Engineering Structures 300 (2024) 117166

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

A modal-based method for blast-induced damage assessment of reinforced

concrete columns: Numerical and experimental validation
Yanchao Shi a, b, Shaozeng Liu b, Ye Hu b, Zhong-Xian Li a, b, *, Yang Ding a, b
a
Key Laboratory of Coast Civil Structural Safety of the Ministry of Education, Tianjin University, Tianjin 300350, China
b
School of Civil Engineering, Tianjin University, Tianjin 300350, China

A R T I C L E I N F O A B S T R A C T

Keywords: Reinforced concrete (RC) columns play a crucial role in the overall performance of modern RC frame buildings
Reinforced concrete column under blast loading. The RC columns under blast loads might experience various irreversible consequences such
Blast-induced damage assessment as concrete cracks, crushing and spalling, changing the status of both modal properties and load-carrying ca­
Modal-based damage assessment method
pacity. In this paper, the blast-induced damage assessment of the RC columns involving the measurement of
Finite element analysis
Field explosion test
modal parameters was introduced and validated. Firstly, a modal-based method for blast-induced damage
assessment of RC columns was proposed, in which the damage is characterized as the reduction of axial bearing
capacity and further expressed as the explicit function of the vibration frequencies and the mass-normalized
displacement mode shapes. Then, the finite-element analysis program LS-DYNA was employed to establish the
numerical model of RC columns under axial compression loads and lateral blast loads. The residual load-bearing
capacities and modal properties of the RC columns were illustrated. By using the numerical results collected in
the present study, the validity of the proposed damage assessment method was discussed and proved numeri­
cally. Further, field explosion tests and experimental modal tests were conducted, and the results showed that the
modal-based damage assessment method exhibited remarkable agreement with the blast damage determined by
residual bearing capacities. Based on the numerical and experimental results, the proposed modal-based damage
assessment method is applicable for the non-destructive evaluation of blast-induced damage of RC columns.

1. Introduction experimental data for the validity of numerical modelling methods given
the drawbacks of explosion tests. To date, numerical simulation is still a
Accidental explosions and terrorist bombings are continuous threats common method to investigate the effects of various influencing factors
to the safe maintenance of modern cities. Reinforced concrete (RC) on blast loads and blast-induced behaviours of RC members [5,8–12].
frame structures are commonly designed as the buildings to resist Among the published literature, assessing the blast-induced damage
possible blast loads because of the huge mass and high blast resistance. of RC columns has exuded fantastic appeal. Pressure-impulse (P-I) dia­
RC columns are considered critical components of modern RC frame grams which depict iso-damage curves of RC columns under combined
buildings when subjected to blast loads from explosion events. peak overpressure and resultant impulse were widely established and
Convincing evidence has been observed that severe damage caused by played a significant role in blast resistance design [13–17]. Meanwhile,
blast loads will result in the malfunction of supported columns, leading empirical equations with scaled distance and charge mass as inputs were
to a structural state of partial failure or even progressive collapses [1–4]. also proposed by many researchers through numerical parametric
To mitigate casualties and property losses, experimental and nu­ analysis [11,18–20]. It should be recognized that both the P-I diagrams
merical studies have been carried out for years to investigate the dy­ and empirical equations are only applicable when the explosive infor­
namic behaviour and load-carrying capacity of RC columns subjected to mation or the blast loads on structural facades are available. However, it
lateral blast loading [5–12]. However, owing to the huge cost and safety is still a challenge to derive the reliable information on explosive or blast
requirements, field explosion tests were occasionally performed and loads after the explosion disasters. To overcome the defects of P-I dia­
sometimes failed to acquire desirable outcomes. In most cases, near-field grams and empirical equations with scaled distance and charge mass as
or close-in explosion tests were carried out to provide reliable inputs, researchers have been devoted to developing empirical

* Corresponding author at: Key Laboratory of Coast Civil Structural Safety of the Ministry of Education, Tianjin University, Tianjin 300350, China.
E-mail address: [email protected] (Z.-X. Li).

https://doi.org/10.1016/j.engstruct.2023.117166
Received 20 July 2023; Received in revised form 22 October 2023; Accepted 10 November 2023
Available online 26 November 2023
0141-0296/© 2023 Elsevier Ltd. All rights reserved.
Y. Shi et al. Engineering Structures 300 (2024) 117166

