Vous êtes sur la page 1sur 12

Engineering Structures 22 (2000) 13781389

www.elsevier.com/locate/engstruct

Assessment of structure damage to blasting induced ground


motions
Guowei Ma a, Hong Hao a,*
, Yingxin Zhou b

a
School of Civil and Structural Engineering, Nanyang Technological University, Singapore 639798, Singapore
b
Lands and Estates Organization, Ministry of Defense, 1 Depot Road, Singapore 109679, Singapore

Received 13 February 1998; received in revised form 27 July 1999; accepted 27 July 1999

Abstract

Empirical allowable vibration levels for surface structures generated by underground explosions are usually given in terms of
peak ground particle velocities. Those allowable vibration criteria do not take the structural type, site conditions and ground vibration
frequency contents into consideration. This paper performs a damage assessment of surface structures subjected to ground stress
wave excitation on the basis of a nonlinear dynamic analysis of structures. The ground excitations are generated numerically by a
commercial software autodyn. The validity of the numerical simulated ground motions is proved by comparing the attenuation
law of their peak particle velocities (PPV) with those obtained independently in field tests. A damage index defined in earthquake
engineering, based on a linear combination of the maximum deformation (ductility) and the hysteretic energy dissipation (loading
history), is used. A parametric study of the effects of structural type, charge distance, site conditions, as well as the explosion
conditions is performed by using three typical reinforced concrete (RC) structural models designed on the basis of British Standard
8110. A safe inhabitant distance for surface structures is estimated and results are compared with those obtained on the basis of
empirical allowable vibration criteria. 2000 Elsevier Science Ltd. All rights reserved.

Keywords: Structural damage; Underground explosion; Ground motion; Damage index; Nonlinear analysis; Reinforced concrete frame

1. Introduction vations of low-rise residential buildings. Although they


are excellent criteria for assessing structural safety, their
It is difficult to identify the characteristics of threshold applications might be limited in practice. This is due
ground motion induced by an underground explosion primarily to the fact that they exclude the effects of
that will cause damage to surface structures. The general structural type, structural conditions, site conditions,
conclusion in mining engineering is that the peak particle ground vibration frequency contents and ground
velocity (PPV) is the best index for assessing damaging vibration duration on structural responses. Moreover, the
ground motions to structures [13]. Allowable vibration empirical criteria can not predict quantitatively the dam-
levels in terms of PPV for residential buildings near an age level of structural components and that of the overall
underground explosion have been suggested, based on structure. In other words, the term damage is not very
many field observations in the past several decades. clearly defined. With the development of computational
Critical PPV values were found to be 23 cm/s for struc- method and structural failure theory, damage indices of
tures sited on hard rock, 11 cm/s for those on weak rock structural components, as well as those of the overall
and 6 cm/s for those on soil [4]. These values are also structure, can be numerically predicted.
adopted by a few design codes as allowable vibration Damage assessment of structures under earthquake
levels for surface structures subjected to underground excitation has been widely performed, and many
explosion-generated ground shock [57]. However, assessing methods have been developed [813]. Damage
those values were obtained based mainly on field obser- indices proposed by Park and colleagues [9,10] define
structural damage indices based on a combination of the
maximum deformation and dissipated hysteretic energy.
* Corresponding author. The damage indices predicted by this method were

0141-0296/00/$ - see front matter 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 1 - 0 2 9 6 ( 9 9 ) 0 0 0 7 2 - 3
G. Ma et al. / Engineering Structures 22 (2000) 13781389 1379

