Christopherson
wrote his seminal report on the structural effects of air attack, the information
having been collected during the Second World War by the Research and
Experiments Department, Ministry of Home Security, and much of it coming
from experiments carried out on behalf of that Department by the Building
Research Station and Road Research Laboratory. Contributions were also
drawn from the work of the National Defence Research Committee in the
USA. Christopherson’s report, entitled Structural Defence, covered every aspect
of the subject from the theory of blast waves in air, earth or water to the
general theory of structural behaviour and the design of protective structures
of all types. Christopherson was educated at Oxford, was a fellow at Harvard
and then a postgraduate at Oxford before joining the Research and Experiments
Department of the Ministry of Home Security in 1941. He left, after writing
Structural Defence in the space of two or three months after victory in Europe,
to join the Engineering Department at Cambridge University in 1945. Since
then he has pursued an illustrious academic career and has been a major
influence in the teaching of Engineering subjects in the universities, as well as
in the management and control of universities and colleges.
The form of the overpressure/duration relationship for a high explosive or
nuclear explosion in air is shown in Figure 1.1, where p is overpressure (or air
blast pressure) and t is time. In the figure the decay of pressure after the first
instantaneous rise is expressed exponentially. There are other ways of denoting
the form of the pressure/duration relationship, as we shall see later, but for the
purposes of this chapter Figure 1.1 will be adequate. The value of the peak
instantaneous overpressure p0 will depend on the distance of the point of
measurement from the centre of the explosion. Using imperial units, in a TNT
explosion p0 might be 200 or 300 psi at the point of burst of the explosion, but
would rapidly diminish with distance. In a nuclear explosion equivalent to 1000
tons of TNT, the peak overpressure would be 2000 psi at 30 metres from the
centre of the explosion. The duration of the positive phase is t0 units of time.
Theoretically, for a perfectly spherical charge in air, the relationship between
p0, the distance of the point of measurement from the centre of the explosion
(R), and the instantaneous energy release (E), takes the form
p =KE/R3, 1.1)
0
3
so that an important non-dimensional parameter is p0R /E. In imperial units E
is measured in ft lb, and in SI units in joules. Experiments show that the explosion
p =K W/R3. (1.2)
0 1
where
p0=peak pressure in psi
1/3
z=R/W (R in feet, W in lb).
The relationship should only be applied when 160>p 0 >2 psi, and
1/3 1/3
20>R/W >3 ft/lb .
1/3
Note that the parameter R/W enabled the results to be applied to any
detonated explosion, conventional or nuclear, as long as the equivalent weight
1/3
of charge in TNT were known. As we saw in the Introduction, R/W is an
important scaling factor, first noted by Hopkinson in 1915. Modern versions
1/3
of Eq. (1.3) give the pressure in bars and z in metres/kilogrammes .
The relationship in Eq. (1.3) is very similar to the logarithmic plot of peak
instantaneous pressure versus scaled distance given by Kennedy [1.11] at the
end of the Second World War, when he summarized the results of free-air
blast tests on cast TNT in the US and the UK. It was not realized at the time
that the shape of the charge could have a significant influence on results, or
that the pressure-measuring instruments could noticeably affect tests by altering
the air flow behind the shock front. Nevertheless, Kennedy’s summary was a
useful indication of the pressure/distance relationship.
The pressure/distance characteristics discussed above only apply to a truly
spherical charge in air, but in many practical circumstances the shape of the
charge is cylindrical, or a plane sheet, or a line source such as detonating cord.
For line charges it is known that the shock front expands cylindrically, so that
1/2
p0 is a function of R(L/W) , where L is the length of the source and L»R.
Generally, blast waves from non-spherical sources exhibit a less rapid decay
of pressure with distance. Theoretically predicted values of p0 have been
proposed by Lindberg and Firth [1.12], and are shown in Figure 1.2.
1/3
The vertical axis is p0/pa, where pa is atmospheric pressure, and E is replaced
by the characteristic dimension R0, where
R0=[E/(paL(3–v)]1/v. (1.4)
(1.5)
8 lb cylinders, Pentolite,
(1.6)
4 lb cylinders, TNT,
(1.7)
(1.8)
1/3
For most of these formulae p0<2 psi and z>20 ft/lb . These different
relationships were empirical and tended to suggest an accuracy of scientific
measurement that did not exist. In fact, when the results are plotted on common
axes, they all lie within a narrow scatterband, as shown in Figure 1.3. Further
information on the behaviour of spherical charges of Pentolite is recorded in
the works of Goodman [1.14], who compiled measurements taken from
experiments over the 15 years from 1945 to 1960.
Figure 1.2 Peak overpressure v Range for various charge shapes (from Lindberg and
Firth, ref. 1.12).
(1.9)
where, as defined above, z=R/W1/3 in ft/lb1/3 and typical values of a1, and b1
were:
Figure 1.3 Peak overpressure v Scaled range for rectangular, cylindrical and
spherical charges (from Stoner and Bleakney, ref. 1.13).
As Figure 1.4 shows, the differences were very small, and all the results lay
within a narrow scatterband. Note that the results summarized in Figure 1.4
lie at the far right end of Philip’s curve, because the charges were small and
the peak pressures very low. Philip’s work still remains as one of the most
authoritative pieces of work on high-explosive bombs, although based on a
dangerously small number of tests. Her results were also given by
Christopherson in reference [1.9]. What they suggest is that the decay of peak
instantaneous pressure with range is exponential, which seems logical. In fact
the lower scatterband on Figure 1.4 is close to the ‘inverse square law’ given
2
by p0=500/z , which is a simple rule still in use in preliminary calculations
(after conversion to metric units).
The 4000 lb high-capacity bomb tested by Philip must have been one of
the ‘blockbuster’ bombs containing a high proportion of charge and a relatively
low containment mass, designed by Barnes Wallis to attack heavily reinforced
concrete protective structures by earth transmitted shock rather than by
penetrative impact. It is interesting to note that during the Gulf War in
1991, when the standard 2000 lb laser-guided bombs dropped on bunkers
by the US Air Force were clearly not entirely demolishing the targets, it was
reported in the press and elsewhere that an accelerated manufacturing
programme was undertaken to produce 4000 lb bombs. Readers who wish
to see what a UK 4000 lb bomb looked like may find examples mounted on
Figure 1.4 Relationship between peak overpressure and scaled range for UK Second
World War bombs (from Philip, ref. 1.15).
plinths at certain RAF stations, but it is best not to stop! According to records,
68000 high capacity versions of this bomb were released over enemy territory
between 1941 and 1945. Barnes Wallis has a further famous place in military
history in connection with the design of bombs to destroy the main structure
of dams, and with the development of the Tallboy and Grand Slam bombs
(see section 7.3).
The characteristics of explosives are nowadays often presented dimensionally
on a log log scale as a relationship between the peak instantaneous overpressure
in bars and the distance in metres in any direction from the centre of the
explosion of a spherical charge. Figure 1.5, due to Lavoie [1.16], gives a
comparison for 1000 tons of TNT, FAE and a nuclear explosion. The peak
overpressures close to the explosion vary considerably, as might be expected,
but in all cases have diminished to 0.1 bar at a radius of 100 or 200 metres.
Figure 1.6 from the same source compares impulses for similar explosions,
and shows that, because of the duration of the positive phase, much more
impulse is available from nuclear and FAE explosions than from TNT.
Some time after the end of the Second World War photographs of the first
atomic explosion in New Mexico were released, and a year or two later, in
1949, a paper on the formation of a blast wave from a very intense explosion,
written by G.I.Taylor [1.17] in 1941 for the Civil Defence Research Committee
t=0.926R5/2ρa1/2E–1/2, (1.11)
when the ratio of specific heats, γ=1.4. As before, ρa was the atmospheric
density and E the energy released by the explosion. There was doubt about
the validity of the assumption that γ=1.4 at all stages of the explosion, because
at the extremely high temperatures of an atomic explosion γ could be increased
due to dissociation and perhaps affected by intense radiation. However, when
the photographs were examined, Taylor was able to compare the time t in
milliseconds with the radius of the shockwave, and found that the relationship
between and log10 t was a straight line:
(1.12)
The assumption that γ=1.4 was constant at all temperatures was vindicated.
The explosion took place only 100 ft above ground, and Taylor pointed out
that the fireball would have reached the ground in less than 1 msec; however,
the photographs indicated that this impact did not affect the conditions in the
upper half of the luminous globe.
As we noted in Eq. (1.1), the instantaneous maximum pressure at any distance
3
R from a perfectly spherical charge is proportional to E/R , and Taylor gave
the relationship as
This raises an interesting fact. The peak pressure at radius R does not depend
on atmospheric density, whereas the time t since the beginning of the explosion
for the shock front to reach a radius R depends on . Taylor used this to
calculate the pressure-time relationship for a fixed point in terms of p/p0, and
t/t0, where t0 is the time for the shock front to reach that point.
By the mid-1950s the electronic computer had begun to transform the
analytical work on blast pressures. It was possible to solve detonation problems
without recourse to experimental data, because the hydrodynamic equations
of motion which lead to non-linear partial differential equations could be
integrated numerically in a relatively short time. Work in this field was
summarized by Erode [1.18], who took a bare sphere of TNT of loading
3
density 1.5 g/cm and produced pressures, densities, temperatures and velocities
as functions of time and radius. His relationship between p, the pressure in
bars above atmospheric, and the shock radius parameter R/W1/3 (where W is
the weight of the charge at atmospheric pressure in kg and R is in m) is given
in Figure 1.7.
A Group Technical Centre paper from ICI Explosives has shown how recent
history has been influenced by the rapid increase in the availability of computer
programs which can calculate the theoretical detonation characteristics of an
explosive. Typical programs are TIGER, from Stanford Research International
in the USA, and IDEX, from ICI Explosives in the UK. Other codes are available
to help predict the practical performance in the field of explosives, depending
on the conditions under which they are used, for example rock blasting.
Correct initiation can be developed by accurate modelling of initiation
behaviour using finite element hydrocodes, such as DYNA, originally from
the Lawrence Livermore Laboratory in the USA. Other codes have been written
linking blast characteristics to fragmentation, vibration and the initiation
sequence. A very simple assessment of the performance of explosives in practical
circumstances, such as in boreholes, can be made by measuring the velocity of
detonation. This can be found by high-speed photography or by a radar
technique based on the use of microwaves. In general, the use of computers in
all stages of the propagation of explosive blast has resulted in a rapid increase
in scientific knowledge and engineering experience.
Eliminating u2 gives
(1.16)
which indicates that when, as is usual, the air in front of the shock wave is at
rest, u1=0 and the velocity of propagation ū can be determined entirely in
terms of the pressures and densities on either side of the discontinuity. As the
compression of the gas in the shock front is very fast, it is reasonable to assume
that compression follows the adiabatic law (changes in pressure and volume
with no change in absolute temperature). Then we may assume that the energy
Figure 1.7 The prediction of Erode for a bare sphere of TNT of loading density 1.5
g/cm3 (from Brode, ref. 1.18).
(1.17)
where v1 and v2 are the volumes occupied by unit mass, i.e. v1=1/ρ1, v2=1/ρ2.
Eq. (1.17) has become known as the Rankine-Hugoniot equation.
The above analysis leads to the general relationship between shock front
particle velocity, ū, and the speed of sound in air (ua), where ua=(8pa/ρa)1/2 and
ρa is the density of the ambient air. This is
(1.18)
(1.19)
where ρ is the density of the air behind the shock front. Taking ua as 1117 ft/sec
and pa as 14.7 psi, Eq. (1.18) can be plotted as a relationship between u¯ and
peak overpressure (p0) as shown in Figure 1.8.
The sudden discontinuous rise to p0 is followed by a continuous decrease
until the pressure returns to atmospheric and p=0. The time between the arrival
of p0 and the return to atmospheric pressure is the ‘positive duration’ (see
Figure 1.1), and for analytical purposes it is useful to represent the pressure-
time curve as a mathematical function. Two functions are often used:
p=p0(1–t/t0), (1.20)
(1.21)
where p is the pressure after any time t. By selecting a value for k (the
wave form parameter), various decay characteristics can be indicated.
Curves with very rapid decay characteristics are typical of nuclear
explosions, and curves with slower decay rates are typical of explosions
with large volumes of product gases. When k=1, the positive and negative
impulses (Figure 1.1) are equal, and the positive impulse is p0t0/e. The
curve given by Eq. (1.21) is often known as the Friedlander curve [1.19],
because it comes from the work of F.G. Friedlander on behalf of the UK
Home Office at the very beginning of the Second World War. Note that p/p0
is non-dimensional and is therefore an intensity characteristic of the blast
wave system.
Figure 1.8 Relationship between shock front velocity and peak overpressure (from
Rankine, ref. 1.3).
The duration of the positive phase, t0, is a function of p0 and the total
energy yield of the explosion. The end of the positive phase has an overpressure
of zero, which is the characteristic of a sound wave, since a sound wave has
no shock front and only infinitesimal changes in pressure amplitude. So the
zero overpressure condition must move away from the centre of the explosion
at the speed of sound in air. This is a lower velocity than the shock front
velocity, ū, and means that the duration of the positive blast wave increases
with distance, reaching a limiting value when p0=0.
Typical values of t0 for high explosives can be found from the formula
(1.22)
1/3
where t0 is in milliseconds and W is in kilograms. As before, z=R/W but in
these units R is in metres.
For nuclear explosions,
(1.23)
compare p0 and R, that led to Eq. (1.9), she also measured positive phase
durations, and from these results proposed the relationship
(1.24)
As Figure 1.9 shows, the scatter is small. Similar curves to Figure 1.9, but for
nuclear explosions were given in many publications in the 1950s.
A knowledge of the instantaneous pressure and the duration of the positive
phase allows us to calculate the blast impulse of an explosion. This is usually
expressed per unit of projected area, and in imperial units is lb sec/in2 or lb
msec/in2. For the three bombs analysed by Philip, she found that a simple
relationship between impulse (I), charge weight and radius took the form
I=K1W2/3/R, (1.25)
Figure 1.9 Relationship between scaled duration and scaled range (from Philip, ref.
1.15).
2
where I is in lb sec/in , W is in lb and R in feet. Values of K1 were
Figure 1.10 shows the scatter of these results, which was noticeably greater
than for pressure or duration.
In Kennedy’s summary of Second World War test results [1.11], he also
gave a logarithmic plot of positive impulse versus scaled distance for explosions
of cast TNT in free air, as shown in Figure 1.11. This is probably the best data
to use in preliminary design calculations, because Philip’s results were not
from explosions in clear air.
Christopherson [1.9] suggested that the values of K1 given above by Philip
1/3
were only appropriate for a range of R/W exceeding 6, and that at lower
1/3
values of R/W the impulse given by Eq. (1.25) would be too high. He quotes
a more pessimistic treatment by Kirkwood and Brinkley [1.20] which suggested
1/3 1/3
that when R/W in ft/lb was equal to unity Eq. (1.25) gave values of I
1/3
about three times too great, and about two times too great when R/W =3.
The true values probably lie somewhere between these extremes.
Figure 1.10 Relationship between scaled impulse and scaled range for UK Second
World War bombs (from Philip, ref. 1.15).
The Kirkwood and Brinkley theoretical air-blast curves for cast TNT have
become a much used standard for free-air explosions in a homogeneous
atmosphere, and have been shown to be suitable for small explosions as well
as nuclear explosions equivalent to 2 million pounds of TNT.
In a nuclear explosion the outward motion of the shock front when a burst
occurs at, or just above, ground level is associated with hurricane-like winds
that blow horizontally, parallel to the earth. The intensity of the wind is related
to the peak overpressure, and rises from zero speed to a high value (perhaps
160 mph, or 256 km/hour) in almost zero time. This is quite different from a
natural hurricane, in which wind speeds usually increase over a relatively
long period. The dynamic pressure, q, due to the winds is given by
(1.26)
where ρ is the air density behind the shock front, and u is the velocity of the
air particles in the shock front. Using the Hugoniot-Rankine equations it can
be shown that
(1.27)
Figure 1.11 Relationship between scaled impulse and scaled range for cast TNT
charges in free air (from Kennedy, ref. 1.11)
(1.28)
Note that the value of q is orders of magnitude greater than the external wind
pressure used in the design of conventional surface structures. At 6 km from
the centre of a 1 megaton explosion the peak overpressure would be 42 Kpa
and the windspeed 81 m/sec; the arrival time of the shock front would be 14
seconds and the duration of the positive pressure pulse 3.0 seconds. At the
same distance from the centre of a 10 megaton explosion the figures would be
135, 200, 12.5 and 4.7. At 12 km the respective sets of figures would be: 1
megaton: 15, 31, 31, 3.7; and 10 megaton: 46, 96, 30 and 6.3.
Nuclear explosions also produce earth shocks in the neighbourhood of
surface or underground structures. These shocks can arrive by direct travel
through the earth from the point of detonation when this point is on or below
the surface. Whether the air induced earth shock front outruns or lags behind
the air blast wave front depends on the relationship between the air blast
wave velocity ū, and the seismic velocity for the soil.
Direct earth shock travels at the seismic wave velocity (cs), which varies
with soil properties as shown in Table 1.1.
At depths of 100 feet or less, and for distances (R) for which p0<200 psi,
the direct earth shock produces low accelerations, whereas air-induced earth
shock has much higher accelerations. Most protective structures are therefore
designed against air-induced earth shock only.
Since the dynamic pressure (q) is proportional to the square of the wind
velocity, it falls to zero some time later than the overpressure, and the dynamic
positive phase is larger than the overpressure positive phase. By the time the
wind stops blowing in a direction away from the centre of the explosion the
overpressure is negative. There is consequently a reversal of wind direction,
back towards the centre of the explosion, but at a relatively low velocity
because it is caused by the negative phase. As a concluding thought to this
Table 1.1 Soil properties and seismic wave
velocity
chapter, readers should be reminded that the formulae presented and the research
described were mainly concerned with the behaviour of explosions at or near
sea level. An examination of the dependence of blast analysis on ambient
pressure and temperature was made in 1944 by R.G.Sachs at the Aberdeen
Proving Ground, Maryland, USA, with particular reference to high altitude
explosions. It was reported in reference [1.21].
1.5 REFERENCES
1.1 Earnshaw, S. (1858) On the mathematical theory of sound, Phil. Trans., cl 133.
1.2 Lamb, H. (1895) Hydrodynamics, Cambridge.
1.3 Rankine, W.J.M. (1870) On the thermodynamic theory of waves of finite
longitudinal disturbance, Phil. Trans., cl 277, p. 530.
1.4 Hugoniot, H. (1887, 1889) Mémoire sur la propagation du mouvement dans
les corps et specialement dans les gaz parfaits, J. de l’ecole polytech., Paris, 57
(1887), and 58 (1889).
1.5 Davis, W.C. (1987) The detonation of explosives, Scientific American, 256(5),
May.
1.6 Taylor, G.I. (1939) The Propagation and Decay of Blast Waves, UK Home Office,
ARP dept, RC 39, October.
1.7 Taylor, G.I. (1940) Notes on the Dynamics of Shock Waves from Bar Explosive
Charges, UK Ministry of Home Security, Civil Defence Research Committee
paper.
1.8 Taylor, G.I. (1941) The Propagation of Blast Waves over the Ground, UK Ministry
of Home Security, Civil Defence Research Committee paper.
1.9 Christopherson, D.G. (1946) Structural Defence, 1945, UK Ministry of Home
Security, Research and Experiments Department, RC 450.
1.10 US Army Fundamentals of Protective Design (Non-nuclear) (1965), Dept of
Army Technical Manual TM5–855–1, Washington.
1.11 Kennedy, W.D. (1946) Explosions and explosives in air. In Effects of Impact
and Explosion, Summary Tech. Rep. DW2, NRDC, Washington, Vol. 1, Chap.
2.
1.12 Lindberg, H.E. and Firth, R.O. (1967) Tech. Rep. AFWL-TR-66–163, Vol. 2,
Air Force Weapons Lab, Kirtland, USA.
1.13 Stoner, R.G. and Bleakney, W. (1948) The attenuation of spherical shock waves
in air, Jour. Appl. Phys., 19(7), 670.
1.14 Goodman, H.J. (1960) Compiled Free-air Blast Data on Bare Spherical Pentolite,
BRL Report 1092, Aberdeen Proving Ground, Maryland, USA.
1.15 Philip, E.B. (1942) Blast Pressure Time-Distance Data for Charges of TNT and
GP Bombs, UK Home Office Report REN 168.
1.16 Lavoie, L. (1989) Fuel-air explosives, weapons and effects, Military Technology,
9.
1.17 Taylor, G.I. (1950) The formation of a blast wave by a very intense explosion, I
theoretical discussion, Proc. Roy. Soc., Series A, 201(1065), March.
1.18 Brode, H.L. (1959) Blast wave from a spherical charge, The Physics of Fluids,
2(2), March/April.
1.19 Friedlander, F.G. (1939) Note on the Diffraction of Blast Waves by a Wall, UK
Home Office ARP Dept, RC(A) July; also (1940) Diffraction of Blast Waves by
an Infinite Wedge, UK Ministry of Home Security, Civil Defence Research
Committee report RC 61, February.
1.20 Kirkwood, J.G. and Brinkley, S.R. (1945) Theory of Propagation of Shock Waves
from Explosive Sources in Air and Water, Div. 2, NRDC Rep. A 318, OSRD
No. 4814, March.
1.21 Sachs, R.G. (1944) The Dependence of Blast on Ambient Pressure and
Temperature, BRL report 466, Aberdeen Proving Ground, Maryland, USA.
ground that air bursts could be detonated (often the height of a tower). Most
bursts were therefore subject to the effects of reflection. Tests during the Second
World War suggested that the results for pressure and impulse were much
more scattered than for air bursts, but nevertheless a most useful data collection
was undertaken in 1946 by Kennedy [1.11]. This work was in connection
with the decision of the US National Defence Research Council to produce an
authoritative report on the Effects of Impact and Explosion, and Kennedy
contributed the second chapter on ‘Explosions and explosives in air’. This
was a historically valuable report, although the data was obtained from tests
with relatively undeveloped instrumentation.
Kennedy’s summary of results for the pressure-distance relationship in ground
burst tests is shown in Figure 2.2, taken from ref. [1.11], and his summary of
the impulse-distance relationships is given in Figure 2.3, taken from the same
reference. He proposed that the factor in Eq. (1.25) should be 0.054. He also
proposed factors of 0.036 for heavily cased charges and 0.081 for bare charges.
Data from ground bursts of hemispherical TNT charges between 5 and 500
tons in weight, and of nuclear charges ranging from 20 tons to 1.8 kilotons
were later analysed by Kingery [2.1], and have been reported in detail by
Baker [4.4]. Experimental results obtained at Suffield, Canada, in the 1960s
on TNT hemispherical ground-burst charges in the 500 ton range have been
Figure 2.2 Relationship between incident pressure and scaled distance for Second
World War bombs and bare charges (from Kennedy, ref. 1.11).
reported by Reisler et al. [2.2]. Typical results from all these tests are summarized
in Figure 2.4. It was generally concluded that for ground bursts a reflection
factor of 1.7 rather than 2.0 was appropriate, so the free-air burst equations
can be used with 1.7 W substituted for W. The impulse-range relationship for
the same series of tests is given in Figure 2.5.
Further test data for 20 ton spherical TNT charges, half buried in the ground,
have been presented by Reisler et al. [2.3]. These results show close agreement
with predictions based on the earlier analysis.
If the detonation of a high explosive occurs on or very near a flat surface,
there will be a local impulsive load delivered to the surface in addition to the
propagation of the blast wave through the air. This impulse, which is influenced
by the size and shape of the charge, produces a shattering effect, or a ‘brissance’.
The total impulse delivered to the flat surface was derived from test data,
and is quoted in [1.10], where it is given as
(2.1)
Figure 2.3 Relationship between scaled impulse and scaled range for Second World
War bombs (from Kennedy, ref. 1.11).
where a and b are the dimensions of the charge in contact with the surface,
2
and h is its height. The ratio ab/h is often called the shape factor.
Figure 2.4 Peak overpressure v ground range for 20 ton hemispherical TNT charges
(from Reisler et al., ref. 2.2).
Figure 2.5 Impulse v. ground range for 20 ton hemispherical TNT charges (from
Reisler et al., ref. 2.2).
The Mach stem is not always a straight line, but is usually assumed to be
so. As it travels outwards it acts in a similar way to the spherical shock front,
in that there is an instantaneous rise in overpressure. The Mach stem is initiated
when the angle of incidence (a) in Figure 2.6 exceeds 45°. In this figure we
have shown the reflected waves as semi-circular in cross section, but this is
not strictly correct because experiments show that the cross section more
nearly approaches the shape of a semi-ellipse.
The reflected shock wave travels faster than the incident shock wave because
the reflected overpressure is greater than the pressure existing in the incident
wave. This is a very fundamental aspect of the interaction of pressure waves
with surfaces which will be important later in the analysis of shock loads on
structures. The peak overpressure at the flat surface of the ground, pr, is related
to the peak incident overpressure, p0, by the formula
pr/p0=2(7pa+4p0)/(7pa+p0). (2.2)
This relationship was first given in the form shown in Figure 2.7 for a value of
pa of 14.7 psi. It applies at zero incidence, i.e. when a=0 in Figure 2.6. The
formula does not change much as a is increased from 0 to 30°, but at higher
values there is a noticeable change, as shown in Figure 2.8. It is intriguing that
for low values of incident pressure (i.e. p0=5 psi), the reflected pressure increases
as the angle of incidence changes from 40° to 55°. This is because blast waves
have finite amplitude, and the features of soundwave reflection (i.e. that incident
and reflected waves have equal strengths) no longer apply.
Figure 2.7 Relationship between peak reflected overpressure and peak incident
overpressure when pa=14.7 psi.
Figure 2.8 Variation of pr/p0 with angle of incidence (from Glasstone and Dolan,
ref. 2.9).
small adjustment of 2% was made to allow for the energy release of hexogen
as 1.02×TNT, although the pentolite results were not adjusted for the energy
release ratio of 1.22×TNT. The computed and experimental reflected shock
pressure ratios are plotted as a zone in the same figure.
where n=3, f is determined by the depth of burial and K is the soil constant
having the dimensions of a modulus of elasticity. The soil constant varies
tremendously, from 2000 psi (loam) to 100 000 psi (saturated clay) and 590
000 psi (limestone).
The form of Eq. (2.3) was proposed by Lampson [2.12] at the end of the
Second World War, and the relationship between p0 and R/W1/3 for a silty clay
Figure 2.9 Pressure ratio behind a regularly reflected spherical incident shock front
(from Dewey, Heileg and Reichenbach, ref. 2.11).
Figure 2.10 Relationship between peak overpressure and scaled range for
explosions in soil (from Lampson, ref. 2.12).
Table 2.1
Table 2.2
Figure 2.11 Relationship between the factor f and scaled cover depth for silty clay
(from Lampson, ref. 2.12).
Lampson’s work was still the accepted criteria for design in the 1960s,
when the US Department of the Army Technical Manual TM5–855–1 on the
fundamentals of protective design (non-nuclear) was published, and in the
1980s when further works on the design of protective structures were written.
Very little fundamental work of an unclassified nature on this subject has
been available to the general consulting engineer since the Second World War.
There is also little experimental data available on the size of the chamber in a
camouflet. Measurements in clay suggest that the chamber is usually spherical,
3
with a volume of about 10 ft per lb of TNT charge, and it is interesting to
note that the products of combustion following an explosion, when expanded
to the volume of the camouflet, would be approximately at atmospheric
pressure.
A study on air blast from underground explosions as a function of charge
burial was reported in 1968 by Vortman [2.13]. He was concerned with two
constituents of an explosion set just below the surface of a soil mass, the first
being the ground shock induced pulse which occurs when the ground shock is
transmitted across the interface between the ground and the air, and the second
from venting gases as the explosion erupts through a mound of rising earth.
The peak instantaneous overpressure from venting gases would clearly reduce
very much in magnitude as the burial depth increased, and by reviewing the
results of many tests covering the period 1951 to 1965, Vortman was able to
1/3
relate overpressure (psi) with scaled burial depth (ft/lb ) at various scaled
1/3
ground ranges. At a scaled ground range of 5 ft/lb the relationship between
ground shock induced peak overpressure (p0 in psi) and scaled burial depth
lay broadly within the band shown in Fig 2.12.
Much attention was given in the 1960s to the possibility of deep underground
nuclear test explosions to avoid fall-out problems, occurring at such a depth
that the only effects at ground level were associated with the propagation of
elastic stress waves through the soil. To support the UK investigations a
theoretical report on deep underground explosions was written by Chadwick,
Cox and Hopkins [2.14]. One of the reasons for their study was that, although
Hopkinson’s size scaling law (which has been used extensively so far in this
chapter) gives satisfactory results in most instances, there is a problem for
scaling the crater size of very large nuclear explosions when the effects of
gravity are no longer negligible. Thus craters from nuclear explosions become
relatively more shallow as the explosion size increases. Further, the onset of
plastic flow in a soil, as described by Coulomb’s law of failure, means that the
resistance to movement of a soil undergoing plastic deformation will increase
with depth.
Table 2.3
1/3
the depth of burial should not be less than 400 feet/KT , and should be increased
even more in media with a substantial water content.
Before leaving the underground detonation of high explosives and nuclear
weapons we must look briefly at three further aspects: positive impulse, the
formation and size of craters, and the vertical and horizontal movements of
the surface material near craters. The positive impulse per unit area of a pressure
wave in soil due to a TNT explosion can be presented empirically in the form
and the maximum positive impulse at a fixed target in the earth is three times
as great as this. This maximum impulse is sometimes written as
The average duration of the positive pressure wave in soil, when the depth
factor f=1, is
The fact that the pressure is maintained for a longer period against a fixed,
rigid target than in free earth is a special feature of pressure waves in earth. It
should be noted that the formulae given in Eqs (2.3), (2.4) and (2.5) refer to
pressures measured at the same depth as the charge. Pressures measured at
the same range, but at different depths, will not be as great. As an example,
for a high explosive charge buried at the optimum depth H=2W1/3 ft, the
pressure at the free soil surface will be 0.6 of that at the level of the charge.
When a charge explodes close enough to the surface to form a crater rather than
a camouflet, the dimensions of interest are the crater diameter (D) and apparent
depth h. The latter is the distance from the free surface to the uppermost surface of
the crater debris, whereas H is the distance below the surface at which the charge
was exploded. There are so many factors contributing to the final shape and size of
the crater that there is considerable scatter in experimental results, and there is also
a general shortage of experimental work in this field. Christopherson [1.9] reported
on British tests carried out during the Second World War, and he began by classifying
the crater types shown in Figure 2.13 for an explosion in clay.
In this diagram the type A crater is formed by an explosion on or very near
the surface of the soil. The crater has a relatively clean surface, clear of rubble
and debris. Type B crater, formed by a deeper explosion, contains much more
debris that has fallen back and often has overhanging shoulders of the original
soil. For a 250 kg charge in clay the depth of the explosion for type B would be
about 10 feet. In type C, explosion depth 20 feet, the whole of the initial chamber
of compression is filled with debris, which will not spread much beyond the
limits of the crater. Christopherson made the point that most of the delay-fused
bombs dropped in Britain during the Second World War penetrated to about 20
feet and produced type C craters. The deeper explosions (30 ft) produced a
camouflet and a chimney, which we discussed earlier in relation to nuclear
explosions, and deeper still (35 ft) the 250 kg charge in clay produced a spherical
camouflet. These crater shapes were originally noted by Walley [2.15] following
trials at Brancaster, where the soil was waterlogged blue clay. Later he surveyed
earth shock effects in other types of soil, including the general classifications of
clay, chalk and sandstone. The results are given in terms of the crater depth
1/3 1/3
parameter h/W , the depth of the explosion parameter H/W , and also the
1/3
crater diameter parameter D/W , in Figure 2.14.
Crater dimensions were also reviewed in Britain by Colonel F.W.Anderson
in 1942 [2.16]. He suggested that the results could be predicted in different
materials by introducing a soil factor to modify W for a given burial depth H.
Taking clay as the basis (factor=1.0), he proposed factors of 2.75, 1.50 and
2.00 for chalk, soft sand and soft sandstone respectively. The British also
looked at crater effects in 1945, when J.S.Arthur [2.17] produced a review of
comparative performance between different types of high explosive. He also
proposed an explosive factor to modify W for a given burial depth H. Taking
TNT as the basis (factor=1.0) he suggested factors of 1.0 for Amatol, 1.20 for
Figure 2.13 Types of crater for explosions in clay, reported by Christopherson (ref.
1.9) in 1945.
RDX/TNT (60/40) and 1.75 for aluminized explosives such as Torpex and
Minol. At about this time an attempt was made by Devonshire and Mott
[2.18] to relate camouflet volume and crater dimensions to the energy available
in the explosion, and their work was later used to support the analysis of deep
underground explosions by Chadwick, Cox and Hopkins that we discussed
earlier.
Lampson included crater dimensions in his chapter on ‘Explosions in earth’
in reference [2.19], when he reviewed the wartime experiments for the National
Research and Development Committee in the USA in 1946. His review included
the British results, so it is not surprising that his proposed relationships between
scaled crater dimensions and scaled depth of burial were very similar to those
Figure 2.14 Crater dimensions related to the depth (H) of the explosion. Second
World War tests reported by Walley (ref. 2.15).
