AdvStructEng HaoHaoLiandChen-1

Article
Advances in Structural Engineering

1–31
Review of the current practices in Ó The Author(s) 2016
Reprints and permissions:
blast-resistant analysis and design of sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1369433216656430
concrete structures ase.sagepub.com
Hong Hao1, Yifei Hao1, Jun Li2 and Wensu Chen1
Abstract
In contemporary society, industrialization and rising of terrorism threats highlight the necessity and importance of structural protec-
tion against accidental and intentionally malicious blast loads. Consequences of these extreme loading events are known to be cata-
strophic, involving personnel injuries and fatalities, economic loss and immeasurable social disruption. These impacts are generated
not only from direct explosion effects, that is, blast overpressure and primary or secondary fragments, but also from the indirect
effects such as structural collapse. The latter one is known to be more critical leading to massive losses. It is therefore imperative to
enlighten our structural engineers and policy regulators when designing modern structures. Towards a better protection of concrete
structures, efforts have been devoted to understanding properties of construction materials and responses of structures subjected to
blast loads. Reliable blast resistance design requires a comprehensive knowledge of blast loading characteristics, dynamic material
properties and dynamic response predictions of structures. This article presents a state-of-the-art review of the current blast-resistant
design and analysis of concrete structures subjected to blast loads. The blast load estimation, design considerations and approaches,
dynamic material properties at high strain rate, testing methods and numerical simulation tools and methods are considered and
reviewed. Discussions on the accuracies and advantages of these current approaches and suggestions on possible improvements are
also made.
Keywords
blast, concrete structures, design, material, numerical simulation, testing methods
Introduction dynamic material properties. The current approaches

in design practice are mainly based on single-degree-
Structural protection against explosive loads draws of-freedom (SDOF) simplification. Besides the simpli-
more and more attentions due to the increasing num- fied approaches specified in various guidelines, experi-
ber of accidental explosions and terrorist bombing mental tests and numerical simulations are often
attacks on important governmental and civilian facili- needed to more reliably quantify the structural
ties. Blast loading has the natures of extremely short responses. Experimental testing including field and
duration but high intensity, which results in different laboratory tests is a direct method to investigate the
structural responses as compared to those from static behaviour of structures or structural elements. The
and less intense dynamic loads such as earthquake, physical testing results can not only demonstrate
wave and wind loads. Therefore, the analysis and
design of structures against static and low-rate
dynamic loads cannot be directly applied onto the 1
Tianjin University and Curtin University Joint Research Center of
design of structures subjected to blast loads. Moreover, Structural Monitoring and Protection, Centre for Infrastructural
structural response and material behaviour under blast Monitoring and Protection, School of Civil and Mechanical Engineering,
Curtin University, Bentley, WA, Australia
loads are usually nonlinear, involving complex stress 2
School of Civil and Environmental Engineering, University of
states due to stress wave propagation in the structure, Technology Sydney, Ultimo, NSW, Australia
and are time dependent, which make some simplified
approaches currently used in design practice not neces- Corresponding author:
sarily lead to reliable structural response predictions. Yifei Hao, Tianjin University and Curtin University Joint Research Center
of Structural Monitoring and Protection, Centre for Infrastructural
To achieve a reliable design and analysis of concrete Monitoring and Protection, School of Civil and Mechanical Engineering,
structures subjected to blast loads, it is important to Curtin University, Kent Street, Bentley, WA 6102, Australia.
understand the characteristics of blast load and Email: [email protected]
2 Advances in Structural Engineering
actual structural behaviours straightforwardly but also critical to understand the blast effects on the structures
be used to validate the fidelity of numerical model. and also the blast-induced structural responses and
However, physical test is often not easy to be carried damages.
out, not only because it is often restricted due to safety
consideration but also because it requires expensive
and highly specialized equipment and instrumentation
Free-field HE explosion
skills to capture and record the very fast blast loading The amplitude and distribution of the blast loads on a
and structural responses. With the advancement of structure are a function of the type of the explosive
computer technology and computational mechanics material, weight and shape of the explosive, distance
techniques, a number of commercial software have and location of the explosive from the structure and
been developed and widely used to conduct numerical the interaction of the blast wave with the ground and
simulations of concrete structures subjected to blast the structure (US Department of Defense, 2008). When
loadings. It is a common practice now to use verified describing blast effects, the relevant parameters are pri-
high-fidelity numerical simulations to predict struc- marily dependent on the amount of energy released in
tural responses subjected to blast loadings. The numer- the detonation (US Department of Defense, 2008). For
ical simulation results can supplement physical testing different HE materials, it is a common practice to con-
data, which are difficult to obtain. They also allow for vert the charge weight W to an equivalent weight of
more detailed observations and recordings of the struc- Trinitrotoluene (TNT). There are many ways to deter-
tural responses and damage which are not possible in mine the TNT equivalence, such as heat of combustion
physical tests. However, numerical simulation can be (related to quasi-static pressure in a confined space),
used only after the model is verified to give reliable heat of detonation (related to peak pressure), detona-
predictions. tion energy and detonation and afterburn energy mea-
This article presents a state-of-the-art review on the sured by blast pressure. These approaches may give
design and analysis of concrete structures against blast different TNT equivalence estimations of explosives.
loads. The characteristics and estimation methods of Table 1 gives some indicative TNT equivalencies of
blast loads are covered in section ‘Blast effects’. selected explosives for pressure and impulse. The blast
Section ‘Current design approaches’ briefly discusses effects are then given as a function of the dimensional
the current design approaches and codes of practice. distance parameter (scaled distance) Z = R/W1/3, in
The properties and popularly used dynamic material which R is the standoff distance from the detonation
models of concrete are discussed in section ‘Concrete (US Department of Defense, 2008). Until now, the
dynamic material properties and models’. Methods for most extensive data about blast load parameters are
blast testing are reviewed in section ‘Blast testing for bare spherical TNT airbursts and bare hemispheri-
methods’. Section ‘Numerical simulation of blast effect cal TNT surface bursts. The blast parameters can be
and structural response’ discusses various numerical scaled and plotted versus the scaled standoff distance
methods and techniques, and conclusions are given in in the form of a series of curves (US Department of
section ‘Conclusion’. Defense, 2008). It is worth noting that these curves
start at a scaled standoff distance about 0.06 m/kg1/3,
and this is because blast loadings from very close-in
Blast effects
detonations are extremely severe and very hard to
Explosions are associated with a rapid release of a measure.
large amount of energy. Based on their nature, explo- As one of the most extensively studied explosion
sions are usually categorized as natural, chemical, elec- type, a typical air blast pressure profile is shown in
trical and nuclear. In contemporary society, the Figure 1. Upon detonation, a pressure front propa-
explosion hazards in protective design are normally gates radially into the surrounding atmosphere as a
referred to the blast overpressure generated by chemi- strong shock wave. At the blast wave arrival time tA,
cal high explosives (HEs) acting on the structures and the pressure increases nearly instantaneously to a peak
human beings. With rising of terrorism activities in value of overpressure, Pso, over the ambient pressure,
recent decades, threats from improvised explosive Po. The pressure then decays to ambient level at time
devices such as vehicle-borne improvised explosive tA + td and then decays further to a negative pressure
device (VBIED) and suitcase bomb are attracting Pso2 before eventually returning to ambient condi-
increasingly more attentions for structure and people tions at time tA + td + td2. The quantity Pso is usu-
protection. These bomb attacks impose threats on per- ally referred to as the peak overpressure or incident
sonnel and assets not only through direct overpressure overpressure.
but also through destroying the structures that shield Design guidelines for concrete structures usually
over the personnel and instruments. It is therefore recommend utilizing only the positive phase of the
Hao et al. 3
Table 1. Indicative TNT equivalencies for common explosive around 45° due to Mach effect. As shown in Figure 2,
materials. waves reflecting off rigid ground can catch and merge
with the incident wave to create a single wavefront
Type of explosive Pressure Impulse
called a Mach front. The point at which the incident
Comp. B 1.11 0.98 and reflected waves merge is known as the triple point.
Comp. A-3 1.09 1.08 The angle of incidence is the angle between the normal
ANFO 0.9 0.9 to the shock wave’s direction of travel and the normal
Comp. C4 1.37 1.19 surface vector of the object with which it interacts.
Cyclotol (70/30) 1.14 1.09
HBX-1 1.17 1.16
Prior to the joining of the incident and reflected wave,
HBX-3 1.14 0.97 two independent pressure peaks exist, while after the
H-6 1.38 1.15 merge there is only one single incident blast overpres-
Minol-2 1.20 1.11 sure (US Department of Defense, 2008).
PETN 1.14 1.15 During the 1950s–1980s, the blast wave parameters
TRITONAL 1.07 0.96
Pentolite 1.42 1.00
of conventional HE materials have been widely investi-
Tetryl 1.07 1.07 gated in a number of studies (Baker et al., 1983; Brode,
TNT 1.00 1.00 1955; Henrych and Major, 1979; Kingery et al., 1984;
Mills, 1987). For practical design purpose, peak over-
ANFO: ammonium nitrate-fuel oil; HBX: high blast explosive; PETN:
pressure Pso and peak reflected overpressure Pr as well
pentaerythritol tetranitrate.
as scaled impulses can be read from UFC 3-340-02 (US
Department of Defense, 2008) as shown in Figure 3. It
is clearly noted that hemispherical explosion on the
ground yields higher blast pressure and impulse than
spherical explosion in the air. For a perfect reflecting
surface, the shock waves from ground explosion would
be instantaneously reinforced by the reflected waves.
This would give the effect of twice the weight of the
actual charge. As the ground is not a perfect rigid
reflector, a ground reflectivity value greater than 1 but
less than 2 would be reasonable.
A comprehensive investigation towards the validity
of different sources including empirical relations,
attenuation curves and computer code for predicting
blast loads was conducted by Low and Hao (2001). It
was demonstrated that different charts and different
empirical formulae available in the literature give dif-
ferent blast loading (peak, impulse and duration) pre-
Figure 1. Typical blast pressure profile (US Department of
Defense, 2008). dictions, especially in the range with scaled distance
Z = R/W1/3 less than 1 m/kg1/3 or larger than 10 m/
kg1/3. These variations in blast loading estimations cer-
blast load by assuming that the negative phase is nor- tainly affect structural response predictions. In a later
mally much weaker and does not affect typical con- study, Hao et al. (2010a) demonstrated that statistical
crete structures. However, it should be noted that if variations in the blast loads have more significant
the positive phase of blast load is not violent enough influence than the uncertainties in the structural para-
to cause significant failure, negative phase of blast load meters on the probabilistic structural responses. For a
should be treated carefully by designers. It might affect more credible prediction of structural responses, the
the overall structural response and damage as in the variations in blast loading predictions from different
case of glass window response to blast loads that approaches should be taken into consideration.
breakage of glass panels and sharp and high-velocity
fragments might be induced by the negative phase of
blast load (Krauthammer and Altenberg, 2000; Wei Industrial accidental explosion
and Dharani, 2005). Blast loads can also be generated from industrial acci-
Upon interacting with the structural surface, the dental explosions such as gas explosion, vapour cloud
incident overpressures are amplified by the reflection explosion (VCE), pressure vessel explosion, condensed
factor. The reflection factors are typically the greatest phase explosion and dust explosion (American Society
when normal incidence occurs except at incident angles of Civil Engineers (ASCE), 2010). Gas explosion or
Figure 2. Unconfined air burst showing formation of Mach front (US Department of Defense, 2008).
Figure 3. Shock wave parameters for spherical TNT explosions in free air (US Department of Defense, 2008): (a) spherical TNT
explosion in free air at sea level and (b) hemispherical TNT explosion on the surface at sea level.
Hao et al. 5
VCE, as a common hazard in the petrochemical indus- or partially confined internal explosions. For the near-
try, usually generates a blast wave with lower peak field explosion or confined explosion, computational
overpressure but longer duration and higher impulse fluid dynamics (CFD) approaches such as FLACS and
than a blast wave generated from HE with the same AUTODYN can be used to predict overpressure. In
energy release. There are two types of blast waves, that structural design practice, blast overpressure caused by
is, pressure wave and shock wave. The pressure wave gas explosion is usually provided by the facility owner
gradually rises to the peak overpressure followed by or from commonly used criteria (ASCE, 2010). For
gradual decay, while shock wave resulting from con- example, two simplified triangular blast overpressures
densed detonations rises to its peak overpressure are considered in the design of structures (Chen et al.,
almost instantaneously. 2015b) 30 m away from a VCE hazard with the
To calculate the blast overpressure parameters, side-on overpressures ranging from 10 to 103 kPa and
three major empirical methods are in use including positive phase duration ranging from 20 to 200 ms, as
TNT equivalence method (TEM) (US Department of specified in the criteria SG-22 (Manufacturing
Defense, 2008), multi-energy method (MEM) from Chemists Association, 1978) and Chemical Industries
TNO 1985 (Van den Berg, 1985) and Baker–Strehlow Association (CIA) (Chemical Industries Association,
method (BSM) (Baker et al., 1983). The strength and 1992).
weakness of the empirical methods are summarized in
Lea and Ledin (2002). Among them, TEM using
pressure–distance curves is most intensively used in Explosions in complex environment
practice nowadays to predict peak overpressure from The current practices estimate blast overpressures on
gas explosion due to its ease of use. The mass of TNT structures based primarily on empirical formulae
equivalence is estimated from the mass of hydrocarbon which are derived from free-field test data that ignore
in the cloud. The method, however, has some draw- wave interaction with structures and surrounding envi-
backs such as requiring a non-unique yield factor ronment. For structures with limited dimension, wave
based on experiment and poor estimation of weak gas diffraction occurs and the dimension and shape of
explosion. The overpressure estimated from TEM structures influence the amplitude and distribution of
gives too conservative predictions, especially for near- blast loadings. Furthermore, when obstacles exist in
field explosion. As suggested by Bjerketvedt et al. the propagating path of blast wave, blast loads on
(1997), TEM is not recommended to predict the gas structures behind the barrier are very different from
explosion overpressure when the overpressure is higher those measured in free-field tests. When confined
than 1 bar (i.e. 101.3 kPa). TEM can predict peak explosion takes place, the shock pressure on structures
overpressure for far-field explosion which is over 30 m is much higher, followed by the gas pressure with
from the gas explosion centre. MEM and BSM are lower amplitude but much longer duration due to slow
more sophisticated methods than TEM. They provide release of energy from limited venting area. In reality
a set of curves based on flame speed or explosion with more complex environment such as city landscape
strength. The dimensionless parameters are selected with dense buildings and streets that form both
from the curves to determine the actual overpressures. complex geometry and confining environment, predict-
MEM can be used by engineers based on a selected ing blast wave propagation and blast load can be very
severity from 1 to 10. The difficult part of using MEM complicated. Progress of recent research works
is to decide the parameters such as explosion source targeting these issues is briefly reviewed.
strength, degree of confinement and the combustion
energy. The choice of charge size and strength has to
be based on the experimental data or conservative Blast wave interaction with structure and blast barrier. When
assumption. MEM has some limitations, such as there are obstacles between the explosion and the tar-
uncertainties in the prediction of overpressures for get structure, blast wave–obstacle interaction must be
large-scale explosion as the method is derived based on considered to estimate the blast load on the structure.
scaled experiments; the method does not account for This could be a very challenging problem as even the
the directional effects of explosions due to localized blast wave interaction with a standalone column is a
confinement and congestion. BSM is a practical alter- complicated process. When impinging the rectangular
native, which considers flame speed and fuel reactivity column, a portion of the blast wave is reflected by the
and the geometry of confinement. front surface while the remainder diffracts around it.
In general, the three empirical methods are reliable In the process of wave diffraction, the incident wave-
for the calculation of overpressure from far-field front is weakened locally and closes in behind the col-
explosion and are in general inappropriate to predict umn, and a number of trailing vortices are formed (Shi
blast pressure from near-field explosion and confined et al., 2007). The load on structural columns with
different geometries can vary significantly in both However, it should be noted that damage of conven-
amplitude and duration as simulated by Gebbeken tional solid protective barrier not only results in loss of
and Döge (2010) using AUTODYN. In the simula- its functionality but also creates secondary fragments
tions, the structural columns were modelled as rigid, that lead to debris hazards. Therefore, blast barriers
and remapping technique was used. normally need to be designed very strong with consid-
The current structural protection technology adopts erable volume and mass to have high stiffness and duc-
the significant influence of obstacles on blast wave tility to resist expected blast load. Besides the high
propagation and distribution. Solid barriers are nor- construction cost, such barriers are not necessarily suit-
mally used to provide standoff distances to protect able for urban areas from the view of city planning and
structures from external explosion as well as mitigate aesthetical appearance.
the blast load. Zhou and Hao (2008) studied the pro- Blast wave diffracts when interacting with obstacles.
pagation of blast wave with a barrier in between the Considering a series of identical obstacles/columns
explosion and structure and distribution of blast load evenly spaced side by side, diffracted waves from adja-
on the structure using AUTODYN. The barrier and cent columns could cancel each other, which would
structure were modelled as rigid, and remapping the significantly weaken the wave amplitude. Ideally, the
hemisphere blast wave from two-dimensional (2D) to wave amplitude due to superposition would be 0 if two
three-dimensional (3D) was used. The effectiveness of same frequency waves with opposite phase interact
barriers for blast load mitigation was examined, and with each other (Zong et al., 2015). The results from a
empirical formulae were derived for estimating the recent experimental study demonstrated that placing
reflected pressure on a structure behind a barrier. The cylindrical bars between the target structure and a
blast wave propagation and load reduction on struc- charge resulted in faster blast wave attenuation and
ture are illustrated in Figure 4. mitigated the damaging effects (Niollet et al., 2015).
Figure 4. Blast wave interaction with protective barrier and comparison of reflected pressures (Zhou and Hao, 2008): (a) pressure
wave propagation and (b) reflected pressure histories.
Hao et al. 7
Figure 5. Interaction of blast wave with column matrix with different shapes (Chaudhuri et al., 2012).
Chaudhuri et al. investigated shock wave attenua- suggested due to the balanced consideration of net
tion with matrix consisting of rigid obstacles with blast load on fence barrier and blast pressure
different geometries as shown in Figure 5 using an mitigation.
in-house compressible Navier–Stokes flow solver. It
was found that reverse triangular columns performed
best in blast wave attenuation, and staggered arrange- Internal explosion – partially confined explosion. The major-
ment of column matrix could bring further enhanced ity of concrete structures are normally designed with
effectiveness (Chaudhuri et al., 2012). The performance frangible elements such as windows and doors. When
of column matrix depends on the net blast load on the internal explosion occurs, frangible elements provide
columns, which is related to the shape of the column limited venting area to leak energy. The initial wave is
cross section. In the design of such type of protective amplified by the inner surfaces of the structure and
barriers, attention must also be paid to the integrity of vented to the atmosphere through limited opening after
individual column, especially when hazards with multi- a finite period of time. Besides the shock wave, gas
ple explosions are considered. Therefore, properly pressure due to confinement of the detonation prod-
arranging the column layout and choosing a column ucts consisting of the accumulated temperature and
cross section that minimizes the net blast load as well gaseous products builds up as a quasi-static pressure
as most significantly attenuates the blast pressure with a significantly longer period (US Department of
behind the column matrix should be considered simul- Defense, 2008).
taneously in the design of this type of blast barriers. In US Department of Defense (2008), the blast load
To achieve the balance between reducing the net on the structures due to partially confined explosion is
blast load on elements of matrix barrier and achieving idealized by shock pressure and gas pressure, and both
satisfactory blast wave attenuation, Zong et al. (2015) decrease linearly with time but with different peak val-
examined the effects of geometry, spacing, dimension ues and duration as illustrated in Figure 7. The peak
and separation distance and number of layers of fence- reflected shock pressure and duration can be estimated
type barriers (Figure 6) using AUTODYN. The col- based on the approaches described in section ‘Free-
umns were also assumed as rigid. Based on parametric field HE explosion’. The peak gas pressure can be esti-
simulations, columns with circular or triangular (with mated according to loading density, namely, the charge
angle facing the explosion) cross sections were weight to free volume from the curve given in Figure 8.
Figure 6. Blast wave interaction with fence-type barriers with columns of different geometries and spacing (Zong et al., 2015).
and Rose, 2006). Intuitively, the characteristics of