proved by experimental study, yielding satisfactory


results in earthquake engineering. For this reason, this
method is employed in the present investigation to assess
the damage of structures subjected to underground
explosion-generated ground motion.
It is known that the properties of ground stress wave
from an underground explosion are different from those
of seismic waves. Ground stress wave usually contains
high-frequency energy and its energy distributes over a
wider frequency band. Because the distances concerned
are of an order of hundreds of meters, its amplitudes are
much higher and the duration is much shorter than those
of seismic waves, for which the distances concerned are Fig. 1. Site configuration of underground explosion.
of an order of tens or hundreds of kilometers. When ana-
lysing the structural responses in earthquake engineer- of an underground chamber in rock mass, the rock mass
ing, vertical ground motion is neglected in most cases and a soil layer above the rock. Explosive TNT is placed
because the dominant frequency of seismic motion is on the chamber floor and will be detonated to simulate
always very small, and it is close around the fundamental stress wave motions on the ground surface. Ground
frequency of a structure and generally activates the first motion is simulated by using the software autodyn
vibration mode in a horizontal direction. Vertical ground [14]. In numerical modelling, air and explosive TNT are
motion generated by an underground explosion, how- simulated by a Euler processor, and are assumed to
ever, can not be neglected because of its large vertical satisfy equations of state (EOS) of ideal gas and Jones
amplitudes and wider dominant frequency band. High- WilkensLee (JWL), respectively. Rock mass and soil
frequency ground motion may activate higher vibration layer satisfy linear EOS and are simulated by a Lagrange
modes of a surface structure including vertical vibration. processor. A piecewise linear MohrCoulomb strength
The vertical component of ground shock could be much model is utilized to calculate the plastic flow of rock
larger than its horizontal component at the area closer mass in the near field from the detonation. A damage
above the detonation. model is developed and incorporated into the strength
In this paper, a parametric study of surface structure model and linear EOS to model the strength and stiffness
response and damage to underground explosion-induced degradation of rock mass after damage. The whole
ground shock is performed. Ground stress waves are domain, including rock, soil, air and TNT, is assumed
generated by a commercial finite difference software, to be axial symmetric. The material constants of explos-
autodyn [14]. PPV of the generated stress waves are ive TNT and air are chosen from the material library of
compared with those of recorded stress waves from inde- the autodyn. The rock mass at the site is granite, and
pendent field tests [15], to verify the validity of the gen- the material constants of rock mass and soil used in
erated ground stress motions. Time histories of stress numerical calculation are listed in Table 1. More detail
waves at various distances from the charge center on the information on the numerical modelling of underground
ground surface are generated and used as input to ana- explosions can be found in a previous paper by the
lyse the responses of reinforced concrete (RC) frame authors [17].
structures. Three typical RC structures, namely a two- The charge weight used in the present study is 100
storey frame, a 24-storey frame and a one-storey factory
are designed according to BS8110 [16] in the study.
Nonlinear responses and damage indices of the three RC Table 1
Properties of rock and soil
frames subjected to generated stress waves at different
locations on the ground surface are calculated. Numeri- Material: Rock
cal results are presented and discussed with respect to
the structural types, surface distances, and site con- EOS: Linear Strength criterion: MohrCoulomb
ditions. They are also compared with those obtained by Reference density 2650 (kg/m3)
Bulk Modulus 3.9 1010 (Pa)
using empirical allowable vibration criteria for surface Shear modulus 2.6 1010 (Pa)
structures.
Material: Soil

2. Ground shock excitation EOS: Linear Strength criterion: Elastic


Reference density 1600 (kg/m3)
Bulk modulus 5 109 (Pa)
Fig. 1 shows the configuration of an underground Shear modulus 3.75 108 (Pa)
explosion considered in the present study, which consists
1380 G. Ma et al. / Engineering Structures 22 (2000) 13781389