From Figure 2.11 f reached a maximum of unity when H/W1/3=2.0 ft/lb1/3, and
the curve of scaled crater diameter for clay also reaches a maximum at this
value. The crater curves for chalk and sand, however, reach a maximum at a
value of H/W1/3 of about 1.5 and 1.0 respectively, which confirms that Figure
2.11 can only be strictly applied to clay soils.
When a crater is formed there are vertical and horizontal movements of
the surface material in its vicinity. The surrounding material is violently displaced
Figure 2.15 Relationship between scaled crater dimensions and scaled depth of
burial of the charge (from Lampson, ref. 2.12).
Up/W1/3=C1(R/W1/3)–4, (2.9)
Vp/W1/3=C2(R/W1/3)–4, (2.10)
Um/W1/3=47.3(R/W1/3)–3+0.216(R/W1/3)–1, (2.11)
Vm/W1/3=12.6(R/W1/3)+0.032(R/W1/3)–1. (2.12)
Table 2.4
The distance from the centre of an underground explosion at which the face of
a tunnel or trench would no longer suffer any disruption is known as the radius of
rupture, Rr. Early British military handbooks on mining suggested that
(2.14)
(2.15)
This equation is identical in form to that for explosions in air, see for example
Eq. [1.25], but the underwater impulse for the same charge is about thirty
times as large.
Kirkwood and Brinkley [2.29] calculated that the energy in the underwater
shock wave diminishes rapidly as the wave progresses outwards, about 30%
is dissipated within 5 radii of the charge and 48% within 25 radii. Energies
found from experimental pressure/time curves are generally 25% lower than
these calculations predict. The effect of substituting TNT with other types of
explosive was examined, and generally it was found that the differences are
Figure 2.17 Comparison of calculated pressures for TNT (from Cole, ref. 2.22).
R=distance travelled by shock wave; a0=original charge radius.
not as great as in air or earth. Table 2.5 lists the equivalent weights for explosives
of the 1940s given by Christopherson [1.9].
The first underwater nuclear test, code-named Baker, was carried out at
the Bikini Atoll in 1946, when a nuclear bomb having a yield equivalent to 20
kilotons of TNT was detonated below the surface of the lagoon (200 feet
deep). As the shock wave initiated by the expanding hot gas bubble reached
the surface a rapidly expanding ring of darkened water was visible, followed
by a white area as the shock wave was reflected by the surface. A column of
water, the spray dome as it was called, was then formed over the point of
burst. Its upward velocity was proportional to the peak pressure in the shock
wave, and in the test it began to form 4 msec after the explosion. The upward
velocity was initially 2500 ft/sec. A few milliseconds later the gas bubble reached
the surface and a hollow column of gases vented through the spray dome. It is
said in ref. [2.9] that the column was 6000 ft high with a maximum diameter
of 2000 ft, and that 1 million tons of water were raised in the column. As the
column fell back a condensation cloud of mist developed at its base and surged
outwards. This bun-shaped cloud, known as the base surge, developed about
10 seconds after detonation, initially reaching a height of 900 ft. After about
4 minutes the height was double this, and the diameter about 3.5 miles. One
of the first deep underwater explosions (in 2000ft of water) took place in
1955, code-named Wigwam. During the pulsations of the gas bubble the water
surrounding it had a large upward velocity and momentum, and the spray
dome broke through the surface at a high speed. It is interesting to note that if
the surface breakthrough occurs when the bubble is in the ‘below ambient’
phase, the spray column breaks up into jets that disintegrate into spray not
unlike the vision of Moby Dick ‘blowing’.
The analysis of the action of the gas bubble, whether from a nuclear or
conventional underwater explosion, was a fruitful field of investigation during
the 1940s. In 1941 Edgerton photographed the pulsations of the bubbles
when detonator caps were exploded at various depths, and at about the
same time Swift [2.30] and others filmed bubble motion when 0.55 lb of the
explosive Tetryl was detonated 300 ft below the surface. The radius of the
gas sphere for this explosion is related to time in Figure 2.18, taken from the
Figure 2.18 Radius/time curve of the gas sphere for 0.55 lb of Tetryl detonated 300
ft below the surface (from Swift, ref. 2.30).
latter work. Note that the reversal of the bubble motion is virtually
instantaneous and discontinuous. A comparison of radius and period, and
the formulation of a simple analysis, had been made much earlier by Ramsauer
[2.31] in 1923. He determined the position of the gas bubble boundary by
means of an electrolytic probe, when charges of gun cotton weighing one or
two kilograms were exploded at depths up to 30 feet in 40 feet of water.
Although he could only measure the radius of the bubble and the time for
isolated points, nevertheless it was a very useful and innovative study. He
found that the variation of maximum bubble radius with depth could be
predicted by the formula
(2.17)
(2.18)
Figure 2.19 Measured and calculated radius of the gas sphere from a detonator one
foot below the surface (from Herring, ref. 2.33).
Since the total energy associated with the radial flow of water (Y) is given
approximately by
(2.19)
(2.20)
(2.21)
where T1 and T2 are the periods, and r2 and r1 are the fractional energies
remaining after the first and second contractions.
For depths greater than 320 feet T2/T1 is very nearly constant and independent
of depth, and is equal to about 0.6. This gives an energy ratio of 0.36.
T=4.32W1/3/(d+33)5/6, (2.22)
(W in lb, d in feet)
His calculations for peak pressure p0 in psi, duration in seconds, and distance
R from the detonation point in feet are given in Table 2.6.
There are further fundamental areas of physical behaviour that must be
mentioned: the reflection of the underwater shock wave downwards from the
free surface of the water; the reflection upwards from the bottom of the sea,
lake or river; and the generation of surface waves. In considering surface
reflection, it is usually assumed that the explosion is sufficiently deep for the
pressure pulse to behave as an intense sound pulse. This means that the depth
is at least 12 charge diameters. At the surface there will be a transmitted pulse
Figure 2.20 Bubble periods against depth for 0.55 lb Tetryl charges (from Ewing
and Crary, ref. 2.32).
Table 2.6
(into air) and a reflected pulse which must be added to the original pulse.
Because of the large difference in density and compressibility between air and
water the transmitted pulse pressure will be very small and is usually neglected.
The surface is then taken as undisturbed, which means that the reflected pulse
must be equal and opposite to the initial pulse. This makes calculation of the
effect of downwards reflection very simple, since at any point P it is a
combination of a positive incident pressure at a distance r from the centre of
the explosion and a negative rarefaction or ‘tensile’ pulse originating from an
imaginary point which is at the image of the centre of detonation in the air
immediately above this centre (i.e. the concept of images as used in the reflection
of sound waves). The combination can theoretically produce a sharp decrease
or ‘cut-off’ in the shock at point P, although in practice the decrease is more
gradual.
Reflection of a pressure pulse from the seabed can vary considerably,
depending on the state of the bed surface. On a rocky bottom the main pulse
may give rise to a reflected wave having a peak pressure over half that in the
main pulse; on a soft mud base the reflected pressure could be almost negligible.
In general, the effect of the seabed is positive, leading to additional pressure.
An upper limit to this additional pressure can be found by assuming that
complete reflection occurs and that the concept of images can be used to
calculate total pressure at a given distance from the centre of detonation.
Some examples were given by Cole in ref. [2.22]. Apparently the peak
pressure, impulse and energy flux density 60 ft from a 300 lb TNT charge
fired on a bottom of hard-packed sandy mud, was increased by 10, 23 and
47% over the values observed from a charge at mid-depth. These increases
correspond to increasing the weight of a charge in free water by between 35
and 50%, rather than by a factor of 2, so a good deal of shock wave energy
was transmitted and dispersed on the bottom.
Surface waves can be generated by underwater explosions or by air bursts
close to the water surface. In the case of underwater explosions the outward
propagating waves result from the gas bubble breaking through the surface,
and in deep water explosions it is usually assumed that between 2% and 5% of
the yield of the charge is directed to the energy of the train of surface waves.
When the detonation is in relatively shallow water, there is often only an initial,
solitary wave, particularly for very large nuclear explosions. After the 20 kiloton
shallow water explosion at Bikini, mentioned earlier, wave heights and times of
arrival were recorded at increasing distances from the detonation point (surface
zero). At distances of 330, 2000 and 4000 yards the arrival times were 11, 74
and 154 seconds respectively, and the wave heights were 94, 16 and 9 feet.
Characteristic properties of the underwater shock wave from nuclear
explosions have been given by Glasstone and Dolan [2.9]. For water with no
reflections or refractions the relationship between peak instantaneous underwater
shock pressure in psi and the slant range in thousands of yards is given in Figure
2.21. Approximate relationships have been established for surface wave amplitude
(H) and radius (R) from surface zero for nuclear explosions as follows:
1/4 1/4
In deep water, 850 W >d>256 W , where W is yield in kilotons of TNT
equivalent, and d is water depth in feet,
H=40 500 W.54/R, (2.23)
The creation of surface waves by air bursts close to the surface had considerable
military significance, and much of the research and testing during the 1960s
was classified and not available through open publication. This is still thought
to be the case. However, a paper by Kranzer and Keller [2.36] in 1959 was
published in the Journal of Applied Physics, and their theory is now generally
accepted as predicting wave train amplitudes for detonations of known size
and position. They applied their general analysis of the distribution of impulse
acting on the surface, and of the depression or elevation of the surface under
explosive impulse. They suggested that the wave pattern contains a single
1/2
maximum which moves outwards with a velocity 0.42 (gH) , where H is the
height of the explosion above the surface. The wavelength at this maximum is
1/2
4.4 H, and the period is 5.1 (H/g) . The analysis seems to be invalid when
H=0, i.e. the explosion takes place on the surface, for it gives values of velocity
and wavelength of zero. This is clearly not true. The authors discuss this and
admit that the analysis is only suitable for larger values of H.
The work of Cole [2.22], which has been so useful in making this survey, has
been summarized in more recent years by Kaye [2.37], and there is a useful
2.6 REFERENCES
2.1 Kingery, C.N. (1968) Parametric Analysis of Sub-kiloton Nuclear and High
Explosive Blast, BRL Report 1393, Aberdeen Proving Ground, Maryland, USA.
2.2 Reisler, R.E., Keefer, J.H. and Griglio-Tos, L. (1966) Basic Air Blast Measurements
from a 500 Ton TNT Detonation, Project 1.1, Operation Snowball, BRL memo
report 1818, Aberdeen Proving Ground, Maryland, USA.
2.3 Reisler, R.E., Griglio-Tos, L. and Kellner, R.C. (1966) Ferris Wheel Series, Flat
Top event, Project Offices Report, Project 1.1, Airblast Phenomena POR-3001,
Aberdeen Proving Ground, Maryland, USA.
2.4 Mach, E. and Sommer, J. (1877) Uber die Fortpflanzungsgesch windigkeit von
Explosions schallwellen, Akademie der Wissenschaften, Sitzangberichte der
Wiener, Vol. 74.
2.5 Sternberg, J. (1959) Triple shock wave intersections, The Physics of Fluids,
2(2), American Institute of Physics, March/April.
2.6 von Neumann, J. (1943) Oblique Reflection of Shocks, Explosives research
rep. No. 12, Buord, US Navy Dept.
2.7 von Neumann, J. (1943) Collected Works, Vol. 6, 239, Pergamon.
2.8 Dewey, J.M. and McMillin, D.J. (1985) Journal of Fluid Mechanics, 152, 67.
2.9 Glasstone, S. and Dolan, P.J. (1962) The Effects of Nuclear Weapons, US Depts
of Defense and Energy, Washington.
2.10 Satori, L. (1983) Physics Today, March.
2.11 Dewey, J.M., Heilig, W. and Reichenbach, H. (1985) Height of burst results
from small scale explosions, Proceedings (11) of 9th International Symposium
on Military Applications of Blast Simulation, Oxford, England, September.
2.12 Lampson, C.W. (1946) Effects of impact and explosions, Explosions in Earth,
NRDC Washington, USA, Vol. 1, Chapter 3.
2.13 Vortman, L.J. (1968) AirBlast from Underground Explosions as a Function of
Charge Burial, Prevention of and Protection against Accidental Explosion of
Munitions, Fuels and Other Hazardous Mixtures, New York Academy of Sciences,
ed. E.Cohen, Vol. 152, Art. 1.
2.14 Chadwick, P., Cox, A.D. and Hopkins, M.G. (1964) Mechanics of deep underground
explosions, Phil. Trans. Roy. Soc., Series A, No. 1069, Vol. 256, April.
2.15 Walley, F. (1944) Note on Water Formation in Puddle Clay, Brancaster Beach,
UK Home Office Research Report REN 318, January.
2.16 Anderson, F.W. (1942) Crater Dimensions from Experimental Data, UK Ministry
of Home Security, Civil Defence Research Committee, Report RC 344, September.
2.17 Arthur, J.S. (1945) The Comparative Performance of Various Bomb Fillings— Crater
and Earthshock Effects, UK Home Office Research Report REN 520, June.
2.18 Devonshire, A.F. and Mott, N.F. (1944) Mechanism of Crater Formation,
Theoretical Research Report No. 26/44, UK Armament Research Dept (AC
6995), July.
2.19 Lampson, C.W. (1946) Effects of impact and explosion, Explosions in Earth,
Summary Tech. Dept. DW2, NRDC Washington, USA, Vol. 1, Chapter 3.
2.20 Hilliar, H.W. (1919) Experiments on the Pressure Wave Thrown out by
Explosions, UK Admiralty Experimental Station Report.
2.21 Hartmann, G.K. (1946) Taylor Model Basin (TMB), Report, 531.
2.22 Cole, R.H. (1965) Underwater Explosions, Princeton University Press 1948,
and Dover Publications, New York.
2.23 Penney, W.G. (1940) The Pressure-Time Curve for Underwater Explosions, UK
Ministry of Home Security, Civil Defence Research Committee Report RC 142.
2.24 Penney, W.G. and Dasgupta, H.K. (1942) Pressure-Time Curves for Submarine
Explosions (2nd paper), UK Ministry of Home Security, Civil Defence
ResearchCommittee Report RC 333 .
2.25 Kirkwood, J.G. and Bethe, H.A. (1942) The Pressure Wave Produced by an
Underwater Explosion—Basic Propagation Theory, US OSRD Report 588.
2.26 Kirkwood, J.G. and Montroll, E.W. (1942) The Pressure Wave Produced by an
Underwater Explosion—Properties of Pure Water at a Shock Front, US OSRD
Report 676.
2.27 Kirkwood, J.G. and Richardson, J.M. (1942) The Pressure Wave Produced by
an Underwater Explosion—Properties of Salt Water at a Shock Front, US OSRD
Report 813.
2.28 Kirkwood, J.G., Brinkley, S.R. and Richardson, J.M. (1953) The Pressure Wave
Produced by an Underwater Explosion—Calculations for Thirty Explosives,
US OSRD Report 2022.
2.29 Kirkwood, J.G. and Brinkley, S.R. (1945) Theory of the Propagation of Shock
Waves from Explosive Sources in Air and Water, US OSRD Report 4814.
2.30 Swift, E. et al. (no date) Photography of Underwater Explosions, US Navy
Bureau of Ordnance.
2.31 Ramsauer, C. (1923) Ann. d. Phys., 4(72), 265.
2.32 Ewing, M. and Crary, A. (1941) Multiple Impulses from Underwater Explosions,
Woods Hole Oceanographic Institution report.
2.33 Herring (1941) Theory of the Pulsations of the Gas Bubble Produced by an
Underwater Explosion, US NRDC Division 6 Report C4-Sr20.
2.34 Taylor, G.I. (1963) The vertical motion of a spherical bubble and the pressure
surrounding it. In The scientific papers of Sir Geoffrey Ingram Taylor, Vol. 3,
Cambridge University Press, 320.
2.35 Bryant, A.R. (1950) The behaviour of an underwater explosion bubble: further
approximations, Underwater Explosion Research, Vol. 2, The gas globe office
of Naval Research, US Dept of Navy.
2.36 Kranzer, H.C. and Keller, J.B. (1959) Water waves produced by explosions,
Journal of Applied Physics, 30(3), March, 398.
2.37 Kaye, S.H. (1983) Encyclopaedia of Explosions and Related Terms, US R and
D Command, Large Calibre Weapons Systems Laboratory, Technical Report.
2.38 Smith, P.D. and Hetherington, J.G. (1994) Blast and Ballistic Loading of
Structures, Butterworth Heinemann, Oxford.
2.39 Henrych, J. (1979) The Dynamics of Explosion and Its Use, Elsevier, Amsterdam
(translation).
2.40 Makovicka, D. (1994) Influence of short shock load on response of masonry
structure. In P.S.Bulson (ed.), Structures Under Shock and Impact (3),
Computational Mechanics Publications, 53.
2.41 Kingery, C.N. and Bulmash, G. (1984) Airblast Parameters from TNT Spherical
Air Burst and Hemispherical Surface Burst, Tech. Report ARBRL-TR-02555,
US Army Armament Research and Development Center, Ballistics Research Lab.,
Aberdeen Proving Ground, Maryland, USA.
are less than half those of an equivalent TNT explosion, but at greater distances
the peaks are more nearly equal.
A study of the fireball from an exploding rocket was given by High [3.2] in
1968, and he collected experimental results relating diameter and duration to
total propellant weight in lbs; his empirical relationships, derived by a least
squares regression analysis of the data, were
D=9.82W0.32, (3.1)
T=0.232W0.32, (3.2)
as the period 1958 to 1975 there have been over 20 grain elevator dust explosions
in Nebraska and Iowa, and over five feed mill dust explosions in Iowa and
New York. The number of explosions related to total storage capacity during
this period was greatest in Tennessee, Alabama and Nebraska. What is ominous
is that in over 60% of recorded grain elevator and feed mill dust explosions
the location and ignition source of the primary explosion remains unknown.
Grain dust and coal dust sustain similarly sized particles, in the size range
between 1 and 150 micron. The determination of representative linear
dimensions is not easy, because most dust particles are irregular in shape. In a
grain elevator there are four types of dust: dump pit dust, belt-loading dust,
main elevator dust and bean dust. About 95% of dump pit dust consists of
bees’ wings, the belt loading dust is mainly starch dust, and the main elevator
dust is a 60:40 mix of bees’ wings and starch. All these dusts are light, fine and
low in ash; they prefer to stay in suspension and are easy to inflammate. The
presence of so many bees’ wings is intriguing. According to Matkovic [3.4]
they are flat and round and difficult to control. They wander!
Research into the physics of dust explosions has concentrated on experimental
work at laboratory scale and some attempt at theoretical predictions. Tanaka
[3.5] suggests that there are three fundamental conditions to be investigated,
the ignition point, the minimum concentration of dust at which an explosion
can occur, and the flame propagation velocity of the explosion. One of the first
analytical studies of ignition was made in 1936 by Parker and Hottel [3.6], and
more recently, in 1959, Cassel and Liebman [3.7] considered a spherical space
of dust cloud in which the particles were uniformly distributed. The rate of heat
generation from the cloud is a function of the rate of oxidation of the constituent
particles, and the rate of heat leaving the cloud is a function of the coefficient of
transfer from solid to gas, the thermal conductivity of air, particle size and
ignition temperature. There appears to be a most dangerous particle size giving
the lowest ignition temperature, as shown in Figure 3.1 which relates ignition
temperature with particle diameter for coal dust experimentally and theoretically.
The experimental data is due to Kurosawa [3.8].
The minimum requirement for a flame to propagate is that the burning out of
the particle occurs at the same time as the ignition of the next particle to it. Using
this as a basis the minimum explosive limit concentration can be calculated as a
function of particle diameter. Computation and experiments show that for clouds
of plastic particles such as ethyl cellulose or cellulose acetate, the relationship
between concentration in milligrams per litre and diameter in centimetres is
approximately linear, and at a particle diameter of 0.05 cm the minimum explosive
limit concentration is between 400 and 500 mg/litre. Flame propagation velocity
appears to approach a maximum of 40 to 50 m/sec for a number of materials,
reaching this value at a time of between 0.01 and 0.03 seconds from initiation.
Dust cloud explosions, particularly in agricultural products, have been a
problem in Japan. A great deal of grain is imported, mainly from the USA,
and stored in grain elevators at sea ports. Between 0.1% and 1% of bulk
Figure 3.1 Comparison of theoretical and experimental ignition points for coal dust
(from Kurosawa, ref. 3.8).
grain is in the form of powder, which forms a suspended dust with an explosive
strength much higher than that of coal dust. Between 1962 and 1975 there
were 23 agricultural dust explosions, killing 8 people and injuring 49. The
main cause was sparks from welding processes in bucket elevators, chutes,
bins and pipes. Because of this a number of university experimental research
programmes were undertaken, using apparatus to generate uniform dust clouds
over a wide range of concentrations. In one of these studies, reported by
Enomoto [3.9], the apparatus consisted of a cylindrical explosion chamber,
diameter 270mm, 10 litres volume, that could be rotated to make a uniform
dust cloud within it. Ignition at the centre of the chamber was by 0.1 gram of
guncotton, detonated electrically. It was found that a maximum explosion
pressure of about 6.5 kg/cm2 occurred at a dust concentration of approximately
1000 g/m3 for cornstarch, potato starch and wheat flour. Maize and wheat
dust produced explosion pressures between 4.5 and 5.0 kg/cm2 over a range
of concentrations between 1000 and 4000 g/m3. The rate of pressure rise for
wheat dust reached a maximum of 60 kg/cm2/sec at a concentration of about
1500 g/m3, but the average rise was about half this. The relationship between
the maximum rate of pressure rise (dp/dt)m and the ratio of the peak pressure
(pm) and initial pressure (pi) is given for many types of dust by the equation
(dp/dt)m=0.7(pm/pi)2.5, (3.3)
(dp/dt)av=0.3(pm/pi)2.5. (3.4)
reported by Palmer [3.11] in a book on dust explosions and fires. The sugars
were dextrose, sucrose and raffinose, and the rates of pressure rise (dp/dt) for
any given pressure were found to be given by an equation of the form
(dp/dt)=Ap–Bp2. (3.5)
and subsequently to the use of flickering oil lamps and carbide lamps. To help
search for accumulations of methane gas, which is odourless and tasteless,
the miners once took canaries underground, but more sophisticated gas detectors
have been introduced over the years. These included a permissible flame safety
lamp that would not become hot enough to ignite gas, but which would burn
with a blue cap above the yellow flame if gas were present. The best answer to
the gas hazard in mines is an efficient ventilation system, and most mines
have a carefully planned system of passages, ducts, fans and air blowers to
ensure that gases are diluted and driven out.
The accumulation of methane is not confined to coal workings, of course,
and there have been examples of methane explosions in civil engineering
workings in very recent times. A notorious case was the explosion at the
Abbeystead waterworks in the Lancashire fells in May 1984. An invisible
cloud of methane gas rushed out of a tunnel that should have been full of
water, a cigarette was lit and the subsequent explosion killed 16 people.
Apparently the tunnel had been designed without ventilation shafts, and
insufficient allowance was made for the possibility of pockets of methane
finding their way into the tunnel. An important consideration was that under
conditions of high pressure methane can be soluble in water. After the accident
there was much scientific discussion about the source of the gas, and the
Isotope Measurements Laboratory at Harwell found that most of the methane
was over 20 000 years old, and had probably been lurking in rock fissures
along the path of the tunnel under Lee Fell. The safety implications were
profound, and resulted in legal action, discussion of which is beyond the scope
of this book; the structural damage, which resulted in the destruction of the
valve house, has been discussed in the journals of professional institutions.
Methane is a paraffin hydrocarbon, and is a product of the anaerobic bacterial
decomposition of vegetable matter under water, hence the name marsh gas.
The activated-sludge process of sewage disposal also produces methane-rich
gases. It forms an explosive mixture with air when the methane (CH4) content
is between 5.3% and 13.9% by volume, and is the major constituent of
commercial natural gas (after the extraction of gasoline), used for heat and
power in industrial and domestic establishments.
This leads us to coal gas, which was the first domestic gas, produced when
coal is heated out of contact with air to produce a stable residue, coke, and a gas
with a heating value of about 500 BTUs per cubic foot. The coke residue was
sometimes gasified in steam to produce ‘water’ gas. The first practical application
for domestic use was developed by William Murdock at the end of the eighteenth
century, in London. The famous Gas Light and Coke company was established
in 1812. The main constituents of coal gas are hydrogen (about 50%), methane
(about 25%), carbon monoxide (10%) and nitrogen (8%). These quantities
varied a little depending on the precise design of the manufacturing retorts.
Coal gas was first used for cooking stoves by James Sharp of Northampton
in 1830, and about thirty years later the gas stove industry had penetrated the
USA. Gas for domestic lighting was overtaken by the electric light earlier this
century, and the main domestic uses became cooking and heating. With this
growth came the danger of domestic explosions. Eventually coal gas in the
UK was replaced by the distribution of natural gas, and the local gasworks
that were a feature of every country town have become obsolete in the past
forty years or so. Natural gas (94% methane), produced from petroleum
feedstocks, was used extensively in the USA, and began to be used in this
country in the 1960s. It was relatively odourless, and the lack of the distinctive
smell of coal gas meant that gas leaks were less likely to be detected. The
odourization of natural gas from the North Sea is now undertaken at the
storage installations on the mainland.
Explosions of domestic gas in the UK were running at about 100 per year
in the 1970s and causing about 10 deaths per year. An inquiry into serious gas
explosions, held in the late 1970s [3.16], noted that one-third of all explosions
were caused by gas escaping from distribution mains and service pipes on the
outside of buildings, and two-thirds from leaks at meters, appliances or
installations within buildings. External gas escapes were frequently connected
with the fracturing of cast-iron mains due to ground movement rather than
corrosion. Often escaping gas in winter could not be released upwards because
of frosted ground, and entered buildings through cellars or other service ducts.
It was natural at the time for suggestions to arise that natural gas represented
a greater explosive hazard than the old town gas, particularly as the countrywide
distribution of natural gas meant an all-steel high pressure network. This
comparison was investigated by an earlier inquiry [3.17], and it was concluded
that the hazards from the use of both gas systems were similar. Natural gas
3
has double the calorific value in BTU/ft , a much greater distribution pressure
of about 30mbar (about three times), and a lower maximum burning velocity
(about half). Natural gas burner pressures are reduced by governing to about
21 mbar.
A domestic gas explosion is of the deflagration type, with a finite time
from ignition to maximum energy release. The pressure rise is relatively small
and direct injury due to this pressure increase (e.g. ear drum damage) is rare.
Gas explosion fatalities usually occur from flying or falling objects, or by
burns. Research has shown that from a total of 39 incidents, peak dynamic
2
pressures from town gas explosions averaged 13 KN/m , and from natural
2
gas explosions 11 KN/m . The air-rich mixture at which ignition occurs if a
source of ignition is present is about the same for both gases at approximately
7.5% by volume of gas. There are important differences in flame propagation
properties, however, with town gas having a propagation capacity about 5.5
times as great as for natural gas.
The effect of gas explosions in underwater tunnels is important, because
the tunnel walls would usually be designed to resist external loads due to
water and soil. In the event of an internal gas explosion the tunnel walls could
experience a load reversal for which they had not been designed. The pressures
generated by internal gas explosions have been investigated in Holland [3.18]
and from these experiments the pressure-time history has been recorded. The
duration of the overpressure plateau of 6 to 7 bar is governed by the time
needed to vent the overpressure. The peak instantaneous pressure of 25 bar
2
(2.5 N/mm or 362 psi) represents a formidable internal load on the structure.
the leak of flammable material occurs for a sufficient time and rate for a
significant amount to form an explosive mixture with air, then there is a strong
possibility that an explosion will occur if there is a source of ignition. Experience
shows that most major leaks occur in external piping systems, atmospheric
vents, pumps and compressors, and that the most vulnerable part of any system
is where smaller pipes and branches join large diameter, rigid pipes. If an
explosion does not occur there is always the possibility of a flash fire or slow
burning. Factors that influence the size and effect of a vapour cloud explosion
include the amount of material in the cloud, the energy of the ignition source,
turbulence, flame speed and wind direction. If a vapour cloud is above its
upper flammability limit it can burn at a relatively slow rate; if it is below this
limit but above the lower flammability limit an explosion can occur.
Because of the relatively slow pressure rise, and the absence of sudden
shock, it is rather misleading to represent unconfined vapour cloud explosions
by an equivalent amount of TNT. However, this is often done, and the evaluation
of damage from explosions has been used to find equivalent values of peak
instantaneous pressure and positive duration; it has been suggested that
2
representative TNT explosions are 70 KN/m peak pressure, duration 20 ms,
2
or 20 KN/m for a duration of 100ms, depending on the span of the structural
component under consideration. In a large unconfined explosion the pressure
could rise to as much as 8 atmospheres over a period of one second.
There have been a number of catastrophic vapour cloud explosions in recent
times, on land, offshore and at sea. Among the most well known occurred in
1974, when the cyclohexame plant at Flixborough, in the UK, exploded with
devastating results and the death of 28 people. The leakage of flammable
vapour came from damaged pipeworks. A second remarkable catastrophe
was the explosion and fire at the Piper Alpha offshore oil platform in 1987,
again due to the leakage of flammable vapour from pipework. The importance
of accurate design and careful fabrication of structural pipes and their
connections is clear. The inquiry into the Piper Alpha explosion noted the
difficulties in making explosion hazard assessments of offshore platforms,
because the explosion process is so difficult to analyse. Much of the available
experimental data is derived from relatively small-scale tests, and it is not yet
certain how far this can be applied to full-scale explosions. Further, most tests
take place in vented chambers, or in chambers with obstacles, so the ‘free
field’ conditions are not known.
The Piper Alpha technical investigation discussed the primary cause of the
accident and considered in detail two likely explanations. It was clear, however,
that there were two explosions, the first occurring in the gas compression
equipment area. This caused a serious oil fire with large amounts of
hydrocarbons to fuel it, which in turn caused the rupture of a major gas pipeline.
This led to the initiation of the explosion and fireball that destroyed the
installation. The preferred explanation for the first explosion was a gas leak
from a condensate injection pump, caused by defective pipework.
3.6 REFERENCES
3.1 Fletcher, R.F. (1968) Characteristics of liquid propellant explosions, Prevention
and Protection against Accidental Explosion of Munitions, Fuels and Other
Hazardous Mixtures, New York Academy of Sciences, 432.
3.2 High, R.W. (1968) The saturn fireball, Prevention and Protection against
Accidental Explosion of Munitions, Fuels and Other Hazardous Mixtures, New
York Academy of Sciences, 441.
3.3 Chiotti, P. (1977) An overview of grain dust explosion problems, Proc. Int.
Symp. on Grain Dust Explosions, Minneapolis, US Grain Elevator and Processing
Society, 13.
3.4 Matkovic, M. (1977) Dust composition, concentration and its effects, Proc.
Int. Symp. on Grain Dust Explosions, Minneapolis, US Grain Elevator and
Processing Society, 62.
3.5 Tanaka, T. (1977) Predicting ignition temperature, minimum explosive limit
and flame propagation velocity, Proc. Int. Symp. on Grain Dust Explosions,
Minneapolis, US Grain Elevator and Processing Society, 79.
3.6 Parker, A.S. and Hottel, H.C. (1936) Ind. Eng. Chem., 28, 1334.
3.7 Cassel, H.M. and Liebman, I. (1959) Combust Flame, 3, 467.
3.8 Kurosawa, M. (no date) Chem A6, 53, 8584.
3.9 Enomoto, H. (1977) Explosion characteristics of agricultural dust clouds, Proc.
Int. Symp. on Grain Dust Explosions, Minneapolis, US Grain Elevator and
Processing Society.
3.10 Meek, R.L. and Dallavalle, J.M. (1954) Ind. Eng. Chem., 46, 763.
3.11 Palmer, K.N. (1973) Dust Explosions and Fires, Chapman & Hall, London.
3.12 Kjäldman, L. (1987) Numerical Simulation of Peak Dust Explosions, Technical
Research Centre, Finland, Nuclear Engineering Laboratory, Research report
469.
3.13 Essenhigh, R.H. (1977) Problems of ignition and propagation of dust clouds,
Proc. Int. Symp. on Grain Dust Explosions, Minneapolis, US Grain Elevator
and Processing Society.
3.14 Schofield, C. (1984) Guide to Dust Explosion Prevention, Part 1—Venting,
Institution of Chemical Engineers, Rugby, England.
3.15 Bartknecht, W. (1982) Explosions: Cause, Prevention and Protection, Springer
Verlag, Berlin.
3.16 Report of the Inquiry into Serious Gas Explosions (Chairman: P.J.King) (1977)
HMSO, London.
3.17 Report of the Inquiry into the Safety of Natural Gas as a Fuel (Chairman: F.
Morton) (1970) HMSO, London.
3.18 Investigation of Transporting Dangerous Materials through Tunnels (1982)
Rijkswaterstaat, Directie Suizen en Stuwen, Utrecht, The Netherlands.
3.19 Anderson, R.P. and Armstrong, D.R. (no date) Comparison between Vapor
Explosion Models and Recent Experimental Results, AIChE Symposium Series
138, Vol. 70, Heat Transfer Research and Design, 31.
