International Journal of Pressure Vessels and Piping 87 (2010) 493e503

Contents lists available at ScienceDirect

International Journal of Pressure Vessels and Piping

journal homepage: www.elsevier.com/locate/ijpvp

Assessment of blast loading effects e Types of explosion and loading effects

Ian G. Cullis a, *, James Schoeld b, Angela Whitby c
a

QinetiQ, Fort Halstead, Sevenoaks, Kent TN14 7BP, United Kingdom

QinetiQ, Rosyth, Dunfermline KY11 2XR, United Kingdom
c
QinetiQ, Farnborough, Hampshire GU14 0LX, United Kingdom
b

a r t i c l e i n f o

a b s t r a c t

Article history:
Accepted 26 January 2009

There has been a great deal of work undertaken on the modelling and protection of the structure of
buildings to explosive blast, fragment and missile penetration. Buildings and tunnels present interesting
challenges because of the different energy release characteristics of modern explosives and gaseous
mixtures. The paper describes how by combining modelling capabilities in blastestructure interaction
with vulnerability models a capability is formed that has a wide range of potential uses, which links the
many stages of response planning to an attack or more general emergency. From designing survivable
infrastructures and planning protective measures, through training of staff and responders to handling of
actual events, the concept provides a comprehensive approach to the whole spectrum of building and
infrastructure survivability assessment issues for a wide range of environments. The paper describes its
application to buildings and other related infrastructure.
2010 Elsevier Ltd. All rights reserved.

Keywords:
Explosions
Numerical simulation
Structural response
Vulnerability models

1. Introduction
There has been a great deal of work undertaken on modelling
and experiment to protect buildings against explosive blast, fragment and missile penetration.
Buildings and tunnels present interesting challenges because of
the different energy release characteristics of modern day explosives, including gaseous mixtures. In addition the shape of the
charge can produce local effects that are not described by semiempirical models such as CONWEP [1], extensively used by industry.
QinetiQ has an in depth understanding of the detonation process,
the resulting blast wave and its effects concerned with buildings,
particularly in complex environments, where blastestructure
interaction can lead to unexpectedly enhanced damage.
This understanding has been generated through a combination
of experimental trials and detailed numerical simulations
employing both Eulerian and Lagrangian methods. This complexity
of model is vital to accurately assess the consequences of explosions, particularly where the explosive products ingress into
structures and for non-ideal explosions. These models are important in predicting the total energy released, but more importantly
in predicting the timing and location of the release in order to
understand the subsequent structural response.

* Corresponding author. Tel.: 44 1959 514889; fax: 44 1959 516050.

E-mail addresses: igcullis@qinetiq.com (I.G. Cullis), jsschoeld@qinetiq.com
(J. Schoeld), ajwhitby@qinetiq.com (A. Whitby).
0308-0161/$ e see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijpvp.2010.07.003

The ultimate end user, whether military commander in the eld,

or the owner of a building or facility, however, is more interested in
the effects on the structure, its services, any personnel in or around it
and surrounding structures. The requirement here is for a fast
running analytic vulnerability model which can provide an assessment of the consequences of a wide range of possible scenarios.
By integrating these two modelling approaches we have
a potentially powerful tool based on a hierarchical approach of
detailed high delity simulations (physics and chemistry based
models) linking to engineering models and ultimately to lower
delity operational analysis tools, including synthetic environments.
The paper describes how integrated modelling can provide
a comprehensive approach to the whole spectrum of building and
infrastructure survivability assessment issues for a wide range of
environments. The paper describes its application to buildings and
other related infrastructure.
2. Hydrocode modelling
QinetiQ and its predecessors, under UK Ministry of Defence
(MOD) funding, have made signicant investment in the development of advanced Eulerian numerical methods that are state of the
art. The resulting multi-material capability, GRIM, has been
extensively used over the past 25 years to solve problems in detonics, weapon design and their target effects. Related development
work on Lagrangian methods have centred on the Lawrence Livermore National Laboratory code DYNA. Both GRIM2D/3D and
DYNA have been extensively validated and veried against a wide