ton TNT with a 10 kg/m3 loading density (charge weight component is larger than the vertical component if there
per unit chamber volume). The charge chamber is is no soil layer above the rock; however, it is smaller
embedded at Dc = 46.4 m below the rock surface as than the vertical component if there is a 20 or 40 m soil
shown in Fig. 1. The depth of rock cover is designed layer above the rock. Fig. 3(a) shows a comparison of
based on the regulation of underground storage to pre- response spectra between seismic motion and under-
vent explosive gas overflow and spall of the ground sur- ground explosion-induced ground motion at 100 m sur-
face when an accidental underground explosion occurs. face distance. It can be seen that the dominant frequency
Fig. 2 shows some typical calculated acceleration time of the Elcentro earthquake motion is around 1 Hz. How-
histories in both the vertical and horizontal directions ever, the frequency band of ground motion due to the
on the ground surface, where DSD indicates the surface underground explosion distributes along a wide band
distance measured from the point directly above the from tens to hundreds of Hz. Fig. 3(bd) plots some cal-
explosive, and DS indicates the depth of the soil layer. culated response spectra of ground motions on rock and
It can be seen that the duration of the stress wave is very soil surfaces. On the rock surface, both vertical and hori-
short. The frequency contents of those corresponding to zontal response spectra decrease with the increase of the
different site conditions and different surface distances surface distance. The vertical motion response spectrum
are different. Both ground motion amplitude and fre- at DSD = 50 m is much larger than the horizontal. How-
quency content decrease dramatically as the surface dis- ever, it is close to the horizontal motion response spec-
tance from the detonation increases. In general, the stress trum at distance DSD = 150 m. For the case of a soil
wave energy distributes in a fairly wide band in the fre- site, the ground motion velocity response spectra at high
quency domain. The dominant frequencies for those on frequencies decrease with the increase of the soil layer
the soil surface are lower than those on the rock surface, thickness because the vibration frequency of the soil
and the deeper the soil layer, the lower are the dominant layer is relatively low, so that the soil layer filters out
frequencies of ground motions. At the point directly high-frequency ground motion energy. However, the soil
above the explosive with DSD = 0 m, horizontal ground layer amplifies ground motion amplitudes at low fre-
motion is zero. At DSD = 50 and 150 m, the horizontal quencies. For example, ground motion at DSD = 50 m

Fig. 2. Calculated surface ground motions induced by underground explosion.


G. Ma et al. / Engineering Structures 22 (2000) 13781389 1381

obtained from independent field tests [15]. Fig. 4 shows


the comparison of numerical and field test data for the
attenuation of free field PPV. As can be seen, the
numerically simulated PPV agree well with the field
data. This indicates that the calculated stress wave
motions on the ground surface can be used to represent
underground explosion-induced surface ground motions.

3. Nonlinear dynamic analysis and damage model

Nonlinear dynamic structural responses are solved by


a step-by-step integration procedure using the Newmark-
method. At time t, the incremental equation of
dynamic equilibrium for a discrete structural system can
be written as
[M]{u} [C]{u} [Kt]{u} {F(t)} (1)
{Fu}
where, in this study, [M] is a constant lumped mass
matrix; [C] is a viscous damping matrix; [Kt] is a current
tangent stiffness matrix; {F(t)} is the effective
incremental load calculated from ground motions; {Fu}
is an unbalance force calculated at the end of the last
time step; and {u} is the displacement incremental vec-
tor.
The viscous damping matrix [C] is assumed as Ray-
leigh damping and is expressed as
[C] a[M] b[Kt] (2)

Fig. 3. Calculated response spectra of ground motions.

on the surface of a 20 m deep soil layer has the largest


velocity response values when the frequency is less than
20 Hz.
As a calibration of the numerically simulated wave
motion, the simulated PPV are compared with the PPV Fig. 4. Comparison of attenuation of peak particle velocity.
1382 G. Ma et al. / Engineering Structures 22 (2000) 13781389

The proportionality factors, a and b, are determined The yield moment from equilibrium is
by specifying the damping ratio, l1,2 for the first two


modes of vibration,

l1,2
a
2w1,2

bw1,2
2
(3)
My bh2fc (a1 m) 1 d 1 xy
h

3
m 0.5 (6)