3.20 Parkes, E.W. (1968) The inelastic behaviour of aluminium alloy tension members
when subjected to heating on one face. In Heyman and Leckie (eds), Engineering
Plasticity, Cambridge University Press.
pr/p0=2(7pa+4p0)/(7pa+p0), (4.1)
the box and reaches the rear edge. It then spills down the back wall. On the flat
roof and sides the instantaneous pressure rise is to p0 (not pr), but because of the
pressure difference at the junctions of the top and sides with the front face,
vortices will be formed there. See, for example, the book by Norris et al. [4.1].
In determining the loads on the structure it is useful to consider the instant
at which the shock front strikes the front face as t=0, and the time for reflection
effects to clear the front face as t=tc, where
tc=3s/ū. (4.2)
In this equation s is the clearing height, taken as either the half width of the
front face (B/2) or the full height (H), whichever is the lesser, and ū is the
velocity of the shock front. Once the time tc is reached, the pressure on the
front face (taken as uniform) is a combination of the gradually decaying incident
pressure, p, and the dynamic pressure, q. The variation of dynamic pressure
with time is usually given by the equation
(4.3)
which is similar to the expression for overpressure decay but with a different
decay rate. From Eq. (1.28) we note once more that the dynamic pressure in
psi at time t0 is given by
(4.4)
To obtain a true measure of the dynamic loading due to the blast winds on the
structure, it is necessary to multiply q by the drag coefficient, Cd, for the shape under
consideration. So the total front wall pressure (ps) from time tc onwards is given by
ps=p+Cdq. (4.5)
For the front face of a rectangular box structure Cd is often taken as 1.0, although
wind tunnel tests show that the average pressure due to blast winds is rather less
than 1.0 q. Some analysts therefore suggest Cd=0.85 for the front face.
When the shock wave reaches the rear wall of the structure distant L from
the front wall, vorticity again occurs at the edges, causing a reduction in incident
pressure due to suction. According to some authorities the overpressure on
the rear wall reaches a maximum in a time given approximately by
tL=(L+4S)/ū. (4.6)
The sides and top of our simplified box are not fully loaded until the shock
front has travelled the full length of the structure. The average pressure is
therefore considered to be the shock overpressure plus the drag loading at a
distance L/2 from the front face. The loading increases from zero to the average
value in a time L/2ū, so that, for the top and side faces
ps=[p+Cdq]t=L/2ū, (4.7)
Table 4.1
the faces. Drag coefficients for bodies in the subsonic flow range have been
given by Hoerner [4.3] who lists the values shown in Table 4.1 for Cd for some
simple shapes (see also Baker et al. [4.4]).
Values of Cd for many other shapes, including framed structures such as
masts and towers are given in books on the effect of wind on structures.
Glasstone and Dolan [4.5] gave procedures for deriving air blast loading
as a function of time for partially open box structures, framed structures and
cylindrical and arched structures, in addition to the simple box structure
considered above. They define a partially open rectangular box structure as
one in which the front and back walls have about 30% by area of openings or
windows, and have no interior partitions that might influence blast wave
behaviour. The average loading on the front face is influenced by the fact that
1 1
the value of tc is now 3s /ū, where s is the average distance from the centre of
a front face wall section to an edge of wall at an opening. The internal pressure
begins to rise at zero time because the blast wave enters the openings, eventually
reaching the blast wave overpressure. The dynamic pressures are assumed to
be negligible (i.e q=0) on the interior. The average loading on the sides and
top are similar to those for a closed box, but the inside pressures take a time of
2L/ū to reach the blast wave value. For much of the decay period the internal
pressure is greater than the external pressure.
The outside pressures on the back face are similar to those for a closed
1
structure, except that s is replaced by s in the analysis. The inside pressure
reaches a similar value to that of the blast pressure, instantaneously, at a time
L/ū, and then decays. The dynamic pressure is reckoned to be negligible.
When a blast wave meets a cylindrical structure there is a very complex
interaction with the curved surface, but the loading increases from zero to a
maximum when the blast front has traversed one radius; this occurs at a time
D/2ū (where D is the cylinder diameter). It is sometimes assumed that the
Figure 4.1 Blast wave meeting 180° semi-circular arch (from Newmark et al., ref.
4.2).
maximum average pressure on the front of the cylinder is 2p, and that this
decays linearly until a time 2D/ū. The average side loading is given as
p(20D/ū)+Cdq(20D/ū).
Figure 4.2 Blast wave meeting 45° dome (from Newmark et al., ref. 4.2).
side to side. This example illustrates two general principles spelt out by
Christopherson:
(a) An initial peak pressure is not diffracted without loss, so an obstacle casts
a distinct shadow when loaded by high pressures of short duration.
(b) The duration of impulses from a shock wave of small amplitude, acting on
a body having dimensions that are small compared with wavelength, will
be governed by the time taken for the sound wave to travel around the
obstacle, rather than by the duration of the original pulse.
This means that weak shock waves will have relatively little effect on columns,
chimneys or the members of frameworks that can be quickly enveloped by the
diffracted shock front. When the shock wave has a very large amplitude the
first principle (a) is still true, but the second will not apply.
The blastwave loading on open structures such as a bridge truss or a lattice
tower is caused almost entirely by the drag pressures due to the high-velocity
blast winds, and it is usual to neglect the overpressure loading due to the
shock front because of the very short duration of its effect on individual truss
members. As before, the drag pressure, pa, is given by pd=Cdq. A frame structure
made from round tubes has a lower drag coefficient on individual members
than a structure made from structural sections that consist of assemblies of
flat webs, flanges or plates. In general the latter have values of Cd in the range
1.8 to 2.2, whereas the round tubes have values of Cd in the range 1.0 to 1.2.
When structural members are in close proximity, shielding has an effect on
loading, as illustrated by the shielding of the leeward truss by the windward
truss in a bridge. Figure 4.4 gives the ratio of leeward to windward drag
coefficients in terms of ‘solidity’ ratio. This is the ratio between the total area
of individual members in cross section to the area of a completely solid cross
section having the same overall dimension, and it assumes that the members
are set normal to the blast wind direction.
The effect of shielding on the blast loading of circular cylinders was
investigated by Mellsen [4.7] in 1973. Two cylindrical cantilever beams,
each 1.5 inches diameter, were subjected to a blast wave of 20 psi incident
overpressure and 25 msec positive duration. The blast wave travelled in a
direction so that one cylinder exactly shielded the other. The distance
between the cylinders was increased from 2 to 8 diameters in a series of
tests, and ratios of peak strains were measured. The ratio of rear cylinder
peak strain to front cylinder peak strain increased from 0.4 to 0.8 over a
range of 2 to 8 diameters, but for a greater distance apart of 16 diameters
there was only a small further increase in peak strain ratio to about 0.9.
The tests also indicated that the loading ratio depended on the value of
peak incident blast pressure. When this pressure was decreased incrementally
from 20 psi to 13.5 psi the ratio rose linearly from 0.8 to 0.95 for a cylinder
spacing of 8 diameters.
Figure 4.4 Relationship between drag coefficient ratio and solidity ratio for two
parallel trusses.
If the soil surrounds the structure and rests at its natural angle of repose, its
presence will considerably reduce the value of the coefficient of drag on the
front face and at the same time buttress the structure against lateral movement.
Underground structural loading is also influenced by the flexibility of the
structure in relation to the properties of the surrounding soil and the soil-
structure interaction effects that occur as the structure begins to deflect under
shock pressures. The structure and its surrounding soil form a composite body,
the loading on which can be influenced by soil-arching and the disturbance to
the natural properties of the soil.
Much of the early work in this field was driven by the need to assess nuclear
weapon effects, and there are good summaries of the problem in the reports
by Newmark and others, and by Glasstone and Dolan, discussed earlier in
references [4.2] and [4.5]. Using their work as a basis, we will discuss first the
structural loading that comes from blast-induced ground shock. We saw earlier
that there are two types of air-blast-induced ground motion. In that where the
velocity v of the shock front exceeds the dilational seismic velocity (u) of the
soil, there is a sloping shock front below ground. If v=u, the ground shock
front becomes almost vertical near the surface, and if v=u, the shock front in
the ground outruns that in the air. The relationship of pressure with duration
will attenuate as greater depths of soil are considered, and Figure 4.5, from
the work of Newmark [4.8] illustrates this. As the depth increases the rise
time is longer, there is a lower peak pressure and a longer decay. The total
impulse (i.e. area under the pressure-time curve) remains about the same, and
it was established experimentally that the rise time at any depth was about
half the time of transit of the shock front to the point considered. The changes
in shape of the pressure pulse were a direct result of the stress-strain
characteristics of the soil in unidirectional compression. A simplified empirical
equation was proposed by Newmark and Haltiwanger [4.9], which gave
Figure 4.5 Attenuation of pressure wave with depth (from Newmark, ref. 4.8).
reasonable agreement with test results obtained by Newmark and Hall [4.10].
The relationship between peak vertical stress, σz, and depth z, and the peak
surface overpressure on the surface, ps, was given by
sz=aps, (4.9)
a=(1+z/Lw)–1, (4.10)
where
(4.11)
VL=(ML/ρ)1/2, (4.12)
Figure 4.6 Bilinear stress-strain relationship for soil (from Hendron and Auld, ref.
4.11).
where ρ is the mass density of the soil. The strain recovery ratio, ML/MU,
could well lie in the range 0.3 to 0.7 and to take account of this greater range
of conditions Hendron and Auld proposed an amended equation
(4.13)
where VL was in ft/sec. For a given peak overpressure and strain recovery
ratio a scaled as z/w1/3 · VL. A further discussion of wave propagation in soils
and a review of elementary wave theory was given at about the same time by
Allgood [4.12].
It is generally accepted that when structures are buried very deeply and
subjected only to direct ground shock, the loading action due to this shock is
insignificant when compared with the loads from the dead weight of soil
cover. However, if a huge megaton range nuclear bomb were to burst directly
over a deep structure, it could cause extensive damage at depths of 500 feet
and over.
The ground shock wave from a very deep underground nuclear explosion
gradually weakens into a train of seismic waves that can cause ground motions
and earthquake-type damage a long way from the point of detonation. Structural
loading is influenced by the type of soil in which the structure is buried, and
by the soil-structure interaction effects associated with lateral ground motion.
As with all analytical work on buried structures, care must be taken when
using measurements taken under static loads that they are still applicable in
dynamic and vibrational conditions.
Design guides for underground structures under blast and shock loads, see
for example ref. [4.2], often distinguish between shallow-buried, surface flush
and earth-mounded structures. Surface flush structures have soil covers (h) over
the highest point of their upper surfaces in the range 0<h<0.2l, where l is the
clear span between supports of the roof slab, crown arch, dome or shell as the
case may be. Shallow-buried structures have covers in the range 0.2l<h<1.5l,
which is reckoned to be deep enough for dynamic soil arching to influence the
loads on the structure. Unfortunately, although there is considerable experimental
and analytical knowledge on soil arching under static conditions, much less is
known about dynamic arching. Knowledge on static soil arching has been
discussed by the author in an earlier book [4.13], and it has been shown that for
most soils shear strength is the key factor. The high shear strength of a well-
compacted granular soil leads to considerable static arching, whereas the low
shear strength of soft clays results in negligible arching. Under dynamic conditions,
experiments by the author [4.14] suggested that static shear strength was no
longer a factor, and that granular soils behaved very similarly to soft clay soils.
The static redistribution of loads away from flexible members to their stiff
supports can considerably reduce load effects and lead to increased structural
resistance. Under dynamic conditions, however, shear stresses cannot develop
quickly, and soil arching cannot redistribute loads until a rarefaction wave travels
from the rigid to the flexible areas of the structure. Because of the general
uncertainty it has been thought conservative to ignore soil arching when assessing
the dynamic loads on shallow-buried structures from large and distant explosions.
This can lead to uneconomic designs, but research to investigate dynamic arching
more thoroughly has been limited by a reduction in funding due to a lesser
nuclear threat in recent years.
The loading of shallow-buried structures is also influenced by the reflection
of the incident shock wave under soil-structure conditions. We saw earlier that
the surfaces of above-ground structures cause large reflection pressures, pr, that
can be double the incident pressure, p0. In underground situations the reflected
stress, sr, can be influenced by movement of the structure in relation to the soil,
by the rarefaction wave travelling from the rigid to the flexible areas of the
structure, and by the possible reflection of the same waves downwards by the
soil/air surface. These factors affect the time after the initial shock meets the
structure that the pressure-duration curve returns to incident pressure decay
values (i.e. a time td corresponding to time tc in the above ground structural
analysis). For the particular case of a reinforced concrete flat slab roof, having
a thickness and a seismic velocity in the concrete of vc, it has been suggested that
td=12T/vc, because the tensile wave reflecting back from the underside of the
roof will influence the conditions of reflectivity at the upperside.
Surface-flush underground structures (h<0.2l) have a slightly different loading
condition, because the duration, td, of reflected stress is smaller. This is because
a relief shock wave can quickly be reflected downwards from the ground
surface, and td now becomes the smaller of the values h/vs or 12T/vc for flat
concrete slabs, where vs is the seismic velocity in the very thin soil cover.
Shallow-buried and surface-flush structures are normally buried deeply
enough to avoid dynamic wind loads, but this is not the case for mounded
structures. The drag coefficients for mounded structures with 1 in 4 slopes on
the berms are about 1.0 for the slope facing the direction of the blast, and –
0.2 on inclines facing directly away from the blast, so significant drag and
reflection can occur for this gentle slope. Nevertheless, in the initial stages of
structural design a 1 in 4 slope is often considered sufficiently shallow to
ensure that surface blast winds do not load the structure, and the loading is
restricted to the attenuation of the instantaneous overpressure p0 by the small
thickness of soil above the roof or crown of the structure.
In all underground structures the position of the water table is important.
For shallow-buried surface-flush or mounded structures it is not advisable to
set the structure under the water table, as the saturated soil will result in
higher loads. In assessing structural loading the possibility of a high water
table coinciding with the detonation of a large explosion should be the subject
of a risk analysis. In mounded structures it is likely that the soil conditions
will change within the depth of the structure due to the change from backfill
soil to undisturbed soil to waterlogged soil. Underground structures often
have surface structures above them, and in certain circumstances the presence
of such structures can affect the peak incident overpressure reaching the
underground structure for a large air burst.
It is clear that the loading on underground structures is influenced by the
shape and flexibility of the structure. For example, because of the support
that can be mobilized from well-compacted soil, it is possible to use thin-
walled metal construction for underground structures or for the linings of
temporary trench shelters or earthworks. Here, the structure is formed from
thin sheet elements that in themselves have low structural strength, but which
can support large overpressures if properly embedded in soil. Corrugated steel
culvert material used in civil engineering is a good example. The limitation to
load-carrying ability is often the inward buckling of the walls of the structure,
or the tensile rupturing of sheeting due to large membrane forces, rather than
collapse by the formation of ultimate moments in structural members.
For heavier, thick-walled construction, typical of reinforced concrete, there
are broad, approximate quasi-static loading conditions that can be used for
initial design calculations without recourse to complex analysis or personal
computers. For example, in a shallow-buried rectangular structure, with a
cover depth less than one-half the span of the roof slab, it can be assumed
roughly that the roof and floors are loaded with the ground surface peak
overpressure, p0. In a dry soil the effect of attenuation might limit the side
pressure on the walls to 0.5 p0, but in saturated conditions with a high water
table this would increase to p0. The loading conditions on roof elements would
be the full overpressure, p0, plus the dead load of the roof structure, the soil
cover, and any debris that might fall on the surface. For side wall elements the
static earth and water pressure would need to be added to the blast overpressure.
Examples of reinforced concrete underground shelters to withstand unclear
explosions, designed on this rough quasi-static basis, were given in the UK
Home Office publication, Domestic Nuclear Shelters [4.15]. It is interesting
to note that in the UK domestic nuclear shelter design of the 1960s and 1970s
the loading overpressures were limited to three atmospheres (315 kpa or 45
psi), because at pressures above this the effects of Initial Nuclear Radiation
would probably kill the occupants.
Perhaps we should pause for a moment and consider radiation doses, since
they are a fundamental part of nuclear explosions. They are measured in various
units such as roentgen, gen, rad or rem, depending on the precise kind of radiation.
We will only consider the roentgen here. Acute radiation sickness is produced
by a brief exposure of the whole body to 50 roentgens; for exposures between
50 and 200 there could be weakness and fatigue in addition to sickness, but
only about one person in twenty would need medical attention. Between 200
and 450 roentgens there would be moderately severe illness, and perhaps one
quarter of the people so exposed could die. Exposures of over 600 roentgens
would lead to death in less than 14 days in almost every case. Exposures in
excess of several thousands causes severe brain damage and death within hours.
Table 4.2
The intensity of Initial Nuclear Radiation falls off rapidly with distance from the
centre of the explosion, and because of this the relationship between blast shock
pressure and radiation is governed by bomb size. For low-yield explosions the lethal
range of radiation extends beyond the range of lethal blast, and it is possible for a
person in the open to survive the shock and blast winds but be killed by radiation.
For very large megaton explosions people can be killed by blast winds when only
receiving a negligible radiation dose. This is illustrated by Table 4.2.
As a general rule, underground structures designed to protect the occupants
against peak overpressures of 11 psi (77 kpa) and above from a nuclear bomb
should also be checked against radiation. In most cases the thickness of cover
over the structure to give sufficient protection against the blast and shock is
also sufficient to reduce radiation to acceptable levels for the occupants.
compared with the natural period of the structure, which is more likely when
the structure is very flexible, then the basis of damage assessment would be
the total impulse of the shock load.
We have already seen that the pulsating bubble phenomena associated with
underwater explosions result in the mass movement of water as well as pressure
pulses. The pressure pulse from a bubble at its minimum is greatest when
bubble migration is small. According to ref. [4.16] this meant for example
that for a 300 lb charge of TNT in a mine, the mine should be moored 14 feet
above the seabed. The mass motion of water that accelerates radially outwards
from the expanding bubble is an important factor when considering close-
range underwater damage, but for larger, distant explosions its effect is
considerably diminished. This is because the kinetic energy of the outflowing
water falls away as the fourth power of distance, whereas the energy in the
shock wave is reduced according to the second power of distance.
Christopherson [4.6], in summarizing the work in underwater explosions
in the UK during the Second World War, made the point that underwater
structures such as submarines are virtually unsupported, in that there is no
rigid support to ensure that the natural period of oscillation of the vessel is
short in comparison with the duration time of a high incident pressure. He
quoted the work of G.I. Taylor [4.17] who considered the oblique impact of
an underwater shock wave on targets with a variety of supporting forces. For
the simplest condition, in which a shock wave strikes normally a completely
unsupported underwater plane surface having mass m per unit area, he showed
that on the surface of the plate the total pressure (p) is given by
p/p1=e–t/θ+φ(t), (4.14)
where p1 is the pressure in the incident shock wave, t is the time measured
from the moment of arrival, and θ is the time for the pressure to fall to half its
initial value. If the unsupported plane surface remains in contact with the
water in the incident wave it is possible to formulate its equation of motion,
and from this to show that
(4.15)
(4.16)
and at this moment the water begins to exert a retarding force on the plane
surface. The plane surface then loses contact with the water and cavitation
occurs. Substituting for t/ρ in [4.15], and remembering that the net velocity
–t/θ
(V) at the target plane is found from the equation ρcV/p1=e –φ(t), leads to
the expression
(4.17)
(4.18)
and if this is available to load the target then the extent of damage can be
predicted.
Christopherson suggested that underwater loading (and hence damage)
will result from the following sources:
(a) The initial shock wave, acting for a very small time of perhaps 1 msec.
(b) The following kinetic wave, which becomes important after the initial
1/3
shock wave ceases to act, but only lasts about 0.02W sec (W in lb).
(c) The second and later bubble expansions, which are only important when
the bubble approaches the plane surface between expansions.
4.5 REFERENCES
4.1 Norris, C.H. et al. (1959) Structural Design for Dynamic Loads, McGraw Hill,
New York.
4.2 Newmark, N.M. et al. (1961) Design of structures to resist nuclear weapon
effects, Manual of Engineering Practice, No. 42, American Society of Civil
Engineers.
4.3 Hoerner, S.F. (1959) Fluid-Dynamic Drag, pub. by author, New Jersey, USA.
4.4 Baker, W.E. et al. (1983) Explosion Hazards and Evaluation, Elsevier, New
York.
4.5 Glasstone, S. and Donlan, P.J. (1977) The Effects of Nuclear Weapons, 3rd edn,
US Depts of Defense and Energy.
4.6 Christopherson, D.G. (1946) Structural Defence (1945), UK Ministry of Home
Security, Civil Defence Research Committee paper RC 450.
4.7 Mellsen, S.B. (1973) Effect of Shielding on Blast Loading of Circular Cylinders,
Defence Research Establishment, Suffield, Alberta, Canada, Memorandum No.
13/72, February.
4.8 Newmark, N.M. (1984) Opening address, Proc. Symp. on Soil-Structure
Interaction, Univ. of Arizona, Tucson.
4.9 Newmark, N.M. and Haltiwanger, J.D. (1962) Air Force Design Manual:
Principles and Practices for Design of Hardened Structures, USAF Special Weapons
Center, Kirkland AFB, Report AFSWC-TDR-62–138.
4.10 Newmark, N.M. and Hall, W.J. (1959) Preliminary Design Methods for
Underground Protective Structures, USAF Special Weapons Center, Kirkland
AFB, Report AFSWC-TR-60–5.
4.11 Hendron, A.J. and Auld, H.E. (1967) The effect of soil properties on the
attenuation of air blast induced ground motions, Proc. Symp. on Wave
Propagation and Dynamic Properties of Earth Materials, Univ. of New Mexico,
Albuquerque.
(5.1)
where ρ is the mass density of the explosive and ux is the outburst speed.
The total contact explosion impulse is equal to the area of the impulse
pattern, and for a prismatic charge with b>2H this is:
(5.2)
Figure 5.1 Triangular and trapezoidal impulse patterns (from Henrych, ref. 5.1).
I=b2Hρuxb/6H. (5.3)
Eq. (5.2) and (5.3) are also valid for a cylindrical charge placed with its base
in contact with the rigid surface. For a hemispherical charge of radius r,
(5.4)
I=blHρux. (5.5)
In applying the above formulae care must be taken over units. The value of
the outburst speed, ux, is approximately equal to (2Q)1/2, where Q is the work
done when the explosive gases expand from their original volume to an infinite
volume.
When the explosion occurs at a short distance from a rigid obstacle the
pattern shape changes. Consider, for an example, a spherical charge of
radius r exploding at a distance a from a flat surface, where a<10r. At a
time t after the detonation the outburst pattern will be spherical, and as
Figure 5.2 shows, reflection will be occurring at the same time that the
central region of the explosion is still stationary. If the velocity of the
Figure 5.2 Spherical charge exploding close to flat surface (from Henrych, ref. 5.1).
(5.6)
(5.7)
I=pA0W. (5.9)
To use all the above equations we have to know the value of Vs and ux. As
discussed above, ux=(2Q)1/2, and Vs is the velocity of the shock front. From
experiments it has been reported by Henrych that Vs has the following values
in metres/sec for TNT and PETN: 7100 and 8450. The values given for ux for
the same two explosives are 6450 and 7700 metres/sec. Care must be taken
when applying the above analysis that the units are correct.
Explosive charges close to the ground are often assembled in an array, to
distribute total explosive mass over a large area and give some uniformity of
Figure 5.3 Schematic view of part of a line charge array (from Baker et al., ref. 5.2).
surface blast loading. This type of arrangement has a military use connected
with the rapid clearance of a lane through a minefield. Experiments on the
blast output from a pair of parallel Primacord line charge arrays have been
reported by Baker et al. [5.2], in which reflected pressures and ground impulses
were measured at various distances from the arrays. Relationships between
charge weight, line spacing and stand-off distance were analysed and scaling
laws developed. In Figure 5.3, taken from the report, two identical line charges
are shown, distance R above the ground, with a charge spacing of S. The
model test conditions ranged from S=1.5 in, R=2 in, using 40 gram/foot
Primacord, to S=9 in, R=16 in using 400 gram/foot Primacord. The greatest
loads in the region below the charges occur when a Mach stem forms and
charge spacings and stand-offs to achieve this formation were measured. The
results are summarized in Figure 5.4, which compares scaled stand-off distance
1/3 1/3
(R/W ) and scaled charge spacing (S/W ) at which a fully formed Mach
stem is produced. At lesser stand-offs or spacing the pressure/time curve at
Figure 5.4 Position of parallel line charges to produce fully formed Mach stem on
the ground surface (from Baker et al., ref. 5.2).
ground level exhibits a ‘double peak’ with the instantaneous incident pressure
pulse followed by a separate reflected pressure pulse. Figure 5.4 suggests an
approximate linear relationship of S/R=0.5. Scaled reflected pressure curves
1/3
show that reflected pressure increases very quickly when R/W <0.1 and scaled
1/3
reflected impulse curves show a rapid increase in impulse when R/W <0.07.
(5.10)
(5.11)
V0=pi/(ρc), (5.12)
(5.13)
Figure 5.5 Peak stresses for underground explosions in various types of soil (from
Drake and Little, ref. 5.5).
Table 5.1
and that near the source of the explosion peak particle velocities in soil (V0)
were virtually independent of soil properties.
A large number of ground shock experiments were monitored by the South
West Research Institute, Texas, USA, in the early 1980s, and a summary of
the findings was reported by Westine and Friesenhahn in 1983 [5.6]. The
buried detonations were from the explosion of charges in mortar and artillery
projectiles. The authors divided their analysis of results into unsaturated and
saturated soils, and developed empirical equations for the relationship between
scaled maximum pressures and the scaled distance from the explosion.
Table 5.2
Table 5.3
The general empirical equation for predicting free-field ground shock pressure
from the detonations of buried explosive was:
(pmax/ρc2)(4.35+y/d)–1(0.25+0.75 tanh(0.48ρ1/3c2/3d/E1/3))
(5.14)
=0.0175(E1/3/(ρ1/3c2/3Re))3.42,
where pmax=maximum pressure, ρ=mass density of the soil, c=seismic velocity
for soil, d=depth of explosion below ground surface, y=depth of point at which
the pressure is required, E=energy release of the explosive, and Re=effective
slant range between the point under consideration and the centre of the
explosion, taking account of the charge geometry and orientation.
Re was given by
(5.15)
Re/l=(M–N2–0.25)1/3
×[(N+0.5)/(M+N)1/2–(N–0.5)/(M–N)1/2]–1/3,
where
2 2 2
M=(z/l) +(x/l) +(y/l) +0.25,
N=(y/l) cos θ+(z/l) sin θ,
and
z = horizontal distance of point under consideration to the vertical
plane in which the length of the charge lies
x = transverse distance of the point under consideration from the
centre of the charge
l = length of explosive charge
θ = angle of orientation of the bomb (θ=0° is vertical)
The format of Eq. (5.15) allows scaled pressure (p/ρc ), which is non-
2
3 1/3
/ E ), which is also non-dimensional, since the units of energy (E) are force×
length. Figure 5.6 shows a plot of Eq. (5.15) in terms of scaled pressure and
scaled effective stand-off, with parallel lines indicating the spread with one
standard deviation for normal distribution of a large number of test results
conducted at the South West Research Institute. This plot also gave very good
predictions of the scatter of experimental results from tests with mortar and
artillery shells at the Waterways Experiment Station and with C-4 explosive
charges fired at White Sands, USA. These tests were carried out in silt and
sandy soils, well clear of any water table. When Figure 5.6 was compared
with test results for 500 lb and 2000 lb bombs carried out at Fort Knox, USA,
in saturated soft brown clay with the water table at a depth of 10 feet, the
scaled pressures were an order of magnitude greater than would be predicted
by Figure 5.6. This was mainly because the seismic velocity in soft clay of
1200 ft/sec was increased to the speed of sound in water (4800 ft/sec). In this
instance it is more accurate to use data we discussed earlier from the work of
Cole [2.22], to give the following equation for shock pressures in water:
(5.16)
p/4.35ρc2=0.04224(ρ1/3c2/3R/W1/3)–1.16. (5.17)
This equation gives good agreement with the Fort Knox tests, and as we might
expect indicates that free-field ground shock pressures are considerably
influenced by the state of saturation of soils.
All the experimental results depend on the accurate measurement of free-
field stresses, and measurements are influenced by the degree to which buried
stress gauges disturb the soil in their immediate vicinity. The influence of
Figure 5.6 Scaled pressure v scaled effective stand-off distance from tests on bombs,
shells and charges (from Westine and Friesenhahn, ref. 5.6).
gauge installation procedures in rock and soil has been discussed by Florence,
Keough and Mak [5.7]. In particular they investigated theoretically the effect
of bonded and unbonded interfaces between the gauge and the surrounding
soil. They assumed that a bonded interface would transmit compression and
shear, and an unbonded interface compressive loads only. Their calculations
show that lack of interface bonding produces a non-uniform stress distribution
at the gauge which produces inaccuracies at low stresses. At higher stresses
plastic flow in the soil increases, and causes the stress distribution to be more
uniform. This improves accuracy.
Figure 5.7 Underwater cylinder subjected to pressure pulse (from Haxton and
Haywood, ref. 5.9).
which the wave path length must be measured around the surface of the cylinder.
The following equations were derived for p1:
(5.18)
T is the decay time constant, c is the velocity of sound in water, en=2 when n=0
and en=1 when n=1 where n is the circumferential modal index for the deflections
of the cylinder wall.
A long time before analytical work of the above type was attempted, the
problems of close-in explosions had been discussed by Christopherson [4.6].
In 1945 no satisfactory measurements had been made of the pressure inside
a gas bubble, and when an underwater target was very near the source of
the explosion there was no general theory of loading. There had been tests
during the Second World War at the UK Road Research Laboratory which
suggested that a charge of TNT of weight W lb in contact with a lightly
reinforced concrete slab of thickness t ft underwater would produce heavy
1/3
damage (severe cracking and bowing) if t/W <1.33. This compares with a
1/3
value of t/W <1.1 for a charge confined in the earth, and suggests that the
underwater charge is about 1.75 times as effective as an underground charge
against a similar target, for this type of damage.
It was found that the contact charge of TNT needed to cause a complete
breach in a masonry or mass-concrete gravity dam of thickness t, at the point
of attack, was given by
W=6.10–3t3.75 lb, (t in feet) (5.19)
and that if the charge exploded underwater at a distance from the dam wall
equal to t then the charge would need to be greater by a margin of 5 or 10 to
produce similar damage. This information was used in the preparation for the
well-known operations against the Möhne dam, where it was shown that the
experimental predictions at small and medium scales were verified at full scale.
Figure 5.8 Peak pressure from a charge exploding inside a tunnel (from Philip, ref.
5.10).
There were two problems, the first resulting from an explosion in the tunnel,
the second due to the entry of blast from an explosion outside. A selection of
Philip’s results for explosions within a long straight tunnel was given by
Christopherson [4.6] and some are shown in Figure 5.8. They refer to brick-
lined tunnels, in which pressure decay is likely to be quicker than in smoother
tunnels of concrete or steel. The rate at which the peak pressures decayed was
1/3 1/2
influenced by the parameter W /S , where S is the cross-sectional area in
square feet, and W is the weight of the charge of TNT in lb. Although the
peak pressures gradually decayed along the tunnel it was interesting to note
that at short distances from the explosion the blast impulse had increased
slightly, and for very small charges this increase could still be measured at
long distances from the point of explosion, as indicated in Figure 5.9.
It was clear that in long straight tunnels the decay was slow enough to suggest
that damage due to flying debris might occur at long distances from the point of
the explosion, and this led to the policy of interrupting long straight runs with
‘blast traps’. Christopherson reported estimates of the effect of a variety of trap
shapes based on trials at the Road Research Laboratory, and a selection of these
estimates is shown in Figure 5.10. The factor Kp should be applied to the pressures
in straight tunnels given in Figure 5.8, if a trap of the geometry indicated is
interposed between the charge and the point under consideration. The factor
K2 gives similar information for impulse, and is applied to the impulses from
Figure 5.9. The estimates from the Road Research Laboratory suggested that a
simple right-angle bend reduced the peak pressure by 30%, and a uniform
Figure 5.9 Blast impulse from a charge exploding inside a tunnel (from Philip, ref.
5.10).
diameter tee junction by about 50%. The reduction in a tee can be increased
slightly if the branching pipe that forms the stem of the tee has either half or
double the cross-sectional area of the main pipe.
When figures were given in 1965 in ref. [1.10], there were some discrepancies.
In this reference the reduction at a simple right-angle bend was given as 50%.
The system of four right-angle bends shown in Figure 5.11 was said to reduce
the pressure by 85%, and if extensions or pockets were introduced at the
Figure 5.11 Four right angle bends with blast pockets (from Christopherson, ref.
4.6).
bends (shown dotted), the reduction was 92%. The Road Research Laboratory
figures were 65% and 77%, but the latter was increased to 86% if the pocket
depth was double the pipe diameter. The introduction of a rectangular pipe
system (Figure 5.12) was said by the US Army Technical Manual to lead to a
reduction of 85%, whereas the Road Research Laboratory figure was 62%.