494

I.G. Cullis et al. / International Journal of Pressure Vessels and Piping 87 (2010) 493e503

range of dynamic response scenarios as a result of impact and

explosions.
The success of any numerical scheme however is totally
dependent on the material equation of state and constitutive
models employed within it. If we are to be able to predict the effects
of explosions and blast waves on structures then we have to have
accurate and validated material models. A crucial parallel research
activity, therefore, has been the development of advanced material
algorithms [2] that can describe the response of a material over
a wide range of strain, strain rates and temperature. This includes
both path dependent deformation behaviour and failure [3,4].
Associated experimental programmes have developed tests to
characterise materials of interest including metals, geological
materials [5] and concrete and polymer composites.
We have long recognised the limitations and deciencies of TNT
equivalency and blast-scaling, and its application in determining
structural deformation for modern explosives, and have developed
a number of methods which can be judiciously chosen to represent
the appropriate blast assessment methodology.
It is well known that for most explosives the energy released by
the detonation process is less than the heat of reaction. It arises
because the products generated by the explosion continue to react
as they expand and cool, releasing or absorbing additional energy.
TNT for example is 74% oxygen decient, which implies that
considerable additional energy is potentially available if there is
sufcient air for the products to mix and react with.
The actual importance of this effect was recognised by the
weapons design team at QinetiQ over 10 years ago, which showed
that although this after-burn energy does not contribute to the
initial air shock, it does affect the blast wave impulse/pressure in
the mid-distance range from the charge, which in turn affects the
TNT equivalence [6,7]. Fig. 1 shows the ability of GRIM to reproduce
the measured pressure-time histories for Ammonium Nitrate Fuel
Oil (ANFO), an oxygen balanced commercial explosive.
3. Combustion models
In order to account for the after burn energy release additional
models have to be incorporated into the numerical scheme. We
have developed a range of models of varying degrees of sophistication to represent this energy release.
3.1. Simple combustion model

much longer times compared to that predicted by GRIM. Numerical

simulations of the blast eld produced by cylindrical charges,
whilst able to predict the initial peak pressure accurately, were
unable to predict the arrival time of the reected ground shock and
its magnitude to the same degree of accuracy. This suggested that
the velocity of sound in the products was too low, consistent with
the persistence of the reball in the experiment.
The secondary combustion energy was introduced into the
GRIM simulation by adding additional energy into the products
over a period of time, specied by the user. In the case of the
explosive DEMEX [7], the secondary combustion energy was estimated as 8.7 MJ kg1, compared to a detonation energy release of
5.1 MJ kg1. Experimental data, for a 2.2 kg spherical charge,
showed the reball to undergo a rapid expansion between 100 ms
and 1 ms after detonation, suggesting this was the time over which
after-burning occurred.
Using this simple approach Figs. 2 and 3 show the agreement
between GRIM and experiment for TNT charges of differing masses.
Fig. 2 compares overpressure and Fig. 3 compares impulse. The
agreement is excellent and within the experimental error. This
implies we can use our Euler code to model a wide range of energetic
sources and predict the blast wave energy, velocity and magnitude.
In the case of a 23.7 kg charge adding the secondary combustion
energy between 100 ms and 30 ms, based on the cube root scaling
law, we were also able to bracket the experimentally measured
impulse-distance behaviour [7]. Thus provided one has access to
experimental reball behaviour data this simple algorithm is
surprisingly effective.
Whilst the simple algorithm is remarkably successful it does not
capture the physics and chemistry of the combustion process and,
being based upon free-eld experiments cannot account for
scenarios where the oxygen available for combustion is limited, e.g.
a conned space. An advanced algorithm was therefore developed
to reproduce these more complex blast scenarios.
3.2. Advanced combustion model
There are a number of competing physical processes that can be
relevant to the combustion of detonation products expanding in air.
As is well known the interface between the explosive products and
the air is naturally RayleigheTaylor unstable and the resulting
secondary combustion energy release is through turbulent
combustion mixing of the products and oxygen in the air [8]. Whilst

The initial combustion model was based upon free eld explosion experiments and the observation that the reball persisted for

100
CONWEP Spherical
CONWEP Hemispherical
6kg TNT Shots
0.75kg TNT Shots
0.75kg PE4 Shots
6kg GRIM
0.75kg GRIM (platform)
0.75kg GRIM

Over Pressure (Atm)

10

0.1
0

Scaled Distance (m/kg(1/3))

Fig. 1. ANFO free eld blast comparison of experiment with GRIM.

Fig. 2. Blast Overpressure for TNT: Comparison of Simulation with Experiment.