where w1,2 are circular frequencies of the first two


modes. In this study, 5% critical damping is assumed
d1


for the first two modes of vibration. It should be noted xy
that the effect of viscous damping is not important dur- d1 xy d 1 h
a2
ing the loading phase due to the very short excitation h 3 h d1
1 xy
duration. In fact, viscous damping is usually neglected h
in dynamic response analysis of structures subjected to
impulsive type loading. In the present study, however, where b is the width of the cross-section. For a bilinear
it is included in the numerical modelling since its effect monotonic curve, the ratio My/fy plays the role of an
on the free vibration phase could not be ignored. As the effective elastic rigidity (EI)1 of the cross-section, which
vibration characteristics of a structure during the free is important for the calculation of the elastic defor-
vibration phase are dominated by its lower modes, the mations and the dynamic characteristics of a structure.
first two vibration frequencies are used to determine the The post-yield or hardening slope (EI)2, often expressed
mass and stiffness proportional coefficients of the Ray- as a percentage of the effective elastic rigidity (EI)1, can
leigh damping. only be determined from a detailed momentcurvature
The most important task is to determine the nonlinear analysis of the type of the fibre model, or if a detailed
stiffness matrix [Kt] which depends on the M f momentcurvature curve is available. In the absence of
relationship of structural components, i.e. the moment such information the hardening ratio is often arbitrarily
and curvature relationship of each structural element. selected. The present study uses a consistent hardening
The monotonic M f relationship is often taken as ratio of 0.1 to identify the post-failure property of each
bilinear. Usually the coordinates of the effective yield component. Fortunately, as referenced by the state of the
point are established in a simplified way as the bending art report [18], for dynamic analysis the results are not
moment and curvature at yield, which are calculated on very critically dependent on the exact value of the hard-
the basis of the plane-sections assumption. The present ening ratio used.
study uses the formulation suggested by CEB [18], The hysteretic model is required by structural dynamic
which gives the value for the normalized (to the total analysis. Roufaiel and Meyer [19] suggested a hysteretic
height h of the cross-section) depth of the compression momentcurvature relationship based on the bilinear
zone at yield, xy, as monotonic curve shown in Fig. 5. In unloading it


includes stiffness degradation with the maximum inelas-
fc
xy (a1 a2 m) (a1 a2 m)2 (4)
Ecey

2
Ecey
fc
1
d1
h
(m a1) a2
d1
h
2

where a1 and a2 are the mechanical ratios of the tension


and compression reinforcement, respectively; Ec and fc
are, respectively, the Youngs modulus and the uniaxial
compressive strength of concrete; ey is the yield strain
of the reinforcement; d1 is the distance of the reinforce-
ment from the nearest extreme fibre; and m = N/Acfc is
the normalized axial force with compression considered
as positive, where N is axial force and Ac is the concrete
area of the cross-section. From the value of xy the yield
curvature fy is obtained from compatibility
ey


fy (5)
d1
1 xy h
h
Fig. 5. Hysteretic momentcurvature relationship.
G. Ma et al. / Engineering Structures 22 (2000) 13781389 1383

tic deformation. An auxiliary unloading branch is drawn pated energy increment. The ultimate curvature fu is
parallel to the elastic branch of the bilinear skeleton dependent on the curvature ductility ratio of a compo-
curve, until it intersects a line through the origin parallel nent. The average curvature ductility ratio at failure of
to the strain-hardening branch of the skeleton curve. The 32 specimens tested was found to be 14.5 [20]. In the
line connecting this latter point of intersection to the present study, the ductility ratio reported by Banon et
point of extreme previous deformation in the opposite al. [20] as well as the calculated yield curvature is used
direction defines the end of the unloading branch on the to determine the ultimate curvature fu for each compo-
horizontal axis and determines the unloading stiffness of nent. By using a regression equation obtained from
(EI)3. Reloading denoted as the stiffness of (EI)4 in the experimental results with 400 reinforced concrete col-
model is not always straight to the point of the most umns and beams, the strength deterioration parameter b
extreme previous post-yield excursion in the current is found experimentally [9,10]. The values of b for the
reloading direction, but it may include pinching, usual reinforced concrete sections are usually very small;
depending on the shear span ratio, M/Vh, where V is the and it is of the order of 0.05 for reinforced concrete
shear force of the cross-section and h the component members.
depth. Pinching is accomplished by directing the Damage in reinforced concrete structures is usually
reloading branch first towards a point on the elastic concentrated in small areas. Even in the case of signifi-
branch of the skeleton curve at an ordinate equal to that cant damage, only a few joints are affected. Hence, if a
of the intersection of this branch with the line of straight global damage index is to be obtained as an average of
reloading to the most extreme previous deformation the local damage, it is necessary first to define the
point, times m 1. The second part of the reloading locations at which the local damage will be evaluated,
branch leads towards this latter extreme deformation and then to use an appropriate weighting function so that
point. Parameter m assumes the following values more weight is given to the more critical areas. In the
present study, the global damage is obtained as a
m 0 for M/Vh 1.5 (7) weighted average of the local damage at the ends of each
m 0.4M/Vh 0.6 for 1.5 M/Vh 4 (8) element, with the dissipated energy as the weighting
function. The definition of the global damage index is
m 1 for M/Vh 4 (9) given by [10]

