Professional Documents
Culture Documents
Underground explosions:
estimating the safe distance
1 Jörg Zimbelmann Dr.-Ing. 2 Conrad Boley Univ.-Prof. Dr.-Ing.
Project Manager, Boley Geotechnik Consulting Engineers, Munich, Director, Department for Soil Mechanics and Soil Engineering, Faculty
Germany (corresponding author: j.zimbelmann@boleygeotechnik.de) for Civil Engineering, University of the German Armed Forces, Munich,
Germany; Chief Executive Officer, Boley Geotechnik Consulting
Engineers, Munich, Germany
1 2
Underground explosions may contain an even more devastating potential than comparable free explosions in air,
because they can produce significantly greater pressures and stronger momenta and the explosive waves may decay
more slowly with increasing distance. For civil applications such as clearance of explosive ordnance, explosive
tunnelling and surface mining in the vicinity of buildings, a tool is desirable, with which the safe distance rhor,safe,
beyond which a building remains undamaged, may be evaluated in a fast and simple manner. The procedure
to derive such a tool is exemplarily demonstrated for near-surface, underground explosions with a charge mass of
125 kg of trinitrotoluene, dry sand as foundation soil and a building with shallow foundation and no basement.
Numerical simulations were carried out with the commercial hydrocode Autodyn. To track the explosion-induced
wave up to sufficiently large distances, the modelling strategy called the ‘moving window’ was developed.
With respect to the damage patterns, ‘air blast’, ‘foundation-induced excitation of the structure to vibrations’
and ‘inadmissible inclination of foundations due to subsidence’, the numerical time courses of the relevant
wave parameters were evaluated. The result is one chart for the safe distance as a function of the explosive’s
charge depth.
67
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
explosions in air and depending on the determining factors such the underground consisting of dry sand and shallow-founded
as explosive charge weight, charge depth and on-site soil buildings without a basement.
conditions, explosions below the soil surface may produce
significantly greater pressures and longer exposure times (i.e. The whole article is based on the thesis by Zimbelmann (2015).
stronger momenta), and the pressures and particle movements
may decay more slowly with increasing distance (Drake and Theoretical basics of shock waves
Little, 1983; Laine, 2012).
Equation of state and simplified one-dimensional
approach
Explosions and their effects on buildings are simulated by means
Explosion-induced waves are shock waves, at least in the vicinity
of hydrocodes. These explicit solvers are used to simulate
of the explosive, where geometric and material damping is of
unsteady, dynamic problems by simultaneously solving the
minor influence. Shock waves are always compression waves.
conservation equations of mass, momentum and energy while
They are characterised by extremely high pressures and a very
taking into account the initial and boundary conditions (Grujicic
steep wave front. Thus, on the passage of the shock front, the
et al., 2006). Examples of such simulations in the relevant
particle velocity v and the state variables pressure p, density r (or
literature are
specific volume V = 1/r) and specific internal energy e change
abruptly – the material jumps from the preshock state to the
■ the construction of underground shelters designed to resist
shocked state almost instantaneously. However, the balance
underground explosions in soil and rock (Bessette, 2008;
equations for mass, momentum and energy must always be met.
Laine, 2001; Lu et al., 2005; Wang et al., 2005)
Thus, shock deformation is a thermodynamic process, and a
■ the scenario of a terrorist attack with a contact explosion at
correct description of the material’s behaviour must therefore
ground level or an explosion slightly above the ground level
incorporate the specific internal energy e. In general, this is done
(Lu and Wang, 2006; Wu and Hao, 2005, 2007).
by means of an equation of state (EOS) p = (r, e) expressing
isotropic pressure p as a function of density r and specific internal
Hydrocode simulations are usually very time-consuming and
energy e. The state surface given by the EOS defines the locus of
require a certain level of expertise in the field of short-time
any state of thermodynamic equilibrium (Figure 1). The abrupt
dynamics. In civil applications such as clearing of ordnance,
change of state during transition of a shock wave is described by
tunnel blasting and surface mining, buildings are usually located
the Rayleigh line. This non-equilibrium path outside the state
in the vicinity of the underground explosion. Therefore, a tool for
surface connects the initial state (p0, V0, e0) and the shock state
evaluating the safe distance rhor,safe, beyond which an
(p1, V1, e1). The Hugoniot curve describes all shock states that
underground explosion (characterised by the type of explosive,
may be reached from the initial state.
the charge weight and the charge depth) is likely to cause no
structural damage to a building, is desirable.
In a simplified one-dimensional (1D) model, a fluid without any
shear strength is assumed to be overrun by a discontinuity (the
This paper introduces a way to generate such a decision aid. The
shock front). The latter propagates with the shock speed vS and
tool consists of a diagram with which the safe distance rhor,safe for
separates the unshocked material in the initial state ‘0’ from the
undamaged buildings may be identified in a fast and simple
shocked material with state ‘1’ (Figure 2). For a volume initially
manner. The procedure is demonstrated for shallow, underground
at rest (v0 = 0), the conservation of mass, momentum and energy
explosions with a charge mass of 125 kg of trinitrotoluene (TNT),
may be written in the form of Equations 1 to 3. These are the
Rankine–Hugoniot equations. A detailed derivation may be
looked up in the books by Meyers (1994) and Hiermaier (2008).
Pressure p
Dexterous rearrangement of Equations 1 and 2 and insertion in
1
Isothermal
compression
Hugoniot curve vS
curve
p1, ρ1, e1 p0, ρ0, e0
Rayleigh line vS
c
Specifi A
al
intern v1 v0
e r g y e
Spec 0 en
ific v
olum
eV vS
Figure 1. State surface, Rayleigh line and Hugoniot curve; Figure 2. Riemann problem – planar shock wave in a fluid
according to Hiermaier (2008) (uniaxial state of strain)
68
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
Equation 3 lead to the Hugoniot curve (Equation 4), which is a Preconditions for the formation of shock waves and
formulation of the conservation of energy that depends solely on hydrodynamic behaviour
thermodynamic quantities. From elasticity theory, it is known that in space, compression
waves propagate with the elastic wave speed cel,P (Equation 6). In
a uniaxial state of stress – as is the case for a bar with a constant
1. r0 vS ¼ r1 ðvS − v1 Þ
cross-section – the compression wave advances with a wave
speed cel,L according to Equation 7.