blast wave can become much more complicated as
the number of buildings interacting with the blast
wave increases. When explosion occurs in a street,
the surrounding buildings form geometrical con-
straints and confining effects take place. It was
demonstrated that the layout and width of street and
height of buildings all affect the blast wave character-
istics. On the other hand, apart from the interaction
of blast wave with buildings in the urban area, the
Figure 7. Shock and gas pressures (US Department of complexity of characterizing blast load can be more
Defense, 2008). pronounced when fac xade failure of the structures
under blast load is taken into consideration.
Consequently, empirical relations to predict blast
load for structural protection in urban landscape
might become invalid, and CFD analyses may be the
only satisfactory approach for such blast loading
predictions (Smith and Rose, 2006).
Current design approaches

Structural protective designs against blast loads usu-
ally have two design considerations, namely, the design
of structural members to be blast resistant and the
design of structures for integrity after blast scenario,
Figure 8. Chart to determine peak gas pressures (US
that is, prevention of structural progressive collapse.
Department of Defense, 2008).
Blast loads with certain magnitude and distribution
can cause damages to structural components, and
The scaled impulse of gas pressure can be obtained these components which are usually in close proximity
from charts in US Department of Defense (2008), to the explosion can be properly strengthened to resist
based on which the impulse and duration of the gas the given blast loads. However, in practice, reliable
pressure can be derived. prediction on the explosion location and blast loads’
magnitude and distribution are usually impossible. It
is therefore impractical and uneconomical to design all
Explosion in complex geometry/cityscape. For scenarios the key and load-carrying structural components as
with detonation in urban landscape, on one hand, the blast resistant. In such cases, structures should be
interaction of blast wave with buildings might result in designed to meet the integrity requirements with suffi-
significantly higher overpressure and impulse (Smith cient strength and redundancy so that loss of one or a
Hao et al. 9
few components does not lead to a disproportionate For other loading conditions which are not defined in
collapse of the remaining structure. UFC 3-340-02, the equivalent load must be calculated
In blast-resistant design analysis, three approaches, with the deflection shape corresponding to the assumed
namely, equivalent static loads (ESLs), SDOF method deflection shape of the structure under the same load-
and numerical simulations, are commonly adopted. ing distributions.
ESL approach simplifies the dynamic analysis to an
equivalent static one. It is a straightforward approach
that is easy to be applied (Kang et al., 2001), but it Response mode and deflection shape. The response mode
does not explicitly consider the dynamic effects nor the of a structural component under blast loads can be
inertial effects. Therefore, the ESL approach is suitable very different from that under static loads depending
only when the blast loading duration is longer than the on the blast loading amplitude and duration. If the
fundamental period of the structure so that the ratio of the blast loading duration to the structural
response can be modelled as quasi-static. Numerical natural vibration period is very small, the structural
methods have been developed to reliably predict struc- response is mainly governed by stress wave propaga-
tural responses and damage. However, even with mod- tion; when this ratio is relatively large, the response is
ern computer power, numerical simulation of dynamic; when the blast loading duration is longer
structural response under blast loads is still time- and than structural vibration period, the response is quasi-
resource-consuming which is mainly due to the high- static and mainly governed by structural stiffness. It is
intensity short duration nature of blast loads. noted that for the first instance, no global response is
Moreover, the accuracy of numerical predictions activated, and SDOF analysis, which is mainly derived
depends on the availability of detailed dynamic mate- by global response, is no longer applicable. For the
rial models, which are sometimes not available. It also other instances, the structural response could be gov-
requires profound understandings of the computa- erned by direct shear, diagonal shear or flexure
tional mechanics, damage mechanics and structural response. In such cases, deflection shapes for SDOF
dynamics. Detailed discussions about the material analysis should be carefully selected. Otherwise, the
models and the numerical methods handling dynamic SDOF analysis cannot yield reliable predictions. In
structural response under blast loads will be given later general, the larger the ratio of the blast loading dura-
in this article. SDOF analysis has been the most com- tion to the structural natural period, the more
monly used approach for designing and analysing likely the structural response will be governed by
structural components subjected to blast loads and is flexural mode.
discussed in detail as follows. Furthermore, because of the short duration of the
blast loads, the peak response usually takes place in
the free vibration phase. This makes the selection of
SDOF approach and its limitations deflection shape for SDOF analysis even more diffi-
SDOF method was initially introduced by Biggs cult. During the blast loading phase, the response
(1964). It idealizes the structural component as an mode can be localized and during the free vibration
SDOF system with an equivalent stiffness, mass and phase, the response quickly changes to global mode as
dynamic load based on assumption of a structural structural vibration continues. For such situations, the
deformation shape. Despite SDOF analysis widely SDOF method developed using deflection shape based
adopted in blast design guidelines such as UFC 3-340- on either the local element or global response mode
02 (US Department of Defense, 2008), it does not will not give a complete representation of the true
necessarily lead to accurate structural response predic- characteristics of structural responses.
tions for all scenarios. A comprehensive discussion on
the limitations of SDOF analysis is given by Hao
Other limitations. Typically, the mass and load factors
(2015). Here, some major limitations of SDOF method
given in the current design guides such as UFC 3-340-
are discussed.
02 for deriving the equivalent SDOF system are
obtained by assuming flexural response mode, and
Loading assumption. In SDOF method, the load factor is they cannot give good predictions of structural
derived by setting the external work done by the equiv- responses when the responses are governed by other
alent load on the equivalent system equal to the exter- response modes. Moreover, the SDOF system is diffi-
nal work done by the actual load on the actual element cult to handle responses including compression and
deflecting to the assumed deflected shape. The load tension membrane effects, arching from axial loads
factors given in the charts in UFC codes are derived and component response that includes secondary
based on uniform or concentrated loading conditions. moments from axial loads, that is, ‘P-delta’ effect. It is
Table 2. Summary of design manuals and reports.
Guidelines Application scope and notes
ASCE Manual 42 (Agbabian, 1985) Design of structures to resist nuclear weapon effects
Spall Damage of Concrete Structures, WES/TR/ Compiled test results and empirical charts are provided to assess possible
SL-88-22 (McVay, 1988) spall damage under free-air or contact explosions generated by cased or bare
explosives
ASCE Structural Design for Physical Security Covers a range of topics including threat assessment, load estimations under
(Conrath et al., 1999) typical attacks, structural response and the corresponding design, and so on. It
also contains the design of windows and doors under blast loads
Unified Facilities Criteria Manual (UFC 3-340-01) Design of hardened structures to resist the conventional weapons’ effects
(US Department of Defense, 2002)
FEMA 428 Manual (Federal Emergency Defines minimum standoff distances to ensure structural safety under
Management Agency (FEMA), 2003a) terrorism attacks
FEMA 426 Manual (Federal Emergency Provides the starting points for people who wish to learn how to design
Management Agency (FEMA), 2003b) protective structures
Unified Facilities Criteria Manual (UFC 4-023-03) Provides guidelines to protect buildings from progressive collapse
Unified Facilities Criteria Manual (UFC 3-340-02) This manual contains step-by-step analysis and design procedures, including (1)
(US Department of Defense, 2008) blast loading predictions, (2) principles of nonlinear dynamic analysis and (3)
reinforced concrete and structural steel design
ASCE design of blast-resistant buildings in Design of industrial blast-resistant structures, especially on petrochemical
petrochemical facilities (ASCE, 2010) facilities
ASCE/SEI 59-11 (ASCE, 2011) Provides the guideline for the design of structures to resist the effects of
explosions from external and internal sources
Unified Facilities Criteria Manual (UFC 4-010-01) Design of structures to meet minimum antiterrorism standards for buildings
GSA Alternate Path Analysis and Design Guidelines to protect buildings from progressive collapse. Both tie force and
Guidelines for Progressive Collapse Resistance alternate path methods are considered
(GSA, 2013)
typically non-conservative to neglect the P-delta effect The other widely used manual is ASCE/SEI 59-11
for blast-resistant design. (ASCE, 2011). In this design guide, element damage
levels are classified as superficial, moderate, heavy and
hazardous as shown in Table 4. Compared to UFC 3-
Design criteria and guidelines 340-02, the angle of rotations corresponding to differ-
The current design approaches in practice are mainly ent damage levels is slightly conservative. Moreover,
based on the procedures given in the design manuals for superficial damage level, instead of the angle of
or reports for blast-resistant structures. A few design rotation, ductility ratio limited to 1.0 (limited to elastic
guides and some textbooks are available that give the deformation) is used. Another difference between these
procedures for analysis and design of blast-resistant two design guides is that double-reinforcement design
structures, as summarized in Table 2. is always required in UFC 3-340-02 for resisting
Among all these design manuals and reports, UFC rebounds while in ASCE/SEI 59-11 single- and double-
3-340-02 appears to be the most widely used guideline reinforced elements are separately considered.
for design and analysis of structures with regard to Dynamic increase factor (DIF) for concrete and
explosive safety applications (US Department of reinforcing steel under high strain rate is considered in
Defense, 2008). One reason for its widespread use is both UFC 3-340-02 and ASCE 59-11. While UFC 3-
that it is approved for public release with unlimited 340-02 lists DIFs for far-range and close-in explosion
distributions. This manual defines ‘close-in’ and ‘far’ scenarios, in ASCE/SEI 59-11 only DIFs for far-range,
design ranges for predicting the response mode of that is, elements are more likely to have flexural fail-
structures. The response limits for structural compo- ure, are provided. For design of RC structures sub-
nents such as slabs and roofs are given as support rota- jected to close-in explosion, it is suggested by ASCE/
tions. The damage levels are divided into three types of SEI 59-11 that explicit dynamic finite element (FE)
cross sections, namely, types I, II and III, correspond- analysis should be conducted.
ing to the angle of rotations as described in Table 3. It should be noted that all the above-reviewed
This manual also provides design consideration when approaches are based on the SDOF analysis, and the
the reinforced concrete (RC) element is shear failure equivalent SDOF system is derived with the assump-
critical. tion that flexural mode governs the structural response
Hao et al. 11
Table 3. Types of damaged cross sections in RC element defined in UFC 3-340-02 (US Department of Defense, 2008).
Damage Angle Description
Type I 0°–2° Concrete is effective in resisting moment. The concrete cover over the reinforcement on both
surfaces of the element remains intact
Type II 2°–6° Concrete is crushed and not effective in resisting moment. Compression reinforcement equal
to the tension reinforcement is required to resist moment. The concrete cover over the
reinforcement on both surfaces of the element remains intact
Type III 6°–12° Concrete cover over the reinforcement on both surfaces of the element is completely
disengaged. Equal tension and compression reinforcement which is properly tied together is
required to resist moment
Table 4. Damage levels of RC elements defined in ASCE/SEI 59-11 (ASCE, 2011).
RC elements Expected element damage