correlations for the blast-induced damage assessment based on

post-blast measurements. For example, Bao and Li [21] established an
expression of the residual bearing capacity of RC columns after short
distance explosion, which demanded the inputs of relative residual
deflection, axial loading ratio, reinforcement ratio, stirrup ratio and
slenderness. Cui et al. [22] also presented an empirical correlation be­
tween residual bearing capacity and relative residual deflection of RC
columns under close-in explosions. Xu et al. [23] attempted to employ
the embedded piezoceramic smart aggregates (SAs) to detect
blast-induced cracks in concrete columns. The proposed
active-sensing-based approach was more sensitive to internal or through
cracks than surface damage, however, the sensors should be embedded
in the structure in advance, which might not work in the real blast
scenario. Fig. 1. Sketch of damage mapping onto RC column.
It is well known that RC columns will experience various irreversible
consequences such as concrete crack, crushing and spalling under blast respectively; KY and KE are the correction factors used to eliminate the
loading, whereby the modal properties including vibration frequencies, errors arising from the simplifications related to the assumptions of
damping ratios, and mode shapes, would be affected accordingly [24, homogeneous damage to the concrete material. KY = 1.0 and KE = 1.0
25]. In this paper, the correlation between blast-induced damage and have been demonstrated to be reasonable in blast-induced collapse
modal parameters of the RC column is established and validated. First, a analysis of RC frame structures and will be further confirmed in the
modal-based damage assessment method was proposed, in which the present study [4].
blast-induced damage was expressed as the explicit function of vibration ( )
frequencies and mass-normalized displacement mode shapes. Then, the fdmg = KY ⋅fc ⋅ 1 − Dc (2)
finite element analysis program ANSYS/LS-DYNA was employed to ( )
model the dynamic behaviours of RC columns subjected to axial Edmg = KE ⋅Ec ⋅ 1 − Dc (3)
compression loads and lateral blast loads. The effect of blast-induced
Then, the blast-induced damage of RC column defined in Eq. (1) can
damage on residual bearing capacities and modal characteristics of RC
be rewritten as Eq. (4), and the homogeneous damage of concrete ma­
columns were illustrated. The blast-induced damage was calculated
terial can be expressed as Eq. (5). In Eqs. (4) and (5), As and Ac are the
using the proposed modal-based damage assessment method and
cross-section areas of longitudinal rebars and concrete respectively; fs is
showed significant agreement with the blast-induced damage based on
the ultimate strength of longitudinal rebars.
axial bearing capacity. Further, experimental results of four full-scaled
RC columns under close-in blast loads were presented and the validity P KY ⋅fc ⋅(1 − Dc )⋅Ac + fs ⋅As
D = 1− = 1− (4)
of the proposed modal-based damage assessment method was confirmed P0 fc ⋅Ac + fs ⋅As
again. The numerical and experimental results in the present study
( )
showed that the proposed modal-based damage assessment method is 1 fc ⋅Ac + fs ⋅As
Dc = 1 − 1− ⋅D (5)
sufficient to capture the blast-induced damage of RC columns. KY fc ⋅Ac
Consider a foresaid damaged RC column laterally subjected to the
2. Modal-based damage assessment method
unit uniform pressure as depicted in Fig. 2. The bending moment of the
RC column can be calculated as Eq. (6), where Me is the unknown
The common definition of blast-induced damage of RC columns is
bending moment at both ends.
expressed in Eq. (1), where P0 and P are respectively the column’s pre-
blast and post-blast axial bearing capacity [11,13]. M(x) = (L − x)x/2 − Me (6)
P Note that no rotation is allowable at both column ends and suppose
D = 1− (1)
P0 that the column deforms elastically, then Eq. (7) can be established,
where θ(x) is the sectional rotation of RC column; Dm (x) = d if
Previous studies have indicated that RC columns subjected to blast
x ∈ damage zone, otherwise, Dm (x) = 0. d is called the sectional damage
loads generally exhibit three kinds of global damage modes: flexural
considering the effect of concrete damage on flexural rigidity of
damage, shear damage and flexural-shear damage [13,26,27]. To
damaged cross-section, and can be written as Eq. (8), where Ic and Ise
quantify the blast-induced damage of RC columns, the damage mapping
represent the moments of inertia of concrete material and longitudinal
methodology of RC columns proposed by Shi et al. [4] is adopted, which
bars about centroidal axial of cross-section, respectively; Es is the
supposes that the damage merely occurs in the concrete material and is
Young’s modulus of the longitudinal bars.
limited to several damage zones. The location of the damage zones de­
pends on the damage modes of the RC column: if an RC column is
damaged mainly by the shear damage, two identical damage zones at
both ends are presumed, as depicted in Fig. 1(a); if the damage mode is
primarily flexural damage, only one damage zone at the column-mid is
presumed, as depicted in Fig. 1(b). In each damage zone, the concrete
damage is assumed to be homogeneous, and the length of the damage
zone is assumed to be a fraction of the column height, i.e., αL (L is the
column height). α = 1/5 was recommended by Shi et al. [4].
The uniaxial compressive strength and the Young’s modulus of the
damaged concrete are defined as Eqs. (2) and (3) respectively, where Dc
is the homogeneous damage of concrete material; fc and Ec respectively
denote the uniaxial compressive strength and the Young’s modulus of
the undamaged concrete; fdmg and Edmg are the uniaxial compressive
strength and the Young’s modulus of the damaged concrete, Fig. 2. Bending moment under unit uniform pressure.