Later research suggested that the above figures were over optimistic, and
more recent publications in the USA and elsewhere propose more conservative
design rules. Much of this information still seems to be restricted, but open
recommendations are given in a book by Henrych [5.1] published in English in
1979. He gives information on the propagation of a shock wave in cranked and
branched channels in which he notes that the factor Kp changes as the peak
entering pressure increases. Figure 5.13, taken from [5.1], gives a range of values
of Kp for several branch geometries. At the higher pressures the reduction at a
simple right angle bend is only 10%, so that Kp=0.9. For successive right-angle
bends his results show that the same coefficient is applied each time, so that for
3
three successive bends the overall value of Kp is 0.9 =0.73. His results suggest
n
that for n successive bends the overall factor is 0.9 , but there is no analytical
proof of this. What is very apparent from a study of all the experimental
information is that the smoothness of the inside of the tunnels has an important
bearing on the results, and as this information is not always given in experimental
reports it is dangerous to correlate the conclusions of separate experimenters.
Figure 5.12 Rectangular pipe system blast trap (from Fundamentals of Protective
Design, TM5–855–1, 1965).
Figure 5.13 Propagation of a shock wave in branched channels (from Henrych, ref.
5.1).
We can perhaps assume that the low values of Kp given in the UK data of the
1940s were measured in tunnels with very rough walls, whereas later and
more conservative results from the USA and Europe were gathered from smooth-
walled tunnels. The 1986 edition of the US Army Technical Manual TM-
n
855–1 gives Kp=0.94 , when the initial pressure (p0)=50 psi.
The entry of blast from a ‘free field’ into a side-on tunnel often needs to
be assessed for design purposes. When a pressure pulse having a peak initial
pressure of p0 passes transversely over a side-on tunnel, the actions are similar
to those for the T-branch tunnels given at D in Figure 5.13. At low values of
p0, Kp=1, but at higher values Kp is reduced to about 0.3. This value of Kp is
probably a little unconservative, but is satisfactory for preliminary
calculations. Curve D in Figure 5.13 can also be represented approximately
by the equation
Kp=n(p0)n.
(5.20)
When the centre of the explosion is directly in line with the mouth of the
tunnel, as shown in Figure 5.14, Henrych quotes experiments which show
that it is sufficient to substitute aR for the distance R of the entrance from the
explosion centre, in order to determine the value of the peak pressure p0 in the
entrance of the tunnel (assuming no reflecting surface around the inlet opening).
2
The relationship between a and log p0, where p0 is in kg/cm , is shown in
(5.21)
where f is the hydraulic coefficient of friction, r is the pipe radius, and there is
no reflecting surface at the inlet opening. The value of f can be found from the
approximate formula
Figure 5.14 Entry of blast into a tunnel from an explosion in line with the tunnel
mouth (from Henrych, ref. 5.1).
that durations of the positive phase tend to be longer, and that the overall
effect on impulse is to reduce it by a rough figure of 50% at any range.
A review of blast wave behaviour in tunnels was reported in 1968 by Taylor
[5.11], in which he made the point that there is a staged progression of a shock
wave moving through a 90° turn in a tube. The expanding shock will produce
different pressures at all points across a cross section drawn one-half diameter
downstream from the bend; then the blind wall reflection forms and moves
across the same cross section to produce a second jump in pressure. Measurements
in a shock tube have shown that pressure fluctuations one diameter downstream
from a 90° turn can exceed the input pressure; in other words a turn amplifies
pressure locally before attenuating it. Taylor shows the measurements of peak
pressure at 2 and 28 diameters downstream from the junction, and indicates
that at 2 diameters the peak pressure is higher than at 28 diameters. Downstream
shock pressure measurements at Y and T junctions were also reported by Taylor,
and pressures at 10 diameters downstream were given by him. As we saw earlier,
pressures were not halved when the routes are doubled.
If the diameter of a straight tunnel is suddenly increased there will be a
reduction in peak pressure. Relationships were given for incident and transmitted
overpressures for area ratios of 7.3, 18.5 and 66.
Most of the experimental work of the 1960s was carried out in relation to
explosions of long duration by the US Army Ballistic Research Laboratory and
the US Air Force Weapons Laboratory, but by the 1980s interest had returned
to the attenuation of short duration air blast in entranceways and tunnels. Tests
were reported by Britt [5.12] in 1985, conducted at the US Waterways Experiment
Station, that modelled the effects of HE bombs in the 100 to 1000kg range. The
results were used to give formulae suitable for design. Some 99 tests were
conducted, in which small spherical charges of C-4 explosive were detonated
outside and inside the entrance to tunnels of circular and square cross section.
The model tunnels were 30 cm diameter and between 4 and 24 diameters long;
blast pressures were measured along the tunnels when charges were detonated
from end-on and side-on positions relative to the tunnel mouths. The results
were compared with earlier work on tunnel pressures in which the pressures
were relatively much greater—almost by an order of magnitude. These were
reported in 1977 by Itschner and Anet [5.13]. Other earlier test results from
Germany had been given by Gurke and Scheklinski-Gluck in 1980 [5.14].
The peak pressure attenuation was given by the equation
px=pi/(1+tan[(p/2)x̄/(x̄+Ē)]), (5.23)
where
px = pressure at a distance x down the tunnel
pi = pressure at tunnel entrance
x¯ = x/A1/2, where A is the cross-sectional area of a circular or square tunnel
E¯ = dimensionless decay parameter.
For end on bursts (i.e. in line with the entrance to the tunnel), Ē was given by
2/3 0.4
Ē=Ke(W /A)(p0/pi) , (5.24)
where
W=charge weight, p0=atmospheric pressure
Ke=0.586 if W is in kilograms and A in square metres, or
Ke=3.72 if W is in pounds and A in square feet.
For side-on bursts, in which the charge explodes at a lateral distance from the
entrance to the tunnel, E¯ was given tentatively as
2/3 0.8
Ē=Ks(W /A)(p0/pi) , (5.25)
where
Ks=2.26 for W in kilograms and A in square metres, or
14.4 for W in pounds and A in square feet,
but further experimental work was said to be needed to confirm this.
These equations were stated by the author to be valid within the following
ranges:
1/3 1/2 1/3
1<W /A <4 kg /m; 0.3 MPa<pi<40 MPa (end-on);
0.6 MPa<pi<11 MPa (side-on)
Britt also examined values of impulse, which was virtually constant along the
tunnel, and he gave the following relationships,
1/3 1/3 1/2 1/2 0.6
I=I0EW (W /A ) (pi/p0) (end-on) (5.26)
where
2
I0E=0.042 for I in MPa · msec, W in kg and A in m ,
or
2
I0E=7.43 for I in psi · msec, W in lb and A in ft .
For side-on bursts
1/3 1/3 1/2 1/2 0.2
I=I0sW (W /A ) (pi/p0) , (5.27)
(5.28)
(5.29)
Te and Ts are static temperatures at points e and s; Tte and Tts are stagnation
temperatures. A(1) and A(2) are empirical coefficients that depend on the
entrance geometry and position of the explosion, and Ms is the flow Mach
number outside the duct. The shock wave attenuation due to internal friction
1/2
in the ducting is represented by a friction factor, given in terms of (h/D) , as
discussed earlier, where h is the average wall protrusion height and D is the
duct diameter. Fashbaugh used the equation
t=0.016+0.00491(h/D)1/2+0.258(h/D), (5.30)
Figure 5.15 Relationship between the stagnation pressure ratio and the surface flow
Mach number for side-on ducts (from Fashbaugh, ref. 5.15).
ps=a(L/D)–b, (5.31)
where ps is the pressure at a distance L along the tunnel given by the parameter
L/D, where D is the tunnel diameter. Values of the exponent b are functions of
the weight of the explosive charge and the distance from the point of the explosion
to the centre of the tunnel entrance. The coefficient a was found for a range of
charge weights from 100kg to 1500kg. Typical results for a charge weight of
1500 kg set at various distances and angles from the tunnel mouth are given in
Figure 5.16, in which the value of peak pressure at a distance L/D=2 along the
tunnel is shown for four values of the stand-off angle a (0°, 30°, 60° and 90°).
Note that when the distance A to the charge=4m, the value of ps changes with
orientation according to Table 5.4 (given for a range of charge weights).
Figure 5.16 Relationship between peak pressure, distance A and stand-off angle (a)
for a charge of 1500 kg, when L/D=2 (from Binggeli and Anet, ref. 5.16).
reflected from all faces, and the reflections will continue until the energy of
the explosion is expended in heat and perhaps by some form of absorption by
the confining walls. If the raising of the internal temperature is the only way
of expending energy the final equilibrium pressure will be higher than the
original pressure within the space. This permanent pressure rise would be a
gas pressure rather than an instantaneous peak pressure at a shock front.
An unyielding perfectly confined space is unusual. Most chambers are
connected to the atmosphere by ducts or tunnels, or are deliberately vented to
reduce the damaging effects of an internal explosion. Chambers in aluminium
structures such as aircraft shells have boundaries that can deflect under pressure,
so that the elasticity of the containing structure cannot be ignored. These and
other problems meant that in the early days of research into confined explosions
the exact analysis of blast pressure was very difficult, and at the end of the
Second World War the recommended design loads for internal explosions in
Table 5.4
(a) incident and reflected blast pulses could be represented by triangular shapes
with an abrupt rise;
(b) the duration of these pulses could be adjusted to preserve the correct
impulses;
(c) initial internal blast loading parameters are always the normally reflected
parameters, even for oblique reflections from the walls.
Assumption (c) is virtually correct up to the angle for limit of regular reflection,
which we saw earlier was about 40° for strong waves and 70° for weak waves.
For chambers approaching a cube in shape shock reflections from the walls
will be regularly reflected in most parts of the cube. Approximate estimates of
reflected wave magnitude were also made in ref. [4.4], where the second shock
was taken to be half the amplitude of the initial reflected shock, and the third
as half the amplitude of the second. Any further reflections were assumed to
have insignificant amplitudes. Furthermore, it was assumed that the time delay
between successive shock loads was equal to twice the time to the initial shock
pulse. These approximations lead to a simplified pressure loading at a point
on the inner surface of the chamber consisting of three successive triangular
pulses with peak pressures pr, pr/2 and pr/4, each associated with a duration of
tr, and the shocks occurring at intervals of ta, 3tA and 5tA.
The development of computer codes to describe the loading actions inside
closed structures began in the early 1970s. It has been pointed out in a recent
paper by Swisdak and Montanaro [5.20] that the original development in the
field was due to Proctor [5.21] in 1972, when he produced a program for the
US Naval authorities called INBLAS. This dealt originally with shock loading
typical of TNT explosions, but his work was later amended by Ward and
Lorenz to include deflagration loading typical of gas or vapour cloud explosions.
In the early 1980s a significant improvement was made by Britt et al. [5.22],
who replaced the original shock calculations with an analysis of greater accuracy
and so formed the BLASTINW code.
This code was designed to run on a mainframe computer, but the rapid growth
of desktop personal computers meant that it was worth updating and combining
the better features in INBLAS and BLASTINW to form a revised code for the
prediction of blast inside closed or vented structures. This code, now called
INBLAST, requires the input of a table of pressure versus material consumption
rate, and a table of total burning area versus the weight of material consumed.
With this information the relationship between pressure and time in all chambers
of the structure can be calculated. The program uses techniques developed earlier
for a low-altitude multi-burst code (LAMB), which is described under NATO
restrictions by Britt and Drake in ref. [5.23], and which predicts the direct and
multiple reflected shock waves present after a closed chamber detonation.
INBLAST has been shown to predict accurately confined gas pressures as a
function of loading density (charge weight/chamber volume), and a comparison
of predictions with test results is shown in Figure 5.18, taken from ref. [5.20].
An earlier paper by Britt and Drake [5.24], presented without restriction in
1985, examined the propagation of short duration blast into chambers from
high explosive charges detonated at the entrance to openings to the chambers.
This work combined the ducting and tunnel technology with the chamber
analysis, and resulted in a useful and progressive comparison between theory
and experiment.
Immediately after the explosion the instantaneous increase in external
pressure causes a high-velocity jet to enter the chamber through an opening of
area A. Eventually the internal and external pressures become equal, and
Kriebel [5.25] had shown in 1972 that the time in milliseconds was equal to
V/2A in units of feet, where V is the chamber volume. At about the same time
pmax/p0=0.65(1–0.25α)(R/D)–1.35, (5.32)
where R is the distance of the centre of the opening to the pressure gauge in
the chamber and D is entrance tunnel diameter. a is the angle in radians between
the normal to the centre of the opening and the gauge location. For small
angles it was sufficient to ignore a and within the units of experimental scatter
to simplify the above equation to pmax/p0=0.65(R/D)–1.35. The values of pmax
are due to the first side-on peak at the gauge as the result of incident pressure
and do not take account of internally reflected waves. If information about
reflected waves is required, then a more complex analysis involving a shock
diffraction model is needed. Path lengths of ‘rays’ of successively higher order
reflections are generated, and arrival times calculated; then shock wave
attenuation with distance is found. The procedure is fully described in ref.
[5.24]. The combined pressure pulse in a chamber, found in this way, can be
calculated using the code CHAMBER, originally for a mainframe computer
but converted to run on a desktop computer. Predictions for the initial peak
reflected shock were shown to agree well with test results.
It is clear that the levels of pressure from an internal explosion, whether due to
a high-explosive detonation or a vapour or dust cloud deflagration, can be reduced
by the judicious use of vents. The design of venting systems is governed by the
pressure reductions that are required, so there has been a considerable research
effort in this area. Going back to the work of Christopherson [4.6], which we
mentioned earlier, we noted that he analysed the effect on the pressure in a rigid
rectangular chamber if one end were vented. The ‘partial enclosure’ was the subject
of an empirical relationship in which the results of a small number of observations
were recorded in terms of the charge weight of TNT, the volume (V) in cubic feet,
and the face-on blast impulse on the venting side wall of a cubical enclosure (AT),
in lb m sec/in2. His results are plotted in Figure 5.19 on the axes AT/W1/3 and V/
W(the volume of the chamber per lb of charge). For initial design purposes, the
relationship took the approximate form (Imperial units):
5.6 REFERENCES
5.1 Henrych, J. (1979) The Dynamics of Explosion and Its Use, Elsevier, New York.
5.2 Baker, Q.A., Baker, W.E. and Spivey, K.H. (1989) Blast loading from arrays of
parallel line charges, Proc. of 4th International Symp. on Interaction of Non-
nuclear Munitions with Structures, Panama City Beach, Florida, USA.
5.3 Vaile, R.B. (1952) Underground Explosion Tests at Dugway, Stanford Research
Institute, Stanford, California, March.
5.4 Sachs, D.C. and Swift, L.M. (1955) Small Explosion Tests—Project MOLE, US
Air Force Special Weapon Center contract 291, Stanford Research Institute,
Stanford, California, December.
5.5 Drake, J.L. and Little, C.D. (1983) Ground shock from penetrating conventional
weapons, Proc. Symp. on Interaction of Non-nuclear Munitions with Structures,
Colorado, USA.
5.6 Westine, P.S. and Friesenhahn, G.J. (1983) Free-field ground shock pressures
from buried detonations in saturated and unsaturated soils, Proc. Symp. on
Interaction of Non-nuclear Munitions with Structures, Colorado, USA.
5.7 Florence, A.L., Keough, D.D. and Mak, R. (1983) Calculational evaluation of
the inclusive effects on stress gauge measurements in rock and soil, Proc. Symp.
on Interaction of Non-nuclear Munitions with Structures, Colorado, USA.
5.8 Hicks, A.N. (1986) Explosion induced hull whipping, Advances in Marine
Structures, ARE Dunfermline, Scotland.
5.9 Haxton, R.S. and Haywood, J.H. (1986) Linear elastic response of a ring stiffened
cylinder to underwater explosion loading, Advances in Marine Structures, ARE
Dunfermline, Scotland.
5.10 Philip, E.B. (1944) Blast Pressure and Impulse in a Tunnel—a Note on the
Latest Piezo-gauge Measurements, UK Home Office, Research report REN 359.
5.11 Taylor, W.J. (1968) Blast wave behaviour in confined regions, Prevention and
Protection against Accidental Explosion of Munitions, Fuels and Other
Hazardous Mixtures, New York Academy of Sciences, p. 339.
5.12 Britt, J.R. (1985) Attenuation of short duration blast in entranceways and tunnels,
Proc. 2nd Symp. on the Interaction of Non-nuclear Munitions with Structures,
Panama City Beach, Florida.
5.13 Itschner, D. and Anet, B. (1977) Entry and attenuation of shock waves in tunnels,
Proc. 5th Int. Symp. on Military Applications of Blast Simulation, Stockholm.
5.14 Gurke, G. and Scheklinski-Gluck, G. (1980) An Investigation of Blast Wave
Penetration into a Tunnel Entrance, Report E7/80, Ernst-Mach-Institut der
Fraunhofer-Gesellshaft, Freiburg, Germany.
5.15 Fashbaugh, R.H. (1991) Computer code SPIDS/Shock propagation in ducting
systems utilising α PC computer, Proc. of the 5th Int. Symp. on the Interaction
of Conventional Munitions with Protective Structures, Mannheim, Germany,
April, p. 218.
5.16 Binggeli, E. and Anet, B. (1991) Experimentelle Untersuchung der Ausbreitung
Iconventioneller Luftstösse in Tunnelsystemern, Proc. of the 5th Int. Symp. on
the Interaction of Conventional Munitions with Protective Structures, Mannheim,
Germany, April, p. 209.
5.17 Gregory, F.H. (1976) Analysis of the Loading and Response of a Suppressive
Shield when Subjected to an Internal Explosion, Minutes of 17th explosive
safety seminar, Denver, Colorado.
5.18 Baker, W.E. (1960) The elastic-plastic response of thin spherical shells to internal
blast loading, Journal of Applied Mechanics, 27(1), Series E, March.
5.19 Kingery, C.N., Schumacher, R.N. and Ewing, W.O. (1975) Internal Pressures
fromExplosions in Suppressive Structures , BRL Interim Memo Report No 403,
Aberdeen Proving Ground, Maryland, USA, June.
5.20 Swisdak, M.M. and Montanaro, P.E. (1991) INBLAST—α new and revised
computer code for the prediction of blast inside closed or vented structures,
Proc. of the 5th Int. Symp. on the Interaction of Conventional Munitions with
Protective Structures, Mannheim, Germany, April, p. 196.
5.21 Proctor, J.F. (1972) Internal Blast Damage Mechanisms Computer Program,
NOL Tech Report TR 72–231, August.
5.22 Britt, J.R. et al. (1986) BLASTINW Users Manual, ARA 5986–2, Applied
Research Associates Inc., Vicksburg, Miss, USA, April.
5.23 Britt, J.R. and Drake, J.L. (1987) Blast loads from internal explosions and other
reflected shock waves, Proc. of Int. Symp. on the Interaction of Conventional
Munitions with Protective Structures, Mannheim, Germany, March, p. N54.
5.24 Britt, J.R. and Drake, J.L. (1985) Propagation of short duration air blast into
protective structures, Proc. 2nd Symp. on the Interaction of Non-nuclear
Munitions with Structures, Panama City Beach, Florida, USA, April.
5.25 Kriebel, A.R. (1972) Airblast in Tunnels and Chambers, US DASA 1200-II
Supplement 1, Defense Nuclear Agency, Washington, October.
5.26 Coulter, G.A. (1972) Blast Loading in Existing Structures—Basement Models,
BRL MR 2208, US Army Ballistics Research Laboratory, Aberdeen Proving
Ground, Maryland, August.
Figure 6.1 BRL side-on piezo-electric blast gauge (from Baker, ref. 6.1).
Figure 6.2 Gauge by Baker and Ewing to be flush-mounted in the surfaces of blast
loaded aerofoils (from Baker and Ewing, ref. 6.3).
slits to focus on particular parts of the image to be studied, and the spark
shadowgraph, which is useful for the photography of shock fronts. At the end
of all this gauging and recording work comes the reductions of data for
comparison with analysis and for codes of practice.
The measurement of shock pressures in soils has been a fruitful research
field in the past thirty years. Before that time most soil pressure gauge work
related to the measurement of static pressures, and many comprehensive reviews
of static developments were written in the early 1960s. Examples are the
paper by Selig in 1964 [6.8] and a review by Abbott et al. in 1967 [6.9]. By the
mid-1980s experimental work on dynamic measurement had resulted in two
types of pressure gauge for the recording of shock pressures in soil, the piezo-
electric gauge similar in form to the air blast piezo-electric gauge, and the
diaphragm deflection gauge with a diameter/thickness ratio of about 10. A
new type of gauge reported at that time was the Polymer pressure gauge,
described by Bur and Roth [6.10]. This is basically a piezo-electric gauge
which uses polyvinyliden fluoride (PVDF) as the transducer material, and
which is also pyro-electrically active. The substance becomes active when
polarized by a large electric field (2 MV/cm at room temperature). The gauge
consists of two layers of PVDF sheet, each containing an active area of 10mm
diameter within an overall gauge diameter of 15 mm. Aluminium and Indium
are deposited over the surfaces of the circular sheets to form electrodes, and a
thermocouple is inserted between the sheets. The sheets are glued together
with an epoxy and are covered with protective polycarbonate. The overall
thickness of the gauge is 0.35 mm. For a sandy soil having a modulus of about
4 6
5.7×10 psi and a gauge modulus of about 2×10 psi, a modular ratio of 35 is
obtained. The aspect and modular ratios satisfy the requirement for accurate
stress measurement in soil, and tests show that a linear relationship exists
between the input dynamic stress and the output voltage signal. For certain
time ranges the gauge needs an active temperature compensation circuit because
of the pyro-electrical characteristics of the material (see ref. [6.10]).
The use of polymer and other relatively recently developed materials for
shock pressure gauges has been the subject of many papers in the on-going
series of conferences on the Military Application of Blast Simulation, and the
reader is referred to the proceedings of these conferences (MABS) for further
information (e.g. ref. [6.11], Turner).
development and use of the shock tube. Wright [6.13] in his book on the
subject tells us that Reynolds [6.14] used a shock tube to produce waves of
known strength in order to calibrate the early piezo-electric gauges described
in the previous section. This was necessary because static calibration was not
possible. By 1945 Smith [6.15] had used photography of the shock front to
study wave reflection, and then Bleakney [6.16] used an interferometer in
conjunction with a shock tube to examine the diffraction of a shock wave
around an obstacle.
The most simple shock tube consists of a rigid cylinder divided at a point
along its length by a transverse gas-tight diaphragm. One half, the compression
chamber, contains gas at a pressure well in excess of the pressure in the other
half, the expansion chamber. When the diaphragm is suddenly ruptured, or
caused to shatter, the shock wave heads down the expansion chamber and a
rarefaction wave travels back into the compression chamber. When the shock
front meets the far end of the expansion chamber it is reflected. If the end is
closed the reflected wave will be a shock wave; if the end is open it will be a
rarefaction wave.
Elementary shock tube theory was set down by Taub [6.17] in 1943. If it is
first assumed for simplicity that the diaphragm has no effect on the system
once it has ruptured, then the shock and rarefaction waves can be said to
form instantly. The shock strength is p1/p0, where p0 is the pressure in the
expansion chamber ahead of the shock, and p1 is the instantaneous pressure
at the shock front. The maximum theoretically achievable strength ym is given
approximately by
(6.1)
where γ0 is the gas constant for the expansion chamber, γ3 is the gas constant for
the compression chamber, and c0 and c3 are the speed of sound for the gases in
the two chambers. If the gas in both chambers is air c0=c3 and γ0=γ3=1.4, so that
ym=42. If the gas in the compression chamber is hydrogen, then c3/c0=3.8 and
γ3=γ0=1.4. When ps/p0=500 for air/air, p1/p0=10; when ps/p0=500 for hydrogen/
air, p1/p0=50, so there is considerable gain in terms of the strength of the shock
front in using hydrogen in the compression chamber. If the hydrogen is heated
c3/c0 >3.8, which increases the strength of the shock still further.
In fact the ruptured diaphragm does influence flow, so the above shock
strengths cannot quite be achieved in practice. To limit the diaphragm effect
as far as possible cellulose acetate sheet, which shatters into small pieces, was
often used in small tubes. In larger tubes metal foil was used because the
stresses in the deflecting diaphragm sheet as the compression chamber pressures
were increased become too great for cellulose acetate. There is also some
reduction in strength due to deceleration of the shock front as it travels down
the expansion chambers. This occurs because of the effect of the boundary
layer between the front and the tube walls. In well-designed tubes the influence
of both of the above effects can be reduced to a level that does not greatly
impair the accuracy experiments.
The properties of shock waves were measured by a number of techniques
during the early years of shock tube development. A number of optical methods
were used, based on the fact that the refractive index of a gas varies with its
density according to the Gladstone Dale Law =1+Kρ, where is the refractive
index, K the Gladstone Dale constant, and ρ the density. The measurement of
at high shock front speeds was often made by using short duration sparks as
the light source and using shadow or Schlieren photography to examine density
discontinuities. The density profile was also found by interferometry, in which
light from a point source was collimated and split at a half-silvered mirror.
One half then passed through the shock tube before the beams were recombined,
and with monochromatic illumination interference fringes could be detected.
There was a linear relationship between fringe shift and density change. High-
speed rotating-mirror cameras were needed to achieve sufficient time resolution
for very fast shocks.
The advent of the piezo-electric pressure gauge enabled pressures rather
than densities to be measured, as long as the gauges were mounted in the
shock tube in ways that eliminated the effect on them of mechanical stresses
transmitted through the walls of the shock tube. This is now the most usual
way of measuring shock wave properties. The instantaneous rise in temperature
across a shock front can also be used to investigate density changes, so there
was some development in the use of resistance thermometers and hot-wire
anemometers. The latter, however, generally have too slow a response time
for shock tube investigations.
We noted earlier that the strength of the shock in a tube can be increased by
introducing combustion into the compression chamber. This method, which
was the subject of much research in the 1950s, burns oxygen with the hydrogen
to raise the temperature of the latter. The best mixture is 1 part by weight of
oxygen to 8 parts by weight of hydrogen, and this leads to a sound speed (c3)
about 1.7 times that of cold hydrogen. For this application the combustion
chamber must be strongly constructed, and gun barrels are frequently used.
Mach numbers of about 10 are obtainable with cold hydrogen, and up to 20
with a combustion shock tube.
The heating process tends to produce non-uniform conditions in the
compression chamber and attenuates the shock wave as it travels down the
expansion chamber. To eliminate this problem multiple diaphragms are
sometimes used, in which an intermediate chamber is introduced at a lower
pressure than the compression chamber, but at a higher pressure than the
expansion chamber. When the first diaphragm is ruptured the shock wave
travels down the intermediate chamber until it is reflected off the second
diaphragm. The reflected pressure is sufficient to rupture the second diaphragm,
but the reflected wave leaves a hot, high pressure region to act as a compression
chamber input which drives the shock front down the expansion chamber
with a gain in strength. Multi-diaphragms result in a noticeable increase in
shock strengths. Each additional chamber is said by White to increase the
shock strength by a factor of 12 or 16 according to whether the gases used
have a constant (γ) of 1.67 or 1.4. These numerical increases are based on the
assumption that the times of rupture of the intermediate diaphragms are long
compared with the time for the reflected shock to travel an appreciable distance.
The strength of the shock is limited by the maximum speed at which gas can
escape from the compression chamber of a cylindrical shock tube (ym=2c3(γ3–1).
This limitation can be circumvented, and the compression chamber sound
speed (c3) increased by using a compression chamber of non-uniform cross
section which converges in diameter as it approaches the expansion chamber.
Alpher and White [6.18] suggested that increases of up to 10% in Mach number
can be achieved by this method. The strengthening effects of convergence are
also attainable by tapered reductions in the diameter of the expansion chamber.
The final interesting development of the 1950s was the replacement of the
‘piston’ of gas in the compression chamber by electromagnetic forces. An
electromagnetic piston, having a velocity limited only by the speed of light,
was established by the introduction of two parallel electrodes, across which a
high voltage was suddenly applied. In argon gas at an ambient pressure of 1
cm of mercury a potential gradient of 100 volts/cm will generate Mach 20
shock fronts, and at very, very short durations it is even possible to achieve
Mach 200.
In using the shock tube to apply instantaneous loading to structures, there
is clearly a problem of scale as well as duration. This was very apparent in the
development of methods of simulating the effect of a nuclear explosion on
civil and military equipment. A very large diameter tube is required to
dynamically load a complete tank hull or ship superstructure, and the duration
of the positive period of the explosion is required to be much greater than that
from a conventional charge. These difficulties have led to a considerable research
interest in blast simulation methods since the 1960s, including the use of vertical
shock tubes.
By the mid-1980s several large air-blast simulators had been built in various
countries as part of the well-financed defence requirement to measure the
blast loads from nuclear explosions on full-sized military equipment such as
tanks, small aircraft and helicopters. Most of these were shock tubes in which
2
cross-sectional areas ranged from 20 to 180 m , and total lengths were up to
100 m, or even 200 m. One of the first large air blast simulators to be built
was at the Atomic Weapons Research Establishment (as it was then called), at
Foulness Island, England, and has been described in a number of papers by
Leys [6.19]. It was completed during the 1970s, and the test section was 20
2
m in area. It was originally built with an explosive driver developed from
two naval guns which fired into a 1.82 m diameter steel tunnel 36 m long.
This was linked by an asymmetric cone to a 2.43 m diameter section 44 m
long, and a further expansion cone connected this section to a 4.86 m diameter
section 66 m long. This was followed by a semi-circular 10.7 m diameter
extension. The development and calibration of the simulator has been reviewed
by Tate [6.20] who noted that the guns were removed and the blast wave was
eventually generated in the 1.82 m tube by the detonation of up to 100 m of
detonating cord (59 g of PETN/m) wound vertically on an expanded polystyrene
former. The use of explosives rather than compressed air meant that the
diaphragms of the early small shock tubes were no longer required. The blast
wave shape was found to be very similar to the ‘Friedlander’ shape discussed
earlier (see Figure 1.1), although for economic reasons the larger tubular sections
were shorter in length than would be normally adopted in shock tube work.
The complete facility is shown in Figure 6.3, taken from ref. [6.20].
We have already noted that a rarefaction wave is produced at the open end
of a shock tube, and considerable research has taken place to support the
development of suitable Rarefaction Wave Eliminators (RWE). In the Foulness
tube the eliminators take the form of metal grids that give partial blockages at
the open end of the final tunnel and at the end of the 4.86 m diameter section.
The blockage due to the grid at any section is about 40% of the total cross-
sectional area, and Tate points out that a restriction less than this would allow
too strong a rarefaction wave to propagate back into the preceding tube; a
greater restriction would limit the peak pressure attainable in the final semi-
circular test section. The ratio of the peak instantaneous pressures in the sections
of the tube are, to a first approximation, inversely proportional to the ratio of
the cross-sectional areas. At the time of Tate’s paper the peak pressure in the
final semi-circular tube was limited to less than 55 psi.
Figure 6.4 Air-driven blast wave simulator, Centre d’études de Gramat, France
(from Gratias and Monzac, ref. 6.20).
use unheated driver gas to achieve smooth wave forms at high pressures, and
to control positive phase duration. It was thought that a rotary drum fast-
acting valve was a practical way to mechanize the wave-shaping technique.
The schematic driver arrangement for the US facility is shown in Figure 6.5.
The German blast simulator was formed by driving a tunnel into the rock of
the Reiter Alpe range of hard chalkstone. The tunnel is 106 m long and semi-
circular in section with a total cross-sectional area of 76 m2 and the general
arrangement is shown in Figure 6.6, taken from a paper by Ackermann and
Klubert [6.27]. The blast wave generator consists of 144 shock tubes, combined
into a battery. The compressed air bottles are clamped horizontally into a frame,
and in contact with a wall so that the thrust resulting from the bursting of the
diaphragms can be absorbed by the rock. Each tube has an interior volume of
Figure 6.6 Shock tube driven blast simulator, Reiter Alpe, Germany (from
Ackerman and Klubert, ref. 6.27).
3
0.4 m , and is safe to 200 bar. The diaphragms are made from hardboard and
fired by plastic explosive. The structure testing area is about 50 m from the
battery of driver tubes, but experiments showed that at this point the side-on
pressure/time relationship is not ideal. A sharp instantaneous pressure rise is
followed by exponential decay, but then there is another sudden pressure increase
lasting for 50 msec. Because of this pressure increase the reflections from the
walls overlap the blast wave and subsequent exponential decay suddenly ends
after a time of 300 msec. During the early stages of use a 0.8 bar shot resulted in
damage to the walls and ceilings of local village houses.
Another blast simulator embedded in rock was constructed in the mid-
1960s at the Swedish fortifications establishment Fort F, Eskilstuna, Sweden.
It has been fully described by Bergman [6.28], and consists of an explosion
chamber, test chamber and exhaust tunnels, as shown in Figure 6.7. Hexotol
charges up to 100 kg in weight can be detonated in the explosion chamber,
which give peak gas pressures of up to 6 or 7 kPa. The pressure wave passes
through a diffuser which reduces the ‘spikes’ in the wave form, and it then
dynamically loads the test specimen. The gas is then gradually evacuated by
regulating the size of the entrance to the exhaust tunnel, and by this means the
duration of the pressure pulse can be controlled. The main tunnel is square in
cross section (2.1 m sides) and has a total length of 220 m. As Figure 6.7
shows, the tunnel eventually discharges to atmosphere through a muzzle system.