I.G. Cullis et al. / International Journal of Pressure Vessels and Piping 87 (2010) 493e503

Fig. 3. Blast Impulse for TNT: Comparison of Simulation with Experiment.

initially mixing of products and air might be driven by diffusion the

subsequent rapid development of the surface instabilities will take
over as the dominant mixing mechanism.
As turbulent mixing is considered the governing mechanism
a key requirement of the numerical scheme, therefore, is the ability
to represent instabilities across interfaces. The advanced interface
tracking scheme in GRIM has this capability.
The combustion model starts with the assumption that equilibrium chemistry is a realistic approximation. We can therefore
write the chemical reaction in terms of the stoichiometric coefcients of the initial detonation products, the fuel, oxygen and the
nal products. Following the work of Kuhl for the detonation of TNT
in air, these coefcients are 1.0, 5.33555 and 1.0 respectively with
molecular weights 227.0, 32.0 and 397.7376. The secondary
combustion energy is then 10.392 MJ kg1.
In the algorithm, once the detonation wave has passed through
the charge, mixed cells containing air and fuel are allowed to react
to produce nal products, releasing the appropriate amount of
energy due to the reaction. The amount of energy released is thus
dependent on the amount of air present and the degree of mixing.
Fig. 4 shows the development of the producteair interface at
100 ms, 500 ms (Fig. 4a) and 1 ms and 5 ms (Fig. 4b) after detonation
for a 1 kg charge of TNT. The second shock wave is clearly observable in Fig. 4b at 5 ms. The energy released as a function of time is

495

shown in Fig. 5. The changes in slope are associated with the release
waves from the producteair interface propagating back towards
the origin and the negative phase in the pressure record. The
bounce of the release wave to form the second shock wave for
example generates the change in the energy release rate record at
about 2.4 ms as it propagates through the productseair interface
increasing the degree of mixing.
Due to a lack of available experimental data the advanced model
was validated against the gas dynamic turbulent combustion model
of Kuhl et al. [8], using a 1 kg TNT charge. Kuhl showed that the
results of his simulations, in terms of the timing and the amount of
energy released were dependent upon the spatial resolution of the
numerical mesh. The GRIM simulations showed a similar trend as
expected since a ne mesh is required to follow the turbulent
mixing of the gases accurately, produce a better degree of mixing
and hence a greater exothermic energy release.
We also recognise that the ability of a 2D axis-symmetrical
simulation to represent the 3D unstable producteair interface is
limited. However, extensive unpublished research in the UK has
demonstrated that the approach reproduces the essential physics.
4. Gaseous explosions
Unlike a condensed explosive, where the energy release is
governed by the detonation wave, a gaseous explosion is governed
by the ame speed, whose velocity is variable and a strong function
of the chemical composition of the gaseair mixture and its physical
surroundings, i.e. connement.
Depending on the initial conditions within the mixture and the
environment containing it the burning rate can vary by several
orders of magnitude. Combustion in ammable gaseous mixtures is
a complex problem. Not all gaseous mixtures are ammable. In the
case of methaneeair the ammability limits are highly clustered
about the stoichiometric mixture. A stoichiometric reaction is
a unique reaction in which all of the reactants are consumed.
The rate of energy release is governed by the ame speed that
results in ignition of the mixture. The slowest ame speeds are
associated with laminar ames and in a methaneeair mixture
these speeds are of order 3.5 m s1. The burning velocity under
these circumstances is about 0.4 m s1, for a stoichiometric
mixture considered here. In complex geometries relevant to real
life scenarios, however, turbulence is the dominant mechanism in
dening the ame speed. The waves propagating away from

Fig. 4. Development of producteair mixing between 100 ms and 5 ms. a) 100 ms (left) 500 ms (right) b) 1 ms (left) 5 ms(right).

496

I.G. Cullis et al. / International Journal of Pressure Vessels and Piping 87 (2010) 493e503
Secondary Combustion Energy Released

4.50E+10

4.00E+10

3.50E+10

2.50E+10

2.00E+10

1.50E+10

1.00E+10

5.00E+09

0.00E+00
0.00E+00

1.00E-03

2.00E-03

3.00E-03

4.00E-03

5.00E-03

6.00E-03

7.00E-03

8.00E-03

9.00E-03

Time (s)

Fig. 5. Secondary combustion energy release with time.

Over Impulse 1kg TNT

4.00E+03

Over Im pul se dynes/cm 2 - s

3.50E+03

After burn
No after burn

3.00E+03

CONWEP

2.50E+03
2.00E+03
1.50E+03
1.00E+03
5.00E+02
0.00E+00
0

0.5

1.5

2.5

Z scaled distance ( m)

Over Impulse 2nd pulse 1kg TNT

6.00E+03

Over I m pul se (dynes/ cm 2 - s)

Energy (Ergs/g)

3.00E+10

5.00E+03
After burn

4.00E+03

No after burn

3.00E+03

2.00E+03

1.00E+03

0.00E+00
0

0.5

1.5

2.5

Z scaled distance (m)

Fig. 6. Brick building response to an internal explosion from a small charge.