2
Y. Shi et al. Engineering Structures 300 (2024) 117166

∫ L

Moment redistribution modification factor M/M0

1 M(x)
θ(x) = dx θ(x = 0) = θ(x = L) = 0 (7)
Ec ⋅Ic + Es ⋅Ise 0 [1 − Dm (x) ]
1.0
M/M0 = 1.07-0.07·e2.26d
KE ⋅(1 − Dc )⋅Ec ⋅Ic + Es ⋅Ise
d = 1− (8)
Ec ⋅Ic + Es ⋅Ise 0.9
By substituting Eq. (5) into Eq. (8), the correlation between sectional
damage d and blast-induced damage degree of RC columns D is given as
0.8
Eq. (9).
[ ( )]
KE fc ⋅Ac + fs ⋅As Ec ⋅Ic
d = 1− ⋅ 1− ⋅D ⋅ (9) 0.7 Shear damage, α=1/6
KY fc ⋅Ac Ec ⋅Ic + Es ⋅Ise
Shear damage, α=1/5
In accordance with the moment-curvature correlation of Shear damage, α=1/4
0.6
Euler–Bernoulli beam theory, the unknown bending moment at column Flexural damage, α=1/6
ends Me can be solved by combining Eqs. (6) and (7). If the RC column is Flexural damage, α=1/5
damaged primarily by shear damage, the unknown bending moment at 0.5 Flexural damage, α=1/4
the column ends is determined as Eq. (10). Otherwise, the unknown
bending moment at column ends is determined as Eq. (11) if the RC 0.0 0.2 0.4 0.6 0.8
column is damaged primarily by flexural damage. It is clearly observed
that the bending moment of the damaged RC column depends on the
Sectional damage d
sectional damage d. Fig. 3. Moment redistribution modification factor with different
[ ( ) )]/[ ( ) ] sectional damage.
1
Me = 1 + − 4α3 + 6α2 − 1 ⋅d 1 + 2α − 1 ⋅d ⋅L2 (10)
12
change of the moment redistribution modification factor for the case of
[ ( ) )]/[ ( ) )]
1 α = 1/4 or α = 1/6 is about 10% compared to that of α = 1/5.
Me = 2 − 2 + α3 − 3α ⋅d 1 − 1 − α ⋅d ⋅L2 (11) Considering the weak dependency of moment redistribution modifica­
24
tion factor on the assumed damage length fraction and damage mode, an
Particularly, for RC member that is damaged primarily by flexural exponentially decay equation is recommended as Eq. (15) for practical
damage mode, the bending moment at RC column-mid Mm is calculated usage, in which only the sectional damage is considered.
as Eq. (12). /
{ [ ( ) )]/[ ( ) )]} M M0 = 1.07 − 0.07⋅e2.26d (15)
1
Mm = 3 − 2 − 2 + α3 − 3α ⋅d 1 − 1 − α ⋅d ⋅L2 (12)
24 The sectional damage d can be determined through Eqs. (14) and
(15) when the curvature ratio ψψ0 is given. There should be three sectional
As shown in Fig. 2, the curvature of the damaged cross-section can be
damage d corresponding to three curvature ratios at column top, middle
estimated in accordance with the moment-curvature correlation of
and bottom, respectively. Consequently, the blast-induced damage of RC
Euler–Bernoulli beam theory as Eq. (13), where ψ is the sectional cur­
column D can be obtained by substituting the most unfavourable one
vature at the middle or the ends of the damaged RC column subjected to
into the reverse form of Eq. (9), as presented in Eq. (16).
the unit uniform pressure; ψ 0 denotes the sectional curvature of the
[ ( )]
identical cross-section of the intact RC column subjected to the unit KY Ec ⋅Ic + Es ⋅Ise fc ⋅Ac
uniform pressure; M can be calculated as Eq. (10) if the RC column is D = 1− ⋅ 1− ⋅d ⋅ (16)
KE Ec ⋅Ic fc ⋅Ac + fs ⋅As
damaged primarily by shear damage while Eq. (12) is employed if the
RC column is damaged primarily by flexural damage; M0 = L2 /12 and As demonstrated above, the curvature ratio ψ0
ψ of RC columns under
M0 = L2 /24 are respectively used when the RC column is damaged unit uniform pressure is the unique input parameter of the proposed
primarily by shear and flexural. blast-induced damage assessment method. The modal data of RC column
can be employed to calculate the required sectional curvatures. The
ψ=
M
=
M/M0 ⋅M0
=
M/M0
⋅ψ (13) transfer function matrix H(ω) is given as Eq. (17) [24], where Z, C and K
(1 − d)⋅(Ec ⋅Ic + Es ⋅Ise ) (1 − d)⋅(Ec ⋅Ic + Es ⋅Ise ) (1 − d) 0 respectively represent the matrices of mass, damping and stiffness of the
Subsequently, the sectional damage d is reorganized as the function RC column; ωr , ξr and φr are respectively the circular frequency,
of curvature ratio ψψ0 as Eq. (14). damping ratio and mass-normalized displacement mode shape of the
r-th mode. Therefore, the flexibility matrix F of the RC column can be
M ψ0 obtained by letting ω = 0 as Eq. (18) [28], where f r = ωr /2π is the vi­
d = 1− (14)
M0 ψ bration frequency of the r-th mode.
Eqs. (10) and (12) show that the coefficient MM0 in Eq. (14) (hereafter ( )− ∑
n
φr φTr
(17)
1
H(ω) = − ω2 Z + jωC + K =
called moment redistribution modification factor) is dependent on the r=1
ω − ω2 + 2jξr ωr ω
2
r
sectional damage d and the assumed damage length fraction α, as
illustrated in Fig. 3. Notably, the moment redistribution modification 1 ∑ n
φr φTr
F = K− 1
= (18)
factor MM0 is always less than 1.0 when the RC column is damaged by 4π2 r=1 fr 2
either shear or flexural damage modes, and the moment redistribution
If measuring points on column façade are placed at equal spacing and
modification factor of shear damage mode is greater than those of
numbered in descending or ascending order along column shaft, the
flexural damage mode especially when the sectional damage is over 0.5.
deflection ui and corresponding curvature ψ i of i-th measuring point on
Moreover, the moment redistribution modification factor will expo­
the RC column under unit uniformly distributed pressure can be ob­
nentially decrease with the sectional damage increasing from 0 to 0.8.
tained, as shown in Eqs. (19) and (20) [28], where φr,j is the
Compared to the sectional damage, the assumed damage length fraction
mass-normalized displace of the r-th mode shape at the j-th measuring
α has relatively small effect on the moment redistribution modification
point; l is the distance between two adjacent measuring points on the
factor, e.g., when the sectional damage is 0.8, the maximum relative
column surface. Since the deflection is inversely proportional to the