By disconnecting the exhaust loop, removing the diffuser and emptying the
test section of equipment, a straight shock tube is obtained. The gas flow
from the entrance can then load large military objects placed on level ground
Figure 6.7 Blast simulator embedded in rock, Eskilstuna, Sweden (from Bergman,
ref. 6.28).
immediately outside the end of the tunnel, where the stagnation pressure can
reach 300 kPa with a duration of about 0.5 sec. However, the side-on pressure
has only about 0.01 sec positive duration.
Instead of detonating a single charge in the explosion chamber, a line charge
can be detonated in the main tunnel. The charge is normally constructed of
cordtex filaments that are either hung on posts or wound on cylindrical formers
of plastic or steel. Two types of cordtex were used in a test programme reported
by Bryntse [6.29], Nobelcord at 10 g PETN/m and E-cord at 5 g PETN/m.
The muzzle at the entrance, which cancels the rarefaction wave, is made as a
hollow concrete block with a number of bricks that can be added or removed
to adjust the size of the opening. In the tests the open area of the muzzle was
2
varied between 0.25 and 2.25 m .
There was an approximately linear relationship between charge length (up
to a length of 100 metres) and positive duration (up to 300msec), but a
Friedlander-type pressure/duration relationship could only be obtained for
higher pressure levels. The lower pressure limit was about 8 kPa reflected
pressure, below which a line of cordtex tends to give a square-formed pressure
pulse; this problem can be alleviated somewhat by helically winding the cordtex
on a 0.3 m diameter cylindrical polystyrene former. The test results were analysed
by Bryntse, who compared the actions following the detonation of a
concentrated charge and a distributed (or line) charge in a tube with smooth
walls. For concentrated charges two exponential waves will propagate along
the tube, after some complex reflection processes, with a gradual change in
the pressure/ duration relationship. For a distributed charge the reflection
interaction with the tube walls will occur over a much greater length of the
tunnel, and the outward moving pressure pulses will be of long duration and
of approximately square wave shape. The wavelength of each pulse will be
equal to the charge length.
The tests indicated that the detonative combustion process for line charges
was somewhat incomplete, and was strongly influenced by imperfections in
the cordtex production process. Although the PETN content per unit length
was thought to vary by no more than ±10%, small variations in packing
density were found to cause large variations in released energy on detonation.
Because of this the relationship between impulse and charge weight contained
a much wider experimental scatter. The relationship was also thought to be
influenced by the changes in shock wave shape, because ripples appear in the
original rectangular-shaped pulse, and these often develop into several peaks.
The final simple exponential profile was only achieved at high-pressure levels.
We have already mentioned the possibility of noise and vibrational
disturbance to the local environment when very large simulators are fired,
and there have been a number of reviews of this problem over the years,
beginning with an investigation by James [6.30] in 1965 and continuing more
recently with studies by Cadet [6.31] in France and Reed [6.32] in the USA.
James concluded that incident over pressure (pi) in psi at a distance d feet
from the mouth of a shock tube, and at an angle θ to the tube axis, was given
approximately by the relationship
2
where A is the cross-sectional area of the tube in m and R the distance in
metres. This assumes an overpressure-distance decay exponent of –1.1. Tests
showed that for a large shock tube (5.8 m diameter) at the Sandia laboratory in
the USA (driven by 500 kg of primacord), and for a tunnel shock tube simulator
2
(cross-sectional area of the tunnel portal of 28.6 m ) at the Nevada Test Site of
the US Defense Nuclear Agency, the distant air blast propagations could be
reasonably predicted by this equation (6.3). Local characteristics, such as woods
and hills, affect these propagations, but the equation gives a useful indication of
overpressures at distances up to 5 or 10 kilometres from the simulators. The
assessment of environmental effects was particularly important at the Sandia
site, because the driver end of the shock tube was open to the east, and pointed
at the Solar Thermal Test Facility. This incorporated 3000 glass mirrors and
was about 1.8 kilometres away! The cross-fertilization of scientific research
does not normally involve years of bad luck through cracked mirrors!
So far we have reviewed the experimental simulation of air blast, but in
assessing the survivability of equipment in nuclear explosions the possible effects
of heat cannot be ignored. Several research projects in the 1980s were aimed at
described in section 6.2 were fitted with test sections in which the travelling
shock front passed over a mass of soil. Various model structural forms could
be built within the mass, and subjected to shock loading via the soil cover. A
typical example of this arrangement was the blast simulator at Foulness Island
in the UK, where a short length of the 8 ft diameter section ran over a tank of
soil. At this point the lower wall of the simulator tunnel was removed so that
the shock front ran transversely over the soil surface. A cross section of the
tube at the test zone was given in an earlier book by the author, Bulson [6.37],
and is reproduced here in Figure 6.9. Similar arrangements were made in
horizontal blast simulators in Europe and the USA.
A second type of test apparatus, for smaller-scale work, was to use a vertically
mounted shock tube, the open lower end of which was set immediately above
a mass of soil. The initiation of a shock wave resulted in the application of an
instantaneous pressure vertically over the surface of the soil. Simple model
structures could be buried in the soil, and the loading on them inferred from
strain and acceleration records. A vertical shock tube facility of this type was
also built at Foulness, and has been described in a report by Clare [6.38]. The
author used this shock tube to apply loads to buried thin-walled cylinders,
and the experiments are described in reference [6.37]. The shock tube was 40
feet long, 2 feet in diameter, and was fired by a coiled charge of cordtex at the
top. Clare showed that the blast pressure-duration relationship closely followed
the Friedlander form. The soil tank introduced under the lower opening of the
tube was 5 feet square, and the test structures were set in various soils so that
they could be viewed through tunnels. High speed cine-film was taken through
the tunnels of each collapse sequence. The firing procedure was first to set the
Figure 6.9 Cross section of shock tube at the test zone (Foulness, UK) (UK Atomic
Weapons Research Establishment, 1980) (from Bulson, ref. 6.37).
Figure 6.10 Large blast load generator at the Waterways Experiment Station, USA
(from Flathau et al., ref. 6.39).
The US Waterways Experiment Station was also the site of one of the most
ambitious soil loading facilities ever constructed, the large Blast Loading
Generator, described in ref. [6.39]. The facility, commissioned in the mid-
1960s, took the form of a central firing station and test chambers, as shown in
Figure 6.10, and was designed primarily to test underground protective
structures subjected to pressures typical of kiloton and megaton nuclear devices.
The structure consisted of a massive post-tensioned prestressed concrete reaction
portal, which was basically a 48 ft×28 ft×28.4 ft deep slab with a 13 ft deep
tunnel cut through it. Into this tunnel fitted the circular test chamber, about
23 ft diameter, containing the soil medium and the structure under investigation.
The chamber was formed by stacking three steel rings, adding a further ring
containing firing tubes, and then adding a top lid or ‘bonnet’. High explosive
was placed in the tubes and a baffle system was used to ensure that the soil
surface was uniformly loaded on detonation. The vertical reaction from the
exploding charges was taken on the concrete reaction structure. It was stated
that peak pressures from 5 to 400 psi with rise times of between 2 and 4 msec,
and positive phase duration times of several seconds could be reproduced in
this generator.
For smaller-scale work the US Waterways Experiment Station also installed
a Small Blast Load Generator, which was designed and constructed by Boynton
Associates [6.40] and is shown diagrammatically in Figure 6.11. The steel
container was a 9/16 in thick cylindrical shell with a dome-shaped top. The
walls were made from stacked rings, all about 4 ft diameter but having different
heights. Combinations of rings were selected to give a range of chamber heights.
The soil mass consisted of a column of sand 4 ft diameter, extending 9 ft
below floor level. For explosive loading primacord was detonated in two firing
Figure 6.11 External view of small blast load generator, Waterways Experiment
Station, USA (from Boynton Associates, ref. 6.40).
Figure 6.12 Blast load simulator at the University of Illinois, USA (from Egger,
1957, ref. 6.41).
6.4 REFERENCES
6.1 Baker, W.E. (1973) Explosions in Air, University of Texas Press, USA.
6.2 Whiteside, T. (1967) Instrument Development Section Notes, Atomic Weapons
Research Establishment, Foulness, UK.
6.3 Baker, W.E. and Ewing, W.O. (1961) Miniature Piezo-electric Gauges for
Measuring Transient Pressures on Airfoils, BRL Memo Report No 1329, Aberdeen
Proving Ground, Maryland, USA.
6.4 Hoffman, A.J. and Mills, S.N. (1956) Air Blast Measurements about Explosive
Charges at Side-on and Normal Incidence, BRL report No 988, Aberdeen Proving
Ground, Maryland, USA.
6.5 Granath, B.A. and Coulter, G.A. (1962) BRL Stroke Tube Piezo-electric Blast
Gauges, BRL Tech Note No. 1478, Aberdeen Proving Ground, Maryland, USA.
6.6 Muirhead, J.C. and McMurty, W.M. (1962) Surface tension gauges for the
measurement of low transient pressures, Rev. Sci. Instr., 33(12).
6.7 Palmer, W.O. and Muirhead, J.C. (1969) A squirt gauge for peak blast pressure
indication, Rev. Sci. Instr., 40(12).
6.8 Selig, E.T. (1964) A review of stress and strain measurement in soil, Proc. Symp.
on Soil-structure Interaction, University of Arizona, Tucson, USA, September.
6.9 Abbott, P.A., Simmons, K.B., Reiff, C.M. and Mitchell, S. (1967) Recent soil
stress gauge research, Proc. Int. Symp. on Wave Propagation and Dynamic
Properties of Earth Materials, University of New Mexico, Albuquerque, August.
6.10 Bur, A.J. and Roth, S.C. (1985) A polymer pressure gauge for dynamic pressure
measurements, Proc. 2nd Symp. on the Interaction of Non-nuclear Munitions
with Structures, Panama City Beach, Florida, April.
6.11 Turner, R.G. (1985) Transducers old and new, Proc. 9th Int. Symp. on Military
Applications of Blast Simulation, St Edmund Hall, Oxford, UK, September.
6.12 Vieille, P. (1899) Comptes Rendus, 129, 1228.
6.13 Wright, J.K. (1961) Shock Tubes, Methuen (London).
6.14 Reynolds, G.T. (1943) A Preliminary Study of Plane Waves Formed by Bursting
Diaphragms in a Tube, OSRD Report 1519.
6.15 Smith, L.G. (1945) Photographic Investigation of the Reflection of Plane Shocks
in Air, OSRD Report 6271.
6.16 Bleakney, W., White, D.R. and Griffith, W.C. (1950) Measurements of diffractions
of shock waves and resultant loading of structures, Journal of Applied Mechanics,
17, 493.
6.17 Taub, A.H. quoted by Reynolds, G.T. in ref. [6.14].
6.18 Alpher, R.A. and White, D.R. (1958) Flow in shock tubes with area change at
the diaphragm section, Journal of Fluid Mechanics, 3(5), February.
6.19 Leys, I.C. (1983) AWRE Foulness nuclear air blast simulator: construction and
calibration of the enlarged facility, Proc. 8th Int. Symp. on Military Applications
of Blast Simulation, Spiez, Switzerland.
6.20 Tate, J. (1985) AWRE Foulness nuclear air blast simulator: development and
calibration of the facility, Proc. 9th Int. Symp. on Military Applications of Blast
Simulation, St Edmund Hall, Oxford, UK, September.
6.21 Clare, P.M. (1979) Methods used to drive the AWRE Foulness nuclear blast
simulator, Proc. 6th Int. Symp. on Military Applications of Blast Simulation,
Cahors, France.
6.22 Ethridge, N.H., Teel, G.D. and Reisler, R.E. (1983) A differential pressure gauge
for measurement of dynamic pressure in blast waves, Proc. 8th Int. Symp. on
Military Applications of Blast Simulation, Spiez, Switzerland.
6.23 Gratias, S. and Monzac, J.B.G. (1981) Le simulateur de Souffle a Grand Gabarit
du Centre d’Etudes de Gramat, Conception, Etudes, Performances, Proc. 7th
Int. Symp. on Military Applications of Blast Simulation, Suffield, Canada.
6.24 Hoffman, G. (1985) Simulation of real weapon-effects in multiple-driver
shocktubes , Proc. 2nd Symp. on the Interaction of Non-nuclear Munitions
with Structures, Panama City Beach, USA, April.
6.25 Gratias, S. (1985) Test of a fast opening valve for the 2.4, shock tube at the
CEG, Proc. 9th Int. Symp. on Military Applications of Blast Simulation, St
Edmund Hall, Oxford, UK, September.
6.26 Osofsky, I.B., Mason, G. et al. (1985) Wave shaping by valve motion, Proc. 9th
Int. Symp. on Military Applications of Blast Simulation, St Edmund Hall, Oxford,
UK, September.
6.27 Ackermann, J. and Klubert, L. (1985) The new large blast simulator of Reiteralpe
Proving Ground, Proc. 9th Int. Symp. on Military Applications of Blast
Simulation, St Edmund Hall, Oxford, UK, September.
6.28 Bergman, S.G.A. (1967) The RSFA underground shock tube facility for testing
fortification equipment, Proc. Conf. on Military Applications of Blast Simulators,
Defence Research Establishment Suffield, Canada, Vol. 1.
6.29 Bryntse, A. (1985) Detonating fuse in the FortF shock tube III to obtain air
shock waves with low amplitude and long duration, Proc. 9th Int. Symp. on
Military Applications of Blast Simulation, St Edmund Hall, Oxford, UK,
September.
6.30 James, D.J. (1965) An Investigation of the Pressure Wave Propagated from the
Open End of a 30×18 in Shock Tube, AWRE Report No 0–60/65, AWRE
Aldermaston, UK, September.
6.31 Cadet, A. (1983) Investigation of the pressure wave and noise measurements
outside the Large Blast Simulator, Centre d’Etudes de Gramat, France, Proc.
8th Int. Symp. on Military Applications of Blast Simulation, Spiez, Switzerland,
June.
6.32 Reed, J.W. (1980) Project PROPA-GATOR—Intermediate range explosion
airblast propagation measurements, 19th Explosives Safety Seminar, Vol. II,
DOD Explosives Safety Board, Washington, DC, USA, September.
6.33 Haasz, A.A., Gottlieb, J.J. and Reid, L.D. (1985) Air-curtain system for blast
wave simulators to remove combustion products from thermal radiation sources,
Proc. 9th Int. Symp. on Military Applications of Blast Simulation, St Edmund
Hall, Oxford, UK, September.
6.34 Pearson, R., Opalka, K. and Hisley, D. (1985) Design studies of drivers for the
US Large Blast/Thermal Simulator; from same source as ref. [6.33].
6.35 Borgartz, B.O. (1985) AWRE Foulness combined blast and thermal radiation
simulator, installation of the TRS in the blast tunnel; from same source as ref.
[6.33].
6.36 Hoffman, G. and Behrens, K. (1983) Simulation of pressure waves and their
effects on loaded objects; Part 1: outlining the problem, description of the
simulation device, Proc. Symp. on the Interaction of Non-nuclear Munitions
with Structures, US Air Force Academy, Colorado, USA, May.
6.37 Bulson, P.S. (1985) Buried Structures—Static and Dynamic Strength, Chapman
& Hall, London.
6.38 Clare, R. (1966) A Face-on Vertical Blast Loading Simulator for Structural
Response Studies, Atomic Weapons Research Establishment, UK, Report
E1/66.
6.39 Flathau, W.J., Dawsey, J.V. and Denton, D.R. (1968) Blast load generator facilities
and investigations of dynamically loaded concrete slabs, Prevention and Protection
against Accidental Explosion of Munitions etc., Annals of New York Academy
of Sciences, 152, Art. 1.
6.40 Boynton Associates (1960) Operation Manual for 250 psi 4 foot Diameter
Dynamic Load Generator, Boynton Associates, La Canada, California, USA.
6.41 Egger, W. (1957) 60 kip Capacity, Slow or Rapid Loading Apparatus, Dept of
Civil Engineering, Univ of Illinois, USA, Report SRS 158.
6.42 Sinnamon, G.K. and Newmark, N.M. (1961) Facilities for Dynamic Testing of
Soils, Dept of Civil Engineering, Univ. of Illinois, USA, Report SRS 244.
7.1 INTRODUCTION
The analysis of dynamic impact is an important feature in the study of explosive
effects on structures, partly because military or terrorist activities frequently
involve the transportation of explosive charges by high-speed projectiles, and
partly because the explosion of the charge is often accompanied by the high-
speed distribution of fragments of disintegrating containers. Sometimes the
disintegration of one part of a bombed structure sends fragments of that
structure over a larger area.
An explosive warhead can be aimed at a structure by a variety of means:
aerial bombs, rockets, artillery, missiles. The degree to which these warheads
penetrate a surface structure, or the earth adjacent to a buried structure before
exploding, has a large effect on the eventual damage. We therefore need to
survey the state of research in the fields of bomb or missile penetration into a
range of materials, from soil to concrete, and from steel sheet to sophisticated
armour plating. We must take account of the possible deformation and
crumpling of the charge casing during the penetrative process, because this
will affect the penetration depth.
The fragmenting casing usually projects small pieces of virtually undeformable
metal at high velocities. When these strike solid structures such as protective
concrete bunkers considerable surface damage may occur. When these solid
fragments strike relatively light structures, like aircraft wings or the cellular
walls of naval craft, the fragments remain in good shape, but the sheet metal is
penetrated and deformed. Certain military weapons like high-velocity cannons
can spray large areas of plated structures with projectiles that produce a pepper-
box type of damage to large areas of unsupported sheeting. If the holes are
irregular in shape, and occur in brittle material, fast-growing cracks can be
formed. These can limit the residual life of the structure under cyclic loading.
The original scientific investigations of dynamic penetration were linked
to the development of military weapons in the eighteenth and nineteenth
centuries, and this section will therefore include a historical reference to the
research work of that era. One of the earliest penetration theories [7.1] was
due to J.V. Poncelet (1788–1867), mentioned earlier in the Introduction, who
was a French military engineer and professor of engineering mechanics at the
military school in Metz. He had a perceptive view of the physics of the
penetrative process, in which he combined the resistance of a medium to the
static penetration by a body with a resistance linked to the speed at which the
body was moving. He considered that the dynamic resistance in a material
such as soil or sand should be analogous to resistance to fluid dynamic flow.
He therefore assumed that the force resisting penetration would be given
by the equation
F=–A(a+bV2), (7.1)
where
F = resisting force
A = cross sectional area of the projectile
V = impact velocity
a, b were constants to be determined by experiment.
If the mass of the projectile=m, the equation of motion is
(7.2)
(7.3)
His very early investigations of the effects of dynamic loads were of great
fundamental value. He appreciated the importance of ductility in the absorption
of kinetic energy and he was the first to demonstrate that a suddenly applied
load produces double the stress of the same load applied gradually. He also
demonstrated the danger of vibrational resonance and of fatigue. There is
little doubt that he was a genius, and his capacity for uncovering the roots of
structural and mechanical behaviour have been of great value in many fields.
(7.4)
where
p = depth of vertical penetration in feet
Wp = total projectile weight in lb
A = cross-sectional area of the projectile in in2
K = constant depending on target material
V = striking velocity in feet/sec.
Further research, which has been summarized by Young [7.4], showed that
the form of Poncelet’s relationship, as presented by Petry, would be improved
if a factor were introduced to take account of the shape of the nose of the
projectile, if penetration was taken proportional to the square root of the
1/2
projectile pressure (Wp/A) , and if the relationship between penetration and
2
V was replaced at higher striking velocities with a relationship between
penetration and V. On this basis, Young discussed the following equation for
penetration into soil:
(7.5)
The units are reconciled in the constant terms 0.53 and 0.0031.
S is the soil constant and N a nose-performance coefficient, which takes
account of the nose shape, ranging from a flat nose (N=0.56), to a cone shape
having a length of cone equal to three times the diameter (N=1.32). Note
that, contrary to the theories of Poncelet and Perry, penetration in soil is
1/2
proportional to (Wp/A) rather than Wp/A. Values of S and N were found
experimentally, and linked to broad ranges of soil type thus:
Rock: S=1.07
Dense, dry silty sand: =2.5
Silty clay: =5.2
Loose, moist sand: =7.0
Moist clay: =10.5
Wet silty clay: =40
Soft wet clay: =50.
Thus, all other things being equal, penetration in soft wet clay is about 50
times as far as in rock, and about 20 times as far as in dense, dry, silty sand.
Typical values of N are:
Flat nose N =0.56
Tangent Ogive 2.2 CRH =0.82
Tangent Ogive 6 CRH =1.00
Cone (L/D=3) =1.32.
Care must be taken not to give these figures a greater scientific accuracy than
they merit, since the scatter of penetration test results is notoriously wide.
Experimental data exists on penetration depths up to 220 feet and in Young’s
equations there is apparently no upper limit to the value of p for penetration
into homogeneous soil. There is a lower limit, however, and it is suggested that
the equations apply as long as the total depth of penetration is equal to ‘three
body diameters plus one nose length’. At lesser depths the mechanics of penetration
are not fully activated. Further, if the nose length is more than one-third of the
total penetrator length there is insufficient length of cylindrical section to ensure
stability, because the centre of gravity of the penetrator is too far aft.
It is useful to note that the practical range of Wp/A is fairly limited, and
ratios greater than 15 to 20psi are difficult to achieve. For solid steel, to take
an extreme example, a billet having a diameter 4 in and a length of 60 in, has
a value of Wp/A=17. There is also a practical range for velocity, V, in Young’s
equations. He suggests that at impact velocities less than 105 ft/sec, the
penetration depth is too shallow for reliable analysis.
© 1997 Philip Bulson
Penetration into soil, stone and rock 145
(7.6)
where γ is a positive constant. This has been pointed out by Allen, Mayfield
and Morrison of the US Naval Ordnance Test Station, China Lake, California
[7.8]. Poncelet’s equation, for example, corresponds to Eq. (3.6), with ß=0,
and a and γ positive constants. The earlier work of Robins [7.9] and Euler
[7.10] took a and ß equal to zero, and γ as a positive constant. Allen et al.
related the coefficients of the penetration formulae to the drag coefficient
used in the analysis of a projectile passing through a continuous fluid having
a density ρ. If the projectile has a mass m, and a presentation area on a plane
normal to the flight line of A, the analysis of fluid dynamics gives
(7.7)
where V is the velocity of the projectile. Although Cd can vary with velocity, it
is thought to be reasonable to assume that it remains virtually constant over a
wide velocity range when the medium is granular rather than a fluid.
Reference [7.8] also summarizes the results of a test programme, in which steel
projectiles, 0.511 in diameter, 5.11 in long, were fired from a Browning aircraft
machine gun into dry quartz sand. The ends of the projectile were conical in shape,
and a range of cone angles were investigated, from 180° (flat end) to 10° (very
sharp). By means of a 0.404 in diameter hole drilled inwards from the base of the
projectile, the centre of gravity could be moved forwards, and it was found that to
ensure stability after entry into the sand the length of the internal hole should be
such that the centre of gravity of the projectile coincided with the base of this hole.
A comparison of penetration versus striking velocity suggested that at lower
velocities penetration would be sufficiently accurately described by the Poncelet
equation, whereas at higher velocities conventional penetration theories did
not predict the results adequately. For the higher range (V>96.5 m/sec), the
relationship was closer to the form
(7.8)
(a) For a flat nosed projectile in a uniform stream of discrete particles, with
perfectly elastic impact, Cd=4.
(b) For a flat-nosed projectile (2φ=180°) in a uniform stream with perfect
inelastic impact, Cd=2.
(c) For the circumstances of the test, 1.97>Cd>1.80 when 2φ varied between
180° and 100°. For 2φ<100, 1.80>Cd>1.0.
Artillery shells, and later bombs, were the subject of much testing, particularly
during the run up to the Second World War, and during the early years of that
conflict. The position in 1946 was summarized by Stipe [7.11] in the USA and
by Christopherson [7.12] in the UK. The former pointed out that early wartime
tests on projectiles and bombs were conducted over a limited range of velocities,
and very often the dimensions and weight of the missiles, their striking velocity
and the soil characteristics were not recorded. This was unacceptable to the
Chief of Engineers, US Army, who asked for a full study of the terminal ballistics
of soil to be made. The lack of information was also recognized in the UK,
and firing trials were organized by the Road Research Laboratory to investigate
the scientific principles in a methodical way.
The US Report noted that similar projectiles would penetrate two to three
times as far in a rich clay than in coarse sand, and that soil stratification could
affect the depth of penetration. Penetration into cohesive soils formed conical
craters, wide at the entrance and gradually tapering for the length of the
penetration path. Displaced soil was compacted, and sometimes pulverized.
It was noted that projectiles striking the soil surface at high angles of obliquity
often followed a curved path, and that aerial bombs sometimes took a J-
shaped trajectory underground, curving forward in the direction of the flight
line of the aircraft. There was a tendency for projectiles to be unstable in end-
on motion, so that they ‘tumbled’ into a side-on attitude, which often curved
the trajectory. Projectiles came to rest with their noses pointing back to the
point of entry, and blunt nosed missiles were found to ‘topple’ less quickly
than sharp-nosed varieties.
The American work was summarized in the form of data sheets, which
eventually found their way into the US Army technical manual ‘Fundamentals
of Protective Design’, TM5–855–1 (7.13). From this manual we have taken
Figure 7.1, which gives the correction term to be applied to calculated vertical
penetration when oblique penetration occurs. This term is a function of the
striking velocity, V, and it can be seen from the figure that if the cosine of the
angle of obliquity were used as the correction term, the vertical component
would be overestimated.
On the assumption that the J-shaped penetration path was a valid
approximation, its length was given as part of weapon data in ref. [7.13].
This indicated that, in general, the straight part is about two-thirds of the
total length and the radius of the curved part is between one-fifth and one-
third of the total length.
Figure 7.1 Correction factor v angle of obliquity for two values of striking speed
(from Stipe, ref. 7.11; US Army TM5–855–1, 1965).
The penetration length, L, which is the total length of the J-path, not the
final vertical depth, p, was found from the tests to be linked to striking velocity
through the parameter , where Wp is the total projectile weight, not
the weight of the explosive charge (W). The curves shown in Figure 7.2 are
taken from ref. [7.13], and refer to four types of soil and projectiles having an
‘average’ nose shape. Detailed information on other nose shapes are given in
the reference.
The vertical penetration of a range of general purpose bombs, with total
weights in the range 100 to 2000 lb, in three different soils, is given in Figures
7.3, 7.4 and 7.5. The curves were said by Stipe to agree with the scatter of
available data to within ±20%, and a rough interpolation between curves
could be made for other soil types. Because the tests were mainly carried out
with free-dropped aerial bombs, the striking velocity has been replaced by the
altitude from which the bombs were dropped in level flight. In each experiment,
delay fusing prevented explosion until full penetration had been achieved,
and the results confirm the earlier statement that penetration into clay is about
Figure 7.2 Length of the J-shaped penetration path (from Stipe, ref. 7.11; US Army
TM5–855–1, 1965 and 1986).
double that into well-compacted sand. More recently, doubt has been cast on
the assumption of a J-path, and on the relevance of analysis based on this
concept.
It was realized that the stability of projectiles in soil needed further
investigation, and since stable underwater rockets had been developed during
the Second World War, it was suggested that these might also be stable in
earth. Whether this line was ever followed is not clear now.
In the UK, Christopherson’s report [7.12] was as always a very precise and
valuable analysis of the fundamentals. He classified the soils into three types,
plastic materials like clay, granular materials like sand or gravel, and the softer
rocks such as chalk and sandstone. He assumed that for soil penetration the variables
were separable, so that the relationship between total track length, L, the diameter
of the projectile, d, and striking velocity, V, could be written in the form
(7.9)
If the velocities given to the material are largely independent of diameter, and
if , then
(7.10)
Figure 7.6 Penetration track length in clay of steel balls and flat-ended cylinders
(UK Road Research Lab., 1944; Christopherson, ref. 7.12).
shown in Figure 7.7. The relationship was given in terms of vertical penetration, p,
divided by the cube root of bomb weight (), and the number of bombs per thousand
recovered that were found at greater depths than that indicated. The mean penetration
(exceeded by half the bombs (500 out of 1000)) was given by , and
the average striking velocity was 850 ft/sec. There was a 99% certainty that, in spite
of scatter, penetration of German bombs never exceeded .
In dry sand no permanent crater was visible at the entrance hole, and the
projectiles usually turned side-on before coming to rest. The mean track length
at 850 ft/sec initial velocity was 2.33 , and the mean vertical penetration was
1.95 , giving a p/L ratio of 0.8.
In chalks and sandstones it was found that track length and penetration
are about equal. The best fit to experimental results was to take n in Eq. (7.9)
as , and curves on this basis are given in Figures 7.8 and 7.9. It was generally
felt that by 1946 a much clearer idea of bombs and shell penetration existed,
but much research remained to be done. It is doubtful whether much has been
accomplished of equal scientific importance since those days, although much
analytical work has been attempted.
If we may include pack-ice as a geological target, then it is relevant to refer
to test work in the late 1960s and early 1970s by the Sandia Laboratories.
Four penetrators with conical noses were air dropped on to pack-ice in the
penetrators were fired from a Davis gun and impacted the rock at a speed of
520 m/s, as described in a report by Longcope and Forrestal. It was noted that
sliding friction helped to increase penetration resistance.
Further work was carried out by two of the authors of ref. [7.15] at high
impact velocities by firing simulated soft sandstone targets at penetrators,
using a gas gun. Striking velocities between 0.2 and 1.2 km/s were obtained
on impact with 20.6 mm diameter penetrators, and it was found that over a
Figure 7.8 Penetration of small projectiles into sandstone (from Christopherson, ref.
7.12).
Figure 7.9 Penetration of projectiles into chalk (from Christopherson, ref. 7.12).
range of striking velocities between 500 and 1200 km/s, the axial resultant
nose force (F) was given by
F=109V1.29, (7.11)
Figure 7.10 Penetration curve for spheres into dry sand (from Stilp et al., ref. 7.18).
Figure 7.11 Absolute mass loss of steel spheres on impact with sand (from Stilp et
al., ref. 7.18).
Figure 7.12 Penetration experiments on cylinders, L/d=1 (steel cylinders into sand)
(from Stilp et al., ref. 7.18).
Figure 7.13 Penetration experiments on cylinders, L/d=5 (steel cylinders into sand)
(from Stilp et al., ref. 7.18).
Figure 7.14 Penetration experiments on cylinders, L/d=10 (steel cylinders into sand)
(from Stilp et al., ref. 7.18).
Figure 7.16 Penetration simulation by SAMPLL (from Schoof et al., ref. 7.19).
p=0.117KSN(Wp/A)0.5(V–30.5) (7.12)
2
(when p is in cm, Wp in kg, A in m , V in m/s).
They were unhappy about the mass scaling factor, K, which they considered had
no theoretical basis, and they also questioned the dimensional correctness of the equation
and the exponent 0.5 used with the mass/area ratio. We have discussed earlier the
variation between Young’s work and other early theories with regard to this exponent.
They also felt the soil penetration index, S, was rather qualitative—a problem already
understood and discussed at the time when Young’s work was first presented.
As a result of about 80 centrifuge model tests, the authors produced the
following ‘best fit’ equations:
Dense Ottawa Flintshot sand,
p=0.00282(Wp/A)0.915, (7.13)
The values of S, buried in these formulae, were given as 0.137 for dense sand,
and 0.30 for loose sand. Note that the exponents are considerably greater
than the 0.5 originally given by Young. The authors went on to investigate
whether all tests could be accommodated by the use of one formula, if the Wp/
A term were replaced by a dimensionless parameter. They proposed pDr, where
pDr=(Mpg)/(ApltV2Dr0.5). (7.15)
Mp and Ap are the mass and area of the projectile, lt is the total soil density,
and Dr is the relative density of the soil, which is a measure of the density in
relation to its minimum and maximum values.
The equation for penetration depth in metres for the Ottawa sands then
becomes
p=21.427(pDr)0.894 (7.16)
(p in metres).
In recent years there has been considerable interest in the use of rock rubble
to help limit the penetration of aerial bombs and missiles, when these are aimed
at underground facilities. At the time of writing much of the work has a security
classification, certainly in terms of performance figures, but it has been generally
accepted that layers of rock rubble give a very high protection against penetration
by general purpose bombs, but are less effective against armour-piercing bombs.
A 20 foot thick rock rubble field is reckoned to be an effective countermeasure
against advanced designs of penetrating weapons, but this is a very expensive
option in terms of raw material and excavation. There is some doubt whether
conventional air bases are fully protected in terms of penetration, and it is thought
they could be vulnerable to advanced weapons.
Recently an analytical method of solving the problem of penetration through
rock rubble has been put forward by Gebara, Pau and Anderson [7.21]. The
authors make the point that although rock rubble overlays are considered a
good alternative to brick and expensive burster slabs for protecting underground
installations from high kinetic energy penetration projectiles, the modelling
of rubble systems has proved difficult. The Waterways Experiment Station of
the US Army have devoted research funds to the problem, as illustrated by the
work of Nelson, Ito, Burks et al. [7.22] and Gelman, Richard and Ito [7.23].