Superficial Moderate Heavy Hazardous
Ductility ratio Angle Angle Angle
Single-reinforced slab or beam 1 2° 5° 10°

Double-reinforced slab or beam without shear reinforcement 1 2° 5° 10°
Double-reinforced slab or beam with shear reinforcement 1 4° 6° 10°
RC: reinforced concrete.
subjected to the blast load. The corresponding design under complex stress state, result in the significantly
criteria as listed in Tables 3 and 4 are also associated complicated nonlinear behaviour of the concrete mate-
with the flexural response. Therefore, these design rial. Generally, concrete material exhibits higher
approaches and criteria given in the current design strength, modulus of elasticity (MoE) and strain at
guides are not suitable for modelling structural maximum stress when the loading rate increases as is
responses subjected to contact or close-in explosions demonstrated by the laboratory test results shown in
where structural response is governed by localized Figure 9.
damage due to concrete crushing and spalling, or gov-
erned by direct or diagonal shear response. In such
situations, high-fidelity numerical simulation or modi- Strength. Under quasi-static or low-velocity impact,
fied SDOF analysis is needed to predict structural only one or two macro-cracks can be observed. When
responses in the design analysis. concrete specimen is rapidly loaded, failure at a much
higher stress level is observed because the well-
developed and widely spread cracks are forced to pro-
Concrete dynamic material properties pagate through regions of higher resistance, as well as
and models a larger number of micro-cracks, which need more
energy, are required prior to the formation of a
Concrete dynamic material properties continuous fracture surface, as illustrated in Figure 10.
The properties of concrete material under dynamic The DIF, defined as the dynamic strength normal-
loading are different from those under static loads. ized by its counterpart under quasi-static loading, was
Together with high strain rate effects, under dynamic considered as a material property that can be used to
loading, concrete might also be subjected to multiaxial model concrete materials in computer codes, as well as
stress state due to stress wave propagation. In the local to design and analyse the responses of concrete struc-
area that the impact and blast load is applied, high tures subjected to impulsive loadings for simple appli-
hydrostatic pressure might be generated. Moreover, as cation by structural engineers (US Department of
a heterogeneous and anisotropic composite material, Defense, 2008). Different empirical DIF versus strain
initial defects such as micro-cracks and air voids exist rate relations have been proposed in a number of con-
in concrete materials which affect the dynamic material crete material models, as will be reviewed in section
properties. The formation, development and propaga- ‘Concrete material models in hydrocodes’, for
tion of these natural defects, as well as macro-cracks numerical simulation as shown in Figure 11.
Figure 9. Dynamic stress–strain curves under compression and tension: (a) stress–strain curves of mortar under dynamic
compression (Grote et al., 2001) and (b) stress–strain curves of concrete under dynamic tension (Yan and Lin, 2006).
Figure 10. Fractures and cracks of concrete under impact load at different strain rates (Chen et al., 2013).
Figure 11. DIF for compressive and tensile strengths in different models (fc = 40 MPa).
In spite of the trend that DIF increases with the striving to investigate and quantify specific influencing
strain rate, significantly scattering DIF models can be factors so that the true dynamic strength of the con-
observed in Figure 11. Besides the variations in testing crete material can be derived. Studies of possible influ-
conditions such as specimen, material and equipment encing factors on concrete dynamic strength are
in different tests which lead to the variations in testing summarized and discussed here.
data, it is well acknowledged that there are several crit-
ical factors that might also have significant influences Influence of lateral inertia confinement. The concrete
on the test results. Following the critical review by strength is dependent on the level of confining pres-
Bischoff and Perry (1991), many researchers have been sure. With dynamic deformation, the specimen
Hao et al. 13
Figure 12. Lateral inertia confinement in impact tests: (a) uniaxial compression, (b) uniaxial tension and (c) splitting tension.
inevitably undergoes lateral expansion. The specimen et al., 2013b). The improved DIF relations for com-
will not be able to expand freely in the lateral direction pressive strength are
because of the inertial restraint, resulting in lateral
stresses that will act as a form of confinement, as illu- CDIF = 0:0419(log e_ d ) + 1:2165 for e_ d ł 30=s ð2Þ
strated in Figure 12.
CDIF = 0:8988(log e_ d )2 2:8255(log e_ d )
The influence of lateral inertia on dynamic strength ð3Þ
of the concrete in impact testing has been the most + 3:4907 for e_ d .30=s
controversial among researchers until very recently
The improved DIF relations for tensile strength are
that some general consensus has been reached. Some
researchers (e.g. Cotsovos and Pavlović, 2008) believed TDIF = 0:26(log e_ d ) + 2:06 for e_ d ł 1=s ð4Þ
the inertial force is the major reason of the strength
increment in dynamic testing while some others TDIF = 2(log e_ d ) + 2:06 for 1=s\_ed ł 2=s ð5Þ
believed that the influence from inertia is trivial and TDIF = 1:4431(log e_ d ) + 2:2276 for 2=s\_ed ł 150=s
can be neglected (e.g. Lu and Li, 2011). Other research-
ers acknowledged the contribution of inertia effect, but ð6Þ
did not rule out the material strength increment at high
strain rate as a material property (e.g. Hao and Hao, Influence of end friction in dynamic compression. In
2016; Li and Meng, 2003). Hao et al. (2010b) con- dynamic compressive tests, the frictional force at the
firmed that the strength increment at high strain rate interfaces between specimen and testing apparatus is
was specimen size dependent. A method to remove the another important structural effect that might contrib-
contribution of lateral inertia confinement from the ute to the observed dynamic strength increment. The
test data to obtain the real material property was pro- frictions at both ends of the specimen constrain the
posed by Hao and Hao (2013a). The true strain rate deformation along the lateral direction (Figure 13),
effect DIFe_ of the concrete material can be simply and thus provide a certain level of confining pressure
derived based on the following equation to the concrete specimen and contribute to the dynamic
strength increment under impact loadings (Bertholf
DIFe_ = DIFA DIFi + 1 ð1Þ and Karnes, 1975).
Impact tests on rock materials by Hakaleht (1969)
where DIFA is the apparent DIF directly obtained confirmed that the influence of confining effect became
from dynamic tests and DIFi denotes the contribution more prominent when specimen length was shorter. Li
of inertia confinement obtained from numerical simula- and Meng (2003) performed numerical simulations
tions of tests without considering any strain rate effect. and found that when the friction coefficient was larger
Based on such methodology, the improved DIFs than 0.2, the influence of end-friction confinement
for compressive and tensile strengths were proposed in became prominent for concrete specimens with aspect
Hao and Hao (2011) and Hao et al. (2012), respec- ratio of 0.5. Zhang et al. (2009) suggested kinetic fric-
tively, and were validated by the experiments (Hao tion coefficient be used to analyse the end-friction
strength than mortar, leads to higher concrete

dynamic strength. The effects of coarse aggregates on
concrete dynamic strengths have been investigated by
a number of researchers (e.g. Yan and Lin, 2006).
The contribution of coarse aggregates to concrete
dynamic compression strength with respect to strain
rate was quantified by Hao and Hao (2011) as
t = 0:1074(log e_ ) + 1:207 for 1=s ł e_ ł 30=s ð8Þ

t = 0:0917(log e_ ) + 1:4998 for 30=s ł e_ ł 90=s
ð9Þ
t = 0:1998(log e_ )2 0:8349(log e_ ) + 2:1889
ð10Þ
for 90=s ł e_ ł 1000=s
where t is the dynamic compressive strength ratio

between the concrete and mortar matrix.
The dynamic strength ratio between the concrete
and mortar matrix in tension is quantified as (Chen
et al., 2014)
z = 0:543(log e_ )2 + 2:123 log e_ 1:78

ð11Þ
Figure 13. Illustration of end-friction confinement in impact tests. for 20=s ł e_ ł 150=s
A higher stress level is needed to break coarse

confinement effects in Split Hopkinson Pressure Bar aggregates under high-speed impact. Being an impor-
(SHPB) tests. This was further supported by experi- tant constituent in concrete mix, the contribution of
mental and numerical studies by Hao et al. (2013a), coarse aggregates to the dynamic strength increment
based on which modification equations to remove the of concrete material at high strain rate should be
influence of end-friction confinement on concrete DIF treated as a material property. Therefore, the sugges-
obtained directly from laboratory tests were suggested. tions that ‘‘strength increment of concrete material at
The ratios of DIFs obtained without end friction high strain rate was mainly due to inertia effect and
(DIFm = 0 ) to those with end friction (DIFm6¼0 ) are should not be considered in the material model’’, are
derived as a function of friction coefficient (m), L/D not exactly correct. Studies of the influence of coarse
ratio and strain rate (_e) as aggregates on concrete dynamic strength are still lim-
ited. In some numerical studies, granite coarse aggre-
DIFm = 0 gates were considered. In practice, source of
= exp½3:0524 3 10 (L=D)0:2494(m)0:1043(log e_ ) + 0:1563
2
g=
DIFm6¼0 aggregates for mixing concrete can be broad and dif-
for 0:5 ł L=D ł 2:0, 0:0 ł m ł 0:5 and 10=s ł e_ ł 600=s ferent aggregate materials lead to different concrete
dynamic material properties. Coarse aggregates nor-
ð7Þ
mally originate from crushed granite, limestone,
sandstone, marble, gravel and so on, and the
Influence of coarse aggregates. Concrete consists of mechanical properties of these rocks can be signifi-
mortar matrix, aggregates and the interfacial transi- cantly different. The crushing strength, MoE and
tion zone (ITZ) between them. ITZ is the weakest porosity of aggregates can vary from 80 to 530 MPa,
section among the three components, followed by 45 to 160 GPa and 0 to almost 40%, respectively
mortar matrix, and aggregates are normally the (Neville, 1995). Researches that investigate the varia-
strongest. Under quasi-static loading, normally the tions in the mechanical properties of aggregates in an
crack initiates at the weakest sections, that is, ITZ, in-depth manner are needed for a better understand-
and propagates via the mortar matrix and through- ing of the aggregate effects and concrete material
out the specimen to cause the fracture. However, properties under dynamic loadings.
when the load is applied faster, the deformation is
also very fast that cracks are formed and propagate Influence of moisture. Impact tests on concrete speci-
in coarse aggregates as can be seen in Figure 10. mens with 0%, 50% and 100% relative humidity (RH)
Damage of aggregates, which usually have higher have demonstrated an important role of free water
Hao et al. 15
content in concrete material, but as long as the water