2. p1 − p0 ¼ r0 vS v1 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1−n E K þ ð4=3ÞG
cel,P ¼ ¼
ð1 þ n Þð1 − 2n Þ r r
1 sffiffiffiffiffi
p1 v1 ¼ r v v2 þ r0 vS ðe1 − e0 Þ ES
3. 2 0 S 1 ¼
6. r
1 1 1
e1 − e0 ¼ ðp1 þ p0 Þ − sffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 r0 r1 E ð1 þ n Þð1 − 2n Þ
1 cel,L ¼ ¼ cel,P
¼ ðp1 þ p0 ÞðV0 − V1 Þ 7. r 1−n
4. 2
σxx σxx
∂σxx t1 t2 > t1
∂εxx σB B
cBpl,L cpl,L
σB σB
Plastic
wave
∂σxx cA
pl,L cA
pl,L
σA ∂εxx σA
σA
σy σy
εxx x
Elastic Plastic
(a) (b)
69
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
tangent bulk modulus K (Equation 8) and therefore the wave of (2/3)sY (the lower part of Figure 4(a)). In the case of s1 ≫
speed continuously increase with stress. Hence, plastic wavelets sHEL, the difference between the sx(ex) and the p(q) compression
triggered at higher stresses catch up with plastic wavelets initiated curve vanishes (the upper part of Figure 4(a)). The influence of
at lower stresses. Superposition of these plastic wavelets causes a shear modulus G becomes negligible, and the solid behaves like a
steepening of the wave front and the formation of a shock wave fluid. Thus, Equation 6 turns into Equation 10; that is to say, the
(Figure 4(b)). At stresses above the imaginary intersection of the compression wave propagates with the bulk sound speed cB.
elastic and the plastic compression curve (point B in the upper
part of Figure 4(a)), even the elastic precursor is outrun. Besides E 2 2
the convexity of the compression curve, a sufficiently rapid s1 ¼ e þ s ¼ Ke1 þ sY
3ð1 − 2n Þ 1 3 Y 3
loading is imperative for the formation of a shock wave. 2
Otherwise, the slower wavelets have already been damped before 9. ¼ sm þ sY
3
they can be outrun by faster wavelets.
∂p ∂p ∂p K ∂p ∂p
K¼ ¼ −V ¼r c2B ¼ ¼ ¼ −V 2
8. ∂q ∂V ∂r 10. r ∂r ∂V
The maximum principal stress in a uniaxial state of strain, when Due to this hydrodynamic behaviour at high pressures, it is useful
reaching the yield point, is called the Hugoniot elastic limit sHEL. and common practice to formulate the compressive behaviour and
Assuming elastic perfectly plastic material behaviour and the von the shear behaviour with separate constitutive equations according
Mises yield criterion s1 − s3 = sY, the maximum principal stress to the decomposition of the stress tensor (Equation 11). The
may be calculated according to Equation 9 (see, for example, compressive behaviour is therefore described by an EOS p(r, e).
Zukas (2004)). With respect to the hydrostatic compression curve A strength model is used to connect deviatoric stresses Sij and
p = f(q), the s1 curve is therefore shifted upwards by a magnitude deviatoric strains eij0 .
σ or p σ or p
Strong
shock
t1 t2 > t1 t3 > t2
σB B
σ1 = f (ε1)
Elastic-plastic
behaviour
σA A
Hydrostate
p = f(θ)
Elastic
σHEL
x
ε1 = θ
Elastic Shock
precursor wave
σ or p
σ1 = f (ε1) (b)
p = f(θ)
(2/3)σy
σHEL Y
(2/3)σy
Es 1−v
E
E (1 + v)(1 − 2v)
K
3(1 – 2v)
ε1 = θ
(a)
70
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
Ground shock
The charge depth dTNT largely determines the allocation of the
Basic information released energy of explosion in the ground and the air. The greater
Explosions generally lead to extremely high pressures. the charge depth, the more explosive energy is transmitted into the
Additionally, explosive waves are very fast, thus leading to a rapid ground and the greater the direct ground shock. For sufficiently
loading and, subsequently (at least for a short period of time), a small depths of burial, the overlying soil is thrown out, and the
uniaxial state of strain. Despite phenomena such as crushing and released energy propagates by way of an air shock wave. Due to
rearrangement of particles, the compression curve for a uniaxial the interaction of the air shock wave with the ground surface, an
state of strain on the whole is convex for nearly all kinds of soil. indirect ground shock wave is generated. In comparison to a free
Thus, explosions in soil do fulfil all necessary preconditions for explosion in air of the same charge weight, the duration of the
the formation of shock waves. Therefore, explosions in soil and pressure phase increases with increasing charge depth, as time is
the related phenomena are called ground shock. needed for the formation of a cavern. Thus, the charge depth also
affects the frequency content. In this regard, the charge weight is
The maximum pressure p1,max transmitted into the ground depends of great importance as well. The greater the charge weight, the
on the impedance ratio of the explosive’s gaseous reaction more low frequencies are stimulated (Heinze, 1987).
products and the surrounding soil. In general, the pressure p(t) and
the specific impulse I are used to characterise the explosive Damage patterns due to underground explosions
loading. Both parameters mainly depend on the following factors Figure 5 shows five principle scenarios for structural damage
(Laine, 2012; Streitkräfteamt Abt V Infrastruktur, 2007) caused by shallow underground explosions
■ type and form of the explosive ■ flying debris of dispersed soil material
■ soil properties – for example type of soil, density, void ratio, ■ cratering and soil rupture
saturation and grain-size distribution (Omidvar et al., 2012) ■ air blast and forced vibrations
3
4
1
2 Cratering and soil rupture 4 Vibrational stimulation of the footing and the structure
Inadmissible angular distortion of foundations due to
5 subsidence (sagging)
71
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
■ foundation-induced excitation of the structure to vibrations numerical results with empirical, permissible values. It has to be
■ inadmissible inclination of foundations due to subsidence. mentioned that decoupled analyses are always accompanied by a
loss of information at each interface. However, time expenses for
To date, no publications are known from the relevant literature numerical simulations could be significantly reduced as the radial
that consider impairment of overground buildings initiated by symmetry of the free-field simulation could be taken into account.