The latter took the rubble as a uniform mesh of octagons, which is a rather
simplified fracturing geometry. Ref. [7.21] presents a solution using the Finite
Block Method, put forward earlier by Chen and Pau [7.24], in which the
penetrator may burrow its way into the mesh of blocks without fragmentation,
or where the penetrator, on impacting one block, may cause it to fragment. In
the latter case the Voronoi construction for fragmentation was chosen, in
which the block area is partitioned by perpendicular bisectors between random
Figure 7.17 The process of fragmentation and penetration into rock rubble by the
finite block method (from Gebara et al., ref. 7.21).
points. This produces fragments that are reasonably equal in length and breadth,
rather than heavily elongated, and is thought to be well representative of
experiments. Figure 7.17, taken from ref. [7.21], shows a penetrator hitting a
single block, causing it to fragment. The size of rocks in a rock rubble fabrication
is about three times the diameter of the projectile, so the fragmentation of
individual rocks is highly likely. This results in less deflection from the direction
of the projectile path than would be the case in a non-fragmenting block
analysis.
Earlier mention of the SAMPLL computer code is a reminder that
considerable efforts have been made in recent years to establish computer
software for the prediction of penetration into soils and other materials. In
1987 a paper by Schwer, Rosinsky and Day [7.25] presented a computational
technique for earth penetration, which included direct coupling between a
deformable target and a deformable penetrator. In their review of recent
computational work they called attention to the code PENCO, which was
established at the US Army Waterways Experiment Station in the early 1980s,
and reported on at the time by Creighton [7.26]. This code treats multi-layered
targets of hard or soft rock materials, where the target resistance is specified
by the unconfined compressive strength, or by the S factor in Young’s equations
(see Eq 7.5). It does not treat the interaction between penetrator and target
for deformable penetrators, and appears to be limited in its treatment of very
thin targets having thicknesses of only one or two penetrator diameters.
Reference was also made to the possible use for soil penetration, using rigid
penetrators, of the codes HULL [7.27] and TRIFLE [7.28], which were Eulerian-
based finite difference codes developed mainly for fluid dynamics problems.
The Lagrangian penetration grid consists of a narrow tunnel that coincides
with the line of the penetrator trajectory, and a typical target grid, taken from
ref. [7.25] is shown in Figure 7.18. The use of the tunnel is to reduce computation
Figure 7.18 Typical Lagrangian penetration grid with narrow tunnel passing
through the target (from Schwer et al., ref. 7.25).
Figure 7.19 Penetration data for a projectile entering Thirsty Canyon tuff (from
Amini and Anderson, ref. 7.31).
Table 7.1
the penetration path length and penetration normal to the face of the composite
specimen. The tests were carried out with either Polyester Polyurethane or a
blend of Polyether Polyurethane as the base material, and with aggregates of
crushed limestone, crushed basalt or river gravel. The percentage of polymer
by weight was about 9% for gravel and 7% for limestone and basalt. The
percentage of rock aggregate by weight was about 60%, and the rock size
was in the range 26.5 to 37.5 mm.
An indication of the results is given in Table 7.1, which gives the mean
normal penetration, and the maximum percentage of voids in the cast specimens
for the Polyether Polyurethane (the preferred material).
ratio between 70 and 80%, and a weight range starting at 2000 lb and going
upwards. They destroyed buildings by blast and ground shock rather than by
penetration and explosion.
The fusing was important to the type of loading that was imparted to targets.
Medium capacity bombs fused instantaneously, so that no penetration occurred
before detonation, and these were used when the fragmentation of the casing
was required to be the major threat—particularly against soft skinned vehicles
or against the light metallic structures of aircraft. Fragments were not thought
to be very important when general damage to surface structures was required.
In addition to the above categories, very large bombs were designed for use
against particular targets. Examples that received much publicity were the
‘Tallboy’ bomb (12 000 lb) designed to attack reinforced concrete German U
Boat pens on the Atlantic coast of France (and also used against V-weapon
sites), and the Grand Slam bomb (22 000 lb). Barnes Wallis (much later to be
knighted) was clear that it was necessary when attacking important buried
targets to ‘inject the largest possible explosive charge to the greatest possible
depth in the medium’. In spite of the accent on penetration, the first Grand
Slam bomb was in fact dropped in March 1945 in operations against the
Bielefeld Viaducts in Germany. The first Tallboy bomb was dropped in June
1944, and it is interesting to note that near misses up to 40 feet were more
damaging than direct hits, particularly when attacking concrete, brick or
masonry bridges and viaducts. The extensive damage was caused by the effect
of earth shock on the foundations.
Forty-one Grand Slam bombs were dropped in the last two months of the
war, with the following penetration figures: Height of drop 16 000 to 18 000
feet, penetration into sand or chalk 60 to 75 feet, penetration into clay 90 to
100 feet. The Tallboy bombs, dropped from the same height, penetrated to
depths about two-thirds as deep.
For a given striking velocity of 1000 ft/sec, the average depth of penetration
of these bombs was linearly related to the cross-sectional density of the bomb
2
(W/A) in lb/in , as indicated in Figure 7.20. The Tallboy bombs used to attack
U Boat pens penetrated the reinforced concrete roof by more than 6 or 8 feet
before exploding, and subsequent blast from the exploding charge caused a
further 10 feet of penetration. The general view of the time was that a Tallboy
bomb was completely effective against capital ships, viaducts and large bridges,
tunnels with up to 50 feet of overburden, and reinforced concrete roofs 10
feet thick.
From the extensive range of wartime data on the penetration of all types of
bomb into reinforced concrete, the British Road Research Laboratory produced
an empirical formula, as follows:
(7.17)
where
p=penetration (in)
Wp=projectile mass (lb)
d=projectile diameter (in)
)
sc=crushing strength of concrete (lb/in2
V=striking velocity (ft/sec)
c=maximum aggregate size (in).
Later, after extended investigations involving larger ranges of diameter and
aggregate size, and after noting that the power to which v was raised was a
function of crushing strength, a final version of the UK equation, due to Whiffen
(7.34), became
(7.18)
where n= .
The experimental ranges on which the equation was based were: sc:800 to
2
10 000 lb/in ; W:0.3 to 22 000 lb; d:0.5 to 38 in, d/c:0.5 to 50; V:0 to 3700 ft/
sec; ogival projectiles with nose shapes between 0.8 and 3.5 calibre radius.
The formula filled experimental data within a scatter band, slightly less than
±15%. The experiments indicated that when V=1750 ft/sec the depth of
Figure 7.20 Average depth of penetration in feet for Tallboy bombs (from
Christopherson, ref. 7.12).
(7.19)
The American wartime work in this field was reported by Beth [7.35]. Smalland
model-scale experiments were carried out at Princeton University, and large-
scale tests were made at the Aberdeen Proving Ground. A number of National
Defence Research Committee reports during the period 1942–45 contributed
towards the final paper by Beth, in which he presented a penetration theory
leading to the equation
(7.20)
The term ⌬l was a nose correction, which took account of the assumption
that the total volume removed by the penetration was more important than
the length of the penetration path. For a flat-ended projectile ⌬l=0, but for a
spherical-ended projectile ⌬l=d/6.
Figure 7.21 Penetration of projectiles into concrete (from Whiffen, ref. 7.34).
(7.21)
The units were similar in dimensions to those in the British formula, and as in
that formula, the accuracy was said to be within ±15%.
During the Second World War it was realized that penetration into concrete
was affected by the amount of cratering that took place at the point where the
projectile first made contact with the slab, and by the amount of rear face
spalling that might occur when the slab was relatively thin. The penetration
formulae so far discussed assumed that the slab thickness was much greater
than the penetration length of the projectile.
The front face spalling or cratering was estimated from the results of tests
as follows:
crater depth (in)=2.75×projectile diameter (d)
crater diameter (in)=11.6d (c/d)0.125,
(7.22)
and the rear face spalling or ‘scabbing’ was estimated for high explosive bombs
and shells in terms of the minimum thickness of slab (tp) to resist full penetration.
If p was the calculated penetration in a thick slab, then tests showed that a
reasonably accurate simple formula was
(7.23)
If tp was less than this, the rear face spalling caused full penetration of the thin
slab to occur.
The analysis of projectile penetration into mass-concrete or reinforced
concrete structures continues to attract sponsored research. The methods of
analysis vary from entirely empirical to attempts to model the behaviour of
the medium using finite blocks or elements. To take an example, a recent
empirical study has been made by Forrestal et al. [7.36], who examined the
penetration depth of ogive-nosed projectiles into concrete targets, and who
presented an equation for penetration depth in terms of an increased unconfined
compressive strength of the concrete. In their work they call attention to a
review of empirical equations presented earlier by Brown [7.37].
A second recent study has been made by members of the Picatinny Arsenal
(US Army) on the resistance of reinforced concrete structures to penetration
by Copper and Titanium spherical nosed projectiles launched from a gun.
The targets were right circular cylinders of concrete, and steel plates were
2 3
linear relationship between t/D and 2V Wp/D , at the upper limit of scatter,
and a power relationship at the lower limit. For the perforation of metal
plates with a Brinell Hardness Number between 250 and 300, and a projectile
striking with zero obliquity with t/D in the range 0 to 2.0, the lower scatter
2 3 3/2
band relationship was found to be WpV /D =(t/D) .
Early publications on metal penetration, taking account of the plastic
deformation of the projectile were due to G.I.Taylor [7.42, 7.43]. An
approximate theory of armour penetration, taking account of energy dissipation
as the result of plastic deformation, and also considering the heating of the
interface between the projectile and the plate, was established by Thomson
[7.44] in 1955. During the 1960s and 1970s there was considerable analytical
activity, and during this period ballistic perforation dynamics was examined
by Recht and Ipson [7.45]. They analysed the relationship between the energy
lost to deformation and heating and the change in kinetic energy of the projectile.
Their equations gave good agreement with post-perforation velocities when
thin plates were perforated by blunt cylindrical fragments. Perhaps the best
documented work during this period was the extensive research carried out in
the USA under the guidance of Goldsmith [7.46], whose work has been
summarized in a number of fundamental papers. A typical review is contained
in reference [7.47], written in the late 1970s.
Much of the fragmentation that causes penetrative damage comes from the
explosion of tubular bombs, and it was during the Second World War that the
physics of the fragmentation process was first investigated, again by G.I.Taylor
[7.48]. He showed analytically that the distribution of stress within the wall of
a tube containing detonating explosive was such that longitudinal cracks were
likely to form at the outer surface. These cracks penetrate through to the inner
wall and cause fragmentation when the internal pressure becomes equal to the
tensile strength of the material. The longitudinal cracks in the steel casing of a
bomb start close to the detonation wave point, and open out rapidly as the tube
expands. Although the cracks are wide, they do not allow the pressurised content
to escape until the tube has expanded to double its initial diameter.
The maximum distances that fragments can be driven outwards from
explosives depend, of course, on their initial velocities, and these are a function
of the size of the explosion. The maximum radial horizontal distance in metres
1/3
(V) is often given as 45W , where W is the equivalent weight of TNT charge
in kilograms. This was discussed by Kinney and Graham [7.49]. For an ejection
2 2
angle of a with the horizontal, r=V sin a/g, so that the maximum range occurs
1/2
when a=45°, and at this range the maximum fragment velocity=(rg) .
Observations of large explosions show that the number and mass of fragments
are related exponentially.
During the 1980s much of the analytical work was directed to providing
finite element solutions for the action of penetrators on metal targets, and using
computer initiated display techniques to plot the successive deformation patterns.
The fundamental science was unchanged but the pictorial demonstration was
Ep/(sud3)=(42.7/10.3)(H/d)2+(1/10.3)(s/d)(H/d), (7.24)
where Ep is the perforation energy of the plate (Nm), su is the ultimate tensile
strength of the plate (N/m2), d is the projectile diameter (m), s is the unsupported
span of the plate and H is the target thickness (m). This is known as the SRI
equation, dating from 1968, originally formulated by Gwaltney [7.58].
A second equation is
Ep/(sud3)=1.4×109(H/d)1.5/su, (7.25)
and the most recent relationship from Wen and Jones [7.54] is
Ep/(Ksad3)=(p/4)(H/d)2+A(s/d)a(H/d)ß, (7.27)
Much of the foregoing work has been connected with missiles versus military
armour, missiles versus the protective shields of nuclear reactors, or fragments
from exploding jet engines impinging on metal protective shields. A further
area of great interest is the effect of missiles and fragments on the thin aluminium
sheeting of aircraft structures. A good deal of work in this area has been reported
in the publications of the Advisory Group for Aerospace Research and
Development (AGARD) of the North Atlantic Treaty Organization (NATO).
In an overview of the problem Avery, Porter and Lauzze [7.59] considered three
realistic threats: (a) from explosive penetrators, (b) high explosive (HE) projectiles
and (c) warheads. They indicated that projectile damage can range from dents,
cracks and holes to large petalled areas accompanied by extensive out-of-plane
deformation. The measurement that has proved most useful is lateral damage,
which is defined as the diameter of a circle that just encloses the limits of fracture,
material removal or material deformation (Figure 7.22). Since damage is limited
to sheet thickness, projectile speed and impact angle, it is possible to establish a
damage regime diagram, as shown in Figure 7.23, which refers to a 0.30 calibre
armoured piercing round impacting aluminium sheet of alloy 7075–T6. The
variation in damage size with projectile velocity for sheet 0.125 inches thick
damaged by 0.30 calibre ball ammunition is shown in Figure 7.24, taken from
the work of Avery, Porter and Lauzze. It can be seen that the maximum lateral
damage occurs just above the penetration limit, and that further increases in
Figure 7.22 Lateral damage circle (from Avery et al., ref. 7.59).
Figure 7.23 Damage type regime diagram for 0.30 calibre armoured piercing round
impacting aluminium (from Avery et al., ref. 7.59).
Figure 7.24 Variation of damage size with projectile velocity (from Avery et al., ref.
7.59).
projectile speed do not produce a significant change in damage size. The maximum
lateral damage is a circle of radius 6 metres, when the impact velocity is just
over 1000 feet/sec.
Damage prediction techniques for fragments have only been properly
developed for high-velocity compact fragments of low density, typical of impacts
from surface/air missiles or high explosive projectiles. An expression proposed
by Avery, Porter and Lauzze is:
(7.29)
where L is the lateral damage size in inches; θ is the impact angle measured
between the flight path and a normal to the target surface; D is the maximum
projected frontal dimension of the fragment and t is the target thickness in inches.
7.5 FRAGMENTATION
Fragmentation was discussed in the previous section (7.4) in relation to the break-
up of the casing of conventional bombs, but the history of fragmentation analysis
is an interesting field in its own right. As before, students of the subject are indebted
to Christopherson [7.12], who dealt with advances in fragmentation prediction
during the Second World War. In 1943 Mott produced three research papers for
the UK Armament Research Department which included the theoretical treatment
of shell fragmentation (refs [7.60] to [7.62] inclusive). Mott argued that the number
of fragments (N) between mass W and W+dW should be given by
(7.30)
where M=W1/2 and MA, B are constants for a given weapon. B was related to
the total fragmenting weight, W0, by the equation , and the
5/6 1/3
parameter MA was given by MA=Gt d (1+t/d), where t=thickness of casing
(in) and d is the internal diameter (in). Experiments showed that for a TNT
filling inside a casing of British Shell Steel (carbon content approximately 0.4
to 0.5%), G=0.3. The fragment weights were measured in ounces.
It was pointed out by Christopherson that equation (7.30) and the formula
for MA were incompatible, but both had experimental support. The limited
range of weapons that could be examined by Mott’s theory was a greater
problem, and this was overcome by Payman [7.63]. His formula related the
weight of fragments (W) each of weight greater than w, to W0 by the equation
–cw=log10W/W0, (7.31)
–2.45 –0.55
where the fragmentation parameter (c) was given by c=Kt d . The
coefficient K depended on the nature of the casing steel and the type of explosive.
In 1943 Payman [7.64] showed that all fragments from an exploding bomb
casing containing a relatively feeble explosive, had nearly the same initial
velocity, given by:
V2=6.95[(E/C)1/3–0.43]107, (7.32)
where E and C are the weights per unit length of the charge and the casing respectively
in the cylindrical portion of a bomb. This indicated that the velocity would be zero
when E/C<=0.0795. Christopherson suggested that this lower limit for E/C should
be about 0.2 for less feeble explosives, and proposed the formula:
V2=8.22×107[1–e–0.69E/C]. (7.33)
Once the fragments have passed through the shock front (because of rapid
deceleration of the front) they are travelling in free air. Then they are subject
to a retarding force proportional to the square of the velocity. Experiments in
the UK in 1943 suggested an equation for velocity, when the velocity was
greater than the speed of sound. This was:
where W was the fragment mass in ounces, and a was the coefficient of area,
–2/3
given by the equation a=AmQ . Am is the mean area presented by the fragment
as it rotates during flight, and a is its volume. When the fragment is an exact
cube, a=1.5. For velocities below the speed of sound,
(7.34)
The above formulae gave information on risk zones, within which people
needed protecting or evacuating.
Christopherson presented calculations to show that the range of fragmentation
of a medium case bomb, in which the initial velocity of the fragments would not
exceed 7500 ft/sec, was about 3000 ft, which coincided with the generally accepted
danger area for a 1000 lb medium capacity bomb. It is interesting to note that
12 or 15 inches of concrete was found to be sufficient to resist the fragmentation
of a 500 lb bomb detonated at a range of 50 ft. The corresponding thickness of
steel plating was between 1.5 to 1.75 inches.
The fragment velocity equations reviewed by Christopherson, and the simple
1/3
formula for fragment range (r in metres=45W , where W is kilograms of
TNT) were used throughout the period 1950 to 1990, particularly by engineers
who were asked to check that an explosion in an ammunition storage unit
would not cause a sympathetic explosion in an adjacent unit. In these
circumstances, the fragments would be pieces of reinforced concrete, most of
which would be propelled at a common velocity. Exploding ammunition stores
can also shower human beings with high-velocity fragments, causing injury
or death. The Societal risks from this type of disaster have been reviewed by
Williams and Ellinas [7.65] in recent years. Explosive storage regulations in
most countries give safe distances between bunkers.
An area of interest has been the combined effect on structures of blast and
fragment loading. Observations on the effect of simultaneous loadings have
shown that combined loading leads to more severe damage than would be
expected if the damage levels from separately occurring effects were added.
This is particularly interesting when the spalling of concrete is augmented by
blast. Research at the US South West Research Institute in the 1980s indicated
that the damage from the combined blast and fragmentation loading of a
steel plate was particularly severe. The combined action analysis was discussed
by Marchand and Cox in 1989 [7.66].
the Second World War, Vietnam and elsewhere have emphasized the poor
performance of conventional bombing against targets such as bridges. Many
hundreds of sorties against a bridge, using conventional bombs, could still fail
to destroy the target. The solution is to add ‘guided bomb units’ to conventional
weapons. During the Gulf War conventional bombs could not properly engage
Iraqi bunkers, often set as far as 30 m below ground with reinforced concrete
roofs many metres thick. The US Air Force therefore designed a bomb case
with a penetrator manufactured from a 203 mm diameter gun barrel, giving a
total weight of 2132 kg and carrying 295 kg of Tritonal explosive. Apparently
this weapon could penetrate 6.5 m of steel-reinforced concrete.
Data from the US suggests that a guided bomb is over 100 times as cost
effective against point targets as an unguided bomb, and without precision
bombs it would not have been possible to destroy 54 main bridges over the
Euphrates and the Tigris during the Gulf War. Further discussion of these points
is contained in a review by Niemzig and Steffen [7.67], who indicate that the
cost of using a guided bomb is about the same as using a cruise missile. Precision
bombing systems need enhanced accuracy, and this was first made possible by
the introduction of digital computing techniques based on solid state hardware.
High accuracy navigation data could be provided by systems based on inertial
navigation sensors, which are basically a combination of gyroscopes and
accelerometers controlled by a microprocessor. This configuration provides
aircraft position, velocity and other information from which navigation and
steering data can be computed. From this data, in conjunction with weapon
ballistic information, release and pull-up points can be calculated. In the mid-
1980s a system was introduced based on a miniature gimballed inertial platform
used in conjunction with a laser range finder. This produced accurate navigation,
weapon control and a means of precision designation.
As well as the kinetic energy penetrator, which is a solid component projected
at high velocity by high-pressure guns and which relies on kinetic energy to
pierce the target, the shaped charge is also used to pierce armour. This well-
known type of charge consists of a copper-lined cone embedded in an explosive
cylinder. When a fast fuse detonates the high explosive charge, the detonation
wave squeezes the liner into the form of a long, high-speed jet, which generates
very high pressures on a steel target when contact takes place. These pressures
are an order of magnitude greater than the yield strength of armour plate,
forcing the material to flow in the hydrodynamic manner of soil or mud under
the influence of a strong water jet. The effectiveness of the shaped charge is
independent of the speed of delivery of the projectile, and because of this it is
suitable for missiles and rocket launchers with relatively low missile velocity.
Although we are only concerned with the structural loading and penetrative
action in this chapter, it is interesting to look quickly at the analysis of shaped
charge action and at methods to combat this effectively. The analysis was
established by G.I.Taylor and Schardin (ref. [7.68, 7.69]) who showed that
the maximum penetration and hole size can be calculated by the conservation
7.7 REFERENCES
7.1 Poncelet, J.V. (1829 and 1835) Cours de mécanique industrielle, Paris.
7.2 Timoshenko, S. (1953) History of Strength of Materials, McGraw-Hill.
7.3 Petry (1910) Monographies de systèmes d’artillerie, Brussels.
7.4 Young, C.W. (1972) Empirical Equations for Predicting Penetration Performance
in Layered Earth Materials for Complex Penetrator Configurations, Report
SC-DR-72–0523, Sandia Laboratories, Albuquerque, December.
7.5 Goodier, J.N. (1965) On the mechanics of indentation and cratering in solic
targets of strain-hardening metal by impact of hard and soft spheres, Proc. 7th
Symposium on Hypervelocity Impact, Vol. 3, p. 215.
7.6 Norwood, F.R. (1974) Cylindrical Cavity Expansion in a Locking Soil, Report
SLA-74–0261, Sandia Laboratories, Albuquerque, July.
7.7 Yew, C.H. and Stirbis, P.P. (1978) Penetration of projectile into terrestrial target,
Journal of the Engineering Mechanics Division, ASCE, EM2, paper 13703,
April, p. 273.
7.8 Allen, W.A., Mayfield, E.B. and Morrison, H.L. (1957) Dynamics of a projectile
penetrating sand, Journal of Applied Physics, 28(3), March.
7.9 Robins, B. (1742) New Principles of Gunnery, London.
7.10 Euler, L. (18th century) Neue Grundsätze der Artillerie, Berlin.
7.11 Stipe, J.G. (1946) Terminal Ballistics of Soil, Effects of Impacts and Explosions,
Summary Technical Report of Division 2, NDRC, Vol. 1, Washington.
7.12 Christopherson, D.G. (1946) Structural Defence (1945), UK Ministry of Home
Security, Civil Defence Research Committee paper RC 450.
7.13 US Army (1965) Fundamentals of Protective Design (Non-nuclear), US Army
TM5–855–1, Department of the US Army, July.
7.14 Young, C.W. and Keck, L.J. (1971) An Air-dropped Sea-ice Penetrometer, Report
SC-DR-71–0729, Sandia Laboratories, Albuquerque, New Mexico, USA,
December.
7.15 Forrestal, M.J., Longcope, D.B. and Lee, L.M. (1983) Analytical and experimental
studies on penetration into geological targets, Symp. on Interaction of Non-
nuclear Munitions with Structures, US Air Force Academy, Colorado, May.
7.16 Bishop, R.F., Hill, R. and Mott, N.F. (1945) The theory of indentation and
hardness tests, Proc. of the Physical Society, 57(3), May.
7.17 Young, C.W. (1969) Depth prediction for earth penetrating projectile, Journal
of Soil Mechanics and Foundation Division, ASCE, May.
7.18 Stilp, A., Schneider, E. and Hülsewig, M. (1985) Penetration of fragments into
sand, Second Symp. on Interaction of Non-nuclear Munitions with Structures,
Panama City Beach, Florida, April, p. 333.
7.19 Schoof, L.A., Maestas, F.A. and Young, C.W. (1989) Numerical method to predict
projectile penetration, Fourth International Symposium on Interaction of Non-
nuclear Munitions with Structures, Panama City Beach, Florida, April.
7.20 Taylor, T. and Fragaszy, R.J. (1989) Implications of Centrifuge Penetration Tests
for use of Young’s equation, Fourth International Symposium on Interaction of
Non-nuclear Munitions with Structures, Panama City Beach, Florida, April.
7.21 Gebara, J.M., Pau, A.D. and Anderson, J.B. (1993) Analysis of rock-rubble
overlay protection of structures by the finite block method, Proc. of Sixth
International Symp. on Interaction of Non-nuclear Munitions with Structures,
Panama City Beach, Florida, May, p. 12.
7.22 Nelson, R.B., Ito, Y.M., Burks, D.E. et al. (1983) Numerical Analysis of Projectile
Penetration into Boulder Screens, US Army Engineer Waterways Experiment
Station Report WES.MP.SL.83.11.
7.23 Gelman, M.D., Richard, B.N. and Ito, Y.M. (1987) Impact of AP Projectile
into Array of Large Caliber Boulders, US Army Engineer Waterways Experiment
Station.
7.24 Chen, W.F. and Pau, A.D. (1991) Finite element and Finite block methods in
Geomechanics, Proc. of Third Int. Conf. on Constitutive Laws of Engineering
Materials, Tucson, Arizona.
7.25 Schwer, L.E., Rosinsky, R. and Day, J.A. (1987) Lagrangian technique for
predicting earth penetration including penetrator response, Third International
Symposium on Interaction of Non-nuclear Munitions with Structures, Mannheim,
Germany, March.
7.26 Creighton, D.C. (1982) Non-normal Projectile Penetration in Soil and Rock:
User’s Guide for Computer Code PENCO2D, US Army Waterways Experiment
Station Technical Report SL.82.7, Vicksburg, September.
7.27 Matuska, D., Durrett, R.E. and Osborn, J.J. (1982) HULL User’s Guide for
Three Dimensional Linking with EPIC3, Orlando Technology Inc. Report
ARBRL-CR-00484, Shalimar, Florida, July.
7.28 Ito, Y.M., Nelson, R.B. and Ross-Perry, F.W. (1979) Three Dimensional Numerical
Analysis of Earth Penetration Dynamics, US Defense Nuclear Agency Report
DNA 5404 F, January.
7.29 Yankelevsky, D.Z. (1983) Projectile penetration through a narrow drill in soil,
Int. Journal of Impact Engineering, 1(4), p. 377.
7.30 Young, C.W. and Young, E.R. (1985) Simplified Analytical Model of Penetration
withLateral Loading , Sandia National Laboratories Report SAND-84–1635, May.
7.31 Amini, A. and Anderson, J.B. (1993) Modelling of projectile penetration into
geologic targets based on energy tracking and momentum impulse principle,
Proc. of Sixth International Symp. on Interaction of Non-nuclear Munitions
with Structures, Panama City Beach, Florida, May, p. 6.
7.32 Pariseau, W.G. and Fairhurst, C. (1967) The force penetration characteristic
for wedge penetration into rock, Int. Journal of Rock Mech. and Min. Science,
4, p. 165.
7.33 Anderson, W.F., Watson, A.J., Johnson, M.R. and McNeil, G.M. (1983)
Optimization of rock/polymer composites to resist projectile penetration,
Materiaux et Constructions, 16(95), 343.
7.34 Whiffen, P. (1943) UK Road Research Laboratory Note No. MOS/311.
7.35 Beth, R.A. (1946) Effects of impact and explosion, Terminal Ballistics of Concrete,
Chap. 7, Vol. 1, Summary Tech Rep DW2, NRDC, Washington.
7.36 Forrestal, M.J., Altman, B.S., Cargile, J.D. and Hanchak, S.J. (1993) An empirical
equation for penetration depth of ogive-nosed projectiles into concrete targets,
Proc. of 6th Int. Symp. on Interaction of Non-nuclear Munitions with Structures,
Panama City Beach, Florida, USA, May.
7.37 Brown, S.J. (1986) Energy release protection for pressurised systems, Part 2. Review
of studies into impact/terminal ballistics, Applied Mechanics Review, 39.
7.38 Gold, V.M., Pearson, J.C. and Turci, J. (1993) Resistance of concrete and
reinforced concrete structures impacted by CU and TA projectiles, Proc. of 6th
Symp. on the Interaction of Non-nuclear Munitions with Structures, Panama
City Beach, Florida USA, May.
7.39 Tate, A. (no date) A theory for the deceleration of long rods after impact, J.
Mech. Phys. Solids, 15, 387.
7.40 Peters, J., Anderson, W.F. and Watson, A.J. (1994) Determination of penetration
parameters from the stress pulse generated during projectile penetration into
cementitious materials, Structures Under Shock and Impact III, Comp. Mech.
Pub., June, p. 235.
7.41 Dancygier, A.N. and Yankelevsky, D.Z. (1994) Comparative penetration tests
of fibre reinforced concrete plates, Structures under Shock and Impact III, Comp.
Mech. Pub., June, p. 273.
7.42 Taylor, G.I. (1948) The formation and enlargement of a circular hole in a thin
plastic sheet, Quarterly Journal of Mechanics and Applied Maths, 1, p. 103.
7.43 Taylor, G.I. (no date) The use of flat ended projectiles for determining dynamic
yield stress, Proc. Roy. Soc. London, Vol. 194, Series A, p. 289.
7.44 Thomson, W.T. (1955) An approximate theory of armour penetration, Journal
of Applied Physics, 26(1), January.
7.45 Recht, R.F. and Ipson, T.W. (1963) Ballistic perforation dynamics, Journal of
Applied Mechanics, September.
7.46 Goldsmith, W. et al. (1965) Plate impact and perforation by projectiles,
Experimental Mechanics, December.
7.47 Backman, M.E. and Goldsmith, W. (1978) The mechanism of penetration of
projectiles into targets, Inter. Jour. Eng. Sci., 16, Pergamon Press, London.
7.48 Taylor, G.I. (1944) The Fragmentation of Tubular Bombs, Paper written for the
Advisory Council on Scientific Research and Technical Development, UK Ministry
of Supply.
7.49 Kinney, G. and Graham, K.J. (1985) Explosive Shocks in Air, Springer Verlag,
New York.
7.50 Ravid, M. and Bodner, S.R. (1983) Dynamic perforation of viseoelastic plates
by rigid projectiles, Int. Eng. Sci., 21, 577.
7.51 Goldsmith, W. and Finnegan, S.A. (1986) Normal and oblique impact of
cylinders—conical and cylindrical projectiles on metallic plates, Int. J. Impact
Engng, 4, p. 83.
7.52 Dehn, J. (1989) Mathematical models in penetration mechanics, Structures Under
Shock and Impact (1), Elsevier/C M Publications, p. 203.
7.53 Nurick, G.N. and Crowther, B.G. (1994) The measurement of the residual velocity
and deflection of a projectile passing through a thin plate, Structures Under
Shock and Impact (3), CM Publications, p. 261.
7.54 Wen, H.M. and Jones, N. (1992) Semi-empirical equations for the perforation
ofplates struck by a mass , Structures Under Shock and Impact (2), Thomas
Telford and CM Publications, p. 369.
7.55 Neilson, A.S. (1985) Empirical equations for the perforation of mild steel plates,
Int. J. Impact Eng., 3, p. 137.
7.56 Jowett, J. (1986) The Effect of Missile Impact on Thin Metal Structures, UKAEA
SRD R378.
7.57 Wen, H.M. and Jones, N. (1992) Experimental Investigation into the Dynamic
Plastic Response and Perforation of a Clamped Circular Plate Struck Transversely
by a Mass, Impact Res Centre ES/85/92, Dept of Mechanical Engineering,
University of Liverpool.
7.58 Gwaltney, R.C. (1968) Missile Generation and Protection in Light Water-cooled
Power Reactor Plants, Report URNL-USTC-22, Oak Ridge National Lab.,
Tennessee, USA.
7.59 Avery, J.G., Porter, T.R. and Lauzze, R.W. (1975) Structural integrity requirements
for projectile impact damage—an overview, Specialist meeting on Impact Damage
Tolerance of Structures, AGARD Conf. Proceedings, No. 186.
7.60 Mott, N.F. (1943) Fragmentation of H E Shells: a Theoretical Formula for the
Distribution of Weights of Fragments, UK Armament Research Department,
AC 3642 SD/FP85, March.
7.61 Mott, N.F. (1943) A Theory of Fragmentation of Shells and Bombs, UK Armament
Research Department, AC 4035, SD/FP 106, May.
7.62 Mott, N.F. (1943) Fragmentation of Shell Casings and the Theory of Rupture in
Metals, UK Armament Research Department, AC 4613 SD/FP 139, August.
7.63 Payman, W. (1943) Empirical Relationships for Use in Comparing Results of
Fragmentation Trials, I—Velocities of Fragments, UK Safety in Mines Research
Station, RC 387, June.
7.64 Payman, M. (1943) Empirical Relationships for Use in Comparing Results of
Fragmentation Trials, II—Weight Distribution of Fragments, UK Safety in Mines
Research Station, RC 388, July.
7.65 Williams, K.A.J. and Ellinas, C.P. (1989) Societal risks from detonations in
explosive storage structures, Structures Under Shock and Impact, Elsevier and
CM Publications, p. 253.