content in concrete reaches a certain level, further
increasing RH seems to have insignificant influence on
the impact strength (Reinhardt, 1982; Reinhardt et al.,
1990). Rossi (1991) intended to attribute the signifi-
cantly higher strength of wet concrete than dry ones to
the viscosity, namely, Stéfan effect, due to the presence
of free water on the micro-scale. However, the theory
of Stéfan effect is only valid for strain rates below 1/s.
At strain rates higher than 1/s, both wet and dry con-
cretes exhibit significant strain rate dependency based
on SHPB test data presented by Ross et al. (1996) and
Cadoni et al. (2001). The influence of moisture condi-
tion is, to some extent, proven by the difference in
dynamic strengths of wet and dry concretes. However,
Figure 14. DIFs for compressive strength and MoE.
it cannot fully explain the mechanism of strength incre-
ment at high strain rates. It should be noted that stud-
ies prior to those by Ross et al. (1996) and Cadoni
less than 50 1/s. Therefore, the reliability of DIF rela-
et al. (2001) only considered micro-concrete without
coarse aggregates to investigate the influence of free tions for MoE in fib model code needs further investi-
water. Previously discussed lateral inertia and end- gation. DIF relations for MoE were also studied by
friction confinement effects and the contribution of Hao and Hao (2013b) and are illustrated in
coarse aggregates to strength increment at high strain Figure 14.The DIF relations for MoE (DIFE) by Hao
rates should be responsible for the significant incre- and Hao (2013b) are
ment in strength of dry concrete at high strain rates.
DIFE = 0:0075(log e_ ) + 1:03 for 104 =s ł e_ ł 64:8=s
ð12Þ
MoE. Many dynamic tests also observed the increase
in MoE of concrete with the strain rate. Reinhardt DIFE = 1:2817(log e_ )2 4:5854(log e_ ) + 5:1441
ð13Þ
et al. (1990) attributed the increase in MoE to the vis- for 64:8=s ł e_ ł 200=s
cous resistance of free water content. Besides this,
Yan and Lin (2006) believed that the resistance of
coarse aggregates also contributed to delaying the Strain at maximum stress. Despite the influence of con-
creation and propagation of micro-cracks, leading to crete quality, type of aggregates and testing methods
an increase in MoE. Inertia due to micro-cracking is and conditions, Bischoff and Perry (1991) believed that
another possible reason for the delay (Ožbolt et al., the critical strain should increase with strain rate. This
2014). If the stress wave propagation effect is taken is reasonable based on the observations from tests that
into consideration, according to John et al. (1987), a large number of well-developed and widely spread
the crack velocity can only reach a few percent of the micro- and macro-cracks are formed as illustrated in
stress wave speed in concrete material. With such a Figure 10. Grote et al. (2001) showed that the critical
huge difference in velocities of cracks and stress compressive strain of mortar specimens increased
wave, delayed creation and propagation of micro- slightly with strain rate (Figure 9(a)). Critical strain
cracks with respect to high-speed stress wave results under dynamic tension was demonstrated by Yan and
in a delayed strain response from test data. In other Lin (2006) to increase with strain rate (Figure 9(b)).
words, at a given stress level, strain decreases while SHPB tests by Chen et al. (2013) further confirmed the
MoE increases with strain rate. increase in critical strain with strain rate as shown in
Empirical relations that estimate the effect of stress Figure 15. Empirical relations that estimate the effect
and strain rates on the MoE are given by the fib model of strain rates on the critical strain in tension and com-
code (fib, 2013). It is generally acknowledged that the pression are given by the fib model code (fib, 2013).
MoE increases with strain rate, but with a less pro- Attention should be paid to derive the material
nounced increment ratio compared to that of strength. properties from dynamic test results. While it is under-
However, as can be seen in Figure 14, compared to the stood that significant amount of cracks and damage to
DIF for compressive strength, DIF for MoE in fib coarse aggregates are responsible for the strength gain
model code is more prominent when the strain rate is under impact load, structural effects such as lateral
material model and incorporated it to AUTODYN

(HZ (Hao and Zhou) model).
The strength criterion, equation of state (EOS),
plastic flow and damage, and strain rate effects are the
main aspects in hydrocode material models. While the
strain rate effects and influencing factors have been
reviewed in section ‘Concrete dynamic material proper-
ties’, the remaining aspects of concrete material model
commonly used in hydrocodes are briefly reviewed in
the following.
Strength criterion. Dynamic material models are often

derived from static models with dynamic effects taken
Figure 15. Critical strain under dynamic compression (Chen
into consideration, for example, by multiplying the sta-
et al., 2013).
tic strengths by DIF as widely adopted. In numerical
simulation, when a plastic damage model is used for
inertia and end-friction confinements induced during
concrete material, yield criterion is necessary. The
impact tests should be carefully treated for a more
adopted yield criterion is normally based on the failure
accurate determination of the material properties such
criterion. Some models including RHT model employ
as strength, MoE and energy dissipation capability.
caps on the meridians at a certain value of pressure.
Influence of moisture condition has been demon-
The yield criterion is often considered damage depen-
strated, but it seems that the influence is only signifi-
dent in a typical plastic damage model.
cant when concrete material totally loses its free water
The tensile strength of the concrete material is much
content, namely, completely dry with 0% RH. A fully
lower than its compressive strength. Moreover, the
dry concrete structure is less likely to exist in reality.
material strength is sensitive to hydrostatic/confining
However, further studies are deemed necessary to
pressure. The strength or failure criterion for concrete
examine the threshold RH that makes the moisture
is generally defined as a function of the state of stress
ratio important on the properties of concrete under
under triaxial stress states. The actual mechanical
dynamic loadings.
properties and behaviour of the concrete materials
under multiaxial stress states are very complicated.
Concrete material models in hydrocodes Therefore, in many concrete models, the strength cri-
teria idealizations have to be introduced. Based on the
Conducting full-scale tests to predict the performance experimental results from static tests, many strength
of concrete structures to blast and impact loading is criteria were proposed such as Mohr–Coulomb
often beyond affordability. With the advancement in Criterion, Drucker–Prager Criterion, Willam–Warnke
computer technology and the marked development of five-parameter criterion, Ottosen criterion, Hsieh–
numerical techniques, computer simulations have Ting–Chen criterion and unified twin shear strength
become a viable tool to conduct numerical tests for model.
quantitative understanding of structural performance
and prediction of damage levels. Hydrocode is based
on the conservation of energy, mass and momentum EOS. In most of the available numerical models, con-
and is popularly used to simulate materials and struc- crete is considered as a porous material with porosity
tures subjected to blast and impact loads. The JH of 10%–15% and its volumetric portion in hydrocode
(Johnson and Holmquist) concrete model was devel- is modelled by porous EOS. The relationship between
oped by Johnson and Holmquist (1994). Based on the density r and pressure P is illustrated in Figure 16. The
framework of JH model, K&C (Karagozian & Case) concrete material with initial density, r0, starts to have
model was published in 1996 with several updates plastic deformation (compact) when the initial compac-
(Malvar et al., 1997). RHT (Riedel, Hiermaier and tion pressure Pe is reached. The dotted curve defines
Thoma) model (Riedel et al., 1999) was developed in the compaction path of the material between the initial
1999. Gebbeken’s model was developed in 2000 compaction pressure Pe and the fully compacted pres-
(Gebbeken and Ruppert, 2000) based on which sure Ps. In this region, unloading and reloading paths
improvements were made in HPG (Hartmann, Pietzsch need to be determined. Air voids completely vanish
and Gebbeken) model in 2010 (Hartmann et al., 2010). from the material when the pressure exceeds the fully
Hao and Zhou (2007) developed a plastic damage compacted pressure, namely, P . Ps.
Hao et al. 17
used to simulate concrete material behaviour subjected