underground explosions. Hereafter the damage scenario ‘flying
debris of dispersed soil material’ as well as the near-field failure Numerical simulations with Autodyn
mode ‘cratering and overlapping with the area of influence for In the work presented in this paper, numerical simulations were
soil rupture’ will not be considered any further. done using the commercial hydrocode Autodyn, version R15. To
allow for large deformations, in particular for the explosive’s
Approach for the generation of a decision aid gaseous reaction products and the surrounding soil, and to capture
Evaluation of the ground shock and air blast parameters as well as the blowout of the explosive and further expansion of the air
their effects on a structure such as deformations, state of stress shock wave, all simulations were done using the multimaterial
and degradation is not a trivial problem. By means of empirical Euler processor.
models, the ground shock parameters may only be determined
approximately. The influence of the explosive’s type and form as Autodyn: a short description
well as the effects of different soils and their parameters, the Autodyn is a commercial hydrocode that uses finite-element,
interaction of the explosive wave with the ground surface and a finite-difference and finite-volume techniques to solve highly
possible air shock wave may be captured in much more realistic time-dependent problems with geometric non-linearities such as
manner with numerical simulations. large strains and deformations as well as with material non-
linearities (plasticity, failure etc.). The expression hydrocode is
The decision aid was intended to be valid for a large variety of entailed historically due to the earlier mentioned hydrodynamic
buildings with different sizes, shapes, structures and stiffnesses. behaviour under extreme pressures. Various numerical processors
Therefore, the interactions of the air and ground shock waves have been implemented, such as the Lagrange solver for solid
with the structure according to the regarded damage patterns had continua and different types of Euler processors for modelling
to be considered separately. Thus, a decoupled analysis was done fluids, gases and large distortion, as well as arbitrary
according to the following procedure (Figure 6). The explosion, Lagrange–Euler, smooth particle hydrodynamics (a mesh-free
the interaction between the explosive’s gaseous reaction products method) and shell processors. The solution for all these solvers is
and the surrounding soil, and the spreading of the shock wave and based on explicit time integration.
its interaction with the soil surface were taken into account by
means of numerical simulations. From these, the time courses of As stated earlier, the balance equations for mass, momentum and
relevant wave parameters according to the considered damage energy must always be met. While Equations 1–3 represent the
patterns – air pressure p1, horizontal particle velocity v1 and shear conservation of mass, momentum and energy for a gas and the 1D
strain gxy in the soil – were recorded. These free-field solutions case, the partial differential Equations 12–14 are the conservation
were further processed as shown in the last section of this article. equations for the general case. In Autodyn, the partial differential
In the last step, the interactions of the air and ground shock waves equations are replaced by finite-difference equations. Then the
with the structure were encompassed by comparing the processed, whole problem that needs to be solved consists of the
72
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
Direct calculation
dvi 1 ∂sij
¼ þ fi
Nodal velocities and Zone volumes and 13. dt r ∂xj
displacements strain rates
de p ∂vi 1
Integration Material model ¼− þ S e_
14. dt r ∂xi r ij ij
fi
Nodal accelerations
Zone pressure and ai ¼
stresses 15. m
Conservation of momentum
Forces/mass
r0 V0
r¼
Nodal forces 16. V
Boundary and/or interactive forces The preceding paragraph is based on the AUTODYN Theory Manual
(Century Dynamics Inc., 2005a) and the ANSYS Mechanical User’s
Figure 7. Lagrange computation cycle; according to Century Guide (Ansys Inc., 2013). Further information concerning the
Dynamics Inc. (2005a) implemented routines may be looked up in these references. The
relevant portions of these manuals are also summarised in the
documents by Fišerová (2006) and Showichen (2008), which are
conservation equations together with a material model and a set of available for free. Anderson (1987), Zukas (2004) and Wilkins
initial and boundary conditions. Figure 7 shows the series of (1999) give an introduction to the general theory of hydrocodes.
calculations that are carried out in every time step for a Lagrange
discretisation. First, the boundary and/or interactive forces are Material models
updated and combined with the forces for inner zones that were Air was described as an ideal gas. In Autodyn, the associated
calculated in the previous time step. By using the momentum EOS is implemented in the form of Equation 17. For the
equation in Equation 15, the nodal accelerations may be simulation, the default values r0 = 1·225 × 10−3 g/cm3, T =
computed. Further integration leads to the nodal velocities and 288·15 K (= 15°C) and p0 = 101·325 kPa were used. An
displacements. In Lagrange models the mesh moves and deforms isentropic exponent k = 1·4 consequently results in the specific
with the material. Thus the new volumes and strain rates can be internal energy e0 = 206 785·7 J/kg. Since the air pressure is the
directly computed from the new velocities and displacements. only relevant parameter, the pressure offset was set to pshift = p0 =
Additionally, the conservation of mass is automatically satisfied 101·325 kPa. Thus, no unwanted velocities at the soil surface
and the new density can be calculated using Equation 16. By were generated and only the overpressure was recorded.
using the material model in combination with the conservation of
energy, the new pressures, stresses and energies may be 17. p1 ¼ ðk − 1Þr1 e1 þ pshift
computed. Conservation of momentum finally gives the new
nodal forces fi that are used in the next time step. p Asymptote (pore space
completely collapsed)
In the Euler processor, the governing conservation equations are pTMD
solved by using a control volume method. The finite-difference
equations are solved in two steps: in the first step, the Lagrange
discretisation is used, and the mesh moves and deforms. In the Plastic compression
D
B,TM
second step, the updated variables from the Lagrange solution are curve
~c
ρs0
While the Lagrange solver uses a central difference method which Release paths
is second-order accurate, the first-order upwind differencing Curved (reality)
scheme is implemented for the Euler processor. Linear (cBU at onset of unloading = original Sand model)
Linear (averaged over entire unloading path = modified model)
73
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
TNT was the explosive of choice. As it is a standard explosive choice when modelling non-cohesive soils, as it accounts for
and conversion tables exist (at least for the parameters of free plastic compression with different loading and unloading paths
explosions in air), the decision aid might be converted to other (Figure 8). This is the crucial property of granular media
types of explosive. Simulations were done using the JWL EOS by that affects damping. Compression is described by a plastic
Jones, Wilkins and Lee (Equation 18), which is a standard EOS compaction curve p(r) and a solid asymptote for the fully
for high explosives and is widely used. The constants A, B, R1, R2 compacted material with densities greater than the theoretical
and w need to be determined experimentally depending on the maximum density rTMD (Figure 9). Elastic unloading and
type of explosive. Alternatively, these constants may as well be reloading is described by the density-dependent bulk sound speed
chosen from tables. For large volumetric strains of the explosive’s cBU = f(l). Yet here the parameter l describes the density of the
gaseous reaction products, the first and second terms in Equation completely relieved material at zero pressure. For the work
18 become negligible and the JWL EOS approaches the ideal gas presented in this paper, the cBU(l) curve was modified based on
equation. For volumetric strains eV = (V − V0)/V0 > 10 – with V0 the data set published in annex ZA of the work by Laine (2012).