3
Y. Shi et al. Engineering Structures 300 (2024) 117166

square of the vibration frequencies and the vibration frequencies of the modal test results can be obtained as listed in Table 1. It should be noted
third order and above are much larger than those of the first two modes, that the transducers on column façade is numbered from column top to
the first two vibration modes are sufficient to ensure the calculation column bottom in ascending order.
accuracy and involvement of higher order vibration modes is STEP 3: Calculation of curvature ratio and sectional damage degree.
unnecessary. Curvature ratio at column top: ψ 0/ψ = (u1/l2) / (u’1/ l2) = 0.185/
⎛ ⎞ 1.402 = 0.132.
∑n ∑n ∑n

∑n
φr,i φr,j φ1,i φ1,j φ2,i φ2,j Curvature ratio at column middle: ψ 0/ψ = [(u3 + u5- u4)/l2)] / [(u’3
l ⎜ ⎟
u’4)/l2)]
l
(19) u’5- = (0.808 +0.791–2 ×0.924) /
j=1 j=1 j=1
ui = 2 ≈ 2⎜ + ⎟ +
4π r=1 fr 2 4π ⎝ f1 2 f2 2 ⎠
(3.374 +3.203–2 ×3.634) = (− 0.249/− 0.691) = 0.360.
Curvature ratio at column bottom: ψ 0/ψ = (u7/l2)/ (u’7/ l2) = 0.172/
ui+1 + ui− 1 − 2ui 1.008 = 0.171.
ψi = (20) Given Eqs. (14) and (15), i.e., d = 1 − MM0 ψψ0 & M/M0 = 1.07 −
l2
As mentioned above, the procedure of the proposed modal-based 0.07⋅e2.26d , by iterative calculation the sectional damage at column top,
damage assessment method is depicted in the flowchart as shown in middle and bottom can be determined as 0.935, 0.753 and 0.910.
Fig. 4. Firstly, modal analysis on the intact and damaged RC columns is STEP 4: Calculation of blast damage degree of reinforced concrete
implemented to determine the first two vibration frequencies and mass- column.
normalized displacement mode shapes. Then, flexibility matrices of the From the last step, it is obviously observed that the most severe
intact and damage RC columns are established, and the sectional cur­ damaged region is located at column top, which produces a sectional
vatures at the column top, middle and bottom can be calculated through damage degree of 0.935. When assign the real numbers to the parame­
[ ( )]
Eqs. (19) and (20). Finally, the sectional damage corresponding to the ters of Eq. (16), i.e., D = 1 − KKYE ⋅ 1 − Ec ⋅IEc +E
c ⋅Ic
s ⋅Ise
⋅d ⋅fc ⋅Afcc +f
⋅Ac
s ⋅As
, the blast
column top, middle and bottom is determined using Eqs. (14) and (15), damage degree can be determined as follows:
thereby the blast-induced damage of RC column can be obtained by Ise = 4 × 0.25 × 222 × 3.14 × (250/2 − 25 − 22/2) =
taking the most unfavourable sectional damage into Eq. (16). 3
31682.64mm4 ; Ic = 250×250 − Ise = 3.25 × 108 mm4 ;
To better understand the procedure of the proposed modal-based 12

damage assessment method, a step-by-step example to work out the Ec ⋅Ic + Es ⋅Ise 210⋅31682.64
= 1+ ≈ 1.0007;
blast damage of reinforced concrete column is presented below. The Ec ⋅Ic 30⋅3.25 × 108
column has a three-dimensional size of 250 mm × 250 mm × 3000 mm
As = 4 × 0.25 × 222 × 3.14 = 1519.76mm2 ; Ac = 250 × 250 − As =
and is reinforced by four steel rebars with a diameter of 22 mm and a
60980.24mm2 ;
concrete cover of 25 mm. The unconfined concrete compressive strength
is 30 MP and the ultimate strength of longitudinal rebars are 500 MPa. fc ⋅Ac 40 × 60980.24
= = 0.76247.
The Young’s modulus of concrete and steel are 30 GPa and 210 GPa fc ⋅Ac + fs ⋅As 40 × 60980.24 + 1519.76 × 500
respectively.
[ ( )]
STEP 1: modal analysis and modal parameter extraction. 1.0
D = 1− ⋅ 1 − 1.0007⋅0.935 ⋅0.76247 = 0.713
Modal tests are required to performed on the reinforced concrete 1.0
column before and after explosion, thereby the pre-blast and post-blast
3. Numerical validation

3.1. Finite element model

Using explicit finite element (FE) hydrocode LS-DYNA to model the

transient behaviours of RC structures subjected to blast loads has been a
common method among civil engineering communities owing to its high
fidelity. In the present study, LS-DYNA was employed to investigate the
blast loading effects on the modal properties and axial bearing capacity
of RC columns.