1.00E-02

I.G. Cullis et al. / International Journal of Pressure Vessels and Piping 87 (2010) 493e503

497

Fig. 7. Comparison of experimental pressure/impulse with GRIM3D predictions.

a laminar ame can reect from obstacles and boundaries and

hence interact with the ame. These interactions, together with
natural instability processes, lead to the formation of a turbulent
accelerating ame front. Turbulent ame speeds can be of the
order of hundreds of metres per second and under the correct
conditions can ultimately transition into a detonation wave, which
in the case of methaneeair travels with a velocity of 1800 m s1.

Our approach to gaseous explosions is to use a combustion code

to estimate the position of the ame front with time and hence the
volume of the gas consumed by the ame against time. The ratio of
this volume to the total volume of gas is then multiplied by the
specic combustion energy to provide the specic energy release as
a function of time, using the simple energy release algorithm
described above in GRIM.

Fig. 8. Predicted failure of aircraft skin.

498

I.G. Cullis et al. / International Journal of Pressure Vessels and Piping 87 (2010) 493e503

Fig. 9. Response of an airframe to internal blast.

Fig. 11. Murrah Building bomb and the effect of blast wave shape.

5. Structural response
In the case of structural response to internal and external
explosions we use the Lagrange code DYNA to study the dynamic
response of the structure, using the loading curves from GRIM. This
approach has been validated against a range of structures including
civilian aircraft, concrete buildings and concrete lined tunnels and
brick built buildings.
Nash, working with the Frazer-Nash consultancy, demonstrated
the importance of after burning in explosives in being able to
accurately reproduce the experimentally observed damage, Fig. 6,
and the experimental impulse measurements, Fig. 7, and the
subsequent structural response of a building to an internal explosion [9,10].
A similar approach was successfully applied to the problem of
the response of an airframe to an internal blast to explain how the
airframe failed, as illustrated in Fig. 8.
The images show the DYNA simulation of the failure of the skin
by the opening up of a crack. This crack then develops into a letterbox failure as shown in the right hand image of the gure. The
deformation of the structure can be clearly observed, as viewed
from the inside. The failure of this section of the load bearing skin of
the airframe forces the transfer of the in-ight loads to adjacent
panels, which are unable to sustain the additional load and also fail.

This process continues and results in the catastrophic loss of the

aircraft. Fig. 9 shows the experiment that validated the modelling
predictions.

5.1. Room explosion

In the case of an explosion in a conned volume, e.g. a room, the
oxygen supply is limited by its volume. This is illustrated by the
detonation of a 1 kg spherical charge of TNT inside a 4 m square
room, which is large enough that there is sufcient oxygen available to allow the secondary combustion to run to completion. The
GRIM simulation predicts this to occur some 38 ms after detonation. The strong shock wave reections from the walls of the room
at various times, Fig. 10 shows the pressure eld at 5 ms, increase
the mixing of products with the air and hence the rate of secondary
combustion. The complex producteair interface is shown in white
in the gure. The fuel is burned at a much greater rate releasing its
energy over a shorter time frame and hence producing greater
structural loading rates within the room. It should also be apparent
that the amount of oxygen available would also control the ultimate
quasi-static overpressure (QSOP) within the structure.
In free eld the scaling length is the charge size. In the room,
however, it is the volume of the room, since that determines the
amount of oxygen and hence the energy that can be released and is
independent of charge size.

5.2. Charge shape effects

Fig. 10. Pressure distribution within a 4 m room for 1 kg TNT at 5 ms.

The shape of the explosive charge also has an important role to

play, especially close in to it. This is illustrated in the Murrah building
explosion, Fig. 11, where in the aftermath of the explosion there is an
additional bite, circled in the image, taken out directly opposite the
site of the vehicle containing the bomb. Simulations of the explosion
predicted the formation of the non-spherical blast wave and product
cloud, which ingresses into the building to create the bite.
The effects of an explosion on underground structures, especially service pipes, are also of great interest. This in turn requires
an understanding of the pressure waves generated within the
ground by an explosive charge either air-burst or buried, and the
way they subsequently interact with buried structures. The size of
the charge and its coupling efciency with the ground are additional parameters of interest. A distributed charge has a greater
potential to cause increased levels of damage to a structure,
because of its ability to load more of it for a longer time.