to high-rate loadings and have been proven yielding
acceptable predictions. In general, as is stated by Hao
and Zhou (2007), the residual strength needs to be rel-
atively high when penetration problem is modelled
whereas models with lower residual strength are suit-
able to simulate the behaviour of concrete subjected to
explosive loading. Irrespective of the model selected,
proper attention should be paid for reasonable mate-
rial parameters to obtain reliable numerical results.
As illustrated in Figure 11 of DIF relations adopted
Figure 16. Equation of state (EOS) for concrete. in different models, despite the theoretical background
of all the models being similar to a certain extent, these
The plastic compaction path is assumed to be linear models differ from each other because the data used by
in JH model. Herrmann’s P–a EOS (Herrmann, 1969) respective researchers in developing the model are dif-
is adopted in RHT, HZ and HPG models. ferent. Due to the page limit, more detailed review and
comparison of these models as well as their limitation
Plastic flow and damage. Strain hardening is not consid- and applicability are not possible here. Interested read-
ered in JH and HZ models for simplicity. Only yield ers should refer to the references provided in this arti-
surface and residual strength surface are defined. cle for more in-depth understanding of these models.
However, strain hardening is adopted in RHT, K&C
and HPG models by constructing the elastic limit sur- Blast testing methods
face and applying hardening rules. In RHT model, the
elastic limit surface is determined from the linear varia- Experimental testing is a direct method to investigate
tion between scaled tensile (0.5–0.8) and compressive structural response to blast loadings. The experimental
(0.3) strengths. In the high-pressure region, a cap is testing and analysis are often carried out to understand
applied to the elastic surface. Once the stress state the performance of some new structure forms and new
approaches the elastic limit surface, it widens to the structural materials subjected to blast loadings. Blast
failure surface with the assumption of isotropic hard- testing methods consist of field explosive testing and
ening. Damage is defined to accumulate when reaching laboratory testing. Most of laboratory tests are
the failure surface, after which the residual strength non-explosive testing.
surface is constructed. Similar approach is adopted in
K&C model but without cap in the elastic limit sur- Field blasting tests
face. Instead, many other parameters are taken into
account for precise description of the behaviour of Field blasting test is important to investigate blast
concrete material in high-pressure region. In HPG loading and its effect on the structural components,
model, two-third of the fracture stress is defined as the although it is extremely demanding in terms of cost,
elastic limit surface. equipment, time, safety and repeatability of testing. A
Various approaches to determine the damage level number of field blast tests have been carried out
are adopted in different models. In JH and RHT mod- around the world in recent decades. However, only
els, the accumulated damage level is determined by both limited articles can be sourced in the open literature
plastic volumetric strain and equivalent plastic strain. due to the limited release of the testing results. For
The accumulated effective plastic strain governs both the those published in the open literature, the essential
strain hardening and softening in K&C model. In HPG information such as charge weight and standoff is
model, the overall damage is related to damage due to often not provided due to the sensitivity of the subject
shear and compaction. Mazars’ double-scalar damage of blasting, which makes it difficult to compare differ-
model (Mazars, 1986) for materials with unequal tensile ent studies. In the open literature, a few field blast tests
and compressive strengths was adopted in HZ model. were conducted to predict blast loads and its effect on
the structures and structural components such as
beams, columns, slabs and walls. For instance, the
Summary. Similar theory background, that is, theory influences of explosive shape and orientation in uncon-
of plasticity with damage evolution law, is adopted in fined and confined environments on the blast loading
all these material models for concrete. Differences were investigated by Wu et al. (2010a). During 2004
among these models are mainly based on different and 2007, a series of field blast trials were carried out
hypotheses and idealizations. All these models can be in Woomera, Australia. Among them, two large
explosion trials (i.e. 5000 kg TNT and 500 kg ANFO) model was conducted by Woodson and Baylot (1999)
were conducted in 2006. The test results such as blast to study the dynamic response of the central column of
pressures and the performance of various structures a two-storey RC structure. A 7.1-kg C4 HE of hemi-
and structural components subjected to blast loading sphere shape (with equivalent TNT weight of 8 kg) at
were observed and analysed (Ngo et al., 2012). a standoff distance of 1.07 m was detonated in the test.
Schenker et al. (2008) conducted full-scale field explo- The testing results have been intensively utilized to
sion tests of 1000 kg TNT on the protected and unpro- calibrate the numerical model of RC structures (Chen
tected concrete slabs with 3-m span in Israel. et al., 2015a; Shi et al., 2008). A series of blast tests on
RC slabs, fibre-reinforced polymer (FRP) retrofitted
RC structural members and aluminium foam protected
Physical similarity. As full-scale field blast test is usually RC slabs were conducted (Wu et al., 2007, 2010b). It
expensive and complex. Blast experiment at reduced should be noted that the tested slabs might be sub-
scales was desirable and often carried out (Neuberger jected to the wave pressure reflected from the ground,
et al., 2007; Wang et al., 2012). Proper scaling needs to which does not represent the real scenario.
be validated by the physical similarity. The experiments
on a small-scale model are usually required before a
large-scale prototype is built. The principles of scaling Measurements and instruments. The typical measure-
between the small-scale model and the full-scale proto- ments in a blast test include pressures, displacements,
type model were stated by Jones (2012). The relation- accelerations and strains at specific locations. Two
ships between the prototype model (superscripts P) and kinds of measurements, that is, single-value measure-
the scaled model (superscripts M) with scaling factor ment and time-dependent measurement, were applied
(S) are given in the following. The linear dimensions in the testing (Schenker et al., 2008). Single-value mea-
(xPi = xM P M
i S), characteristic times (t i = t i S) and surement devices such as displacement ‘comb’ devices,
deformations at geometrically scaled locations for the displacement painted tube device, ‘Bikini’-type pres-
corresponding scaled times (dPi = dM P M
i S at t i = t i S) sure gauges were used for the measurement of the
are proportional to the scale factor (S). The angles maximum values (Neuberger et al., 2007; Schenker
(aPi = aM P M
i ), densities of materials (r i = r i ), stresses et al., 2008). Time-dependent measurements include
P M P M
(si = si ), strains (ei = ei ), angular deformations pressure transducers, displacement transducers, accel-
(vPi = vM P
i ) and loads at scaled locations (Fi = Fi at
M
erometers, strain gauges, high-speed data acquisition
P M
xi = xi S) are the same. It should be noted that some system and high-speed camera. Pressure transducers
phenomena cannot be scaled according to these princi- can be used to measure side-on and reflected overpres-
ples. For example, gravitational forces cannot be sure. In the study by Wu et al. (2010a), nine pressure
scaled according to the principles of geometrically simi- transducers were placed on the top surface of the slab
lar scaling. However, the gravitational force is not sig- to measure the overpressure histories and one pressure
nificant and can be neglected due to high acceleration transducer located about 2 m from the centre of the
involved in blast tests. The strain rate sensitivity in a explosion was used to measure the side-on or incident
small-scale model has scale factor times larger than overpressure. The sampling rate for the pressure trans-
that in a full-scale prototype. Since the actual scale fac- ducers is 2 MHz (Li et al., 2015; Wu et al., 2010a).
tor is not large, the material properties are assumed to The water-cooling pressure sensors with a range of
be scale independent. In addition, the phenomenon of 220 to 150 kPa were used to prevent overheating (Li
fracture cannot be scaled according to the principles of and Hao, 2014). Data acquisition instrument was
geometrically similar scaling. When scaling spherical sampled at a frequency of 100 kHz per channel (Li
blast wave phenomena, the most common scaling and Hao, 2014). The sampling rate can be up to
method is Hopkinson or ‘cube root’ scaling law stated 2 MHz for each channel using National Instruments
by Baker et al. (1983). The Hopkinson’s method was data acquisition system. The linear variable differential
used to calculate the corresponding charge weight (W) transformers (LVDTs) was sampled at the rate of
for the scaled model as W M = W P =S 3 . 10 kHz (Li et al., 2015). To record the time-dependent
A few blast tests have been conducted on the scaled displacement at very high rates, torsion measurement–
models. For example, scaled one-way square RC slabs based displacement gauges can be used (Schenker
subjected to close-in blast loadings were tested to et al., 2008). High-speed cameras filmed at the rates of
investigate the effect of scale factor on the dynamic 2000–30,000 Hz depending on the quality of camera
response (Wang et al., 2012). It was found that the spe- have also been used for measurements. To measure the
cimen with smaller-scale factor experienced more imparted impulse, a pendulum impact testing metho-
severe local damage than the slab with larger-scale fac- dology was developed by Gabauer et al. (2010) for the
tor. A series of experiments using 1/4-scale RC frame evaluation of two-post sections of strong post w-beam
Hao et al. 19
Figure 17. Left: wall specimen testing in UCSD blast simulator (Freidenberg et al., 2014); right: UCSD blast simulator and tested
column (Gram et al., 2006).
barrier. The generated impulse imparted on the plates loads ranging up to 55 kPa s can be applied on the
was measured by means of two-cable pendulum structure using hydraulic actuators and six combined
(Hanssen et al., 2002) or four-cable ballistic pendulum hydraulic/high-pressure nitrogen energy source called
system (Langdon et al., 2012; Nurick et al., 2009). The blast generators (BGs) via a computer-controlled colli-
blast pressure exerted on the specimen can be esti- sion. The duration is controllable between 0.5 and
mated according to the oscillation amplitude and rota- 5 m s. The peak pressure and impulse are also control-
tion angle of the pendulum. lable. Three high-speed cameras and data acquisition
system of 1 MHz sampling rate are employed for the
Laboratory testing and facilities measurements. It is a unique apparatus for conducting
full-scale testing of blast effects on structures without
With the results of field blast testing, theoretical and the use of actual explosives. Since 2006, more than 500
analytical models can be validated and the verified experiments have been conducted. A range of struc-
model can predict structural response and damage in a tural components such as concrete columns, structural
real blast situation. However, field blast tests are usu- steel columns, RC and concrete masonry walls were
ally expensive, time-consuming and often beyond tested in full-scale to demonstrate how UCSD blast
affordability. Therefore, some facilities and methodol- simulator results compare to the corresponding field
ogies for the laboratory testing have been developed, testing results using actual explosives. It was reported
such as physical blast simulator, shock tube, uniform that the laboratory tests exhibited similar loading and
impulsive loading simulator and other blast simula- damage modes as field tests (Crawford and
tors. Laboratory tests can be better controlled and in Magallanes, 2011). The limitation is that no tension
general cost less as compared to field blast test. force can be used to pull back the specimen towards
the BG to simulate the negative phase of overpressure.
University of California San Diego blast simulator. A blast
simulator funded by US Federal Government was
developed in University of California San Diego Shock tube facility. Shock tubes use a controlled detona-
(UCSD) (2006) to simulate blast-like scenario (Stewart tion or sudden release of compressed gas at one end of
et al., 2014), as shown in Figure 17. The impulsive a highly reinforced tube to apply a load to a specimen
Table 5. List of shock tubes.
Shock tubes Descriptions and capability
BakerRisk’s, USA (Dusenberry, 2010; Irshidat et al., 2010) Peak pressure of 310 kPa and impulse of 6.9 kPa s; testing
size of 2.4 m2; 4.25 m for extension section
Vicksburg, USA (Robert and Johnson, 2009) Peak pressure of 218.9 kPa and impulse of 494.4 Pa s;
capability to simulate negative phase
Large Blast/Thermal Simulator (DTRA), USA (Rinehart et Testing size with a radius of 10 m and 167-m-length tunnel
al., 2010)
University of Rhode Island, USA (Tekalur et al., 2008) Testing size of 70 mm diameter
SIMLab, Norway Testing region of 0.3 m2
NFPBS, Australia Pressure up to 100 kPa; capability to simulate negative phase
DTRA: Defense Threat Reduction Agency.
at the other end. It can generate blast wave without of 2.6. The tested specimen has a square cross section
using HEs to test structural components. The shock of 0.3 m 3 0.3 m. In Australia, a pressure-driven
tube is a less costly and much more controllable shock tube would be commissioned at the University
method to replicate the field conditions in the labora- of New South Wales in 2017. The facility is capable of
tory. The structural components can be tested with a generating a peak pressure up to 100 kPa and negative
variety of blast pressure and impulse combination phase. It should be noted that the blast load generated
using the shock tube. There are a few shock tube facili- by the shock tubes without using HEs should be vali-
ties available worldwide. In the United States, the dated against air blast test to ensure the blast wave
available shock tube facilities include BakerRisk’s profile is the representative of real scenario.
shock tube test facility (Dusenberry, 2010; Irshidat
et al., 2010); blast load simulator (BLS) in Vicksburg, Uniform impulsive loading simulator. For the far-field
MS, at the U.S. Army Engineer Research and explosion or gas explosion, the structure can be
Development Center (ERDC) (Robert and Johnson, assumed to be subjected to uniform impulsive loading.
2009); Large Blast/Thermal Simulator (LB/TS), a large Uniform impulsive loadings can be generated in the
shock tube in New Mexico, sponsored by Defense laboratory by various means as given in Table 6. The
Threat Reduction Agency (DTRA) (Rinehart et al., testing methods are simple and achievable in the
2010); shock tube facility at the University of Rhode laboratory, although the impulsive loading duration is
Island (Tekalur et al., 2008) and so on (Table 5). The larger as compared to the typical blast-induced
BakerRisk’s shock tube can generate a peak pressure loading.
up to 310 kPa and peak impulse of 6.9 kPa s. The
pressure and impulse applied to a test specimen can be
varied independently. The size of testing specimen can Other blast simulators. As given in Table 7, some other
be up to 2.4 m2 at the normal section or 4.25 m2 at the blast simulators such as underwater BG, land mine
extension section. No negative pressure can be gener- charge simulator and VCE simulator are available to
ated. It is reported that a range of structural compo- simulate the shock impact.
nents have been tested using BakerRisk’s shock tube
(Dusenberry, 2010). The BLS in Vicksburg can gener- Numerical simulation of blast effect and
ate a peak pressure of 218.9 kPa and impulse of
structural response
494.4 Pa s. The blast simulator can generate the pres-
sure with initial rise to a peak reflected pressure fol- The numerical procedures of blast prediction consist
lowed by a decay to zero pressure and then a negative of load determination and response determination,
phase (Robert and Johnson, 2009). The LB/TS and both can be divided into first-principle and semi-
(DTRA) has a cross section with a radius of 10 m and empirical methods. First-principle methods solve
167-m-long tunnel (Rinehart et al., 2010). In Europe, a motion equations based on the basic laws of physics.
shock tube at Ernst Mach Institute (EMI) in Freiburg, Semi-empirical models utilize extensive data from past
Germany, and a Shock Tube Facility in Norwegian experiments, and therefore, they require less computa-
University of Science and Technology (NTNU), tional effort. However, due to a lack of accessibility to
Norway, are available. The shock tube in NTNU is past experimental results and most of the semi-
designed to release pressures up to 17 MPa, which empirical numerical tools being restricted for public
results in an incident shock wave with a Mach number distribution, structural engineers and researchers are
Hao et al. 21
Table 6. List of uniform impulsive loading generators.
Uniform impulsive loading generators Authors Capability
Inflated airtight chamber Mostaghel (2003) A plate was dropped onto the membrane from various
heights to achieve the specific impulse amplitude and
duration
Inflated airbag Chen and Hao (2014) Peak pressure (114 kPa); duration (0.1 s)
Wang et al. (2015) Peak pressure (829 kPa); duration (0.04 s)
Plastic explosive placed Yuen and Nurick (2005) To generate uniform loadings by detonating explosive and
on polystyrene foam pad directing the blast through a tube towards the target
Table 7. List of other blast simulators.
Other blast simulators Descriptions and capability
Underwater shock simulator, Cambridge Peak pressures (15–70 MPa); decay times (0.1–1.5 ms); the peak pressure and
University (Deshpande et al., 2006) pulse duration can be adjusted
WBWG, Portugal (Pereira et al., 2015) To test out-of-plane walls under dynamic loading; the shock wave in water is
4.5 times faster than in air, and the pressure–impulse for the shock wave in
water is 15–20 times higher than in air
A laboratory-scale buried charge simulator, Internal diameter of 28 mm and height of 80 mm; to generate a flow of sand
Cambridge University (McShane et al., 2013) to mimic the buried explosive; the impact velocity of the sand can be up to
39 m/s
DLG test, BakerRisk, USA (http:// To produce blast waves with a gradual rise to peak pressure of up to 206 kPa,
www.bakerrisk.com/) which is representative of the blast wave from a VCE
The UK air blast tunnel facility, Essex (Clubley, 200-m-long tunnel with two test regions (4.9 m diameter and 10.2 m
2014) diameter); to generate the blast wave with the maximum blast load of 110 kPa
and long positive phase duration of 0.2 s, which is used to simulate unconfined
VCE
WBWG: underwater blast wave generators; DLG: deflagration load generator; VCE: vapour cloud explosion.
more relying on the first-principle methods dealing stable and the results are usually very accurate.
with such problems. However, implicit methods require the inverse of the
In recent decades, with development of CFD, com- stiffness matrix at each time step which substantially
putational structural dynamics (CSD) and also com- increases computation time for blast-induced struc-
puter power, numerical tools which usually utilize the tural response analysis in which the stiffness of the
first-principle or coupled semi-empirical and first- element changes with time due to nonlinear material
principle methods are available for predicting blast and structural behaviour.
effects and the corresponding structural response under Explicit solution techniques, unlike an implicit
blast loads. This section discusses the characteristics, method, calculate response at each time step based on
limitations and uses of various advanced modelling equilibrium at the previous time step. Explicit methods
techniques. therefore do not require inverse of the stiffness matrix
for each time step. However, since equilibrium is not
satisfied precisely at each time step, explicit solutions