3.1.1. Model construction

As the first step of FE model construction, the selection of element
types and mesh size is dependent on the accuracy and computation cost.
In this paper, the concrete was modelled by 8-node reduced integration
solid element while the longitudinal and transverse reinforcing steels
were modelled using the Hughes-Liu beam element. To prevent hour­
glass associated with reduced-integration solid elements, the Flana­
gan–Belytschko-based hourglass control option available in LS-DYNA
was activated with an hourglass coefficient of 0.1 [11,29,30]. Conver­
gence tests on mesh size declared that a mesh size of 25 mm is suffi­
ciently accurate for modelling concrete and steel bars of RC columns,
and a finer element size will lead to a huge increase in CPU time and risk
of memory overflow [11,31]. Hence, a maximum mesh size of 25 mm for
all elements was utilized in this paper.
Previous studies [32,33] have demonstrated that perfect bond
modelling between steel bar and concrete is adequately accurate in
modelling RC structures under blast loads, especially when the
debonding failure mode is not dominant. Two ways are available to
Fig. 4. Procedure of the proposed modal-based damage assessment method. ensure the perfect bonding connections between concrete material and

4
Y. Shi et al. Engineering Structures 300 (2024) 117166

Table 1
Pre-blast and post-blast modal test results.
Pre-blast modal result Post-blast modal result

Vibration frequency (Hz)

f1 101.74 f2 262.27 f ’1 49.76 f ’2 109.50
Mass-normalized displacement of mode shapes
φ1,1 0.098 φ2,1 -0.239 φ’1,1 0.170 φ’2,1 -0.335
φ1,2 0.273 φ2,2 -0.461 φ’1,2 0.318 φ’2,2 -0.412
φ1,3 0.428 φ2,3 -0.372 φ’1,3 0.410 φ’2,3 -0.260
φ1,4 0.489 φ2,4 0.013 φ’1,4 0.442 φ’2,4 0.031
φ1,5 0.420 φ2,5 0.387 φ’1,5 0.390 φ’2,5 0.300
φ1,6 0.257 φ2,6 0.455 φ’1,6 0.288 φ’2,6 0.412
φ1,7 0.091 φ2,7 0.218 φ’1,7 0.123 φ’2,7 0.249
∑n ∑n ∑n ∑n
j=1 φ1,j 2.056 j=1 φ2,j 0.001 ′
j=1 φ 1,j 2.141 ′
j=1 φ 2,j -0.015

STEP 2: Calculation of lateral deflection under unit uniformly distributed pressure

reinforcing steel material. The first method enables separate construc­ recommended for practical usage [11,29,30].
tion and meshing of different materials and then employs the *CON­
STRAINED_LAGRANGE-IN_SOLID keyword to interact concrete with 3.1.3. Loading sequences
reinforcing steels [29]. The second is to mesh the concrete and re­ A schematic illustration of loading sequences is presented in Fig. 5.
inforcements by simply sharing common nodes on the interface [34,35]. Five successive loading stages are included: initial axial loading and
In the present study, the second method was employed to simulate the stabilizing stage, pre-blast modal exciting loading and free vibrating
perfect bond between reinforcing bars and concrete. The boundary stage, blast loading and free vibrating stage, post-blast modal exciting
conditions should be modelled following the actual support conditions. loading and free vibrating stage, and residual axial loading stage.
In the present study, the column bottom was modelled as fixed ends. The At the initial axial loading stage, uniformly distributed pressure is
column top was modelled without any translational restraint in the axial applied to the top of the RC column to simulate the dead weight of the
direction to ensure the free transmission of the axial loading while other superstructure and remains constant during the whole simulation pro­
degrees of freedom were restrained. cess except for the residual axial loading stage. The residual axial
loading is prepared for the determination of the residual bearing ca­
3.1.2. Constitutive material models pacity of damaged RC columns and blast-induced damage degree. In this
Appropriate material models and material parameters are pivotal in stage, the displacement-controlled compression on the column top holds
the reliable prediction of structural behaviours under transient loads. In until the overall failure of the RC columns.
this paper, the Karagozian & Case Concrete (KCC) model was chosen to The pre-blast and post-blast modal analyses are conducted for com­
simulate the concrete material, and the *MAT_PLASTIC_KINEMATIC parison of modal properties of RC columns before and after blast dam­
(MAT_003) plasticity model was employed to model the reinforcing age. Fig. 6 depicts the configuration sketch of the pre-blast and post-
steels. The major advantage of KCC model is that it allows a single input blast modal analysis. As depicted in Fig. 6, a modal exciting force with
of unconfined compressive strength, while the remaining material pa­ a peak of 20 kN and a duration of 1.0 ms is exerted on the front surface
rameters are automatically generated through a built-in algorithm and of RC column at the quarter-point. The velocity time histories at seven
can be modified by the users. Relevant investigations have elucidated points (S1~S7) on the rear side of the RC column are sampled at a
that the KCC model and the *MAT_PLASTIC_KINEMATIC (MAT_003) sampling rate not less than the Nyquist frequency, which is twice the
material model are reliable in modelling RC columns subjected to blast frequency range of interest [24]. Spectrum analysis software and
loading [13,14,30]. single-degree-of-freedom (SDOF) frequency-domain identification
The strain rate effect of concrete merely employed the dynamic in­ method are used for extracting modal parameters since the coupling
crease factor (DIF), and the empirical functions given by CEB-FIP [36] effect among adjacent modes is negligible for RC structural members
and Malvar & Crawford [37] were employed for the compressive and [24].
tensile strength of concrete respectively. As presented from Eqs. (21) to There are three methods of applying blast loads on RC columns in LS-
(26), fc and ft respectively represent the dynamic compressive strength DYNA, namely the multi-material arbitrary Lagrange Euler (ALE)
at strain rate ε̇c and dynamic tensile strength at strain rate ε̇t ; fc0 and ft0 method, built-in load blast enhanced (LBE) method, and user-simplified
denote the static compressive and tensile strength respectively. pressure-time history method [38]. The ALE method is capable of
/ { modelling blast waves propagation and interaction with surrounding
(ε̇c /ε̇c0 )1.026α if ε̇c ≤ 30s− 1 structures. However, this method demands the detailed modelling of the
DIF = fc fc0 = ε̇c0 = 30⋅10− 6 s− 1 (21)
γ s (ε̇c /ε̇c0 )1/3 if ε̇c > 30s− 1 explosive charge, air domain, and the building of interest, significantly
/ { resulting in a much higher computation cost. In contrast, the LBE and
(ε̇t /ε̇t0 )δ if ε̇t ≤ 1s− 1 user-simplified pressure-time history methods are cost-effective since
DIF = ft ft0 = ε̇t0 = 1⋅10− 6 s− 1
(22)
κ(ε̇t /ε̇t0 )1/3 if ε̇t > 1s− 1 they employ empirical or semi-empirical blast loading models. The
built-in LBE model is realized by adding *LOAD_BLAST_ENHANCED
log γ s = 6.156α − 2 (23)
keyword based on the CONWEP - Conventional Weapons Effects Pro­
gram, while the user-simplified pressure-time history method simply
α = 1/(5 + 9fc0 /f0 ) f0 = 10MPa (24)
applies user-defined loading on finite elements [29]. In the present
log κ = 6δ − 2 (25) study, the blast loads acting on the column head-on surface are modelled
in accordance with the blast loading model proposed by Chen et al.,
δ = 1/(1 + 8ft0 /f0 ) f0 = 10MPa (26) which has been reported to produce blast loading with acceptable ac­
curacy for numerical simulations [10,11,16]. A sketch of exerting blast
For the strain rate effect of steel materials, the dynamic increase loads is presented in Fig. 7, where the column head-on surface is divided
factor DIF of yield strength was defined by the Cowper & Symonds strain into five segments and the blast loading history on each segment is
rate model as 1 + (ε̇/C)1/p , where C = 40s− 1
and p = 5 was calculated separately.