In the present study, a nonlinear structural dynamic DiEi


analysis procedure is developed according to the above i1
Dg (11)
hysteretic model.

n

The objective of the present study is to predict the Ei


strength reserve and dynamic response characteristics of i1

a structure with a specific index of damage to its compo- where Di is the local damage index at location i; Ei is
nents and to the entire structure. The damage index is the energy dissipated at location i; and n is the number
widely used in earthquake engineering, which is defined of locations at which the local damage is computed.
as a value normalized with respect to the failure level, According to this definition, the structural damage index
such that a damage index value of unity corresponds to has a maximum value of one. Park et al. [10] calibrated
the subjectively defined failure [12,13]. The local dam- the damage model by analysing many buildings that had
age index denotes the damage levels of components such undergone past earthquakes. On the basis of the statisti-
as columns and beams. In the present study, the damage cal results, they concluded that the global damage index
index for structural components proposed by Park and defined in Eq. (11) has the following definitions: Dg
Ang [9], which was validated by experimental obser- 0.4 represents reparable damage; Dg 0.4 represents
vation, is employed to assess damage indices of surface damage beyond repair; and Dg 1 represents total col-
structures subjected to underground explosion-induced lapse.
stress wave excitations. It includes two terms that reflect
the influence of the maximum deformation and the
absorbed hysteresis energy as
4. Three typical structures

D
fm fy
fu fy

b
fuMy
dE (10) Structures are idealized as plane frames subjected to
simultaneous horizontal and vertical ground excitations.
where fm is the maximum curvature of a component; fu All the mass of the structure is lumped at the ends of
is the ultimate curvature under monotonic loading; fy each element. The P effect is taken into account
and My are yield curvature and moment, respectively, by adding the geometric stiffness to the column stiffness,
calculated by Equations (5) and (6); and dE is the dissi- using the axial forces produced by the static loads. The
1384 G. Ma et al. / Engineering Structures 22 (2000) 13781389

shear deformation of the element is, however, neglected. m thick soil layers are analysed. Fig. 6 shows the three
Three structural models, namely a two-storey RC build- structural models.
ing, a single-storey factory and a 24-storey RC high-rise The two-storey frame (Fig. 6) consists of six columns
building designed according to BS8110 [16] and located and four beams. All the beams are designed as 230
on a rock site, and on soil sites with 10, 20, 30 and 40 400 mm2 with 2.5% tensile reinforcement and 1.5%

Fig. 6. Three moment resistant RC frame models.