Implicit and explicit analysis can become numerically unstable if time steps are not
The numerical analysis of blast effects and structural small enough.
responses employs either implicit solution technique or Until now, for analyses of structural response to
explicit solution technique to solve the equation of blast, explicit methods are overwhelmingly chosen
motion for the structural system. Both procedures have because the benefit gained by not having to inverse the
advantages and disadvantages that make them suitable stiffness matrix during each time step outweighs the
for different simulation cases. requirement of small time steps. In addition, element
Implicit algorithm solves the motion equation failure and removal from the analysis is achieved and
based on the equilibrium of the external, internal and easy to be used in explicit computational codes, while
inertial forces in the structural system at each and such algorithm remains a computational challenge for
every time step. As a result, implicit solutions are implicit codes.
Table 8. Computer codes used for predicting blast loads. The first method models both the explosive and the
air elements using multi-material arbitrary
Computer code Developer Lagrangian–Eulerian (ALE) formulations. The multi-
CTH Sandia National Laboratories material ALE formulation is adequate for air blast
FEFLO SAIC simulations as an element can contain two or more dif-
FOIL Applied Research Associates, ferent materials, such as the air and explosive gases
Waterways Experiment Station generated by the explosion. This modelling strategy
SHAMRC Applied Research Associates, Inc. has been found to model correctly the blast–structure
Air3D Royal Military of Science College,
Cranfield University interaction and wave propagation through a medium
HULL Orlando Technology, Inc. such as air.
LS-DYNA Livermore Software Technology The second method to model an explosion in
Corporation LS-DYNA only requires a mesh of the air domain.
FLACS GexCon AS The explosive can be included within the air mesh by
ANSYS/FLUENT ANSYS Inc.
AUTODYN Century Dynamics specifying an initial volume fraction of the explosive in
ABAQUS/CEL ABAQUS Inc. the air domain through the *INITIAL_VOLUME_
FRACTION_GEOMETRY option in LS-DYNA.
This option is used in conjunction with the ALE
multi-material formulation. The explosive geometry
Numerical simulation of blast wave propagation can be specified to be of a sphere, a cylinder or a cube.
Numerical modelling of detonations includes model- The location and time of detonation can also be
ling of the detonation process and the subsequent blast defined by the user.
wave passage through the surrounding medium, and The third modelling technique is semi-empirical
ultimately the interaction between the blast wave and method using CONWEP which is built in LS-DYNA
structure. First-principle CFD codes for blast model- to generate pressure histories. The CONWEP algo-
ling are listed in Table 8. While most of these codes are rithm is based on the empirical data of Kingery and
restricted from public release, in the following section, Bulmash (1984). This method can model free-air blast
commercially available and most widely used codes of a spherical charge and surface blast of a hemispheri-
including LS-DYNA, AUTODYN, ABAQUS/CEL cal charge. This semi-empirical method cannot be used
to predict close-in detonation process, shock wave pro-
and Air3D are introduced.
pagation and interaction with structures.
Blast modelling in Air3D. Air3D is a CFD Eulerian code Blast modelling in AUTODYN. AUTODYN is an explicit
developed by Ritzel and Mathews (1997). With mass, 3D program for numerical analysis of nonlinear
initial energy and density of the explosive as the input, dynamic problems involving solids, fluids and gases. It
Air3D models the explosive as a balloon with high contains Lagrangian solver for structures modelling
pressure and temperature. Air3D calculates the pres- and Eulerian solver for modelling fluids and solids that
sure histories at locations defined by the user. It can undergo large deformations. Remapping technique of
model one-dimensional (1D) (radial symmetry), 2D AUTODYN enables the output of a high-resolution
(Cartesian coordinates) and 3D problems and has initial detonation stage to be remapped as initial con-
remapping capabilities. Air3D models structures as ditions for the subsequent calculation stage. This
rigid blocks and perfect reflectors. The key advantage allows modelling of the detonation process with very
of modelling detonations in Air3D is that it requires high grid resolution without increasing the computa-
limited input and is user-friendly. However, it is not tional demand. AUTODYN is the only CFD code
capable of modelling overpressure in close-in range to capable of modelling the afterburning effect of the
the explosive and it cannot handle the fluid–structure detonation. Addition of afterburning energy into the
interaction. numerical simulation results in the gases expanding
more rapidly, and a much higher temperature (Ritzel
and Matthews, 1997). Afterburning also has an effect
Blast modelling in LS-DYNA. LS-DYNA is probably the
of reducing the near-field peak overpressure and
most widely used computer code for solving high strain
increasing the impulse in the far field.
rate problems such as blast and impact. A detonation
problem with shock wave propagation can be modelled
in a number of ways in LS-DYNA. Here, three major Blast modelling in ABAQUS/CEL. Coupled Eulerian–
implementations including both the first-principle and Lagrangian (CEL) capability is a relatively new feature
semi-empirical methods are discussed. of ABAQUS/Explicit. It models the air as Eulerian
Hao et al. 23
Figure 18. Comparison of numerical study on a concrete slab with test results (Li and Hao, 2014).
mesh and imparts the blast wave (pressure) as a and also the coupling algorithm. Sensitivity
boundary condition into the air mesh, and then having analyses (convergence test) must be performed.
the blast wave propagates into the structure with 2. Material models and EOS should be adequately
Lagrangian mesh. The Lagrangian structure can be representing the real explosive and medium.
positioned arbitrarily within the Eulerian domain to For example, in CFD analysis, if large explosive
achieve any angle of incidence that is desired mass is used or close-in detonations are studied,
(Mougeotte et al., 2010). This new method overcomes the temperatures in the vicinity of the explosive
the need for determining the reflected pressures for are much higher than default value for ideal
oblique surfaces. This approach remains to be vali- gas, and the assumption of a constant value for
dated against test data, but thus far the results look the ratio of specific heats for air is incorrect in
promising (Carlucci et al., 2010). this region.
3. Charge shape and orientation effect should be
considered in the simulation, especially when
Challenges and suggestions in blast effect modelling. Despite non-spherical explosives are used in close-in
many CFD and semi-empirical codes being available range. Charge shape influences substantially
for the modelling of blast detonation, propagation and the overpressure distributions in the near-field
interaction with structures, challenges remain for struc- regions. The incident overpressures and
tural engineers and researchers who want to warrant impulses generated by a cylindrical charge are
their simulation results are feasible for design and guid- significantly greater than those generated by a
ing purpose. spherical charge. A risk of underestimating the
As discussed above, semi-empirical codes are devel- blast effect can take place when ignoring these
oped based on extensive test data. Their modelling effects in the analysis.
results, therefore, are preferable because the models
include validated empirical data. However, semi-
empirical modelling is limited to blast scenarios similar
Structural response analysis through numerical
to the test data from where the code is developed, and
they cannot handle problems with complex geometries. methods
First-principle models (CFD codes) solve the phy- Under blast loading condition, structural response
sics equations of detonation process. They are more involves a rapid transient phase during which high
widely used than semi-empirical codes. However, their strain rate effect, nonlinear material behaviour and
results need to be validated from the available test time-dependent structural deformation all take place.
results. In addition, a reliable CFD simulation should Failure of structural component can be caused either
address the following concerns: by the material fracture under stress wave effects (for
detonation in close-in range) or large structural defor-
1. CFD results are highly dependent on element mations (for detonation in far-range). Besides all these
size, and their accuracy varies with time step uncertainties, prediction of blast loads distribution
along with a deforming structure makes the response adequate strain-based element erosion technique, con-
analysis an even more complex issue. crete cracks can be clearly and reliably simulated.
Analysis carried out by hydrocodes can be categor- Despite convenience of use, it is worth noting that
ized as coupled analysis and uncoupled analysis. In an such erosion algorithm lacks solid physics background,
uncoupled analysis, the blast loads are calculated by and massive element deletion breaches the mass con-
assuming blast wave acting on a rigid structural surface servation. A comprehensive discussion about erosion
and then the calculated blast loads are applied on the limit for concrete is presented by Luccioni et al. (2013).
responding deformable structural model. Obviously, in To avoid deleting or eroding the elements and preserve
an uncoupled analysis, the blast loads calculated in the mass conservation, Xu and Needleman (1994) used
program are over-predicted, especially when significant interfacial elements along potential fracture paths in
deformation happens quickly on the target structure. FE model to simulate fragmentation process. The
In a coupled analysis, the blast load simulation is interfacial element model was later extended to use
related to the structural response. The blast load pre- cohesive elements along weak links in the structure
diction through CFD is solved simultaneously with (Zhou et al., 2005). These techniques avoid deleting the
structural response obtained from CSD. Since the elements in traditional FE model. However, the draw-
structural deformation and failure are accommodated back is that the final crack pattern and the fragment
in the coupled analysis, the acting blast pressure can be size distribution are restricted by the predetermined
more accurately predicted. AUTODYN, ABAQUS/ interfacial elements. Camacho and Ortiz (1996) devel-
Explicit and LS-DYNA are three representative soft- oped a 2D adaptive meshing technique with cohesive
ware packages that are capable of coupling blast pres- law. This method does not require predefining the
sures with structural response. weak surface. Therefore, it will provide more realistic
The numerical methods handling the structural predictions of concrete crack propagation and frag-
response under blast loads can be classified into the mentation process. In a later study, a fully 3D FE
categories discussed in the following sections, and each model with cohesive fracture surfaces capable of track-
of these approaches has distinctive features and can ing 3D crack propagation and interaction was pro-
handle different situations with varying accuracy. posed by Ortiz and Pandolfi (1999). More recently,
cohesive FEs have been used to simulate structural
dynamic fragmentation (Clayton, 2005). Despite the
Traditional FE method. FE method is the most common merits of realistic modelling of fragmentation and
choice for analysing the structural response under blast cracks, cohesive element method is computationally
effects. Its popularity not only stems from its easy inefficient as it requires a lot of remeshing since con-
accessibility but also because of its versatility in allow- crete structures under blast loads often break into
ing for coupling with a fluid solver so that the loading many small fragments.
environment can be more realistically reproduced.
Traditional FE model with damage mechanics has
often being used to model structural response under Meshfree methods. Besides interfacial element method,
blast loads. This method requires small time steps and various meshfree methods have been developed to
small element size to achieve a stable and accurate avoid the numerical problems with mesh distortion
simulation. In addition, although the first-principle under extreme deformations in Lagrangian solver.
CFD and CSD methods are developed by solving Belytschko et al. (1994) developed element free
physical equations, their results can only be preferably Galerkin (EFG) method and used this method to ana-
used after validation with the existing experimental lyse the dynamic fracture in concrete. Adopting EFG
results. method, Rabczuk and Belytschko (2006) investigated
A key issue with the Lagrangian grid-based FE high-velocity impact penetration on concrete struc-
model of structures subjected to large blast loading is tures. Another modelling technique which can handle
the large element deformations, which can cause singu- the mesh problem under large deformation is the mate-
lar Jacobi matrices, leading to high inaccuracy and rial point method (MPM) (Sulsky et al., 1994). After
ultimately the computational overflow. To overcome years of development, MPM is now capable of han-
this problem, the so-called ‘erosion algorithm’ is dling large deformation and multiple materials, includ-
adopted and elements are deleted when some prede- ing fluid–structure interactions. It is worth noting that
fined failure criterion is reached. In fact, this technique until now, only limited calibrations on the use of
has been incorporated in commercial hydrocodes such MPM in the modelling of concrete structures for
as LS-DYNA and ABAQUS and utilized by many impulsive and blast loading have been reported.
researchers. Figure 18 shows an example of modelling One of the most widely used meshless methods is
of concrete slabs under blast loads; it is noted that with the smooth particle hydrodynamics (SPH) method
Hao et al. 25
Figure 19. (a) MPM mesh and simulation results, after Hu and Chen (2006); (b) SPH simulation results, after Rabczuk and Eibl
(2006).
developed by Lucy (1977). Comparing with the tradi- Another kind of coupling method is based on the
tional FE grid method, SPH method allows tracking use of mixed interpolation functions in a transition
of material deformation and the time-varying beha- zone between particles and FEs. For example, SPH
viour. Complex material model that is used in tradi- particles can be coupled with the FE elements by a slid-
tional FE model can still be allocated to SPH model. ing interface. When an SPH particle is linked to FE ele-
Figure 19 shows the modelling of concrete slabs under ment, the equation of motion is governed by the force
blast loads using MPM and SPH methods. Because of from surrounding SPH particles and FE elements. If
its good performance, commercial software such as the SPH particles and FE elements are not joined, the
AUTODYN and LS-DYNA has incorporated SPH particles will slide along the surface of the FE elements,
into their solvers which further promote its wide and in this case, a special sliding interface algorithm
application. must be used (Attaway et al., 1994). The simulation
A generalized particle algorithm (GPA) was devel- shown in Figure 20 adopted a coupled SPH–FE model
oped by Johnson et al. (2000). It models the structure for fragments’ simulation.
by continuum FE first, and the elements are converted Finally, coupling method based on the use of
into meshless particles when specific failure or erosion Lagrange multipliers with or without an overlapping
criteria are met. However, similar to MPM and SPH zone was developed and applied (Rabczuk and
method, GPA method cannot reliably predict the frag- Belytschko, 2006). Compared to other coupling
ment size distributions because the fragment size is method, this method maintains good convergence rate.
related to the predefined particle size. Caleyron et al. (2011) adopted this method to study
the impact response of RC slab in which the concrete
is modelled with particles while reinforcement is mod-
Hybrid FE–meshfree methods. Despite wide application elled with beam element.
and good performance, certain limitations exist for Different to the methods discussed above, Johnson
meshless method. Comparing with the conventional and Stryk (2003) proposed a method to resolve the FE
FE method, the computational cost involving meshless mesh distortion problem using coupled SPH and FE
particles is normally higher. However, the Lagrangian method. This method simply converts highly distorted
formulation of SPH enables it to be linked to standard elements into meshless particles during the dynamic
FE model. Until now, there is a wide range of methods response. This approach is suited for problems involv-
for the coupling of meshfree methods with FEs. A ing severe localized distortion such as contact
good overview of the various methods is given by detonation-induced structural response, as in such
Rabczuk et al. (2006). cases the effect of the distorted elements to the remain-
One of the first coupling schemes is based on fluid– ing structure may be rather significant. With a particle
structure interactions. The fluid is modelled with a conversion and the associated contact algorithms such
meshfree method while the structure is modelled with an effect could be well preserved (Lu, 2009).
FEs. During the calculations, forces are computed to
prevent the penetration of the particles into the ele- Other methods. Besides the methods discussed above,
ment faces (master-slave coupling). the discrete element method (DEM) which is also a
Figure 20. SPH–FE model for simulation of cratering and fragmentation (Wang et al., 2005).
Figure 21. LDPM model (Pelessone et al., 2007).
meshfree Galerkin method was proposed by Hart depends on the predefined particle size and the ran-
(1991) and used to model damage and fragmentation dom particle distribution.
under high strain rates (Kun and Herrmann, 1996). It In general, to predict structural response under blast
has been recognized to be a more superior alternative loads, continuum FE model based on Lagrangian
for studying the mechanical behaviours of fractured mesh is either not reliable as they need to adopt ero-
masses. With correct definition of the material failure, sion algorithm or not efficient if interfacial element
DEM can correctly model the energy dissipation dur- model is adopted in which large amount of ‘remeshing’
ing the fracture process and the kinetic energy of each is required. Meshless methods including SPH, MPM,
cell. The main limitation is the computational cost. GPA, DEM and other relevant methods including
Moreover, it may not give a realistic prediction of con- LDPM can well handle the element breaking, but the
crete crack initiation and propagation because they are concrete crack initiation and propagation as well as
controlled by the discretization model. the fragment size distribution are not well addressed.
Lattice discrete particle model (LDPM) is a realistic Some discussions are given in Bonet and Kulasegaram
3D model of concrete meso-structure developed by (2000), Rabczuk et al. (2004) and Vignjevic et al.
Cusatis et al. (2003). It has been extensively calibrated/ (2000) regarding possible approaches and modifica-
validated under a wide range of quasi-static and tions of these methods to guarantee that fracture is
dynamic loading conditions, showing superior capabil- caused due to the physics only.
ities in predicting qualitative and quantitative beha- In a recent study, a two-step stochastic approach is
viours of the concrete. This method has been widely developed based on the theories of continuum dam-
adopted in the simulation of fibre RC under dynamic age mechanics and mechanics of micro-crack devel-
loading environment. Figure 21 shows the typical opment (Wang et al., 2009). A simple algorithm is
LDPM and the predicted fragments of structures used to predict the fragment trajectory and the launch
under impact loads. It is worth noting that in this distance based on the fragment size and the ejection
approach, the crack initiation and propagation again velocity.
Hao et al. 27
Applications of numerical analysis sufficient number of computed points to allow for