7.66 Marchand, K.A. and Cox, P.A. (1989) The apparent synergism in combined
blast and fragment loadings, Structures Under Shock and Impact, Elsevier and
CM Publications, p. 239.
7.67 Niemzig, I. and Steffen, R. (1992) Engagement of hard targets from the air,
Military Technology, November, p. 48.
7.68 Taylor, G.I. (1941) The Analysis of a Long Cylindrical Bomb Detonated at One
End, Paper written for the Civil Defence Research Committee, UK Ministry of
Home Security.
7.69 Schardin, H. (1956) Uber das Wesen der Hohlladung, VDI Journal, 98, 1837.
7.70 Gurney, R.W. (1943) The Initial Velocity of Fragments from Bombs, Shells and
Grenades, US Ballistic Research Laboratory, BRL—Report 405.
7.71 Chou, P.C. et al. (1981) Improved formula for velocity, acceleration and projection
angle of explosively driven liners, Proc. 6th Int. Symp. Ballistics, Orlando.
7.72 Vered, G.R. (1987) Evolution of ‘Blazer’ reactive armour and its adaptation to
AFVs, Military Technology, December, p. 53.
immediately above and below the explosion. The area of demolition fell off
rapidly at lower floors, soon down to 100 square feet. Debris was usually
held by the second and third floor slabs below the explosion, but occasionally
it brought down a succession of floors.
The survey emphasized that the virtual absence of progressive collapse in
well-designed frame buildings meant that floors and walls near the detonation
point were effective as screens. Of equal importance was the ductility and
continuity of the framework, which made the building highly resistant to
explosions. A possible weakness, however, was the lack of ductility and
continuity in bolted or riveted connections. As in many other examples of
structural loading, the crucial factor lay in the strength or weakness of
connections. During the years before the Second World War welded connections
were relatively rare, so there was not a great deal of evidence about the behaviour
of welded frameworks. The authors, however, drew attention to the possibility
of introducing ductility into building structures by clamped connections that
rely on friction to produce the effects of continuity. The resulting structure
could absorb a large amount of energy without collapse.
The following types of damage were noted in single-storey steel-framed
buildings such as hangars, warehouses and workshops: direct damage, primary
collapse and spreading collapse. Direct damage was the cutting of members
by a bomb (before exploding), by fragments or by crater debris; primary collapse
was the collapse of members that depended for their stability on a destroyed
member. Spreading collapse occurred when the forces set up by a primary
collapse were transmitted to adjoining undamaged members, producing
instability in these members. The spreading collapse of a roof could occur
when the cutting of a single member suddenly involved the whole area of a
large structure. The chief risk to roof steelwork arose from the failure of
stanchion to roof-girder connections, which were sheared by violent
displacement or lateral blast, or destroyed by blast uplift.
Dwelling houses, of course, are rarely constructed with a ductile steel
framework, so in assessing the behaviour of houses in explosions, it was
necessary to examine the strength of masonry and brick structures to blast.
Civilian masonry structures are not normally designed to resist explosive
blast. The stone castles and military fortifications of the Middle Ages were
more likely to be threatened by the penetrative action of missiles, and their
wall thicknesses were set empirically by the need to resist damage by local
impact and fragmentation. Civil masonry and brick structures were known
to have relatively little resistance to local explosions, and in earlier times no
attempt was made to predict how they might behave under attack, or what
their residual strength might be. Masonry and stone were not employed much
in the construction of ‘bomb-proof’ shelters once reinforced concrete appeared
on the structural scene. As the Second World War approached, however,
experimental research was initiated to check the behaviour of conventional
building structures when subjected to the general blast from exploding aerial
bombs, and by 1940 there were several UK reports on the behaviour of brick
wall panels.
Most domestic brick walls were 9 in thick, and only a small number of
factory or civic buildings were constructed with 13.5 in thick walls. Christopherson
[8.3] reported that the static pressure to cause collapse of an 8.5 ft square panel
2
of 9 in brickwork was about 2.25 lb/in , and that the ultimate deflection for
walls enclosed in steel channels was about 5 in. For 13.5 in walls the figures
2
were 5.3 lb/in and 8 in respectively. In terms of hydrostatic impulse, it was
shown that for a very wide range of incident pressures the impulse needed to
2
destroy 9 in brickwork was 92±10 lb msec/in . A typical two-storey brick house
of the 1940s had free wall spans between 8 and 10 feet square, and field
observations following air raids on British cities showed that damage involving
the collapse of at least one external wall occurred when the hydrostatic impulse
2
lay between 64 and 84 lb msec/in . Complete demolition of more than 75% of
2
all external walls occurred at impulses between 82 and 160 lb msec/in .
During the Second World War it was necessary to use protective brick
walls to reduce blast damage on industrial constructions, and in the early
years of the war the design criterion was that collapse of a wall could be
avoided if
ρ0Mwu>0.25A2, (8.1)
Figure 8.1 Impulses for the destruction of brick panels (from Christopherson, ref.
8.3).
exterior of the brick shelter was covered by steel mesh, held in position by an
additional 4.5 in brick skin. A typical reinforced brick shelter used brick side
walls combined with a concrete floor and roof. Reinforcing bars were introduced
into the cavity between bricks before the cavity was filled with concrete. When
attacked by blast weapons surface brick shelters were of course much more
vulnerable than buried shelters.
Brickwork was also protected during the Second World War by internal
strutting, which was placed so that there was enough lateral stability to eliminate
the possibility of collapse from unsymmetrical loading. Strutting was also
used to support the load of debris on the roof of the building if destruction of
the walls occurred. Tests on brick shelters during the Second World War were
described in a series of reports by the Research and Experiment department,
Ministry of Home Security, and examples are given in refs [8.7] to [8.9]. The
lateral velocity imparted to a surface shelter by earthshock from a bomb of
250kg at 15 ft horizontally from the shelter wall was found to be about 10 ft/
sec, as measured and reported by Walley [8.10].
The amount of reinforcement required to keep a brick surface shelter intact was
calculated very simply by Christopherson. He supposed that when an unreinforced
shelter breaks up the maximum relative velocity of roof and walls is 5 ft/sec for a
shelter 30 ft×8 ft in plan area with walls 1 ft thick. If the roof is assumed to be 5 in
thick, with a density of 144 lb/ft3, its kinetic energy is (60×25)/64 ft lb/ft2. The
kinetic energy of the roof per foot run of wall is (60×25×240)/64×72 ft lb. If there
2
are A in of reinforcing bars per ft run of wall, having a yield stress of 50 000 lb/
2
in , and if their extension is d ft, then A.d. 50 000=the kinetic energy per foot
run. If, as suggested earlier, a minimum area of reinforcement of 0.06% is used,
then in a 1 ft thick wall, the value of A=0.06×1.44, and d=(60×25×240)/(64×72×50
000× 0.06×1.44)=0.22 in. Thus the separation of wall and roof cannot be greater
than about in , and could well be less. This indicated that the shelter components
would remain virtually intact.
This wartime research and testing was still used as a basis for design rules
over the next thirty or forty years. Thus, when the US Army Technical Manual
on Protective Design (Non-nuclear) appeared in 1965 [8.11], it contained
wartime information on the thickness of reinforced and plain brickwork to
provide protection against fragments and blast from general purpose bombs
detonated at a distance of 40 feet. The required thicknesses are shown in
Table 8.1.
It was also recommended that brick structures designed with these thicknesses
should be spaced 120 ft apart. Blast resistant protective brick walls were also
required to be buttressed at intervals not greater than 10 ft, with the buttresses
reinforced horizontally.
In Britain, the response of brick walls was often based on information
given by a wartime bulletin issued by the Ministry of Home Security [8.12],
which gave the minimum distance in feet at which only slight damage would
occur from surface blast. The required distances are shown in Table 8.2.
Table 8.1
Table 8.2
The stability of masonry walls under the action of blast from conventional
explosives was examined experimentally in a series of tests in the late 1970s
and early 1980s in the USA. Collapse mechanisms were recorded, as well as
the nature and distribution of debris from shattering walls, and a range of
construction was investigated, varying from load-bearing masonry to reinforced
concrete frames with a masonry infill. These tests were summarised by Rempel
and Beck [8.13], who also analysed crushing energy and wall rotation.
Attention was also paid to the design of explosive storage facilities constructed
of masonry, including overpressures to cause incipient failure and safe separation
distances between storage facilities and neighbouring buildings of brick, masonry
and prefabricated blocks. The state of affairs in the early 1980s was reviewed by
Napadensky and Longinow [8.14]. They quoted current US regulations, which at
the time permitted conventional, inhabited buildings to be located at distances of
1/3 1/3
40 W to 50 W ft from an explosives store, where W is the weight of explosive in
pounds. For long-duration pressure/duration blast loads that would be experienced
in a nuclear explosion, incipient failure overpressures (where the structure will fail
under any additional load) was given in terms of probability of failure as follows.
(a) Light commercial buildings with masonry load bearing walls: 2.0; 3.2; 4.9
psi for failure probabilities of 10%, 50% and 90% respectively.
(b) One-storey masonry load-bearing wall: 1.8; 2.8; 4.6 psi for failure
probabilities of 10%, 50% and 90% respectively.
Safe separation distances, peak overpressures and impulses, for a range of
non-nuclear explosions, were tabulated by the authors for a range of targets.
Excerpts from their survey are given in Table 8.3.
The authors compared the damage, pressures and distances with the recorded
effects of the Flixborough vapour cloud explosion in 1974. Damage analysis
Table 8.3
suggested that the blast effects were similar to those produced by 35 300 lb of
TNT detonated at a height of 147.6 ft (45 m). According to US safety standards,
brick buildings farther away than 1312 ft (400 m) would have been safe, but
the damage extended beyond this distance. It was concluded that for this type
of vapour cloud explosion damage is directly related to peak overpressure.
Gas explosions in dwelling houses can cause the outward deformation of
brick, masonry or precast wall panels. Structural loads due to a range of gas
explosions have been measured by the UK Building Research Establishment,
where pressures and wall displacement time histories were recorded during
the demolition of maisonette blocks. The results have been summarized by
Ellis and Crowhurst in two reports ([8.15] and [8.16]). Aerosol canisters
containing butane were ruptured and the contents ignited to produce internal
peak pressures similar to those in domestic gas explosions. Peak pressures in
2 2
the range 2.6 KN/m to 9.0 KN/m were generated, depending on the size of
the canister. Although structural damage to partitions and internal doors was
significant, there was no permanent damage to main structural walls. The
2
largest measured displacement of a wall panel was 0.63 mm for a 9.0 KN/m
3
explosion (750 ml canister). Typical room volumes ranged from 17.5 m (smaller
3
bedrooms and kitchens) to 28.1 m (large second-floor bedroom).
A collation of reports of gas explosion damage to domestic buildings has
been made by Moore [8.17] in which explosions are classed in terms of the
severity of damage, as moderate, severe and very severe. A moderate explosion
can destroy a relatively weak structure, such as a small brick or masonry bungalow.
The effect of blast on dwelling houses of brick was also investigated during
the Second World War by the US authorities. Experiments at full scale were
undertaken at the Aberdeen Proving Grounds of the Corps of Engineers, and
at Princeton University, and the results were summarized in the NRDC technical
report on the effects of impact and explosion published in 1946 [8.18]. One
series of tests concerned a structure, 6.5 ft square, 4 ft high, with a reinforced
concrete floor, roof and columns, enclosed by brick walls, which were bonded
to the framework along their bottom and sides but which were free at the top.
Charges consisting of 22 and 44 grams of tetryl explosive were detonated at
the centre of the enclosed space, and 0.5 and 1.0 lb charges of TNT were
exploded outside the structure, 3 ft from the centre of the brick wall.
The 44 gram charge completely blew out one wall (one course thick), whereas
the 22 gram charge caused enough cracking to reduce structural strength to
zero, without actually blowing the wall outwards. The external detonation of
0.5 and 1.0 lb charges of TNT 3 feet from the centres of already cracked brick
walls showed no appreciable additional effect. When a 0.5 lb charge was
detonated within the structure complete destruction resulted.
At larger scales (20 ft square, 12 ft high), with 12 in thick brick walls, it
was calculated that internal explosions of 1.25 lb TNT would cause serious
wall cracking, that 2.5 lb would blow out the walls without destroying the
frame, and 15 lb would cause complete destruction. 30 lb of TNT detonated
externally a distance of 10 ft from the wall would not seriously damage it. It
should be noted that these early experiments were made with no venting to
reduce the degree of confinement.
Glass and light sheeting in buildings are very damage-prone under the action
of blast loads or penetrators. Unless treated on the surface by the addition of
film or mesh, glass will rapidly shatter and be transformed into high-speed
splintered fragments. Light sheeting will tear and distort if made from ductile
materials, or shatter like glass if made from brittle materials. The loading will
be not unlike that calculated for similar areas of brickwork, but the effects of
the blast will be felt over larger areas.
Damage from terrorist bombs in the commercial areas of cities can be
widespread and expensive to repair. Readers will have seen photographs of
the distortion and tearing of cladding on high rise offices and the shattering of
glass in the windows of shops and public buildings. In analysing the results of
this form of blast loading, it is usually assumed that the instantaneous pressures,
and the pressure decay, are similar to those experienced by flat areas of masonry,
brickwork or reinforced concrete.
Figure 8.2 d’Eauplet railway bridge, 1944 (from Hill et al., ref. 8.19).
The bridge was then used for road traffic by removing rails and sleepers
from one track over spans 1, 2 and 3, and building a raised wooden
roadway across the dip of damaged span 4. It is interesting that aerial
reconnaissance by the Allied forces between May and August 1944 did
not detect the destruction of the pier, and it was thought that all the
bridge damage resulted from the cutting of the top booms. In fact, the
top booms did not buckle until the pier collapsed.
Figure 8.3 Tourville railway bridge, 1944 (from Hill et al., ref. 8.19).
After the destruction of the bridge, Italian and Russian labourers under
German control constructed a diversion bridge immediately on the down-
stream side. This was attacked repeatedly from the air, and during one
of these raids a hit was scored on Pier 2 of the original bridge. Half of
the pier collapsed and span 3 tilted downstream, interfering with the
diversionary bridge. The German engineers therefore demolished the
rest of the pier and dropped span 3 into the river.
(c) Le Manoir bridge (railway) 9.5 miles south of Rouen, Figure 8.4.
The bombing history was as follows:
21 May 1944 A single bomb fell on Pier 1, cutting the end post
and overhead cross girder
25 May 1944 68×500 GP bombs Damage to main booms in spans 1
36×500 GP bombs and 3
26 May 1944 32×500 GP bombs No further damage
27 May 1944 73×2000 GP bombs Span 1 hit near pier. End of pier
destroyed allowing downstream
girder to collapse. Left bank end
of span 2 destroyed, and fell into
river
30 May 1944 77×500 GP bombs No further damage
16×1000 GP bombs
14 June 1944 Final air attack Span 3 cut and end of span near
abutment fell into the river. The
other span rotated about Pier 2,
causing further damage.
The form of this destruction showed that it was possible to destroy a bridge of
this type without completely demolishing a pier, as long as adjacent spans
were completely severed.
Figure 8.4 Le Manoir railway bridge, 1944 (from Hill et al., ref. 8.19).
Figure 8.5 Vacoulers railway bridge, 1944 (from Hill et al., ref. 8.19).
The ten attacks 20 June to 8 August during which 1037 bombs were dropped
failed to damage the single line temporary bridge.
The explosive loading of the bridges described above was the result of direct
hits of aerial bombs on structural members or piers, but it was realized towards
the end of the Second World War that the damage resulting from near misses of
very large bombs could often be more significant than damage due to direct hits
from smaller weapons. Targets such as railway lines, viaducts and bridges were
considered relatively invulnerable because of their extreme tenuity, but a potent
way of destroying or disabling large lengths of bridges was to produce very
large craters that destroyed the bridges foundations. As discussed earlier in 7.3,
the UK Tallboy bomb, weighing 12 000 lb, was designed to attack German U-
Boat pens on the Atlantic Coast of France, and this was followed by the Tallboy
(large) or Ground Slam bomb (22 000 lb). However, the first Tallboy (large)
bomb) was dropped by the UK in operations against the Bielefeld Viaduct, and
it fell 50 ft away from the piers. The loosening of the earth caused the destruction
of six piers and five arches of the viaduct, and although the approximate shape
of the crater was 150 ft diameter, 45 ft deep, the loosening of the soil occurred
over a diameter of 240 ft, and to a total depth of 75 ft. These lessons of the
Second World War were perhaps learned again by the forces of the west in the
Gulf War against Iraq, when the US was reported to be turning away from the
precision bombing of bridges with relatively small bombs to the use of very
heavy bombs that could cause destruction from near misses.
Near miss explosions, as we have seen earlier, can cause high blast pressures
on components such as the relatively thin plating of the webs of deep plate
girders. Damage is then mostly by permanent deformation and distortion, and
removal of the integrity of web members. In certain circumstances this can lead
to shear failure and possible lateral buckling. A final commentary on the damage
and repair of bridges during the Second World War was provided by
Christopherson [8.3], who briefly examined ways of pre-strengthening existing
civil bridges against the possibility of bomb damage. He emphasized the
importance of permeability, ductility and redundancy, and the need to eliminate
the possibility of progressive collapse in viaducts. In general, the results of surveys
in the UK and US suggested that when it came to bombing from aircraft, it was
better to shake down the piers than shoot up the superstructure. The susceptibility
of bridge piers to undermining processes is, of course, well known from studies
of civil bridge structural failure, where it has been found that about half of all
bridge failures is due to scouring under piers and abutments.
The explosive damage to bridges from hand-placed contact charges is also
an important aspect of our review. In 1971, for example, during a local war in
Bangladesh, explosive charge damage was recorded on a number of relatively
small reinforced concrete road bridges, and on one steel truss bridge. The
damage was often in the form of holes in the deck, sometimes involving the
partial destruction of multi-beams supporting the deck slabs, and occasionally
this was severe enough to cause the complete collapse of a bridge span. Damage
was also reported to reinforced concrete piers, which collapsed sideways, and
to masonry abutments. The latter suffered extensive cracking and spalling
and often a partial collapse of the bridge seating.
A famous example of bridge deck damage due to hand-placed explosives
was the attack by German frogmen against the Nijmegen road bridge in
September 1944. This has been reported in detail by Hamilton [8.21], who
recounts that early in the morning of 28 September large explosions were
heard. Over 70 feet of the roadway disappeared, following the actions of the
frogmen, who used naval mines weighing 1200 lb, fitted with float chambers.
These were placed in a necklace around the piers before the float chambers
were released and delay fuses activated; however, the mines released their
impact forces upwards, failing to damage the piers, but disrupting the roadway.
It was necessary to build twin Bailey bridges, each 80 ft in length, across the
gap in the deck in order to continue to use the bridge for military traffic.
More recently there have been a number of demolitions, or part demolitions
of civil bridges during the conflict in Bosnia. Reinforced concrete box section
bridges have been destroyed or partially destroyed, mainly by charges placed
at deck level; but there have been instances of damage to piers by explosives
set at their bases. The problems encountered because of explosive damage to
Bosnian bridges have been discussed by Pelton [8.22].
In order to classify the extent of explosive damage it has been suggested
that the main load carrying components of all bridge structures can be classified
as follows:
(The minimal risk level might well be of the same order as the annual probability
of failure of an elderly undamaged civil bridge under maximum road class
traffic.)
The design loadings for civil bridges were compared in 1979 by the
Organization of Economic Co-operation and Development (Paris) [8.23] who
conducted a study in which the total bending moments caused by live loading
in various codes of practice were calculated for a simply supported bridge.
Calculations were made for two, three and four traffic lanes over spans from
10 m to 100 m, taking account of impact and reductions for multiple lane
loading. Differences in allowable stress levels were taken into account. The
maximum bending moments obtained were converted into an equivalent
uniformly distributed load qL. Consider, as an example, the results for American
AASHTO load specifications applied to road bridges on motorways, trunk
roads and principal highways (A roads), given in Table 8.4.
From these figures it is possible to calculate the maximum live load bending
moment that can be applied to an undamaged bridge under working conditions
( for a simply supported span of length L), or under ultimate conditions
. The corresponding maximum shear forces would be qLL/2 and
1.5qLL/2.
When assessing residual strength, a knowledge of the live loading, qL, is
not enough. It is necessary to know the equivalent dead load per metre
(=qD) due to the self weight of the bridge and its accessories. Various equations
have been proposed for the rough calculation of total dead load, and these
depend on the number of traffic lanes, an estimate of the load classification
and properties of the material from which the bridge is constructed. An
Table 8.4 Values of equivalent uniformly distributed load, qL, for AASHTO loading
Table 8.5
the deck structure and stiffening girders. This figure has also been shown to
apply to prestressed concrete bridges in the 100 m to 200 m span range, and
it has been suggested that the ratio of dead load to total (dead plus live) bending
moments varies with span according to Table 8.5.
If we take the above results to apply in general to civil bridges, they show that
a relatively small amount of explosive damage to a key part of the structure of a
long span bridge will reduce the strength of the bridge such that it is only just able
to cope with its own dead load. In this event, the chances of being able to re-
classify the damaged structure to take even a very light vehicle are not high.
The residual strength of any bridge component, whether class one, two or
three, is often linked to the amount of cross-sectional area that is missing,
where the reduced area is located, and how far the reduced area extends along
the member. It can also be linked to the degree of deformation suffered from
blast due to near misses, which may be in the form of permanent lateral
deflection, and shortening or twist. The residual strength of groups of
components can also be influenced by the complete destruction of a single
component or of a joint between adjacent components.
When the explosive damage results in the loss of cross section, the weakening
may result in a reduction in tensile, compressive, bending or shear strength, or
to combinations of these. The loss of section can, of course, take any shape, but
let us first assume that the width, b, of a rectangular section is maintained, but
that the damage results in a reduction of the depth, d, as shown in Figure 8.6.
The type of component can be linked with the order of damage by the
relationship:
(8.4)
where the index m can be 1, 2 or 3. The first term in Eq. (8.4) is sometimes
m
referred to as the ‘strength ratio’, and the second as the (area ratio) .
For a simple tension member it can be assumed that the local loss of cross
section extends over the whole length of the member, and then m=1, because
tensile strength is directly proportional to section area (neglecting stress
concentration, crack growth or fatigue). For a member such as a girder web,
or the vertical plating of a box girder, when damage results in local section
loss near an area of maximum shear, it is approximately correct to take m=1.
For members contributing to bending strength, it can again be assumed
that the damage extends over the whole of the beam length. In this event m=2,
because the section modulus will generally be proportional to the square of
the depth. For a simple slender compression member such as the top boom of a
Figure 8.6 Relationship between residual strength and initial strength for a given
loss of depth of a rectangular section.
truss, or the pylon of a cable-stayed bridge, assume that the section loss extends
over the entire length. This leads to m=3, because the second moment of area,
required in the calculation of buckling or compressive strength, will be
approximately proportional to the cube of the depth of the component.
For a very stocky compression member such as a masonry pier or abutment,
damage will only lead to m=3 if so much of the cross section is removed that
the remainder buckles as a slender strut. Otherwise, the order is more likely to
be m=1. The above discussion is illustrated in Figure 8.6, and from it we can
form a relationship between strength ratio and area ratio, as shown in Figure
8.7. This figure represents the simplest loading and the most elementary member
Figure 8.7 Relationship between strength ratio and area ratio. Permissible stress
design in the elastic range.
shape, and any variations such as combined loading or unusual cross sections
will result in values of m between 1 and 3 in the strength ratio equation. For
example, a bridge component subjected to combined bending and compression
might be expected to have an m value between 2 and 3. Figure 8.7 indicates
that for damage that reduces the cross section by 25% or less, the strength
ratio is very similar for simple members in bending or compression, but this
ratio could in turn be 25% less than the strength ratio in tension or shear.
This analysis looks to have a logic about it, but is based on rather idealistic
circumstances. Also, it must be remembered that if the foundations of a pier
are undermined, bridge failure could be by overturning—a catastrophic
happening not linked to bridge ductility or loss of section, but to the explosive
destruction of external support from the ground.
The other extreme to the damage configuration assumed in Figure 8.6 would
be the vertical slicing of the section resulting in the reduction of the width
from b to b1. In all the loading cases considered the value of m would now be
1, and the strength ratio would be directly proportional to the area ratio. In
practice cross-sectional damage would probably be somewhere between the
two extremes discussed above.
The assessment of strength reduction from a knowledge of the area reduction
of truss members was pursued during and after the Second World War by the
designers of the military ‘Bailey Bridge’ in the UK. Their results were presented
in the form of a table, given in reference [8.24], which gave a relationship
between the amount of damage to the webs and flanges of I beams and
percentage residual strength of verticals and diagonals in the bridge truss
panels. For example, the complete removal of the web in shorter members led
to a residual strength of 66%, and removal of half the web gave 73%. Complete
removal of one flange gave a residual strength of 24%, and removal of one
flange and half the other reduced the figure to 18%. Similar residual strengths
were given for main chords consisting of two back-to-back channels. The
removal of the web of the channel gave a residual percentage of 63% for the
shorter members, and removal of the webs of both channels gave 61%.
Complete removal of one channel flange and web gave a residual percentage
of only 3%.
The article in reference [8.24] on the damage assessment of military bridges
noted that in the majority of cases where bridges were damaged by enemy air
attack or shell fire, several bridge members would have been reduced in strength.
Each member was then considered separately and the final classification assessed
on the worst case. The article also stated that if a bridge member is struck by
flying metal and is deformed but not holed, it must be ‘carefully watched as
loads cross the bridge’. If further deformation occurs, the member is treated
as severed, and the bridge strength assessed accordingly.
Rapid methods of evaluating residual strength can be illustrated by taking
examples, and three areas where damage might occur to Class One bridge
components are:
Figure 8.8 Damage to the deck slab in a short-span reinforced concrete bridge.
Figure 8.9. If it is intended, for example, that 100 full load class vehicles
should be able to cross the damaged structure before repairs are commenced,
residual strength calculations should be based on a width at the hole of
W–(dW+2) metres, and a maximum width of load of ßW–dW/2–1.0 metres.
If the damaged area lies close to the ends of the span, ie between 0 and
L/8 or 7/8L and L, the remaining shear capacity should be checked. The
working shear strength of the undamaged bridge is qLL/2, and this would be
supplied by a section of full width W. If the reduction in shear area at the hole
for 100 crossings is proportioned to [W–(dW+2)]/W, the shear strength at the
ends of a damaged bridge would be this fraction multiplied by qLL/2.
Figure 8.10 Damage to the central girder and associated deck in a multi-beam
bridge.
Figure 8.11 Example of damage to the deck slab of a single cell box girder bridge.
qD=256 KN/m. The live load moment=80 000 KN/m, the dead load moment
=320 000 KN/m, and the total 400 000 KN/m. Suppose an explosion occurs
on the deck of a single call box girder concrete bridge, leaving a hole 4 m
wide, the centre of which is 14 m from the edge of the footway; then it can be
shown that the residual bending strength for 100 subsequent crossings (level
2 risk) is about 340 000 KN/m. Since the dead load moment is virtually
unchanged, the remaining live load capacity is only 20 000 KN/m. This is
only 25% of the original undamaged capacity.
An examination of explosive damage to bridges by the author has indicated
that a method of rapid field assessment of residual strengths can be derived
and presented as a graphic display on a personal computer. This would show
the damaged area, zone of internal cracking, available residual lane width,
the load class of vehicle that would still use the bridge, crossing speed limitation
and number of crossings that could be safely made.
There are at least 370 different types of bridge structure in the world, and
for each of these there are at least twelve sources of explosion and twelve
records relating the point of the explosion to the structure. The sources influence
the type of damage and the explosion points influence the extent of the damage.
This gives a total of about 53 000 separate groups of calculations, many of
which would be very similar.
To take one example, Figure 8.12 shows the possible explosive sources for
damage to a tower and cable system, such as part of the structure of a cable-
stayed bridge. The extent of the damage from each source can be used to
calculate the safe residual strengths of the main load-carrying components.
We must limit the discussion to these general remarks, because more detailed
information could be of a restricted nature.
It is acknowledged that the major part of section 8.2 has been drawn from
a report by the author to the Defence Research Agency (Chertsey), who
commissioned a study on the assessment of damaged civil bridges.
residual strength and life after damage—all part of the study of the impact
damage tolerance of aircraft structures.
It was realized that existing design guidelines and specifications did not
fully address the projectile damage threat, and to improve design methods
Avery, Porter and Lauzze reviewed structural integrity requirements [8.25]
and resistance to battle damage. The damage in metal sheet, stiffener and
plate structures typical of modern aircraft takes the form of cracks, spallation,
petals, holes, dents or gouges, and the authors pointed out that for a given
target material the type of damage depends on sheet thickness, projectile velocity
and impact angle. A ‘damage regime’ diagram for 0.30 armour piercing
projectiles in plating made from the aluminium alloy 7075–T6, taken from
ref. [8.25] was shown in Figure 7.23. The most useful way of quantifying
projectile damage is by the ‘lateral damage’ notion, which is defined as the
diameter of an imaginary circle that just encloses the limits of fracture or
material removal. A typical variation of damage size with projectile velocity
was shown in Figure 7.24.
In skin and stiffener structures there are several types of damage
configurations, depending on whether the lateral damage is confined to the
skin between stiffeners, spreads across a stiffener with the stiffener remaining
intact, or spreads across a stiffener with the stiffener failing. The latter type is
normally the critical case for vulnerability analysis. If they remain intact the
stiffeners frequently provide a crack-arresting capability which can significantly
improve the residual strength of a battle-damaged structure.
The response of aircraft structures to impact damage was also examined,
for example, by Massmann [8.26], who considered the residual strength of
damaged structures by using finite element analytical methods and fracture
mechanics techniques. He developed a structural strength model, which included
an idealization of the wing of the US F84 aircraft. F84 wings had been shot at
during field tests, giving typical damage patterns of holes and cracks. Wing
plating in which the damage resulted in cracks only was analysed by using the
fracture toughness theory of Griffith, Westerg and associates to determine
residual strength. When the crack ran into a circular hole, the residual strength
was found to be a function of hole radius and crack length.
Test results and analysis showed that hits close to the front spar of the wing
considerably reduced the load capacity, and that wings constructed of stiffened
panels and three spars had a residual strength of over 2.5 times the residual
strength of milled integral panels having two spars (both geometries having
been designed to have similar initial load capacities). The stiffened panel design
was greater in weight than the integrally milled design, but in terms of the
entire weight of the aircraft, the increase in weight was less than 1% and the
decrease in manoeuvrability negligible.
In addition to the possible loss of strength from impact damage, the stiffness
of the aircraft structure may be altered, and stiffness degradation can lead to
aero-elastic problems associated with flutter, control and load redistribution.
Figure 8.13 Critical gross stress for stiffener and skin failures (from Avery et al., ref.
8.27).
hold, and fell from the sky. The loss of this aircraft has led to research to
investigate the blast forces on aircraft stiffened sheeting and structural members
after the blast has travelled through various irregular geometries of spaces,
conduits and ducts on its way from the hold to the vulnerable zones of the
airframe or control equipment. The geometry of the problem lends itself to a
finite element solution, but the basic variations of blast pressure with distance
and configurations still employ the fundamental science discussed in the later
sections of Chapter 5.
Turning now to ship structures, it is well known that below the waterline
marine vessels are particularly vulnerable to torpedoes and mines, and above the
waterline to shells, bombs and guided missiles. Much research has been devoted
to these problems, and in addition there was considerable activity in the 1950s
and 1960s to examine the response of naval vessels to nuclear blast. The work of
Hopkinson and others on anti-torpedo blister structures during the First World
War has already been mentioned, and at the more recent end of the research
spectrum it is noticed that new designs of warships incorporate structural features
that make them less easy to detect by guided bombs and search/strike missiles.
The costly attack/defence development battle is never ending.
The analysis of ship structures under blast loading is very complex, and
much of the knowledge used by designers is drawn from large-scale testing. In
a paper given in 1988, Charles Smith [8.28] described blast testing on deck, side
and bow components, and made the point that in designing warships to withstand
blast elastic design is generally too conservative, and that fairly large inelastic
deformations are tolerable as long as the protection of internal systems can be
maintained. He showed photographs of internal blast damage caused by guided
missiles, superstructure damage caused by air blast, and the breaking of a ships
back amidships by hull bending excited by a vibratory or ‘whipping’ response
of the hull to a non-contact underwater explosion. The latter problem was
examined in a paper by Hicks in 1986 [8.29], and by Jinhua and Zhang Qiyong
in 1984 [8.30]. The latter authors derived formulae for dynamic bending moments
which were shown to agree well with full-scale experiments in China. In these
tests charges varying in weight between 200 kg and 1000 kg of TNT were
detonated at depths between 7.5 and 40 m, and at distances from the vessels
between 7.5 m and 100m. The amplitudes of the modes of vibration were not
unexpectedly found to depend on the length along the vessel of the point of
action of the explosion. If this point was near the middle cross sections the 1st
mode of vibration was predictably the prime mode. If the point of action was
too near the quarter point the 2nd mode of vibration dominated.
Probably the most interesting paper referring to the history of underwater
explosion research in the ship structure field was published in 1961, and was
written by A.H.Kiel [8.31]. He referred first to the history of systematic testing,
beginning with the first reported tests in the USA in 1881 (Abbott, [8.32]),
and to the development between the two world wars of the side protection
systems against torpedoes used in the battle ships Midway (USA), Hood (UK),
Figure 8.14 Internal damage to a hull from a simulated torpedo explosion (from
Keil, 8.31).