I.G. Cullis et al. / International Journal of Pressure Vessels and Piping 87 (2010) 493e503

Fig. 12. Gas explosion velocity eld within the house.

5.3. Gaseous explosion

In 1999 a natural gas explosion occurred in the under-oor
space of a residential house, completely demolishing it and causing
signicant damage to the surrounding buildings.
A 272 mm diameter 6.5 mm thick ductile iron natural gas
(methane) main, laid in approximately 1974, through the front
gardens of the neighbouring houses and parallel to the roadway
operating in the medium pressure range (7.5  103 MPa to
0.2 MPa). The post explosion investigation of this medium pressure
main identied two visible holes in the pipe in the section of the
main that ran through the garden of the site of the explosion.
The holes in the pipe occurred in regions of graphitic corrosion. The
question was raised whether the explosion could have caused the
weakened pipes to fail.
The gaseous explosion within the house and the structural
response of the ground and the pipe was simulated in using the
combustion model described above.
The simulation helped visualise the effects of the waves generated by the explosion as they propagate through the house and the
under-oor void. The simulation identied a series of multiple
reections developing within the house and its foundations, which
interact as they propagate into the soil. The resulting pressure wave
that is then created propagates through the surrounding soil and
loads the gas main. The load on the gas main is therefore temporally and spatially complex. The shape of the house denes the
shape of the charge and hence the loading pulse on the pipe.

499

The gas above oor level effectively vented vertically through

the roof and sideways through the walls generating shock waves in
the air. Very little of the energy contained in these shock waves, as
they propagate away from the house is radiated into the ground to
load the gas main.
The increasing energy released as the explosion progresses,
coupled with the reection of waves inside the under-oor void
and from the foundations leads to further compressive waves that
propagate out into the soil. In addition the strong reections within
the foundations act as secondary sources for pressure waves. The
compressive waves also reect from the free surface as tensile
release waves increasing soil porosity.
As a result the interaction of the pressure waves in the soil with
the gas main is also quite complex. These pressure waves, on striking
the pipe, wrap around it creating pressure maxima on the opposite
side of the pipe. The waves will also be reected by the pipe as waves
with a cylindrical prole, which interact with the incident waves to
form a complex loading pattern. The pressure loading will excite the
various natural modes of vibration of the pipe. The pipe will therefore ex, twist and vibrate as a result of the loading. This load was
shown to be sufcient to cause the areas of graphitic corrosion, with
assumed 6 MPa corrosion product strength, to fail.
The velocity eld of the roof and on the vertical plane through
the house, at 140 ms, is shown in 3D and plane views in Fig. 12. The
plot clearly shows the blast wave in the air that is perceived by an
observer as the bang and the velocity distribution of the roof. The
motion of the pressure waves within the rooms generates the
velocity hot spots in the gure. The ejection of a window, shown
on the left of the lower image in the gure, caused by the blast wave
is also clearly shown.
The above examples demonstrate, given physically based material
models, the ability of modern hydrocodes to reproduce explosions
and their effects. This very detailed assessment of specic damage to
elements of the structure and its contents can be developed through
the use of these complex numerical techniques and then be used to
generate recommendations in the form of safe standoffs for structural, glazing or personnel damage from an explosion.
However, these simulations are complex and in 3D take some
considerable time to run. They readily lend themselves to neither
parametric studies of the explosion source, material or structural
variability nor construction methods. For these we need to employ
other modelling tools.
6. Vulnerability models
As well as the physical properties of different types of blast wave
and their interaction with structures we also need to understand
and assess their effects from a system vulnerability perspective.
Given the complexity of modern systems and structures we
therefore need a simpler approach to provide quick answers to the
questions Have I got a problem?, What if?, etc.
A simpler approach, based on analytic methods can provide
a very rapid assessment of the likely damage from a given blast
threat, for example for use in establishing personnel exclusion
zones around suspect devices. More detailed methods allow
assessment of specic events and locations of interest and advance
the understanding of damage mechanisms. Both of these
approaches have benets, but QinetiQ has introduced and developed the concept of a middle ground of vulnerability assessment
that offers a method that is versatile enough to be applied to
a generic building or infrastructure whilst being rapid enough to
provide results for a large number of possible threat scenarios, their
effects and possible mitigation.
We have spent many years developing a vulnerability strategy
and philosophy applied to military platforms in general and naval