being the initial volume of the unreacted explosive – the EOS is Figure 10 shows the bulk sound speeds cBU,SAND(l) according to
to be changed from JWL to the ideal gas equation by using the original definition and the cBU,MOD(l) curve. The difference
Equation 19 for the conversion of the parameter w to the between the two definitions is depicted qualitatively in Figure 8:
isentropic exponent k (Century Dynamics Inc., 2005b). While for while the original definition cBU,SAND(l) was the bulk sound
Autodyn versions R14 and earlier this change in the EOS had to speed at the onset of unloading according to the respective
be done manually, in version R15 it is an automatic process. tangent moduli of the plastic compression curve, the modified
definition cBU,MOD(l) gives unloading paths that are linearly
wr R1 averaged over the real curved unloading paths.
p¼A 1− exp −
R1 r
wr R Like Compaction EOS linear, the Mo Granular strength model
þB 1− exp − 2 þ wre
18. R 2 r was developed for granular materials such as powders, soil and
sand. It is a modified Drucker–Prager model with a volume
independent flow rule of the Prandtl–Reuss type. The yield
19. k ¼wþ1 surface is both pressure- and density-dependent (Equation 20).
However, only the pressure-dependent part was used with the
definition of the yield surface according to the original Sand
The Sand model (Laine and Sandvik, 2001) implemented in the model (Figure 11). Although Laine and Sandvik (2001) stated that
Autodyn material library was used. It is valid for dry sand with the shear modulus for elastic unloading and reloading GU had
loose to medium-dense degree of density. The Sand model been derived from the measured values of the shear wave
includes Compaction EOS linear, the Mo Granular strength model velocities, it seemed to be calculated by Equation 21 using the
and a failure model. Compaction EOS linear was originally bulk sound speed for unloading cBU and a constant Poisson’s ratio
developed for granular materials and therefore is the appropriate of n = 0·23, which is an appropriate choice for sands. To be
800
Plastic compression curve
Asymptote (fully compacted material)
700
600
Pressure p: MPa
500
400
300
200
100
0
1650 1750 1850 1950 2050 2150 2250 2350 2450 2550 2650 2750
Density ρ: kg/m³
74
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
5000
4500 cBU,SAND
Bulk sound speed for unloading cBU: m/s
cBU,MOD
4000
3500
3000
2500
2000
1500
1000
500
0
1674 1774 1874 1974 2074 2174 2274 2374 2474 2574 2674 2774 2874
Density λ (at pressure p = 0): kg/m³
consistent with the bulk sound speed cBU, the definition of shear criterion simply states that the cell cannot bear shear stresses
modulus GU was modified as well using Equation 21 and n = when pmin is reached.
0·23. In Figure 12 the shear modulus GU,SAND(l) of the original
Sand model and the new, modified GU,MOD(l) curve are depicted. 1D wedge simulations: convergence and calibration of
the soil material
20. Y ¼ Y ð pÞ þ Y ðrÞ Sand has a certain (pressure-dependent) shear strength that is
accounted for by a strength model. Therefore, the material model
for sand is much more complicated than the ones for air and for
3 1 − 2n 3 2 1 − 2n TNT, which consist only of equations of state. Thus, sand was
G U ðl Þ ¼ K ¼ lcBU seen to be the decisive material for a convergence study. To save
21. 2 U 1þn 2 1þn
computation time, convergence was checked using a 1D wedge
model, which is a 1D, spherically symmetric model (Figure 13). It
The failure model used in the Sand model is the hydrotensile is defined by inner and outer radii ri and ra as well as the number
limit, which was set to a minimum pressure of pmin = −1 kPa. The of cells n in the direction of radius (Figure 14). As ri must be
250
225
200
175
Yield stress Y(p): MPa
150
125
100
75
50
75
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
40 000
GU,SAND
Shear modulus for unloading GU: MPa
35 000 GU,MOD
30 000
25 000
20 000
15 000
10 000
5000
0
1674 1774 1874 1974 2074 2174 2274 2374 2474 2574 2674 2774 2874
Density λ (at pressure p = 0): kg/m³
Spherical Radial
symmetry symmetry
ri
ra
Material location
‘Flow-out’ boundary element Material location
Void Void
SAND_mod3a
Gauge points at distances SAND_mod3a
Detonation point
Δx = 2·5 m/12·5 m
TNT-2 TNT-2
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
(a) (b)
76
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
greater than zero, the 1D wedge model represents a hollow 12·5 mm, this seems to be the case for pressures p £ plim =
sphere. The explosive charge weight WTNT is defined by filling 0·3 kPa. As the free surface is stress-free, there will always be
the inner part of the wedge from ri to ra,TNT. Simulations were some inevitable numerical noise with the chosen material model.
done with the following parameters A cell size of dx = 12·5 mm was chosen for the two-dimensional
(2D) simulations, as this seemed be an acceptable compromise
■ inner radius ri = 25 mm between computation time and the necessary accuracy down to
■ outer radius ra = 300 000 mm = 300 m very low pressures and stresses.
■ explosive’s outer radius ra,TNT = 263·634 mm (equivalent to a
charge weight WTNT = 125 kg for the initial density r0,TNT = Due to the fragmentary state of knowledge concerning soil
1·63 g/cm3) behaviour at high strain rates, the numerical results should always
■ variation in the number of elements in the range 2000 £ n £ be compared with experimental data or at least empirical models.