350 kg of TNT, detonating 16 feet below the water line. This ripped a hole 20
ft by 18 ft and flooded damaged spaces with 1140 tons of seawater. The
second test was to detonate 100 kg of TNT near the relatively unprotected
bow and 12 feet below the waterline, and this ripped a hole 19.5 ft by 16.25
ft in the hull. A third test used 150 kg of TNT 20.8 ft below the waterline in
the area near the main gun magazines, producing a hole 19 ft by 17 ft. It is
interesting to note that the large variation in charge size did not produce a
noticeable variation in the size of the hole through the hull.
A further useful test was carried out on the Tosa, when a projectile with an
underwater trajectory was fired at the hull. The projectile was a 40 cm armour-
piercing shell, which struck the ship 11.7 ft below the waterline, as shown in
Figure 8.15. It penetrated the outer hull plate, the inner bottom plate and a
torpedo bulkhead (3 inch high tensile steel plate) before exploding. A section
14 ft by 8 ft was carved out of the double hull, and a total of 2950 tons of
seawater gained access to the damaged area. These tests have been recorded
in the History of Japanese Naval Construction, written in the 1950s by 50 of
Japan’s leading naval architects and naval constructors.
The structural response to explosions of submarines is another area where
much classified research has taken place. The resistance of hulls has been
increased, partly by the introduction of high-strength steels which lessened
the chances of hull splitting and flooding under attack by depth charges. It
was pointed out by Roseborough, also in the discussion of Kiel’s paper, that
damage to internal components and systems was more likely than hull damage.
Such damage could make vital controls inoperative, causing the submarine to
exceed the hull collapse depth or conversely to surface out of control. The
need to preserve the integrity of seawater systems within submarines, and
their hull valves, in the face of explosions was also underlined by Roseborough.
A submarine hull is often a ring stiffened cylinder, which lends itself to an
analytical as well as an experimental evaluation of structural performance, so
Figure 8.15 Tosa armour-piercing shell experiment (from Keil, ref. 8.31).
much time has been devoted over the years to the prediction of elastic response
to transversely exponentially decaying shockwaves. Work in this field was
summarized by Haxton and Haywood [8.35] in 1986.
8.4 REFERENCES
8.1 Thomas, W.N. (1948) The effects of impulsive forces on materials and structural
members, The Civil Engineer in War, UK Institution of Civil Engineers, London,
p. 65.
8.2 Baker, J.F., Leader Williams, E. and Lax, P. (1948) The design of framed buildings
against high explosive bombs, The Civil Engineer in War, UK Institution of
Civil Engineers, London, p. 80.
8.3 Christopherson, D.G. (1946) Structural Defence, UK Ministry of Home Security,
Civil Defence Research Committee paper RC 450.
8.4 Eytan, R. (1992) Response of real structures to blast loadings—the Israeli
experience. In P.S.Bulson (ed.), Structures under Shock and Impact (2), CM
Publications and Thomas Telford.
8.5 Rhodes, P.S. (1974) The structural assessment of buildings subjected to bomb
damage, The Structural Engineer, 52.
8.6 Williams, M.S., Lok, T.S. and Mays, G.C. (1989) Structural assessment of damaged
buildings. In P.S.Bulson (ed.), Structures under Shock and Impact (1), CM
Publications and Elsevier.
8.7 UK Ministry of Home Security (1941) Tests on Brick Surface Shelters at Stewartby,
Report RC 217, UK Ministry of Home Security, R and E Dept, April.
8.8 UK Ministry of Home Security (1941) Report on Tests: Methods of Strengthening
Brick Surface Shelters, Report RC 272, Ministry of Home Security, R and E
Dept, November.
8.9 UK Ministry of Home Security (1942) Report on Tests on Four Factory-type
Brick Surface Shelters, Richmond Park, Ministry of Home Security, Group 6,
Report RC 355, R and E Dept, October.
8.10 Walley, F. (1942) Movement of Shelters due to Earthshock, UK Ministry of
Home Security, R and E Dept, Report RC 287, January.
8.11 US Army (1965) Fundamentals of Protective Design (non-nuclear), US Army
TM5–855–1, November.
8.12 Bernai, J.D. (1942) Bulletin A-4, UK Ministry of Home Security, R & E Dept.
8.13 Rempel, J.R. and Beck, J.E. (1985) Stability of walls against airblast forces,
Proc. 2nd Symp. on the Interaction of Non-nuclear Munitions with Structures,
Panama City Beach, Florida, USA, April.
8.14 Napadensky, H. and Longinow, A. (1985) Explosion damage to urban structures
at low overpressure, Proc. 2nd Symp. on the Interaction of Non-nuclear Munitions
with Structures, Panama City Beach, Florida, USA, April.
8.15 Ellis, B.R. and Crowhurst, D. (1989) On the elastic response of panels to gas
explosions. In P.S.Bulson (ed.), Structures under Shock and Impact (1), CM
Publications and Elsevier.
8.16 Ellis, B.R. and Crowhurst, D. (1991) The response of several LPS maisonettes
to small gas explosions, Conf. on Structural Design for Hazardous Loads,
Brighton, England, April.
8.17 Moore, J.F.A. (1983) The Incidence of Accidental Loadings in Buildings 1971–
1981, UK Building Research Establishment, Current Paper 2/83.
8.18 The Effect of Impact and Explosions (1946) NRDC Summary Technical Report,
Washington.
8.19 Hill, L.G.W. et al. (1944) Report on Material Damage Caused by Bombs to
Bridges, UK Ministry of Home Security Research and Experiments Dept., Report
RC 448.
8.20 Walley, F. (1994) The effect of explosions on structures, Proc. Inst. Civ. Engrs
Structs and Bldgs, 104, August, p. 325.
8.21 Hamilton, H.G.W. (1994) Reminiscences of a Corps SORE II in the British
Liberation Army, 1944, The Royal Engineers Journal, 108(3), December, 289.
8.22 Pelton, J.F. (1993) Bridge inspections in Bosnia—Operation Grapple, The Royal
Engineers Journal, 10(2), August.
8.23 Organisation for Economic Co-operation and Development (1979) Evaluation
of Load Carrying Capacity of Bridges, Road Research Group Report, December.
8.24 Military Engineering, Vol III, published by UK Royal Engineers.
8.25 Avery, S.G., Porter, T.R. and Lauzze, R.W. (1976) Structural integrity requirements
for projectile impact damage—an overview, Proc. Conf. on Impact Damage
Tolerance of Structures, AGARD Conf. Proc. No. 186.
8.26 Massmann, S. (1975) Structural Response to Impact Damage, AGARD Report
633.
8.27 Avery, S.G., Porter, T.R. and Walter, R.W. (1972) Designing Aircraft Structures
for Resistance and Tolerance to Battle Damage, AIAA 4th Aircraft Design, Flight
Test and Operations Meeting, Los Angeles, California, August (AIAA paper
72–773).
8.28 Smith, C.S. (1988) The testing of marine structures, Proc. of Conf.: The Future
of Structural Testing, Bristol University, Computational Mechanics and McGraw
Hill.
8.29 Hicks, A.N. (1986) Explosion induced hull whipping, Proc. of Conf.: Advances
in Marine Structures, ARE Dunfermline, May, Elsevier Applied Science
Publications.
8.30 Jinhua, M.A. and Qiyong, Zhang (1984) The estimation of dynamic bending
moment for a ship subjected to underwater, non-contact explosions, Proc. Int.
Symp. Mine Warfare Vessels and Systems, London, June.
8.31 Keil, A.H. (1961) The Response of Ships to Underwater Explosions, US Society
of Naval Architects and Marine Engineers, Annual Meeting, November.
8.32 Abbott, H.L. (1881) Report upon Experiments and Investigations to Develop
a System of Submarine Mines for Defending the Harbours, Papers of the Corps
of Engineers, No. 23.
8.33 King, R.W. (1959) Modern Weapons and Ship Protection, Paper to Chesapeake
Section, SNAME, February.
8.34 Hollyer, R.S. (1959) Direct Shock Wave Damage to Merchant Ships from Non-
contact Underwater Explosives, Paper to Hampton Roads Section, SNAME,
April.
8.35 Haxton, R.S. and Haywood, J.H. (1986) Linear elastic response of a ring-stiffened
cylinder to underwater explosion loading, Proc. of Conf.: Advances in Marine
Structures, ARE Dunfermline, Scotland, May.
9.1 INTRODUCTION
Although the text of this book deals mainly with explosive loads and their
effects, it would be an omission if we did not consider in a general way the
analytical methods that seek to predict how structures respond to blast. We
should also consider briefly the qualities of structures that are designed to
resist known levels of explosive loading, because cases exist where the behaviour
of the structure under shock loading can change the applied transient loading
during the loading process.
Most hardened land structures are designed for military purposes, although
some thought has to be given to response in the design of grain silos, chemical
and explosive production plants, power stations and offshore structure blast
walls. Any historical review has an early start, because permanent defensive
installations have been an operational requirement since the beginning of
civilization, and we are all familiar with the great defensive ditches, earth
banks and stone walls built throughout the world in earlier ages. It is interesting
to note that the Brits, the Celts and the Saxons tended to dig their fortifications,
whereas the Romans and Chinese built walls.
As the knowledge of materials and structures grew, so the fortifications
became more sophisticated, and stone-built castles became an early example
of a hardened structure capable of protecting the inhabitants from attack for
long periods, and then becoming the springboard for offensive operations.
The design of castles, like all hardened structures, was conditioned by the
type and quantity of the hostile fire expected from the enemy, and it passed
through many stages in the Middle Ages. The single large tower, or keep, was
replaced by a fort within a series of forts; the walls were increased in height
and galleries were added for archers. The coming of gunpowder and cannon
in the fifteenth century raised many new problems, because artillery fire could
be concentrated in one spot, battering down weak masonry. The castle therefore
lost its defensive capacity and was replaced by the bastioned fortress.
The great military engineers of the time developed the ‘star’ fortress, with
five or six bastions each shaped in plan like the ace of spades, so that the
defenders’ artillery could ward off the assault forces and at the same time give
covering fire to adjacent bastions. However, as we saw earlier in the
Introduction, the dominance of this form was threatened in the seventeenth
century by systematic methods of attack developed by Vauban. He developed
fortress designs to withstand the new attack methods, and thus became the
outstanding figure in both siegecraft and in fortress defence of his time. It has
been said that a fortress built by Vauban was impregnable, and that one besieged
by him was doomed. His contribution has been discussed by Harris [9.1].
By the early years of the twentieth century concrete began to be used in
increasing quantities, and concrete roofs of 2 or 3 metres thick were built over
revetments. Major cities in Europe were ringed with fortresses and in the
years leading to the outbreak of the First World War the ideas of the Frenchman
Brialmont were used in the construction of defensive systems. He used steel
armour to protect guns, and devised retractable cupolas and turrets; his ideas
were used by the French in developing their fortress system along the German
frontier.
Reinforced concrete now became the standard material for protective
structures, and was used in conjunction with the application of ballistics to
the design of fortifications. Individual forts were often triangular in plan, and
by 1914 the Brialmont system had been used to build this type of fort at
Antwerp, Namur and Liege in Belgium. At the beginning of the First World
War the Germans attacked through Belgium, using the Krupp 420 mm Howitzer.
Liege fell in eleven days, Namur in four and Antwerp in ten, and apart from
the success of the fortress of Verdun as a very strong point in the battle of the
Marne, it was generally concluded at the end of the war that the cost of the
fortifications was not justified.
In spite of this, as the Second World War approached, the French built the
celebrated but ill-starred Maginot line, the Germans built the Siegfried line,
and the Russians the Stalin line. On the Dutch/Belgian border a fort very
similar to the Maginot line design was built to dominate the River Meuse. It
was thought to be impregnable. On 10 May 1940 this fortress was apparently
taken in a few hours by less than one hundred German glider-borne engineers.
They are said to have landed directly on top of the fortifications and used new
shaped hollow charges to blast into the cupolas, although the accuracy of this
report has been questioned. After the breakthrough the Maginot line was
outflanked, and the huge construction of mutually supporting positions
stretching from Luxembourg to Switzerland, which involved 100 kilometres
of tunnels, 12 million cubic metres of earthworks, 1.5 million cubic metres of
concrete and 150 000 tonnes of steel, was of no value. The fortifications have
been described by Taylor [9.2].
In the period since the Second World War there has been a change in the
protective structure requirements of military operations, brought about by
the development over the past forty years of missile systems, satellite and
electronic surveillance, nuclear weapons, and sophisticated command, control
and communication centres. The structures are now required to protect
personnel and instrumentation against the effects of nuclear, non-nuclear and
chemical weapons, to form the protective shells of weapon silos and anti-
missile installations and the protective shelters of aircraft and naval craft.
There has been a notable increase in experimental and theoretical studies to
examine the behaviour of hardened structures under shock and impact loading,
and to investigate the penetrative characteristics of high-speed missiles into
metal and concrete. Attempts have also been made to formulate design rules
and codes of practice for the development of this type of structure.
Much experimental data has been assembled from all the above sources,
gained over a period of over 100 years, from two world wars, countless minor
armed conflicts or terrorist actions, cold war defence research and weapon
development programmes. A complete review of all data could fill several
books, but as far as the author is aware no unified and unclassified data bank
of experimental results has ever been put together.
The gradual collection of field and laboratory evidence in the early days
naturally led to the development by applied mathematicians, physicists and
engineering analysts of theories to predict structural behaviour, based on the
fundamental criteria of dynamic loading, and requiring input loading functions,
resistance and deformation relationships, mass distribution and material failure
relationships. The analytical solutions for structural response, always tedious
to produce, were later increased rapidly in number by the coming of the
computer and by the development of theories of non-linear and limit state
behaviour. Discussions of the accuracy of the empirical formulae of early
design codes, so useful in making initial assessments of structural performance,
have tended to be superseded by debates on the ‘modelling’ of material
characteristics and structural actions in the all-embracing hydrocodes of modern
computer analysis. The marketing of software to give instant design solutions
for structures responding to explosive loading is now a business activity of
some magnitude, but the assumptions on which the modelling is based are
not always critically reviewed.
To weave through this forest in such a way that all trees can be inspected,
if only briefly, means that a somewhat tortuous path must be followed. As an
entry point we will examine the state of affairs in Britain at the beginning of
the Second World War.
treated as linear elastic and undamped, and it was assumed that it could be
represented by a mass supported by a spring, capable of displacement in one
direction only, and therefore having a single degree of freedom. This meant
that the well-established and familiar governing equation of motion using
Newton’s second law could be used, and was set down in the usual form
(9.1)
where m is the mass, k is the spring stiffness, f(t) is the time-varying force, t is
time and y is mass displacement.
Solutions were developed for a range of ‘forcing functions’, varying from a
simple sinusoidal relationship between force and time to an approximately
triangular relationship with an instantaneous pressure rise followed by an
exponential pressure decay. This was taken as a typical force-time relationship
for a detonated explosion at that time.
After the Second World War the further development of nuclear bombs
resulted in intense target response research, particularly in the USA, and the
basic equation (9.1) was used to provide data on the elastic analysis of simple
systems in order to attempt to explain the results of the large number of response
tests held on mainland USA and on small islands in the Pacific. The ratio
between the displacement (y) of a system calculated from Eq. (9.1) and the
displacement under the same load system applied statically (yst), known as the
Dynamic Load Factor, was easily shown to be related to the ratio of the duration
of the applied load, td, and to the natural period of oscillation of the system
(T). Ranges of values of y/yst for a range of values of td/T for triangular pulses
were calculated and recorded in great detail in the works of the American
authors Biggs [9.4] and Newmark [9.5]
When consideration of the vibration of a one-degree of freedom elastic
system after the first peak of elastic response was required, it was necessary to
include damping in the analysis. On the convenient though not necessarily
accurate assumption of viscous damping, Eq. (9.1) became
(9.2)
However, with the values of C for most structures turning out to be about
onetenth of the critical damping value (C=2mw) the natural frequency of
vibration was hardly changed.
It was soon clear to the research fraternity that most structures suffered
permanent deformation in the face of blast loading, with the deformation
governed by structural ductility as well as by elastic and inelastic properties.
It was therefore necessary to consider the effect of large plastic strains and the
effects of plastic energy absorption. In calculating the response of a single
degree of freedom system to a suddenly applied constant load, F, for example,
it was realized that there were two important and discontinuous stages. There
was elastic response until deflections reached the limit of elastic behaviour,
yel, followed by a fully plastic response until the maximum deflection, ym was
reached. The assumption here was that the response changed from fully elastic
to fully plastic with no intermediate elasto-plastic zone.
1/
In Eqs (9.1) and (9.2) there is a relationship between (spring stiffness/mass)
2
and the natural circular frequency of the oscillations of the undamped spring/
mass system, so that (k/m) =ω. Using this relationship it was shown that for
2
the first stage of loading y=yst(1–Cos ωt) when 0<y<yel, and for the second
stage
(9.3)
when yel<y<ym.
Rm is the structural resistance when y=yel, and t1=t–tel.
The time when the maximum response was reached was given by
(9.4)
and when this was substituted in equation (9.3) an expression was found for
ym.
Charts were produced by Biggs (9.4) and others that plotted values for ym
and tm for ranges of values of Rm/F, and for a variety of loading functions. The
triangular pulse loads were in the form of an instantaneous pressure rise followed
by a linear decay, to represent detonated explosions, or a linear pressure rise
followed by an equal decay, taken as an approximation to the characteristics
of a dust cloud explosion. If the structural resistance was equal to or greater
than the applied load, the response was entirely elastic.
The limitation of the charts was that in practice the behaviour of very few
practical structures or structural components could be represented by a single
degree of freedom system, but to combine multi-degree of freedom analysis
with inelastic response was thought to be an almost impossible task in the
days before computers. Considerable efforts were therefore made to represent
a range of types of structural component by ‘equivalent’ single degree of freedom
systems. This representation relied on the establishment of equivalent values
of peak force (Fe), mass (me) and maximum resistance (Rme) so that the deflection
of a significant point on the real structure was similar to that of the equivalent
single degree of freedom spring/mass system. The period of initial vibration
was also kept similar. This led to the idea of transformation factors, defined
as Km=me/ m, KL=Fe/F and KR=Rme/Rm, and methods were developed to calculate
these for particular structural components, as described by Biggs (9.4).
It is important to note that the equivalent system was first set up on the
basis of kinetic similarity, so that the maximum component displacements
were accurate. Moments and shears, which depend on the derivatives of
deformation patterns, rather than displacements, were less accurate. For the
Figure 9.1 The equivalent bilinear response for a clamped beam (from Baker and
Spivey, ref. 9.10).
conditions. Circular plates and other structural elements have also been treated.
They appear in references [9.4] to [9.8] and in other related official publications.
The paper by Baker and Spivey [9.10] describes a computer program that
solves numerically various dynamic response problems represented by the
above system. The development in the late 1980s of programs like this has
superseded the former tables of transformation factors. An important feature
of the new programs is that dynamic reactions at supports are considered, so
that dynamic shear forces can be checked. These are normally missing from
the single degree of freedom spring-mass analogy. However, these dynamic
reactions have been analysed on the assumption that the deflected shape of a
beam, for example, remains the same after the transition from elastic to plastic
response, which is not necessarily true. A very high, short-duration blast load
could lead to high beam curvature close to the supports, producing high shears.
Care should therefore be taken when applying the single degree of freedom
analogy to explosive forces from very local or close-in detonations.
It has become customary to represent the ultimate strength of dynamically
loaded structures on a pressure/impulse diagram. When the duration of the
blast pressure, td, on an elastic structure is large compared to the natural
period (T), there is very little reduction in the initial peak pressure before
the structure respond elastically and reaches its maximum deflection. For
values of td/T>40, and an approximately triangular pulse, (y/yst)max=2, and
deflection depends only on the peak value of the pressure and the structural
stiffness. It is independent of the duration, td, and the mass of the structure.
However, when the duration is small in comparison with the natural period,
so that td/T<0.4, deflection is proportional to impulse I, where I=pt. Between
these extremes deflection is governed by a combination of pressure and
impulse. The information for a particular structure can be represented non-
dimensionally by dividing the applied pressure by that required to achieve
y/yst=2, and by dividing the impulse by that required to produce a direct
proportionality between impulse and deflection. Similar curves can be
calculated for an elasto/plastic response and for other types of pressure/
duration relationship.
Test results can be plotted on a pressure/impulse diagram so that the form
of the transition can be established for a range of types of structure under
various types of explosive attack. By these means ‘iso-damage’ curves are
produced, which are useful to designers. As well as using test results, theoretical
relationships were established in the mid-1970s by Abrahamson and Lindberg
[9.11] for both elastic and rigid-plastic single degree of freedom response.
The response to explosions of complex multi-compartmentalized structures
of all types seemed to demand representation by multi-degree of freedom
systems, with the structural representation in the form of lumped masses
connected by elastic springs, and this method was given a close examination
in the 1950s. The number of separate types of motion is equal to the number
of degrees of freedom, so deflections were ‘coupled’ and the solution involved
Since {an} is not zero, this equation can be simplified to the eigen-value
relationship
and the eigen values are equal to the squares of the natural frequencies of the
modes.
The response of combinations of elements is not the only circumstance
requiring multi-degree analysis. Structures buried in consolidated soil or rock
create a multi-degree system involving the elastic and elasto-plastic
characteristics of the earth material. It is also possible for the structure to be
surrounded by earth material and then by a further layer of elemental structure
such as a slab or beam grid. Methods exist for analysing this type of
configuration as a multi-degree of freedom system.
Some structures that contain combinations of elements can be analysed
accurately if the elements can be treated as separate single degree systems.
This is possible if the kinetic energy, the internal strain energy and the work
done by both systems when vibrating in a normal mode are equal. Once the
equivalent system has been established, it can be shown that the dynamic load
factor only depends on the applied force and angular velocity, so that the
relationships developed for single-degree systems can be used. Examples were
prepared by Biggs in the early 1960s and are discussed in reference [9.1].
There was a rule that if the ratio of the periods of component elements was
greater than 2, then the components could be treated separately. If the ratio
was less than 2, a multi-degree approach should be used.
Theoretical response methods have been augmented in recent years by the
advent of non-linear explicit three-dimensional finite element codes such as
DYNA-3D from the Lawrence Livermore National Laboratory in California.
Calculations based on these codes have been shown to give good agreement
with experimental data for a range of blast/structure problems, including
structures in soil. In certain circumstances the finite element codes have been
used in a preliminary way to improve the modelling used subsequently in
single or multi-degree of freedom analyses. Typical examples of research in
these areas are contained in papers by Bingham, Walker and Blouin [9.12],
Wright and Hobbs [9.13], Terrier and Boisseau [9.14], and Crawford and
Table 9.1
Table 9.2
expected number of fragments that could hit the structure without ricocheting
using a simple spray pattern (cylindrical or spherical). The third step is to
select a design fragment and find its lethality (based on penetration); using a
formula for concrete as an example, lethality=, where Wf is fragment weight
(lbs) and Vs the striking velocity (feet/sec). This formula is likely to be very
conservative. The next steps are to calculate the probability that no fragments
with a lethality greater than the design fragment will be generated by the
exploding bomb, and to calculate the expected number of lethal fragments in
the total fragment population. The final stages are to select a size of design
fragment so that the desired level of reliability is achieved and there will be no
lethal hits. This fragment is then used in the design of the protective structure.
The method relies on a large amount of reliable experimental data that has
been analysed on a probability basis to provide fragment weight and velocity
distributions, and whether this could be assembled for a full range of weapons
and circumstances is debatable.
All this research has to do with load factors, but before the overall safety
level of a structure can be assessed these factors must be combined with structural
resistance factors. Twisdale and his colleagues looked at this problem for
loads due to projectiles and fragments, and their preliminary findings were
reported in reference [9.28]. They took as an example the design of a reinforced
concrete wall to resist 30 mm cannon fire with a 95% reliability, and used
load factors based on existing databases for small projectiles impacting massive
concrete targets. From these databases 710 records were used, of which 534
gave depth of penetration and 703 gave information on spall and perforation.
A typical striking velocity coefficient of variation was 0.2, and for a reliability
of 95% the specified striking velocity needed to be multiplied by a load factor
(for concrete perforation) of 1.12. The factored striking velocity was then
used to calculate the minimum thickness of wall to prevent perforation. For a
reliability of 95% the calculated wall thickness was multiplied by a resistance
factor of 1.54, so that the overall safety factor was 1.12×1.54=1.8. Thus the
safety margin for 95% reliability in design against projectile perforation looks
to be less than that for a structure under incident air blast, where just the load
factor might be of the order of 2.0. These figures are influenced of course by
the coefficients of variation of the experimental results, so too much should
not be read into them. They should be taken as an illustration of recent thinking
in the use of reliability-based analysis in the design of structures under the
threat of blast or penetrating weapons.
It is interesting to compare these factors with resistance factors proposed
by the author when discussing the design of underground protective structures
subjected to surface air-blast loading, given in reference [9.29]. Using engineering
judgement a load factor of 1.5 was proposed for dynamic pressures due to
explosions on the surface, and a resistance factor of 1.6 for strength variations
due to uneven soil compaction or the presence of ground water. For structures
in which close attention was paid to emplacement it was considered that the
9.4 EVOLUTION
We have now examined the gradual assembly of information on the explosive
loading of structures over two centuries and more. As in many other subjects,
the march of progress has not been gradual and well-conditioned. There have
been periods of intense activity followed by years of rest and recovery, not
unlike the growth patterns of nature which so often consist of steep gradients
and level plateaux. From the analytical and experimental viewpoints, major
progress was made at the time of two world wars and the nuclear cold war,
when money was available for the very costly activity of full-scale field testing.
Theories and analysis are subject to swings of fashion, but the results of well-
conducted tests have a lasting value.
There are psychological problems when a field of research is connected
with an activity considered to be associated with physical danger, or with the
pursuit of war or terrorism. A background of uncertainty can sometimes lead
to a superficial consideration of basic engineering or scientific principles. Much
has been written in this book about the behaviour of engineering structures,
but less about the behaviour of structural engineers. Perhaps future research
will examine this problem in greater depth, and the subject has been explored
by the author in a series of short stories.
Future work on the shock loading of structures is likely to continue at a
steady rate as the train of research moves on to its next stop. On the way it is
expected to journey through fields of risk and reliability, modelling laws, new
instrumentation, and the computer simulation of large-scale testing. There is
also a considerable interest now in the prediction of fragment loading and
damage, and in penetration mechanics.
At the time of writing these closing remarks the 8th International
Symposium on the Interaction of the Effects of Munitions with Structures
has just taken place in the USA. A glance through the agenda shows that,
very approximately, about half the papers dealt with structural response
and protective design, about one quarter with penetration and fragmentation,
and the remaining quarter was divided about equally between blast loading
research and testing or simulation. There were many papers on the application
of hydrocodes and other numerical techniques to the various problem areas
of the subject.
One of the main tasks of the near future will be to bring the fruits of
structural loading research into a form suitable for design codes of practice
for civil buildings, bridges and other land, sea and air structures. It is hoped
that the research history and inspirations from the past retailed in this book
will help towards this work, and at the same time define a background for
the evolution of new scientific insights as well as the development of existing
knowledge.
9.5 REFERENCES
9.1 Harris, A. (1989) A man of mettle, New Civil Engineer, 18 May.
9.2 Taylor, G. (1987) The general design and use of hardened defences in twentieth
century warfare, Royal Engineers’ Journal, 101(2).
9.3 Anon (1939) The Design of Buildings against Air Attack, Part 2, Mathematical
Analysis of the Effect of Blast on Structures, HOPP/18, Civil Defence Research
Committee paper RC 23.
9.4 Biggs, J.M. (1964) Introduction to Structural Dynamics, McGraw Hill, New
York.
9.5 Newmark, N.M. et al. (1961, rev. 1964) Design of Structures to Resist Nuclear
Weapons Effects, American Society of Civil Engineers, Manual of Engineering
Practice No. 42. Republished in 1985.
9.6 US Dept of the Army (1969 and 1991) Structures to Resist the Effects of Accidental
Explosions, US Dept of the Army Technical Manual TM5–1300.
9.7 Crawford, R.E., Higgins, C.J. and Bultmann, E.H. (1974) The Air Force Manual
for Design and Analysis of Hardened Structures, AFWL TR 74–102, AFWL
Kirkland AFB, New Mexico, October.
9.8 US Dept of the Army (1986) Fundamentals of Protective Design for Conventional
Weapons, US Army TM5–855–1, November.
9.9 Baker, W.E. et al. (1983) Explosion Hazards and Evaluation, Elsevier, New
York.
9.10 Baker, W E., and Spivey, K H. (1989) BIGGS—simplified elastic-plastic dynamic
response. In P.S.Bulson (ed.), Structures under Shock and Impact, Elsevier, p.
135.
9.11 Abrahamson, G.R. and Lindberg, H.E. (1976) Peak load—Impulse
characterisation of critical pulse loads in structural dynamics, Nuclear Engineering
and Design, 37, pp. 35–46.
9.12 Bingham, B.L., Walker, R.E. and Blouin, S.E. (1993) Response of pile foundations
in saturated soil, Proc. 6th Int. Symp. on Interaction of Non-nuclear Munitions
withStructures , Panama City Beach, Florida, USA, May.
9.13 Wright, S.J. and Hobbs, B. (1994) Use of the DYNA-3D FE code to quantify the
response of steel plate girders subjected to localised combined blast and fragment
loads. In P.S.Bulson (ed.), Structures under Shock and Impact, Computational
Mechanics Publications, June.
9.14 Terrier, J.M. and Boisseau, J.F.X. (1989) Numerical simulations of reinforced
structure response subjected to high explosive detonation, Proc. 4th Int. Symp.
on Interaction of Non-nuclear Munitions with Structures, Panama City Beach,
Florida, USA, April.
9.15 Crawford, J.E. and Mendoza, P.J. (1985) Combined finite element and lumped
mass techniques for parametric structural analysis of structures, Proc. 2nd Symp.
on the Interaction of Non-nuclear Munitions with Structures, Panama City
Beach, Florida, USA, April.
9.16 Walker, A.C. (1993) Non-linear analysis of blast walls to evaluate safe ultimate
loads, 2nd Int. Conf. on Offshore Structural Design against Extreme Loads,
ERA, London.
9.17 Schleyer, G. and Mihsein, M. (1992) Development of mathematical models for
dynamic analysis of structures, 1st Int. Conf. on Structural Design against Extreme
Loads, ERA, London.
9.18 Louca, L.A., Punjani, M. and Harding, J.E. (1996) Non-linear analysis of blast
walls and stiffened panels subjected to hydrocarbon explosions, Journal of
Constructional Steel Research, 37(2), April.
9.19 Van Wees, R.M. (1993) Design of stiffened panels to withstand explosion loads,
2nd Int. Conf. on Offshore Structural Design against Extreme Loads, ERA,
London.
9.20 Houlston, R. and DesRochers, C.G. (1987) Non-nuclear structural response of
ship panels subjected to air blast loading, Proc. 6th Conf. on Non-linear Analysis,
ADINA.
9.21 Carson, J.M., Morrison, D. and Hampson, R.J. (1984) Conventional High
Explosive Blast and Shock (CHEBS) Test Series: Mark 82 General Purpose
Bomb Tests, Air Force Weapons Laboratory, AFNL-TR-84–27, June.
9.22 Carson, J.M. and Morrison D. (1987) Conventional High Explosive Blast and
Shock (CHEBS) Test Series: Mark 83 General Purpose Bomb Tests, Air Force
Weapons Laboratory, AFWL-TR-86–53, Parts 1–3, January.
9.23 Twisdale, L.A., Sues, R.H., Lavelle, F.M. and Miller, D.B. (1991) Research to
develop reliability-based design methodology for protective structures, Proc.
5th Int.Symp. on the Interaction of Conventional Munitions with Protective
Structures , Mannheim, Germany, April.
9.24 Twisdale, L.A., Sues, R.H. and Lavelle, F.M. (1993) Reliability based design
methods for the Protective Construction Design Manual, Proc. 6th Int. Symp.
on the Interaction of Non-nuclear Munitions with Structures, Panama City
Beach, Florida, May.
9.25 Sues, R.H., Drake, S.L., and Twisdale, L.A. (1993) Reliability based safety factors
for grounds shock loads in protective construction, Proc. 6th Int. Symp. on the
Interaction of Non-nuclear Munitions with Structures, Panama City Beach,
Florida, May.
9.26 Drake, J.L. et al. (1989) Protective Construction Design Manual, ESL-TR-87–
57, Air Force Engineering and Services Centre, Tyndall Air Force Base, November.
9.27 Sues, R.H. and Twisdale, L.A. (1993) How to select a design fragment for
protective structure design with consistent reliability, Proc. 6th Int. Symp. on
the Interaction of Non-nuclear Munitions with Structures, Panama City Beach,
Florida, May.
9.28 Sues, R.H., Hwang, C.W., Twisdale, L.A. and Lavelle, F.M. (1991) Reliability
based design of R/C structures for protection against projectiles and fragments,
Proc. 5th Int. Symp. on the Interaction of Conventional Munitions with Protective
Structures, Mannheim, Germany, April.