500

I.G. Cullis et al. / International Journal of Pressure Vessels and Piping 87 (2010) 493e503

platforms in particular. Much of this research is directly applicable

to strategies for civil utilities and industrial infrastructure and it is
therefore useful to illustrate the level of maturity of our approach in
the sea systems area. In this context platform could be a train,
tunnel or civil infrastructure element.
Platform survivability is important for success in naval operations and essential for ensuring effective force projection. The
political, nancial and prestige implications of the loss of a vessel
would be serious and are equally applicable to public infrastructures
such as transport systems. The survival of Royal Navy ships hinges on
technologies, design standards and build techniques that enable
structures and systems to withstand battle damage. Demanding
modern system requirements prompt novel solutions and improved
methods in working practices. This process has demonstrated that
enhanced resistance to all damage mechanisms is most effectively
achieved by setting well thought out survivability targets.
The key to reducing warship vulnerability is the Vulnerability
Reduction Strategy. This is a 6-step strategy developed by the
Vulnerability Reduction section of UK MoDs Sea Technology Group
and QinetiQ. The 6 steps are:
1. Prevent catastrophic loss e ensure the vessel remains aoat
and largely intact following damage.
2. Reduce critical elements e identify and reduce those elements
which, if damaged or destroyed, would cause the entire function to be lost.
3. Concentrate critical elements e where a function includes one
or more critical elements these should be concentrated in as
small a part of the ship as possible to reduce the volume within
which a weapon has to detonate to affect them.
4. Separate redundant elements e redundant elements should
ideally be located so that the effects of a single hit cannot affect
both (and their associated sources of power and other services).
5. Hardening e if, with the preceding four steps taken, an
adequate level of survivability cannot be achieved, hardening of
ships systems and personnel should be considered.
6. Hit point management e weapons can be steered to hit
selected portions of the ship, which should be designed to
withstand the associated damage.
To these 6 points can be added use best design practice e for
example the use of low value spaces to shield high value spaces
should be routine practice during the layout process.
The principles of vulnerability assessment are well established.
To assess the ability of any vessel to perform its specied mission, it
is necessary to break down the top level functions covering oat,
move and ght aspects into a number of sub-systems such as
full or minimum power levels for move and Anti Air Warfare or

Command and Control functions for ght. Each of these can also
be broken down into a series of lower level systems (which may
affect more than one sub-system) dened so as to inform the
contractor as to what would be required to provide that system. The
functions and threats are combined in a Threat/Function table,
which species the required level of functional survivability for
each threat/function combination. A vulnerability assessment of
the probability of each function surviving any given threat is
compared to the targets to measure compliance. Systems can be
assessed and traded off to give affordable solutions that offer the
best performance possible.
6.1. SURVIVE
This naval platform vulnerability capability is contained within
the vulnerability and assessment tool SURVIVE, developed by
QinetiQ. It has been successful in identifying poor design features
before their development was taken too far (or even before they
formally become part of the design).
The overall vulnerability assessment will determine the probability
of retaining a particular system in each damage case. QinetiQ uses
SURVIVE to undertake this kind of study. The elements required for
each function to operate are laid out in the target description (as
shown for the SURVIVE Type 23 Frigate in Fig. 13), which is then
exposed to threat weapon(s) in any number of scenarios.
Under development since 1995, SURVIVE has been built using
QinetiQs long experience of vulnerability analysis methods. It can
analyse many damage mechanisms and actions applicable to
attacks on buildings and infrastructure, such as kinetic energy,
blast, fragmentation, which can arise from explosions as well as
tailored explosive devices such as shaped charge jets, and assess
effects in terms of re, smoke, casualties and personnel movement.
This capability envelope naturally covers the complete range of
existing and potential threat scenarios likely to be experienced by
any military or civilian system in its widest sense. One of its main
strengths lies in its assessment of systems functionality after
damage, whether from a successful weapon attack or from some
other cause such as accidental damage or simple component
failure. This assessment capability therefore can be easily adapted
to many situations such as complex systems in buildings, infrastructures or to other vehicles such as aircraft. Fig. 14, for example,
shows the use of SURVIVE in a study of a petrochemical plant.
When constructing a system SURVIVE can provide input at all
stages of the design process. At the concept stage, simple advice can
be given on general methods of survivability management, principles of vulnerability reduction and systems layout. Once the rst
layout has been produced, a rapid visual overview can highlight any
potential areas of concern. Optimizing layout at this early stage pays

Fig. 13. SURVIVE Type 23 Frigate model in external and transparent views.