G. Ma et al. / Engineering Structures 22 (2000) 13781389 1385

Table 2
Column size and reinforcement of the two-storey frame

Floor External column Internal column

Size (mm mm) Height (m) Reinforcement Size (mm mm) Height(m) Reinforcement

3rd2nd 200 300 3 2% 250 300 3 2%


2nd1st 200 300 3.5 2% 250 300 3.5 2%

compressive reinforcement. The sizes and reinforce- springs and dashpots to each support. The two low-rise
ments of the columns are listed in Table 2. The single- buildings are supported by shallow foundations, and the
storey factory shown in Fig. 6 has a span length of 20 24-storey high-rise structure is supported by piles. The
m. Its roof is designed as 230 500 mm2 with 2.5% stiffness coefficients of the foundations [21] are calcu-
tension reinforcement and 1.5% compressive reinforce- lated by the formulas given in Table 4, where B is the
ment as well. The columns of the factory are designed half length of a square shallow foundation; n is Poissons
as 230 600 mm2 with a total reinforcement ratio of ratio of soil; G is the effective shear modulus of soil; d
2%. The beams of the 24-storey building are the same is the equivalent radius of pile; l is the pile length; Ep
as those of the two-storey frame; the sizes and reinforce- and Es are Youngs moduli of pile and soil, respectively.
ment ratios of columns are listed in Table 3. The foundation radiation damping coefficients, which
The concrete grade is C30 with fc = 3 107 N/m2 and cannot be obtained straightforwardly, are also given in
the steel yield strength is 4.6 108 Pa. A consistent the latter reference. In the present study, the constants
cover depth of 40 mm is used for each element. Thus, used are: B = 0.5 m; d = 0.226 m; Ep = 2.5 1010 N/m2;
the strength and stiffness properties of the elements such Es = 8.86 107 N/m2; the pile length is the same as the
as My, fy and (EI)i are calculated automatically according soil layer depth.
to Equations (5) and (6) and the assumptions as
described in the preceding part. The mass density of
reinforced concrete is 2400 kg/m3. Masses of beams, col- 5. Local damage indices
umns and floors or roof with 5 m floor length are concen-
trated at the corresponding nodes. The depth of each 5.1. Dynamic response and damage of the two-storey
floor slab of the 24-storey building and the 2-strory building
building as well as the roof of the factory is 175 mm.
Static load applied to the structures is combined by The first two natural frequencies of the two-storey
design dead load Gk and imposed load Qk based on building model are calculated as 4.71 and 16.38 Hz,
BS8110 which is far below the dominant ground vibration fre-
Total load 1.4Gk 1.6Qk (12) quencies as shown in response spectral curves in Fig. 3.
Fig. 7 illustrates the calculated local damage indices
based on the damage model defined by Park and Ang
The dead load includes selfweights of the components [9] for the frame located at different surface distances
like columns, beams and slabs at each floor. The and on sites with different conditions. It should be noted
imposed load of 2.5 kN per square meter is specified. that damage is usually limited to the areas close to the
The soilstructure interaction effect is included in the two ends of an element. The straight line joining the
analysis of structures located on a soil site by adding local damage indices at the two ends of each element
does not mean the damage spreading over the entire
Table 3 element, rather the line only indicates local damage
Column size of 24-storey building occurring at the joints of that particular element.
As can be seen, at a surface distance of 50 m, the
External column Internal column first storey columns are severely damaged and the most
Floor Reinforcement
(mm mm) (mm mm)
significant damage occurs at the bottom of the columns.
25th19th 200 300 250 300 2% Some minor damage also occurs in beams and the
19th16th 200 300 250 400 2% second storey columns, indicating that all the structural
16th13th 200 350 250 600 2% members undergo inelastic deformations. At a surface
13th10th 250 350 250 700 3% distance of 150 m, the damage to the first storey columns
10th7th 250 400 250 800 3%
7th4th 250 500 250 900 4% are still the most severe as compared to the other struc-
4th1st 250 600 250 1000 5% tural members, but the damage indices are less pro-
nounced due to the attenuation of ground motions. When
1386 G. Ma et al. / Engineering Structures 22 (2000) 13781389

Table 4
Dynamic stiffness of shallow or pile foundation

Foundation type kx ky kq

Shallow (square) 9GB 4.54GB 3.6GB


2n 1n 1n
Pile dEs(Ep/Es)0.21 1.9Esd
l
d
2/3
Ep
Es
l/dEp/Es


0.15d 3Es
Ep
Es
0.75

Fig. 7. Local damage indices of components of the two-storey frame.

the structure is located on the site with a 20 m thick soil 5.2. Dynamic response and damage of the single-
layer, damage to its members is most severe. This is storey factory
because of the amplification effect on ground motions
by the soil layer, as shown in Fig. 3. Nevertheless, if the The first two natural frequencies of the factory model
soil layer is 40 m thick, the response of structures will shown in Fig. 6 are calculated as 3.91 and 4.93 Hz,
decrease. This is because of the attenuation of ground which are lower than that of the two-storey frame due
motion amplitudes due to stress wave propagating to its relatively large span. Fig. 8 illustrates the damage
through a thick soil layer. The above observations indi- indices at different structural sections for structures on
cate that a shallow soil layer above the rock mass might different sites. Plastic deformations occur at all the struc-
cause more severe damage to structures due to the soil tural joints. The most severe structural damage occurs at
amplification effect on stress waves at low frequencies. the two supports. Similar to those observed above for
When the soil layer is thick enough, however, the struc- the two-storey structure, the damage level to the struc-
tural damage will be reduced because a very thick soil ture rapidly reduces as the surface distance DSD
layer attenuates stress waves transmitted from rock increases. Damage to the structure sited on the 20 m
mass significantly. thick soil layer is again the most severe. The most sever-
G. Ma et al. / Engineering Structures 22 (2000) 13781389 1387