5
Y. Shi et al. Engineering Structures 300 (2024) 117166

Fig. 5. Schematic illustration of loading sequences.

Fig. 6. Configuration sketch of modal testing.

3.1.4. Verification of FE modelling method

The established FE model needs to be validated through experi­
mental records or theoretical analysis. The field explosion experimental
data collected by Chen et al. [10,11,16] were used as a benchmark to
validate the established LS-DYNA modelling method in the present Fig. 7. Application of blast loading model.
study. The tested column designated with a scaled distance of
0.67 m/kg1/3 and an initial axial loading ratio of 0.2 was chosen for attached to the column rear surface, D1, D2, and D3, as depicted in
comparison. The chosen column had a square cross-section of Fig. 7, as well as the residual compression test data, were used for
200 mm × 200 mm, a total length of 2500 mm, and a concrete cover of verification of numerical results.
25 mm. The reinforcement steels consisted of four 20 mm (in diameter) Comparisons of lateral displacement of RC column under blast loads
longitudinal rebars and 8 mm (in diameter) transverse stirrups with a and residual bearing capacity of the damaged RC column are presented
reinforcement spacing of 150 mm. Hence, the longitudinal and trans­ in Fig. 8 and Fig. 9, respectively. As seen from Fig. 8, during the first
verse reinforcement ratios were 3.14% and 0.34%, respectively. Ac­ vibration cycle, the FE numerical results appear in agreement with the
cording to the laboratory compression test of concrete specimens, the experimental data, especially the moment of trough and crest. However,
uniaxial compressive strength of concrete was 50.16 MPa. The yield and the numerical displacements in the following cycles as well as the re­
ultimate strengths of longitudinal rebar were 483.5 MPa and sidual deflection are slightly higher than those of experimental records.
582.4 MPa, while the transverse stirrups had a yield strength of This is mainly attributed to the damping properties and non-uniform
466.7 MPa and an ultimate strength of 601.3 MPa. In the benchmark blast loads that cannot be precisely applied in the present model. The
study, the blast loads calculated using the blast loading model proposed peak axial forces recorded by the compression test and numerical
by Chen et al. [10] were applied on the front column surface. Lateral simulation are 1050 kN and 960 kN, respectively. The relative error is
deflection of the three linear variable differential transducers (LVDTs)

6
Y. Shi et al. Engineering Structures 300 (2024) 117166

10 yield strength of longitudinal rebars and transverse stirrups are 400 MPa
and 300 MPa respectively, and the corresponding reinforcing ratio are
2.43% and 0.45%. To investigate the effects of initial axial loading and
0
blast loading on modal properties and residual bearing capacities, three
Lateral displacement (mm)

axial loading ratios (ALR), namely 0, 0.2, and 0.3, as well as nine ex­
-10 plosion load cases (B1~B9, c.f., Table 2) standing for different blast
scenarios were employed in this paper.