I.G. Cullis et al. / International Journal of Pressure Vessels and Piping 87 (2010) 493e503

501

Fig. 14. Example of basic SURVIVE model of a petrochemical facility.

real dividends later in the design process, when subsequent layout

changes can be difcult and expensive to achieve. The contribution
of SURVIVE itself can start as soon as a basic layout is complete.
Critical spaces and the main equipment items are all that is necessary to produce an initial assessment of a targets vulnerability.
For SURVIVE to be effective it requires relevant information on
the structure of interest, its surroundings and the layout of internal
systems as well as a series of damage algorithms that readily
describe the effects of various events from re, explosions, impact
by fragments and projectiles over a wide range of velocities (up to
perhaps 10 km s1 in the case of shaped charge penetrators). Many
of these algorithms have been developed from experimental data,
analytical models, hydrocode modelling or a combination of all
three. There are also capabilities that can address the release and
ow through a structure of toxic gases and materials as well as
chemical and biological agents and radiation.
6.2. 3D laser scanning
The assessment of wide urban areas poses questions as to how
to rapidly model such environments at the required level of detail.
QinetiQ has access to a 3D laser scanning capability that allows just
such data gathering. The scanner consists of a portable head unit
that emits a near infra-red laser pulse through a window of 360 in

the horizontal and 90 in the vertical. The time taken for a pulse to
reect back from a solid surface is recorded. Thus a point in 3D
space is calculated, which when combined with all points calculated from a scan, forms a cloud of data of the scanned environment that can be viewed as a 3D point model. A standard scan will
take approximately 4 min to complete and will capture approximately 6 million points. Each point has 3D co-ordinates, RGB colour
and reectivity index. The 3D co-ordinate properties allow an
operator to measure the distance between any two points in the
scan, so the data can be treated as a high detail digital survey. The
reectivity values allow scans to be captured in poor lighting, or in
fog or smoke, where colour values cannot be recorded. A number of
scans can be processed and matched together into a global coordinate system to produce a complete 3D map or point model of
a structure or complex.
A 3D model can be built from the scan data tailored to the
required level of detail. A wire mesh model is constructed and
textured by mapping digital photographic images, captured from
the scene environment. An example is shown in Fig. 15.
Environmental effects can be controlled including the level of
lighting, position of the sun, rain fall, wind direction, smoke and fog
to name a few. The model can then either be rendered to photo
realistic quality, to produce still images or movies, or can be converted into a real-time model.

Fig. 15. SURVIVE models generated from LIDAR scan data.

502

I.G. Cullis et al. / International Journal of Pressure Vessels and Piping 87 (2010) 493e503

Fig. 16. PALETTE blast damage radius (left) and fragment trajectories (right).

Images, movies and real-time computer generated models have

been created for the following security related applications:
 Reconstruction of security and military incidents, viewed from
any perspective to aid an investigator to determine the cause of
events;
 Recording of events, including tracking and recording of
response team or emergency services movements during an
exercise or real life emergency;
 Visualisation of proposed engineering or defence measures;
 Assistance in brieng exercises. Intended actions and paths for
a response team can be visualised in 3D before the start of an
exercise;
 Provision of visual understanding and explanation of potential
threat and outcomes, for example showing the effect of a bomb
blast on a sensitive building;
 Analysis of existing or proposed CCTV camera positions. The
dimensionally correct 3D model can be viewed from any
location allowing the visual eld of existing or proposed CCTV
cameras to be assessed, from which the need for additional
cameras can be identied;
 Training of emergency services and response teams for emergency scenarios in sensitive locations, for example airports,
underground stations, paramilitary buildings;
 Brieng of emergency services and response teams prior to an
incident response;
 Incident response assistance. Using tracking devices a response
team can be monitored and directed by an operator from
a secure location. The operator can view the 3D model clearly,
without the hindrance of smoke or other obstacles that might
be present in the real world environment, and direct the
response team along the most appropriate route;
 Crowd control at public events. For example, during an emergency at a football stadium, an interactive 3D model could be
used to plan the safest evacuation routes.