60 000 (according to cell sizes 150 mm ≥ dx ≥ 5 mm). For the latter, the widely used empirical model by Drake and
Little (1983), according to Equation 22 and Table 1, as well as the
Gauge points for recording the time courses of pressure p(t) were improved version by Drake et al. (1989), given by Equations
set at a regular distance of Dx = 2·5 m in the range of x £ 25 m 23–25 and Table 2, were chosen. For a discussion of both models,
and Dx = 12·5 m in the range of 25 m £ x £ 125 m. To extend the the reader is referred to the work by Laine (2012). As the 1D
period of time during which the time courses could be recorded, wedge model is valid for the entire space (i.e. no boundaries and
the outer edge of the model was chosen to be much greater than reflections etc.), the ground-shock coupling factor f in Equations
the distance x of the outermost gauge point. Additionally, flow-out 22 and 24 was set to f = 1, which simply means that the explosive
boundary elements were used at the outer edge of the model to energy is completely transferred into the soil. Comparison was
reduce unwanted reflections further. done for the peak pressure p1,max. Very good agreement was
achieved for the new model by Drake et al. (1989) and medium-
Figure 15 depicts the calculated peak pressures p1,max in the range dense, dry sand (Figure 17).
of 75 m £ x £ 125 m depending on the cell size dx. For a cell size
of dx = 12·5 mm, the solution seems to have converged.
p1,max N=m2 ¼ 48768f r kg=m3 c ½m=s
However, aside from the peak values, the time courses need to be !−n
r ½m
checked as well. As can be seen in Figure 16, depending on the 27997 p ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
cell size dx, a threshold plim exists below which the time courses 22.
3
WTNT ½kg
start to oscillate. For pressures smaller than plim, the chosen
material model for sand with Compaction EOS linear, the Mo
Granular strength model and the hydrotensile limit failure model 23. p1,max ¼ r0 vS,max v1,max
obviously does not work properly. For a cell size of dx =
450
x = 10 m
x = 17·5 m
400 x = 25 m
x = 75 m
350
Peak pressure pmax: kPa
300
250
dx = 12·5
200
150
100
50
0
200 20 2
Cell size dx: mm
77
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
5 dx = 10 mm
dx = 12·5 mm
dx = 25 mm
4
Pressure p(t): kPa
0
150 250 350 450 550 650 750 850 950
Time since detonation t: m/s
!−n model dimensions in combination with the cell size necessary for
9906 r a convergent solution would have resulted in a number of cells far
v1,max ½m=s ¼ pffiffiffiffiffi f
24. r0 1=3
0155WTNT larger than the available 4 × 106. Simulations within a single
overall model were thus excluded.
Dry sand, loose 1500 180 3·10 Type of soil r0: kg/m3 c0: m/s n
Dry sand, medium-dense 1700 300 2·75
Dry sand, dense 1950 520 2·50 Dry sand, loose 1500 180 2·75
Dry sand, medium-dense 1700 300 2·30
Data according to Laine (2012)
Dry sand, dense 1950 520 2·10
Table 1. Parameters for the empirical model by Drake and Little
(1983) Table 2. Parameters for the empirical model by Drake et al. (1989)
78
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
1·0 × 104
1D wedge model (Autodyn)
Drake and Little (1983) – dry sand, loose
Drake and Little (1983) – dry sand, medium-dense
Drake et al. (1989) – dry sand, loose
1·0 × 103 Drake et al. (1989) – dry sand, medium-dense
Peak pressure p1,max: kPa
1·0 × 102
10·0
1·0
0·3 2·5 25·0
Scaled distance RTNT = r/(WTNT)1/3: m/kg1/3
is the axis of symmetry by default). Each section consists of three prevent reflections at the upper and lower model boundaries.
model areas as shown in Figure 18. The central model area is The element height dx of the run-out areas increases towards
finely meshed according to the convergence study. Sufficiently the respective model boundaries, allowing for a lower amount
long run-out areas are located above and below the central area to of elements. The model areas within a section are coupled to
x
Depth of burial = centre of
(Axis of rotational
Flow-out boundary elements symmetry)
run-out area
explosion (variable)
Upper
nx,o = 200
40 m
Air Air
1·75 m
Coupled nodes
6·25 m
(joins)
(joins)
Sand Sand
ny = 600 cells
40 m
Lower
79
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
Pressure: kPa
1·555 x 103
1·400 x 103
25 20 15 10 5
1·244 x 103
1·088 x 103
24 19 14 9 4
9·328 x 102
23 18 13 8 3
7·772 x 102
6·215 x 102
4·659 x 102 22 17 12 7 2
2
3·103 x 10
1·546 x 102
21 16 11 6 1
–9·968 x 10–1
sand3a_125kg_d1,5m_step2
Cycle 9750
Time 1·782E+001 ms
Units mm, mg, ms
Axial symmetry
Sand_mod3a/125kg/d=1,5m/Y<15m
the mutual nodes. In the simulations carried out, the following ■ run-out areas: respective height Dx = 40 m with nx,o =
provisions with regard to the geometry have been taken 200 cells for the upper run-out area and nx,u = 300 cells for
the lower one
■ section width Dy = 7·5 m and the corresponding element
width dy = 12·5 mm = const. in accordance with the At the beginning of the simulation, only the first section was
convergence study modelled. Shortly before the wave’s arrival at the outer edge of
■ central modelling area: height Dx = 8 m with −6·25 m £ x £ the section, the calculation was interrupted. An additional section
1·75 m and the ground surface at x = 0, element width dx = was then appended, and the common nodes of the first and the
12·5 mm in accordance with the convergence study new section were coupled. After the maximum processable cell
20
Air pressure x = 1·5 m above ground – p(t): kPa
r = 20 m
18
r = 30 m
16 r = 40 m
14
12
10
8
6
4
2
0
50 70 90 110 130 150 170 190
−2
Time since detonation t: m/s
80
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
number was reached – in the present case, with five sections – the
The shock wave parameters corresponding to the failure modes
respective innermost section was deleted before a new section
are air pressure p, horizontal particle velocity vhor and shear
was added (see Figure 18). This enabled a wave section with a
deformation gxy.
width of 30 m, correspondingly four sections, to be traced to the
initially unknown abatement distance. Additionally, in the first
Since the decision aid is intended to apply to different types of
four stages of simulation, the number of elements could thus be
structures, a decoupled analysis is carried out and the structure
kept low and computing time could be saved.
itself is not modelled. The interaction between the shock wave in
air and soil and the structure is registered through permissible
Gauge points were arranged in a regular grid of 2·5 m × 2·5 m.
thresholds for the relevant shock wave parameters, which are
This is where the time courses of the observed damage patterns
based on empirical analyses.
were recorded (Figure 19). A total of four separate simulations
were performed, corresponding to the investigated charge depths
Evaluation of the calculated time courses
dTNT = {0·5 m; 1·0 m; 1·5 m; 2·0 m}.