-20 3.2.1. Residual bearing capacities

The residual bearing capacities of the damaged RC columns are
D1, experiment result presented in Fig. 10. As expected, the residual bearing capacities of the
-30
D2, experiment result RC columns decrease with the intensification of blast loads from B1 to
D3, experiment result B9. Besides, when subjected to the identical blast load, the residual
-40 D1, numerical result bearing capacities of the RC columns increase obviously with the growth
D2, numerical result of ALR, which indicates that the initial axial loading is beneficial to the
D3, numerical result blast resistance of the RC columns under the load cases investigated in
-50
this paper. However, as the scaled distance decreases, the positive effect
0 20 40 60 80 100
of initial axial loading on blast resistance is gradually decreasing. For
Time (ms) blast load case Z = 0.7 m/kg1/3, the residual bearing capacity of the RC
Fig. 8. Comparisons of lateral displacement. column without initial axial loading is nearly the same as those with ALR
of 0.2 and 0.3. This distinct tendency may be attributed to the fact that
the initial axial loading only helps to promote concrete resistance when
1400
Experimental result the blast loads are not destructive. When the blast load becomes
Numerical result destructive, severe damages will occur, leading to the insignificant effect
1200 Initial axial load of initial axial loading. The corresponding blast-induced damage to RC
columns is calculated per Eq. (1), as shown in Fig. 11.
1000
3.2.2. Vibration frequencies and mode shapes
Axial load (kN)

800
Fig. 12 shows the vibration frequencies of the first two vibration
modes of RC columns with different blast-induced damage levels. It is
observed that the first and second-order vibration frequencies decrease
600
as expected with the increase of the blast-induced damage of RC col­
umns. Moreover, the decreasing extent of unit blast-induced damage
400
apparently increases when the blast-induced damage exceeds 0.5, where
the RC columns are located into the severe damage range. Further,
200 nonlinear fitting analysis demonstrates that the correlation between the
first two vibration frequencies and blast-induced damage conforms to
0
0 2 4 6 8 10 12 14 16 18 20
Axial displacement (mm) Table 2
Lateral deflection under unit uniformly distributed pressure (unit: 10-3 mm).
Fig. 9. Comparisons of residual bearing capacity.
Pre-blast modal result Pre-blast modal result

about 8.6%. This limited difference in the residual axial load might be u1 3000/8/4/3.142 u’1 3000/8/4/3.142
ascribed to the fact that crack development and large deformation were × (0.098 ×2.056/101.742- × (0.170 ×2.141/49.762
0.239 ×0.001/262.272) = 0.185 +0.335 ×0.015/109.502)
tough to be accurately simulated in the current FE analysis. = 1.402
In fact, the agreement in the moment of lateral displacement trough u2 3000/8/4/3.142 u’2 3000/8/4/3.142
and crest in Fig. 8 has already validated the fundamental vibration × (0.273 ×2.056/101.742- × (0.318 ×2.141/49.762
frequency in the numerical model. In addition, to validate the modal 0.461 ×0.001/262.272) = 0.516 +0.412 ×0.015/109.502)
= 2.619
parameters calculated from numerical analysis, the rotation constraints
u3 3000/8/4/3.142 u’3 3000/8/4/3.142
at both ends of the chosen intact column were released to model the × (0.428 ×2.056/101.742- × (0.410 ×2.141/49.762
simply supported state. Then, the first three vibration frequencies of the 0.372 ×0.001/262.272) = 0.808 +0.260 ×0.015/109.502)
chosen intact column at the simply supported state were calculated = 3.374
numerically and compared to theoretical solutions [39–41], as listed in u4 3000/8/4/3.142 u’4 3000/8/4/3.142
× (0.489 ×2.056/101.742 × (0.442 ×2.141/49.762-
Table 1. It is obvious that the first two frequencies closely match the +0.013 ×0.001/262.272) 0.031 ×0.015/109.502) = 3.634
theoretical results with an error within 2.0% respectively. The relative = 0.924
error of the third-order frequencies is 7.50%. The slight difference in u5 3000/8/4/3.142 u’5 3000/8/4/3.142
relative errors reveals again the fact that the established FE mode of RC × (0.420 ×2.056/101.742 × (0.390 ×2.141/49.762-
+0.387 ×0.001/262.272) 0.300 ×0.015/109.502) = 3.203
columns is reliable.
= 0.793
u6 3000/8/4/3.142 u’6 3000/8/4/3.142
× (0.257 ×2.056/101.742 × (0.288 ×2.141/49.762-
3.2. Typical numerical results
+0.455 ×0.001/262.272) 0.412 ×0.015/109.502) = 2.363
= 0.485
This section presents the typical numerical results of a RC column u7 3000/8/4/3.142 u’7 3000/8/4/3.142
under different blast loads. The RC column is 3.0 m high with the cross × (0.091 ×2.056/101.742 × (0.123 ×2.141/49.762-
section 250 mm × 250 mm, and it has an unconfined concrete +0.218 ×0.001/262.272) 0.249 ×0.015/109.502) = 1.008
= 0.172
compressive strength of 33 MPa and a concrete cover of 25 mm. The

7
Y. Shi et al. Engineering Structures 300 (2024) 117166

3000 exponential decay function with an adjusted R-square of 0.95.