Fig. 8. Local damage indices of factory structure.

ely damaged joints are N5, N9 and N7. Other joints other stories are only subjected to minor damage or are
experience minor to moderate damage only. without damage. The damage index of the first storey
exceeds 0.40, which is a critical level according to Park
5.3. High-rise RC building and Angs definition [9]. If the surface distance is 150
m, all the storey damage indices are small, namely the
The natural frequencies of the high-rise building are structure is safe. The results shown in Fig. 9 all indicate
much lower than the two low-rise buildings, as 0.56 and that only the low levels of high-rise RC buildings will be
1.53 Hz for the first two modes. Therefore, the response damaged when a 100 ton underground explosion occurs.
of high-rise buildings to ground shock is quite different However, the structure may collapse because the damage
from that of low-rise buildings. Fig. 9 shows the storey at low levels might cause instability to the whole build-
damage indices, which are calculated by the weighted ing.
average of the local damage indices of all columns and For cases when the structure is located on a soil site,
beams on the floor, using Eq. (11) when the building is the response characteristics are different if the soil layer
located at different surface distances and on different thickness is different. The storey damage indices have
sites. The first two stories of the building on the rock rather large values when the structure is located at a sur-
site are damaged when the surface distance is 50 m, but face distance of 50 m. The severe damage in the flexible
1388 G. Ma et al. / Engineering Structures 22 (2000) 13781389

Fig. 9. Storey damage indices of the 24-storey frame.

frame located on a soil site is probably due to the rela- 70 m, only minor structural damage is expected. It
tively low stress wave frequency contents and very should be noted that PPV = 0.23 m/s occurs at about a
high amplitudes. surface distance DSD = 170 m, where structures suffer
minor damages only according to the present numeri-
cal results.
6. Global damage indices When the three structures are located on a soil site,
their damage indices are larger. Depending on the soil
As defined in Eq. (11), the global damage index can layer thickness, the surface distance corresponding to the
be estimated by the weighted function of local damage severe structural damage index varies from 90 to 110 m
indices of structural members. Fig. 10 shows the global for the low-rise structures, and from 110 to 170 m for
damage indices of the three structures located at different the high-rise structure. It should be noted again that only
surface distances and on sites with different conditions. very minor damage will occur to the three structures
The damage index 0.4 corresponding to severe overall when PPV is about 0.06 m/s, which occurs at a surface
structural damage, is also shown in the figure for easy distance of approximately 280 and 250 m when the soil
comparison. As can be seen, the damage indices of the layer thickness is 20 and 40 m, respectively.
three structures decrease rapidly as the surface distance
DSD increases. In general, the 24-storey high-rise struc-
ture suffers the most severe damage, followed by the 7. Conclusions
single-storey factory structure. Damage to the two-storey
RC frame structure is relatively less significant among Nonlinear dynamic structural analysis has been car-
the three structural models. Site conditions also signifi- ried out in the present paper to study the damage of
cantly affect the structures performance. The damage structures subjected to an accidental underground
indices are smallest when the structures are sited on a explosion. Ground motions generated by the commercial
rock surface. The indices of damage to all the three software autodyn were used as input in structural
structural models increase if they are sited on a soil site. response analysis. The numerical results have shown
As can be seen, the damage indices are largest if the soil that, although the high-rise building has a natural fre-
layer thickness is 10 and 20 m. As the soil layer thick- quency far away from the dominant ground motion fre-
ness increases further from 20 m, the damage indices quency, it is most vulnerable due to the concentrated
decrease due to the attenuation of stress waves as dis- damage at low stories, which greatly affects the overall
cussed above. structural stability. Structures sited on a soil site will
When the three structures are sited on a rock surface, usually suffer more damage than those sited on a rock
the surface distance corresponding to the severe struc- surface. Most severe damage occurs to structures site on
tural damage index is about 50 m for the two-storey a 20 m thick soil layer. When the soil layer thickness
structure, and about 70 m for the other two structures. increases above 20 m, however, structural damage levels
If structures are built at a surface distance larger than will be reduced. Structural responses and damage not
G. Ma et al. / Engineering Structures 22 (2000) 13781389 1389