Since the Autodyn simulations were performed only for a charge
mass WTNT = 125 kg, the influence of the distance and the depth of
Assessing the impact of underground
the charge is given in the unscaled values rhor and dTNT.
explosions on buildings
Preliminary remarks Criterion number 1: ‘air blast’
In the following, a method of generating a decision aid is For all four charge depths dTNT = {0·5 m; 1·0 m; 1·5 m; 2·0 m},
introduced, with which the effects of underground explosions the time courses of air pressure p1(t) at a height of x = 1·5 m
70 12
d TNT = 0·5 m (Autodyn) d TNT = 1·5 m (Autodyn)
60 d TNT = 0·5 m (Formula) d TNT = 1·5 m (Formula)
d TNT = 1·0 m (Autodyn) 10 d TNT = 2·0 m (Autodyn)
d TNT = 1·0 m (Formula) d TNT = 2·0 m (Formula)
Peak air pressure p1,max: kPa
50
8
40
6
30
4
20
10 2
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Horizontal distance rhor: m Horizontal distance rhor: m
81
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
90
p1,perm = 5 kPa
80 p1,perm = 10 kPa
p1,perm = 15 kPa
Horizontal, safe distance rhor(p1,max): m
p1,perm = 20 kPa
70 p1,perm = 25 kPa
60
50
40
30
20
10
0
0·50 0·75 1·00 1·25 1·50 1·75 2·00
Charge depth dTNT: m
82
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
(radial symmetry)
vy,max
X axis
Air
Sand
Gauge points
Time course
(different depths) t
—
vAmpl(f ) = vAmpl(f )/ vAmpl(f *)
v1,perm (f )
—
vy,max·vAmpl (f )
f f
f* f*
1400
Reinforced concrete frames (Lu et al., 2002)
Reinforced concrete structures (AASTP-1)
Permissible particle velocity v1,perm (f ): mm/s
800
600
400
200
0
0 20 40 60 80 100 120 140 160 180 200
Frequency f : Hz
83
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
30. vhor,max vAmpl f * £ v1,perm f * consistent with the fact that surface waves experience significantly
less dampening than compression and shear waves and thus
have a longer range and therefore a greater destructive
The following preliminaries need to be carried out for the DFT potential for shallow foundations. Since surface waves can also
induce vibrations in buildings with sublevels, the limit curve
■ smoothing of time courses to reduce noise and unrealistic corresponding to the foundation level x = −1·0 m was used for
frequencies merging the three damage patterns in Figure 29.
■ conversion of the numerically determined time courses to a
constant sampling interval Ts (vhor(t) → vhor(tn), tn = nTs, Criterion number 3: ‘inadmissible inclination of
n Î N) foundations due to subsidence’
■ zero padding: an expansion of the signal sequence vhor[n] by Subsidence due to dynamic stimulation of the soil such as under
appending zeros in order to calculate the Fourier spectrum machine foundations and traffic infrastructure during earthquakes
vAmpl ( fm) = vAmpl[m] ( fm = mDf, m Î N) with a sufficiently or even in the far field of explosions is mainly caused by shear
fine resolution strains on passage of shear waves (Hsu and Vucetic, 2004;
■ multiplying the signal sequence vhor[n] with a suitable Vucetic, 1994). This requires a certain sensitivity to vibration and
window sequence w[n] to reduce leakage effects. a tendency to grain rearrangements in sand. The plastic distortions
start to occur when the linear threshold shear strain gtl is
Figure 24 is a comparison of different permissible particle exceeded. The granular structure collapses completely when the
velocities v1,perm( f ). The values according to DIN 4150-3 shear strain gxy exceeds the volumetric threshold shear strain gtV.
(DIN, 1999) are clearly the most conservative. This is mainly due The latter is subject to a certain spread, depending on the
to the fact that these thresholds preclude any form of damage, plasticity index Ip (Figure 26). For granular soils such as sand and
while the criteria according to AASTP-1 (Nato, 2010) and Lu et gravel, Ip needs to be set to zero.
al. (2002) permit damage corresponding to a loss of rigidity by
40%. In the present case, the DIN values for brick residential The shear strain gxy from the Autodyn simulations are input
buildings were used. values for the evaluation with regard to the damage pattern
160
Depth of gauge point x = −1m
Depth of gauge point x = −3·5m
140
Safe horizontal distance (criterion number 2)
120
100
r hor,safe: m
80
60
40
20
0
0·50 0·75 1·00 1·25 1·50 1·75 2·00
Charge depth dTNT: m
84
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
0·60
γ tl (Vucetic,1994)
γ tv,U (Vucetic,1994)
γ tv,U (Hsu and Vucetic, 2004)
0·50 γ tv,M (Vucetic,1994)
γ tv,M (Hsu and Vucetic, 2004)
γ tv,O (Vucetic,1994)
γ tv,O (Hsu and Vucetic, 2004)
0·40
Plasticity Ip
0·30
0·20
0·10
0 −6
1∙0 × 10 1∙0 × 10−5 1∙0 × 10−4 1∙0 × 10−3 1∙0 × 10−2
Linear and volumetric threshold shear strain γtl and γtv
Figure 26. Linear and volumetric threshold shear strain gtl(IP), gtv(IP)
(Hsu and Vucetic, 2004; Vucetic, 1994)
‘inadmissible angular distortion of foundations due to This corresponds to vertical sagging with a decrease in
subsidence’. For quantitative determination of the subsidence, the the porosity to nmin. For gxy £ gtl, no sagging occurs and
following two assumptions were made. the porosity remains unchanged at the initial value of
porosity n0.