Successful numerical simulation which is based on curve fitting. Unlike analytical solutions, numerical
rational material and structural models can be power- approaches allow complex nonlinear resistance func-
ful tools providing supplement information for experi- tions and complex loading functions to be used.
mental works which are normally expensive, Furthermore, the numerical approach can describe the
dangerous and unrepeatable. The following section behaviour of the P–I curve in the dynamic response
describes some circumstances where a numerical analy- domain accurately.
sis, either prior to a blast loading event or after, is of
particular help. Conclusion
Full-scale structural collapse analysis. The numerical simu- This article presents a state-of-the-art review on the cur-
lation can be used as pre- or post-analysis on full-scale rent available design guidelines and research work on
structure collapse which is impractical to conduct concrete structures against blast loads. Blast phenom-
through experimental analysis. Luccioni et al. (2004) ena and load prediction as well as material dynamic
built a 3D frame structure in hydrocode and tested it behaviour and laboratory and field tests of structures
under blast loads. To study the progressive collapse of subjected to blast loadings are briefly reviewed. Merits
a real eight-storey building, Kwasniewski (2010) pro- and defects of the various approaches and material
posed a detailed 3D model with large number of FEs. models are discussed. The existing numerical tools and
Prior to the full-scale explosive test, Morrill et al. simulation algorithms are also discussed.
(2015) carried out a 3D numerical study to predict the
blast pressure–time histories and the associated struc- Declaration of Conflicting Interests
tural response. The author(s) declared no potential conflicts of interest with
respect to the research, authorship and/or publication of this
Mesoscale simulation. It is well known that concrete is a article.
composite material, containing appropriate portions of
coarse and fine aggregates, cement, water and some Funding
additives if necessary. Obviously, the concrete material
is heterogeneous, and its heterogeneity makes the beha- The author(s) disclosed receipt of the following financial sup-
port for the research, authorship and/or publication of this
viour of concrete under blast loads rather complicated.
article: This study was financially supported by the
Especially, the heterogeneity of concrete affects the Australian Research Council.
crack pattern and the fragment size distribution when
it is under blast loads. Mesoscale model of concrete
material was developed by Zhou and Hao (2009) to References
study the influence of aggregates on crack and frag- Agbabian M (1985) Design of structures to resist nuclear
ment distribution under blast loading environment. weapon effects. In: American Society of Civil Engineers
Recent development of concrete material for protec- (ed.) ASCE Manual on Engineering Practice No. 42. Syra-
tive purpose has witnessed increasingly more usage of cuse, NY: American Society of Civil Engineers (Revised
fibrous material; it was reported that addition of steel sub edition).
American Society of Civil Engineers (ASCE) (2010) Design of
fibres in concrete matrix can provide enhanced
Blast Resistant Buildings in Petrochemical Facilities. 2nd
mechanical performance, ductility and fracture energy ed. Reston, VA: ASCE.
(Li et al., 2015). However, these additives make the American Society of Civil Engineers (ASCE) (2011) Blast
heterogeneity of concrete more prominent. To identify protection of buildings. ASCE SEI 59-11. Available at:
the mechanical behaviour of each component in con- http://ascelibrary.org/doi/book/10.1061/9780784411889
crete matrix, numerical simulation based on mesoscale Attaway S, Heinstein M and Swegle J (1994) Coupling
model is deemed necessary and feasible. of smooth particle hydrodynamics with the finite
element method. Nuclear Engineering and Design 150(2):
Generation of damage curves. The pressure–impulse (P–I) 199–205.
Baker WE, Cox P, Kulesz J, et al. (1983) Explosion Hazards
curves are a series of iso-damage curves that can be
and Evaluation. Amsterdam: Elsevier.
used to quickly assess the structural damage under
Belytschko T, Lu YY and Gu L (1994) Element-free Galer-
blast loading environment (Li and Hao, 2013; Shi kin methods. International Journal for Numerical Methods
et al., 2008). To generate a reliable damage curve, large in Engineering 37(2): 229–256.
number of iso-damage cases involving different pres- Bertholf L and Karnes C (1975) Two-dimensional analysis of
sure and impulse combinations are required. P–I dia- the split Hopkinson pressure bar system. Journal of the
grams can be generated numerically by generating a Mechanics and Physics of Solids 23(1): 1–19.
Biggs JM (1964) Introduction to Structural Dynamics, vol. 3. Clubley SK (2014) Non-linear long duration blast loading of
New York: McGraw-Hill. cylindrical shell structures. Engineering Structures 59(0):
Bischoff P and Perry S (1991) Compressive behaviour of con- 113–126.
crete at high strain rates. Materials and Structures 24(6): Conrath E, Krauthammer T, Marchand K, et al. (1999) Struc-
425–450. tural design for physical security. Reston, VA: ASCE.
Bjerketvedt D, Bakke JR and Van Wingerden K (1997) Gas Cotsovos D and Pavlović M (2008) Numerical investigation
explosion handbook. Journal of Hazardous Materials of concrete subjected to compressive impact loading. Part
52(1): 1–150. 1: a fundamental explanation for the apparent strength
Bonet J and Kulasegaram S (2000) Correction and stabiliza- gain at high loading rates. Computers & Structures 86(1):
tion of smooth particle hydrodynamics methods with appli- 145–163.
cations in metal forming simulations. International Journal Crawford JE and Magallanes JM (2011) The effects of mod-
for Numerical Methods in Engineering 47(6): 1189–1214. eling choices on the response of structural components to
Brode HL (1955) Numerical solutions of spherical blast blast effects. International Journal of Protective Structures
waves. Journal of Applied Physics 26(6): 766–775. 2(2): 231–266.
Cadoni E, Labibes K, Albertini C, et al. (2001) Strain-rate Cusatis G, Bazant Z and Cedolin L (2003) Confinement-
effect on the tensile behaviour of concrete at different rela- shear lattice model for concrete damage in tension and
tive humidity levels. Materials and Structures 34(1): 21–26. compression: I. Theory. Journal of Engineering Mechanics
Caleyron F, Chuzel-Marmot Y and Combescure A (2011) 129(12): 1439–1448.
Modeling of reinforced concrete through SPH-FE cou- Deshpande V, Heaver A and Fleck N (2006) An underwater
pling and its application to the simulation of a projectile’s shock simulator. Proceedings of the Royal Society A:
impact onto a slab. International Journal for Numerical Mathematical, Physical and Engineering Science
Methods in Biomedical Engineering 27(6): 882–898. 462(2067): 1021–1041.
Camacho GT and Ortiz M (1996) Computational modelling Dusenberry DO (2010) Handbook for Blast Resistant Design
of impact damage in brittle materials. International Jour- of Buildings. Hoboken, NJ: John Wiley & Sons, Inc.
nal of Solids and Structures 33(20): 2899–2938. Federal Emergency Management Agency (FEMA) (2003a)
Carlucci P, Mougeotte C and Huidi J (2010) Validation of Primer to design safe school projects in case of terrorist
Abaqus Explicit – CEL for classes of problems of interest attacks: providing protection to people and buildings.
to the US Army. In: 2010 SIMULIA customer conference, Report no. 428, December. Washington, DC: FEMA.
Providence, RI, 25–27 May 2010. Federal Emergency Management Agency (FEMA) (2003b)
Chaudhuri A, Hadjadj A, Sadot O, et al. (2012) Study of Reference manual to mitigate potential terrorist attacks
shock-wave mitigation through solid obstacles. In: 28th against buildings. Report no. 426, December. Washington,
international symposium on shock waves, Manchester, UK, DC: FEMA.
17–22 July 2011. pp. 493–498. Springer. fib (2013) fib Model Code for Concrete Structures 2010. Ber-
Chemical Industries Association (CIA) (1992) An Approach lin, Germany: Ernst & Sohn.
to the Categorization of Process Plant Hazard and Control Freidenberg A, Aviram A, Stewart L, et al. (2014) Demon-
Building Designs. London: CIA. stration of tailored impact to achieve blast-like loading.
Chen G, Hao Y and Hao H (2014) 3D meso-scale modelling International Journal of Impact Engineering 71: 97–105.
of concrete material in spall tests. Materials and Structures Gabauer DJ, Kusano KD, Marzougui D, et al. (2010) Pendu-
48(6): 1887–1899. lum testing as a means of assessing the crash performance
Chen W and Hao H (2014) Experimental investigations and of longitudinal barrier with minor damage. International
numerical simulations of multi-arch double-layered panels Journal of Impact Engineering 37(11): 1121–1137.
under uniform impulsive loadings. International Journal of Gebbeken N and Döge T (2010) Explosion protection – archi-
Impact Engineering 63: 140–157. tectural design, urban planning and landscape planning.
Chen W, Hao H and Chen S (2015a) Numerical analysis of International Journal of Protective Structures 1(1): 1–22.
prestressed reinforced concrete beam subjected to blast Gebbeken N and Ruppert M (2000) A new material model
loading. Materials & Design 65: 662–674. for concrete in high-dynamic hydrocode simulations.
Chen W, Hao H, Chen S, et al. (2015b) Performance of com- Archive of Applied Mechanics 70(7): 463–478.
posite structural insulated panel with metal skin subjected Gram MM, Clark AJ, Hegemier GA, et al. (2006) Laboratory
to blast loading. Materials & Design 84: 194–203. simulation of blast loading on building and bridge struc-
Chen X, Wu S and Zhou J (2013) Experimental and model- tures. Structures under Shock and Impact IX 87: 33–44.
ing study of dynamic mechanical properties of cement Grote D, Park S and Zhou M (2001) Dynamic behavior of
paste, mortar and concrete. Construction and Building concrete at high strain rates and pressures: I. Experimen-
Materials 47: 419–430. tal characterization. International Journal of Impact Engi-
Clayton J (2005) Dynamic plasticity and fracture in high den- neering 25(9): 869–886.
sity polycrystals: constitutive modeling and numerical GSA (2013) Alternate Path Analysis & Design Guidelines for
simulation. Journal of the Mechanics and Physics of Solids Progressive Collapse Resistance. Available at: http://
53(2): 261–301. www.gsa.gov/portal/content/103205
Hakaleht K (1969) The Behaviour of Rock under Impulse
Loads: A Study Using Hopkinson Split Bar Method.
Hao et al. 29
Thesis, Technical University, Otaniemi, Helsinki: Finnish Hu W and Chen Z (2006) Model-based simulation of the
Academy of Technical Science. synergistic effects of blast and fragmentation on a con-
Hanssen A, Enstock L and Langseth M (2002) Close-range crete wall using the MPM. International Journal of Impact
blast loading of aluminium foam panels. International Engineering 32(12): 2066–2096.
Journal of Impact Engineering 27(6): 593–618. Irshidat M, Al-Ostaz A, Cheng A-D, et al. (2010) Nanoparti-
Hao H (2015) Predictions of structural response to dynamic cle reinforced polymer for blast protection of unrein-
loads of different loading rates. International Journal of forced masonry wall: laboratory blast load simulation
Protective Structures 6(4): 585–606. and design models. Journal of Structural Engineering
Hao H and Zhou X (2007) Concrete material model for high 137(10): 1193–1204.
rate dynamic analysis. In: Proceedings of the 7th interna- John R, Shah SP and Jeng Y-S (1987) A fracture mechanics
tional conference on shock and impact loads on structures, model to predict the rate sensitivity of mode I fracture of
Beijing, China, 17–19 October. concrete. Cement and Concrete Research 17(2): 249–262.
Hao H, Stewart M, Li Z-X, et al. (2010a) RC column failure Johnson GR and Holmquist TJ (1994) An improved compu-
probabilities to blast loads. International Journal of Pro- tational constitutive model for brittle materials. High-
tective Structures 1(4): 571–591. Pressure Science and Technology 309(1): 981–984.
Hao Y and Hao H (2011) Numerical evaluation of the influ- Johnson GR and Stryk RA (2003) Conversion of 3D dis-
ence of aggregates on concrete compressive strength at torted elements into meshless particles during dynamic
high strain rate. International Journal of Protective Struc- deformation. International Journal of Impact Engineering
tures 2(2): 177–206. 28(9): 947–966.
Hao Y and Hao H (2013a) Dynamic compressive behaviour of Johnson GR, Beissel S and Stryk R (2000) A generalized par-
spiral steel fibre reinforced concrete in split Hopkinson pres- ticle algorithm for high velocity impact computations.
sure bar tests. Construction and Building Materials 48: 521–532. Computational Mechanics 25(2–3): 245–256.
Hao Y and Hao H (2013b) Numerical investigation of the Jones N (2012) Structural Impact. 2nd ed. Cambridge: Cam-
dynamic compressive behaviour of rock materials at high bridge University Press.
strain rate. Rock Mechanics and Rock Engineering 46(2): Kang B, Choi W and Park G (2001) Structural optimization
373–388. under equivalent static loads transformed from dynamic
Hao Y and Hao H (2014) Influence of the concrete DIF loads based on displacement. Computers & Structures
model on the numerical predictions of RC wall responses 79(2): 145–154.
to blast loadings. Engineering Structures 73: 24–38. Kingery CN and Bulmash G (1984) Air Blast Parameters
Hao Y and Hao H (2016) Finite element modelling of mesos- from TNT Spherical Air Burst and Hemispherical Surface
cale concrete material in dynamic splitting test. Advances Burst. Aberdeen, MD: Ballistic Research Laboratories.
in Structural Engineering. 19: 1027–1039. Krauthammer T and Altenberg A (2000) Negative phase
Hao Y, Hao H and Li Z-X (2010b) Numerical analysis of lat- blast effects on glass panels. International Journal of
eral inertial confinement effects on impact test of concrete Impact Engineering 24(1): 1–17.
compressive material properties. International Journal of Kun F and Herrmann HJ (1996) A study of fragmentation
Protective Structures 1(1): 145–168. processes using a discrete element method. Computer Meth-
Hao Y, Hao H and Li Z-X (2013a) Influence of end friction ods in Applied Mechanics and Engineering 138(1): 3–18.
confinement on impact tests of concrete material at high Kwasniewski L (2010) Nonlinear dynamic simulations of
strain rate. International Journal of Impact Engineering 60: progressive collapse for a multistory building. Engineering
82–106. Structures 32(5): 1223–1235.
Hao Y, Hao H and Zhang X (2012) Numerical analysis of Langdon G, von Klemperer C, Rowland B, et al. (2012) The
concrete material properties at high strain rate under response of sandwich structures with composite face
direct tension. International Journal of Impact Engineering sheets and polymer foam cores to air-blast loading:
39(1): 51–62. preliminary experiments. Engineering Structures 36:
Hao Y, Hao H, Jiang G, et al. (2013b) Experimental confir- 104–112.
mation of some factors influencing dynamic concrete Lea C and Ledin H (2002) A review of the state-of-the-art in
compressive strengths in high-speed impact tests. Cement gas explosion modelling. HSL/2002/02. Buxton: Health
and Concrete Research 52: 63–70. and Safety Laboratory.
Hart R (1991) General report: an introduction to distinct ele- Li J and Hao H (2013) Influence of brittle shear damage on
ment modelling for rock engineering. In: Proceedings of accuracy of the two-step method in prediction of struc-
the 7th ISRM congress, Aachen, 16–20 September 1991. tural response to blast loads. International Journal of
Hartmann T, Pietzsch A and Gebbeken N (2010) A hydro- Impact Engineering 54: 217–231.
code material model for concrete. International Journal of Li J and Hao H (2014) Numerical study of concrete spall
Protective Structures 1(4): 443–468. damage to blast loads. International Journal of Impact
Henrych J and Major R (1979) The Dynamics of Explosion Engineering 68: 41–55.
and its Use. Amsterdam: Elsevier. Li J, Wu C and Hao H (2015) An experimental and numeri-
Herrmann W (1969) Constitutive equation for the dynamic cal study of reinforced ultra-high performance concrete
compaction of ductile porous materials. Journal of Applied slabs under blast loads. Materials & Design 82: 64–76.
Physics 40(6): 2490–2499.
Li Q and Meng H (2003) About the dynamic strength spherical explosions. Part I: air-blast loading. International
enhancement of concrete-like materials in a split Hopkin- Journal of Impact Engineering 34(5): 859–873.
son pressure bar test. International Journal of Solids and Neville AM (1995) Properties of concrete. Longman.
Structures 40(2): 343–360. Ngo T, Lumantarna R and Mendis P (2012) Protective struc-
Low HY and Hao H (2001) Reliability analysis of reinforced tures research at the University of Melbourne. Australian
concrete slabs under explosive loading. Structural Safety Journal of Structural Engineering 13(1): 43.
23(2): 157–178. Niollet J, Yuen SCK and Nurick G (2015) A study to assess
Lu Y (2009) Modelling of concrete structures subjected to the use of cylindrical bars as blast barriers. International
shock and blast loading: an overview and some recent stud- Journal of Protective Structures 6(2): 263–286.
ies. Structural Engineering and Mechanics 32(2): 235–249. Nurick G, Langdon G, Chi Y, et al. (2009) Behaviour of
Lu Y and Li Q (2011) About the dynamic uniaxial tensile sandwich panels subjected to intense air blast. Part 1:
strength of concrete-like materials. International Journal experiments. Composite Structures 91(4): 433–441.
of Impact Engineering 38(4): 171–180. Ortiz M and Pandolfi A (1999) Finite-deformation irreversi-
Luccioni B, Ambrosini R and Danesi R (2004) Analysis of ble cohesive elements for three-dimensional crack-propa-
building collapse under blast loads. Engineering Structures gation analysis. International Journal for Numerical
26(1): 63–71. Methods in Engineering 44: 1267–1282.
Luccioni B, Aráoz G and Labanda N (2013) Defining ero- _
Ožbolt J, Sharma A, Irhan B, et al. (2014) Tensile behavior
sion limit for concrete. International Journal of Protective of concrete under high loading rates. International Journal
Structures 4(3): 315–340. of Impact Engineering 69: 55–68.
Lucy LB (1977) A numerical approach to the testing of the Pelessone D, Cusatis G and Baylot JT (2007) Application of
fission hypothesis. The Astronomical Journal 82: the lattice discrete particle model (LDPM) to simulate the
1013–1024. effects of munitions on reinforced concrete structures. In:
McShane GJ, Deshpande VS and Fleck NA (2013) A Proceedings of the 12th international symposium on inter-
laboratory-scale buried charge simulator. International action of the effects of munitions with structures, Orlando,
Journal of Impact Engineering 62(0): 210–218. FL, 17–21 September.
McVay MK (1988) Spall damage of concrete structures. Tech- Pereira JM, Campos J and Lourenc xo PB (2015) Masonry
nical report SL-88-22 (DTIC Document), June. Washing- infill walls under blast loading using confined underwater
ton, DC: Department of the Army. blast wave generators (WBWG). Engineering Structures
Malvar LJ and Crawford JE (1998) Dynamic increase factors 92(0): 69–83.
for concrete (ANSI Std.). In: Proceedings of the 28th DDESB Rabczuk T and Belytschko T (2006) Application of particle
seminar, Orlando, FL, 10–12 August 1998, pp. 1–17. PN. methods to static fracture of reinforced concrete struc-
Malvar LJ, Crawford JE, Wesevich JW, et al. (1997) A plasti- tures. International Journal of Fracture 137(1–4): 19–49.
city concrete material model for DYNA3D. International Rabczuk T and Eibl J (2006) Modelling dynamic failure of
Journal of Impact Engineering 19(9): 847–873. concrete with meshfree methods. International Journal of
Manufacturing Chemists Association (1978) Siting and con- Impact Engineering 32(11): 1878–1897.
struction of new control houses for chemical manufacturing Rabczuk T, Eibl J and Stempniewski L (2004) Numerical
plants. Safety guide SG-22. Washington, DC: Manufac- analysis of high speed concrete fragmentation using a
turing Chemists Association. meshfree Lagrangian method. Engineering Fracture
Mazars J (1986) A description of micro-and macroscale dam- Mechanics 71(4): 547–556.
age of concrete structures. Engineering Fracture Mechanics Rabczuk T, Xiao SP and Sauer M (2006) Coupling of mesh-
25(5): 729–737. free methods with finite elements: basic concepts and test
Mills C (1987) The design of concrete structure to resist results. Communications in Numerical Methods in Engi-
explosions and weapon effects. In: Proceedings of the 1st neering 22(10): 1031–1065.
international conference on concrete for hazard protections, Reinhardt HW (1982) Concrete under impact loading, tensile
Edinburgh, 27–30 September, pp. 61–73. strength and bond. HERON 27(3), 1982.
Morrill KB, Crawford JE, Brewer TR, et al. (2015) Numeri- Reinhardt HW, Rossi P and van Mier JG (1990) Joint inves-
cal analysis, modeling and testing of a full-scale structure tigation of concrete at high rates of loading. Materials and
subject to VBIED loading. In: Proceedings of the 3rd Structures 23(3): 213–216.
international conference on protective structures (ICPS3), Riedel W, Thoma K and Hiermaier S (1999) Penetration of
Newcastle, NSW, Australia, 3–6 February. reinforced concrete by BETA-B-500 – numerical analysis
Mostaghel N (2003) Blast load simulation system. Patent using a new macroscopic concrete model for hydrocodes.
6536258 B1, USA. In: Proceedings of the 9th international symposium on
Mougeotte C, Carlucci P, Recchia S, et al. (2010) Novel interaction of the effect of munitions with structures, 3
approach to conducting blast load analyses using Abaqus/ May, pp. 315–322.
Explicit-CEL (DTIC document). In: Army Research Devel- Rinehart EJ, Henny RW, Thomsen JM, et al. (2010) DTRA
opment and Engineering Center Picatinny Arsenal, NJ, 2010. weapons effects testing: a thirty year perspective. Defense
Neuberger A, Peles S and Rittel D (2007) Scaling the response Threat Reduction Agency Kirtland AFB NM Test Technol-
of circular plates subjected to large and close-range ogy DIV, 4 October.
Hao et al. 31
Ritzel D and Matthews K (1997) An adjustable explosion- properties. International Journal of Impact Engineering
source model for CFD blast calculations. In: Proceedings 36(6): 808–820.
of the 21st international symposium on shock waves, Great Wang W, Zhang D, Lu F, et al. (2012) Experimental study
Keppel Island, QLD, Australia, 20 July, pp. 97–102. on scaling the explosion resistance of a one-way square
Robert SD and Johnson CF (2009) Blast response of conven- reinforced concrete slab under a close-in blast loading.
tional and high performance reinforced concrete panels. International Journal of Impact Engineering 49: 158–164.
Structures Congress 2009: 1142–1150. Wang Y, Liew JR and Lee SC (2015) Experimental and
Ross CA, Jerome DM, Tedesco JW, et al. (1996) Moisture numerical studies of non-composite Steel–Concrete–Steel
and strain rate effects on concrete strength. Materials sandwich panels under impulsive loading. Materials &
Journal 93(3): 293–300. Design 81: 104–112.
Rossi P (1991) A physical phenomenon which can explain Wang Z, Lu Y, Hao H, et al. (2005) A full coupled numerical
the mechanical behaviour of concrete under high strain analysis approach for buried structures subjected to sub-
rates. Materials and Structures 24(6): 422–424. surface blast. Computers & Structures 83(4): 339–356.
Schenker A, Anteby I, Gal E, et al. (2008) Full-scale field Wei J and Dharani LR (2005) Fracture mechanics of lami-
tests of concrete slabs subjected to blast loads. Interna- nated glass subjected to blast loading. Theoretical and
tional Journal of Impact Engineering 35(3): 184–198. Applied Fracture Mechanics 44(2): 157–167.
Shi Y, Hao H and Li Z-X (2007) Numerical simulation of Woodson SC and Baylot JT (1999) Structural Collapse:
blast wave interaction with structure columns. Shock Quarter-Scale Model Experiments (DTIC document).
Waves 17(1–2): 113–133. Vicksburg, MS: Engineering Research and Development
Shi Y, Hao H and Li Z-X (2008) Numerical derivation of Center, US Army Corps of Engineers.
pressure–impulse diagrams for prediction of RC column Wu C, Fattori G, Whittaker A, et al. (2010a) Investigation of air-
damage to blast loads. International Journal of Impact blast effects from spherical- and cylindrical-shaped charges.
Engineering 35(11): 1213–1227. International Journal of Protective Structures 1(3): 345–362.
Smith PD and Rose TA (2006) Blast wave propagation in Wu C, Huang L and Oehlers DJ (2010b) Blast testing of alu-
city streets – an overview. Progress in Structural Engineer- minum foam–protected reinforced concrete slabs. Journal
ing and Materials 8(1): 16–28. of Performance of Constructed Facilities 25(5): 464–474.
Stewart L, Freidenberg A, Rodriguez-Nikl T, et al. (2014) Wu C, Oehlers DJ, Wachl J, et al. (2007) Blast testing of RC
Methodology and validation for blast and shock testing slabs retrofitted with NSM CFRP plates. Advances in
of structures using high-speed hydraulic actuators. Engi- Structural Engineering 10(4): 397–414.
neering Structures 70: 168–180. Xu X-P and Needleman A (1994) Numerical simulations of
Sulsky D, Chen Z and Schreyer HL (1994) A particle method fast crack growth in brittle solids. Journal of the
for history-dependent materials. Computer Methods in Mechanics and Physics of Solids 42(9): 1397–1434.
Applied Mechanics and Engineering 118(1): 179–196. Yan D and Lin G (2006) Dynamic properties of concrete in direct
Tekalur SA, Shukla A and Shivakumar K (2008) Blast resis- tension. Cement and Concrete Research 36(7): 1371–1378.
tance of polyurea based layered composite materials. Yuen SCK and Nurick G (2005) Experimental and numerical
Composite Structures 84(3): 271–281. studies on the response of quadrangular stiffened plates.
US Department of Defense (2002) Design and analysis of har- Part I: subjected to uniform blast load. International Jour-
dened structures to conventional weapon effects. Report no. nal of Impact Engineering 31(1): 55–83.
UFC 3-340-01, June. Washington, DC: Unified Facilities Zhang M, Wu H, Li Q, et al. (2009) Further investigation on
Criteria. the dynamic compressive strength enhancement of
US Department of Defense (2008) Structures to resist the concrete-like materials based on split Hopkinson pressure
effects of accidental explosions. Report no. UFC 3-340-02, bar tests. Part I: experiments. International Journal of
December. Washington, DC: Unified Facilities Criteria. Impact Engineering 36(12): 1327–1334.
US Department of Defense (2009) Design of Buildings to Zhou F, Molinari J-F and Ramesh K (2005) A cohesive
Resist Progressive Collapse. July. Washington, DC: Uni- model based fragmentation analysis: effects of strain rate
fied Facilities Criteria. and initial defects distribution. International Journal of
US Department of Defense (2012) DoD minimum antiterror- Solids and Structures 42(18): 5181–5207.
ism standards for buildings. Report no. UFC 4-010-01, Zhou XQ and Hao H (2008) Prediction of airblast loads on
February. Washington, DC: Unified Facilities Criteria. structures behind a protective barrier. International Jour-
Van den Berg A (1985) The multi-energy method: a frame- nal of Impact Engineering 35(5): 363–375.
work for vapour cloud explosion blast prediction. Journal Zhou XQ and Hao H (2009) Mesoscale modelling and analy-
of Hazardous Materials 12(1): 1–10. sis of damage and fragmentation of concrete slab under
Vignjevic R, Campbell J and Libersky L (2000) A treatment contact detonation. International Journal of Impact Engi-
of zero-energy modes in the smoothed particle hydrody- neering 36(12): 1315–1326.
namics method. Computer Methods in Applied Mechanics Zong R, Hao H and Shi Y (2015) Numerical analysis and
and Engineering 184(1): 67–85. design of a new fence type blast wall for blast protection.
Wang M, Hao H, Ding Y, et al. (2009) Prediction of frag- In: Proceedings of the 11th international conference on
ment size and ejection distance of masonry wall under shock and impact loads on structures, Ottawa, ON,
blast load using homogenized masonry material Canada, 14–15 May.