7. PALETTE
QinetiQ, under contract to MOD, is also developing an alternative concept to vulnerability modelling called PALETTE (Penetration
and Lethality Effectiveness Toolbox for Target Engagement).
PALETTE is a suite of vulnerability tools representing different
effect, e.g. blast, fragmentation, penetration, cratering, each with
a graphical User Interface which can be tailored to the individual
user requirement. The aim of PALETTE is to bring together and/or
create a number of vulnerability/lethality tools matched to
a specic task. Whilst the research has been geared to support the
development of weapon systems, in particular for the defeat of

buildings and structures, it has obvious non-defence applications

where blast and other explosion effects are of concern.
In terms of explosions and their effects, PALETTE includes blast
tools which can assess: pressure, breach, crater formation, and
quasi-static pressure. The pressure routine provides pressure-time
history, impulse, time of arrival and duration as a function of range
and explosive mass. These can be matched directly to the performance of the specic explosive material. The breach routine predicts
the damage to a reinforced concrete wall from an explosive mass as
a given range. The crater routine provides crater dimensions in
concrete and soil. Finally the quasi-static pressure routine predicts
the QSOP within a room, which can include a vent area.
Blast damage radius (BDR) can be calculated, based on a user
dened pressure of interest. The region where this blast pressure is
exceeded can be displayed visually on a 3D model of the structure, as
shown in Fig. 16.
Tools within PALETTE not only predict damage to individual
components of a target, but also assess the structural collapse of
a building, based on explosive mass and position, including the
shear and exural response of beams or slabs to blast and impact.
PALETTE also provides a fragment modelling tool which can be
used to rapidly predict the fragment trajectories and associated
damage generated by an explosion, Fig. 16.
Target data denition can be captured in a number of ways. For
example building models can be developed rapidly using PALETTEs
on line target generation interface, which includes table entry and
drag and drop. Alternatively, a similar approach can be used to
that described earlier for SURVIVE, generating 3D models from
laser scan data.
PALETTE thus provides a sophisticated toolbox that can be used
as a ready reckoner to assess the likely gross structural effects in
response to an explosion. It naturally complements the SURVIVE
capability and the hydrocode modelling in a hierarchical structure.
8. Conclusions
This paper has described the approach adopted by QinetiQ to
assess the consequences of the explosive event by using a hierarchical approach from detailed numerical simulation and experiment and their integration into analytic models. It has shown that
by combining complex numerical simulation capabilities in
blastestructure interaction with either SURVIVEs or PALETTEs
sophisticated damage assessment and functionality evaluation,
a capability is formed that has a wide range of potential uses, which
links the many stages of response planning to an attack or more
general emergency. This includes designing survivable infrastructures and planning protective measures, through training of staff
and responders to handling of actual events.
The availability of this hierarchical approach for models capable
of dealing with a structural description, functional items and

I.G. Cullis et al. / International Journal of Pressure Vessels and Piping 87 (2010) 493e503

personnel within them allows a whole range of What If.?

questions to be asked and many options for varying building
structure and infrastructure design to be investigated, moving the
tool from a design aid to one that can consider planning, training,
audit and response factors. Wrapping these abilities into an allencompassing package can provide a comprehensive solution to
many of an organisations requirements in exercising and protecting its business and employees from explosions and their effects.

References
[1] CONWEP. USAEWES/SS-R. Vicksberg: US Army Engineering Waterways
Establishment; 1991.
[2] Church PD, Goldthorpe BD, Andrews T. A review of constitutive model
development within DERA. ASME PVP; 1999:113e20.

503

[3] Gould PJ, Goldthorpe BD. A path-dependent constitutive model for gilding
copper. J Phys IV France 2000;10(9):39e44.
[4] Goldthorpe BD. A path dependent model for ductile fracture. J Phys IV 1997;
7(C3):705e10.
[5] Cullis IG, Hinton M, Gilbert S, Church PD, Porter D, Andrews T, et al. Towards
predictive modelling of concrete. Int J Impact Eng 2008;35:1478e83.
[6] Cullis IG, Gilbert J, Greenwood P, Pang T. Blast wave scaling and its importance in structural loading. 16th Int Sym on Ballistics, pp. 173e182, San
Francisco, USA, 1996.
[7] Huntington-Thresher W, Cullis IG. TNT blast scaling for small charges. 19th Int
Sym on Ballistics, pp. 647e654, Interlaken, Switzerland, 2001.
[8] Kuhl AL, Ferguson RE, Oppenheim AK. Gas dynamic model of turbulent
exothermic elds in explosions, advances in combustion science. Prog
Astronaut Aeronaut 173, pp. 251e260.
[9] Nash MA, Dorn MR. The numerical prediction of the collapse of a complex
brick building due to an internal explosion, structures under extreme loading.
ASME PVP Conf; 1997.
[10] Dorn MR, Nash MA, Kesby A, Anderson G. Conict and the modelling response
of complex masonry structures to internal explosions. 20th Int Sym on
Ballistics, pp. 761e768, 2002.