■ The decrease in volume corresponds to the decrease in the
pore volume. The granular structure does not experience a As shown in Figure 27, vertical distortion ex can be calculated in a
deformation as a result of particle rearrangement. given point according to Equation 32. Integration over the depth in
■ A linear relationship exists between the shear strain gxy and a certain horizontal distance rhor from the centre of explosion yields
the porosity n1 after passage of the wave (Equation 31). For sags (Equation 33). In the present case, the integration was carried
shear strains gxy ≥ gtv,M, the soil particles undergo particle out only in the depth range −6 m £ x £ −1 m, corresponding to the
rearrangement and the maximum relative density is reached. depths of the integration points in the numerical model.
∆Vp
n1(γxy) ∆n
Vp,0
n0
Pores
Vp,1(γxy)
Soil particles
1 – n0
1 – n0
Soil particles
VS
VS
(unchanged)
(a) (b)
85
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
E
the considered curve. Thus, for a given charge depth dTNT, the
upmost of the three intersections is in general decisive and
s¼ ex dx
the upper part of the diagram that lies above all three curves
33. x designates safe distances. In the case of soil that is not prone to
subsidence, the upmost intersection of the two curves that
correspond to air blast (criterion number 1) and vibrational
Subsidence, tilting and angular displacements of the foundations stimulation of the structure (criterion number 2) is relevant.
can lead to unacceptable strains and stresses in the structure. The
angular displacements were calculated according to Equation 34 When comparing the air blast criterion with the criterion for
by reference to the differential settlements to the corresponding vibrational stimulation of the structure, it is noticeable that the
horizontal distance. For the considered charge depths dTNT, the air blast criterion at low charge depth is decisive, while for
corresponding curves tan b = f (rhor) are shown in Figure 28. The larger charge depths, the vibrational excitation of the structure
safe distance rhor,safe is reached when the angular rotation falls determines the safe distance. This is logical, as with increasing
below a limit value. According to DIN EN 1997-1, appendix H charge depth, the proportion of the explosive energy that is
(DIN, 2009), the criterion for freedom of damage was assumed as transferred into the soil and that propagates in the form of direct
tan b = 1/500. ground shock waves increases.
0·010
tan β perm = 1/500
tan β (dTNT = 0·5 m)
tan β (dTNT = 1·0 m)
0·008 tan β (dTNT = 1·5 m)
tan β (dTNT = 2·0 m)
Angular displacement tanβ
0·006
0·004
0·002
0
0 20 40 60 80 100 120 140 160 180 200
Horizontal distance rhor: m
86
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
140
120
100
80
Example: d TNT = 1∙35 m → r hor,safe = 120 m
60 (criterion number 2)
40
20
0
0·50 0·75 1·00 1·25 1·50 1·75 2·00
Charge depth dTNT: m
criterion in Figure 29. This may seem surprising at first, since for air pressure as well as the timings of the horizontal particle
the shallowest charge depth dTNT = 0·5 m, the least energy is velocity and the shear strain (both in the subsoil). The mentioned
transmitted to the soil and the direct ground shock wave should parameters are respectively decisive for the three considered
thus have a much shorter range. However, it seems that for the damage patterns air blast, vibrational stimulation of the footing of
shallowest charge depth dTNT = 0·5 m – if compared to the other the structure and inadmissible angular distortion of foundations
charge depths considered in this study – a greater part of the due to subsidence. To keep track of the blast wave over great
explosive energy propagates in the form of Rayleigh waves. distances, the modelling strategy of the moving window was
These are surface waves that tend to decay much slower with developed. The simulations were performed with a charge mass of
increasing distance than compression waves and shear waves. 125 kg of TNT at different charge depths between 0·5 and 2·0 m
Additionally, the effect fits in principle to the statement that air- below ground level and for dry sand.
induced, indirect ground shock waves have greater displacement
and acceleration amplitudes than direct ground shock waves The safe distance rhor,safe at which a building will remain
(Cook et al., 1962). undamaged in the event of a near-surface, underground explosion
can be taken from Figure 29. The diagram thus allows a quick
The diagram in Figure 29 applies to a charge mass of WTNT = and easy answer to the question: how large must the minimal
125 kg and a dry, medium-dense sand as foundation soil. For horizontal distance between a building and the epicentre of an
other load masses and other ground conditions, separate charts explosion – characterised by the explosive, the charge mass and
must be created. depth – be so that the building remains undamaged?
Conclusion Acknowledgements
The aim of this study was to develop a procedure for the creation The material presented in this paper is based on the study ‘Effects
of a decision aid with which the impact of shock waves induced of Artificially Weapon-induced Earthquakes’ supported by the
by underground, near-surface explosions on buildings can be German Federal Ministry of Defence. The authors are indebted to
estimated. Dipl.-Ing. (FH) Heidi Noack for the initiation of this study and
her continuing interest in further work on this topic.
First, the theoretical basics of shock waves were explained.
The influence of the relevant parameters – state of stress and REFERENCES
strain, form of the compression curve and strain rate – was Anderson CE (1987) An overview on the theory of hydrocodes.
demonstrated. International Journal of Impact Engineering 5(1–4): 33–59,
http://dx.doi.org/10.1016/0734-743X(87)90029-7.
Numerical simulations were conducted with the commercial Ansys Inc. (2013) ANSYS Mechanical User’s Guide – Release
hydrocode Autodyn. The aim was to determine the timings of the 15.0. Ansys Inc., Canonsburg, PA, USA.
87
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
Bessette GC (2008) Modeling blast loading on buried reinforced Shock and Impact Loads on Structures (Lok TS, CI-Premier
concrete structures with Zapotec. Shock and Vibrations 15(2): Conference Organisation and Doboku Gakkai Impact Problem
137–146, http://dx.doi.org/10.1155/2008/719215. Committee (eds)). CI-Premier Conference Organisation,
Century Dynamics Inc. (2005a) AUTODYN Theory Manual – Singapore, vol. 4, pp. 361–368.
Revision 4.3. Century Dynamics Inc., Concord, CA, USA. Lu Y and Wang Z (2006) Characterization of structural effects
Century Dynamics Inc. (2005b) AUTODYN Remapping Tutorial – from above-ground explosion using coupled numerical
Revision 4.3. Century Dynamics Inc., Concord, CA, USA. simulation. Computers and Structures 84(28): 1729–1742,
Cook MA, Keyes RT and Ursenbach WO (1962) Air blast and http://dx.doi.org/10.1016/j.compstruc.2006.05.002.
ground shock waves generated at long distances from Lu Y, Hao H and Ma G (2002) Experimental investigation of
demolitions of high explosives. Journal of Applied structural response to generalized ground shock excitations.