References

[1] Dowding CH. Blast vibration monitoring and control. Englewood


Cliffs, NJ: PrenticeHall, 1985.
[2] Nicholls HR, Johnson CF, Duvall WI. Blasting vibrations and
their effects on structures. Bureau of Mines, Bulletin 656, Wash-
ington DC, 1971.
[3] Odello RJ. Origins and implications of underground explosives
storage regulations. Technical memorandum, No. 51-80-14,
Naval facilities engineering command, USA, July 1980.
[4] Gustafsson R. Swedish blasting technique. Gothenburg, Sweden:
SPI, 1973.
[5] Odello RJ. Ground shock effects from accidental explosions.
Technical memorandum, No. TM M-51-76-22, Department of the
Army Picatinny arsenal, USA, November 1976.
[6] North Atlantic Treaty Organization. Document AC/258-D/258:
Manual on NATO safety principles for the storage of ammunition
and explosives. Brussels, Belgium, 1977.
[7] US Department of Defense Explosive Safety Board Ammunition
and Explosive Safety Standards. DOD 6055.9-STD. Washington,
DC, 1996.
[8] Chung YS, Meyer C, Shinozuka M. Seismic damage assessment
of reinforced concrete members. Technical report NCEER-87-
0022, National Center for Earthquake Engineering Research, Buf-
falo, New York, October 1987.
[9] Park YJ, Ang AHS. Mechanical seismic damage model for
reinforced concrete. Journal of Structural Division, ASCE
1985;111(4):72239.
[10] Park YJ, Ang AHS, Wen YK. Seismic damage analysis of
reinforced concrete buildings. Journal of Structural Division,
ASCE 1985;114(4):74057.
[11] Izuno K, Iemura H, Yamada Y. Seismic damage assessment of
RC structures using different hysteretic models. Memorandum of
the Faculty of Engineering, Kyoto University 1993;55(1):118.
[12] Miramontes D, Merabet O, Reynouard JM. Beam global model
for the seismic analysis of RC frames. Earthquake Engineering
and Structural Dynamics 1996;25:67188.
[13] Koyluoglu HU, Nielsen SRK, Cakmak AS, Kirkegaard PH. Pre-
diction of global and localized damage and future reliability for
RC structures subject to earthquakes. Earthquake Engineering
and Structural Dynamics 1997;26:46375.
[14] Century Dynamics, Inc. autodyn user manual. Revision 3.0,
1997.
[15] Murrell DW, Joachim CE. The 1996 Singapore ground shock test.
Department of the Army, Waterways Experiment Station (WES),
September 1996.
[16] British Standards Institution. BS8110: structural use of concrete,
Part 1, 1985.
[17] Ma GW, Hao H, Zhou YX. Modeling of wave propagation
induced by underground explosion. Computers and Geotechnics
1998;22(3-4):283303.
[18] Comite Euro-Internatonal Du Berton (CEB). RC frames under
earthquake loading: state of the art report. London: Thomas Tel-
Fig. 10. Global damage indices for different types of structures. ford, 1996.
[19] Roufaiel MSL, Meyer C. Analytical modeling of hysteretic
behavior of R/C frames. Journal of Structural Engineering
1987;113:42944.
only depend on the peak ground motions, but also on [20] Banon H, Biggs JM, Irvine HM. Seismic damage in reinforced
structural type, ground motion frequency contents and concrete frames. Journal of Structural Engineering, ASCE
1981;107:171329.
site conditions. It is proved that the predictions of safe [21] Gazetas G. Foundation vibration. In: Fang HY, editor. Foundation
inhabitant building distance based on the PPV of ground engineering handbook. 2nd ed. New York: Van Nostrand Rein-
motion alone is very conservative. hold, 1990.