AdvStructEng HaoHaoLiandChen-1

Uploaded by

Copyright:

Available Formats

AdvStructEng HaoHaoLiandChen-1

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

AdvStructEng HaoHaoLiandChen-1

Uploaded by

Copyright:

Available Formats

Article

Advances in Structural Engineering

concrete structures ase.sagepub.com

Hong Hao1, Yifei Hao1, Jun Li2 and Wensu Chen1

Introduction dynamic material properties. The current approaches

and Rose, 2006). Intuitively, the characteristics of

Current design approaches

Table 2. Summary of design manuals and reports.

Guidelines Application scope and notes

Damage Angle Description

Table 4. Damage levels of RC elements defined in ASCE/SEI 59-11 (ASCE, 2011).

RC elements Expected element damage

Single-reinforced slab or beam 1 2° 5° 10°

RC: reinforced concrete.

strength than mortar, leads to higher concrete

t = 0:1074(log e_ ) + 1:207 for 1=s ł e_ ł 30=s ð8Þ

where t is the dynamic compressive strength ratio

z = 0:543(log e_ )2 + 2:123 log e_ 1:78

A higher stress level is needed to break coarse

content in concrete material, but as long as the water

material model and incorporated it to AUTODYN

Strength criterion. Dynamic material models are often

used to simulate concrete material behaviour subjected

Table 5. List of shock tubes.

Shock tubes Descriptions and capability

DTRA: Defense Threat Reduction Agency.

Table 6. List of uniform impulsive loading generators.

Uniform impulsive loading generators Authors Capability

Table 7. List of other blast simulators.

Other blast simulators Descriptions and capability

Figure 21. LDPM model (Pelessone et al., 2007).

Applications of numerical analysis sufficient number of computed points to allow for

You might also like