Meteorology 1: 91–101, http://dx.doi.org/10.1175/1520-0450 Experimental Mechanics 42(3): 261–271, http://dx.doi.org/10.
(1962)001%3C0091%3AABAGSW%3E2.0.CO%3B2. 1007/BF02410981.
DIN (Deutsches Institut für Normung e.V.) (1999) DIN 4150-3: Lu Y, Wang Z and Chong K (2005) A comparative study of buried
Erschütterungen im Bauwesen, Teil 3: Einwirkungen auf structure in soil subjected to blast loading using 2D and 3D
bauliche Anlagen. Beuth, Berlin, Germany (in German). numerical simulations. Soil Dynamics and Earthquake
DIN (2009) DIN EN 1997-1: Eurocode 7: Entwurf, Berechnung Engineering 25(4): 275–288, http://dx.doi.org/10.1016/j.
und Bemessung in der Geotechnik – Teil 1: Allgemeine soildyn.2005.02.007.
Regeln. Beuth, Berlin, Germany (in German). Meyers MA (1994) Dynamic Behavior of Materials. Wiley, New
Drake JL and Little CDJ (1983) Ground Shock from Penetrating York, NY, USA.
Conventional Weapons. US Army Engineer Waterways Nato (North Atlantic Treaty Organization) (2010) Allied
Experiment Station, Vicksburg, MI, USA. Ammunition Storage and Transport Publication 1 (AASTP-1):
Drake JL, Smith EB and Blouin SE (1989) Enhancements of the Manual of NATO Safety Principles for the Storage of Military
prediction of ground shock from penetrating weapons. Ammunition and Explosives. Defence Investment Division,
Proceedings of the 4th International Symposium on the North Atlantic Treaty Organization International Staff,
Interaction of Non-nuclear Munitions with Structures, Brussels, Belgium.
Panama City Beach, FL, USA. Omidvar M, Iskander M and Bless S (2012) Stress–strain
Fišerová D (2006) Numerical Analyses of Buried Mine Explosions behaviour of sand at high strain rates. International Journal of
with Emphasis on Effect of Soil Properties on Loading. Impact Engineering 49: 192–213, http://dx.doi.org/10.1016/j.
Cranfield Defence and Security, Shrivenham, UK. ijimpeng.2012.03.004.
Grujicic M, Pandurangan B and Cheeseman BA (2006) The effect Showichen A (2008) Numerical Analysis of Vehicle Bottom
of degree of saturation of sand on detonation phenomena Structures Subjected to Anti-tank Mine Explosions. Cranfield
associated with shallow-buried and ground-laid mines. Shock Defence and Security, Shrivenham, UK.
and Vibration 13(1): 1–21, http://dx.doi.org/10.1155/2006/ Steidinger M and Krüning B (1991) Mustersicherheitsanalyse
652405. nach §7 Störfall V für eine Sprengstoffabrik. Bundesanstalt
Heinze H (1987) Sprengtechnik – Anwendungsgebiete und für Materialforschung und -prüfung (BAM), Berlin, Germany
Verfahren. Deutscher Verlag für Grundstoffindustrie, Leipzig (in German).
Stuttgart, Germany (in German). Streitkräfteamt Abt V Infrastruktur (2007) Handbuch für die
Hiermaier S (2008) Structures under Crash and Impact. Springer, Grundlagen zur Bemessung von Bauwerken gegen die
New York, NY, USA. Wirkung konventioneller Waffen. Streitkräfteamt Abt V
Hsu CC and Vucetic M (2004) Volumetric threshold shear strain Infrastruktur, Bonn, Germany (in German).
for cyclic settlement. Journal of Geotechnical and Vucetic M (1994) Cyclic threshold shear strains in soils. Journal
Geoenvironmental Engineering 130(1): 58–70, http://dx.doi. of Geotechnical Engineering 120(12): 2208–2228, http://dx.
org/10.1061/(ASCE)1090-0241(2004)130:1(58). doi.org/10.1061/(ASCE)GT.1943-5606.0001461.
Laine L (2001) Numerical simulations of ground shock attenuation Wang Z, Lu Y, Hao H and Chong K (2005) A full coupled
layers for Swedish rescue centres and shelters. In Proceedings numerical analysis approach for buried structures
of the 4th Asia-Pacific Conference on Shock and Impact Loads subjected to subsurface blast. Computers and Structures 83(4–5):
on Structures (Lok TS, CI-Premier Conference Organisation 339–356, http://dx.doi.org/10.1016/j.compstruc.2004.08.014.
and Doboku Gakkai Impact Problem Committee (eds)). Wilkins ML (1999) Computer Simulation of Dynamic Phenomena.
CI-Premier Conference Organisation, Singapore, vol. 4, Springer, Berlin, Germany.
pp. 353–360. Wu C and Hao H (2005) Modeling of simultaneous ground shock
Laine L (2012) Markstötvag. Myndigheten för smhällsskydd and airblast pressure on nearby structures from surface
och beredskap, Karlstad, Sweden, Publ.-Nr. MSB344 explosions. International Journal of Impact Engineering 31(6):
(in Swedish). 699–717, http://dx.doi.org/10.1016/j.ijimpeng.2004.03.002.
Laine L and Sandvik A (2001) Derivation of mechanical properties Wu C and Hao H (2007) Numerical simulation of structural response
for sand. In Proceedings of the 4th Asia-Pacific Conference on and damage to simultaneous ground shock and airblast loads.
88
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license
Geotechnical Research Underground explosions: estimating the
Volume 3 Issue 3 safe distance
Zimbelmann and Boley
International Journal of Impact Engineering 34(3): 556–572, Dr.-Ing. thesis, University of the German Armed Forces.
http://dx.doi.org/10.1016/j.ijimpeng.2005.11.003. Munich, Germany (in German).
Zimbelmann JKR (2015) Beitrag zur Boden-Bauwerk-Interaktion Zukas JA (2004) Introduction to Hydrocodes. Elsevier,
in nichtbindigen Böden infolge hochdynamischer Anregung. Amsterdam, the Netherlands.
89
Downloaded by [] on [02/02/21]. Published with permission by the ICE under the CC-BY license