Fig. 13 shows the variation envelopes of the first two mass-
ALR=0 normalized displacement mode shapes of the RC columns under
2500 ALR=0.2
various blast load cases. In Fig. 13, the mode shapes of the intact RC
Residual bearing capacity (kN)

ALR=0.3
column are denoted by solid lines while the dash lines represent the
2000 envelopes of mode shapes of RC columns with various damage degrees.
It is shown in Fig. 13 that the blast-induced damage has an obvious
influence on the mode shapes of RC columns, leading to the expansion or
1500 shrinkage of mode shapes. Particularly, the variation of the first mode
shapes at the column-mid can reach up to 15% while the maximum
variation of first mode shapes at column-ends is much higher (over
1000
50%). It can also be seen from the figures that the amplitude changes in
the mode shapes are also sensitive to blast-induced damage if compared
500 to the changes of vibration frequencies, which indicates that using the
variation of vibration frequencies alone is insufficient to assess blast-
induced damage.
0
B1 B2 B3 B4 B5 B6 B7 B8 B9
Blast load cases 3.3. Blast-induced damage assessment

Fig. 10. Residual bearing capacities under different load cases.

With the modal data presented in Figs. 12 and 13, the damage de­
grees are obtained in accordance with the modal-based method and
axial bearing capacity and are plotted in Fig. 14, where the abscissa is
1.0
the damage assessment through the proposed modal-based method and
ALR=0
0.9 the ordinate is the damage degree based on residual bearing capacities.
ALR=0.2
Apparently, it is observed that all the scatter data are situated around the
0.8 ALR=0.3
predicting line by the proposed damage assessment method with a co­
0.7 efficient of variation of 17.8%, which confirms the validity of the pro­
posed modal-based method for blast-induced damage assessment of RC
Blast damage

0.6 columns.
It should be recognized that the proposed modal-based method was
0.5
established through analytical derivations and can be applied to
0.4 different RC columns. The proposed method has been validated by the
numerical simulations of the typical RC column with a cross-section of
0.3
250 mm × 250 mm and a height of 3.0 m. Further, three other RC col­
0.2 umns were also examined to extend the application scope of the pro­
posed method. The dimensions and reinforcing ratios of the three
0.1 columns are presented in Table 3. The yield stress and ultimate strength
of the longitudinal rebars are 500 MPa and 630 MPa, while the yield and
0.0
B1 B2 B3 B4 B5 B6 B7 B8 B9 ultimate strength of the stirrup steels are 400 MPa and 540 MPa,
respectively. The load cases B3, B6 and B9 described in Table 2 are
Blast load cases applied with an ALR of 0.2.
Fig. 11. Blast-induced damage under different load cases. Similarly, the blast damages given by the modal-based method
against the blast damage based on residual bearing capacities of the

1st frequency Initial 1st mode shape

2nd frequency Envelope of damaged 1st mode shape
300
Exponential Fit of 1st frequency Initial 2nd mode shape
Exponential Fit of 2nd frequency Envelope of damaged 2nd mode shape
250
3.0
Vibration frequency (Hz)

200 2.5
Equation y = 268.3 + -6.0*exp(4.5*x)
Column height (m)

Adj. R-Square 0.95

2.0
150

1.5
100

1.0
50 Equation y = 105.4 + -3.6*exp(3.7*x)
Adj. R-Square 0.95
0.5

0
-0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.0

Blast damage -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06

Fig. 12. Frequencies of the first two vibration modes. mass-normalized displacement (kg-1/2)

Fig. 13. Variation envelopes of 1st and 2nd damaged mode shapes.

8
Y. Shi et al. Engineering Structures 300 (2024) 117166

1.0 4. Experimental validation

ALR=0 To further validate the proposed modal-based damage assessment

ALR=0.2 method, near-field explosion tests and experimental modal tests of RC
0.8 ALR=0.3 columns were carried out. In this paper, the experimental results of
modal testing were presented and utilized to derive the blast-induced
damage through the proposed modal-based method, which was then
Blast damage degree

validated through the blast-induced damage calculated by the residual

0.6 bearing capacities of the tested RC columns.

4.1. Experimental set-up

0.4 COV=17.8% 4.1.1. Columns description

Four full-scale RC columns under different axial loads and scaled
distances were investigated in this experimental program. The column
image and sketch of external dimensions and reinforcing details are
0.2 shown in Fig. 16. The cross-section and height of the column shaft are
respectively 250 mm × 250 mm and 3.0 m, and the concrete cover is
designed as 25 mm. Commercial concrete was supplied for
manufacturing RC columns. According to the standard compressive tests
0.0
on the concrete specimens, the average unconfined strength was
0.0 0.2 0.4 0.6 0.8 1.0
measured as 42.5 MPa and the average Young’s modulus was 4.3 × 104
Damage assessment through modal-based method MPa. The steel grade of longitudinal and transverse reinforcing was
HRB400 (hot-rolled ribbed bar with the standard yield strength of
Fig. 14. Validation of the modal-based method.
400 MPa). The longitudinal reinforcement consisted of four steel rebars
with a diameter of 22 mm, while twenty-one hoops of 8 mm-in-diameter
Table 3 transverse stirrups were distributed along the axial direction with a
Comparisons of vibration frequencies. spacing of 150 mm around the length of 1800 mm at the column shaft
middle and a spacing of 120 mm around the length of 600 mm at both
Order of Theoretical results Numerical results Relative error
frequency (Hz) (Hz) (%) shaft ends. The main properties of reinforcement steels are listed in
Table 4.
1 67.5 67.9 0.73
2 269.8 265.1 1.79
3 607.0 564.7 7.50 4.1.2. Near-field explosion tests
Fig. 17 shows the testing site and location of the main experimental
apparatus for field explosion tests. In Fig. 17, from the left to the right
three RC columns were also plotted in Fig. 15 for comparison purpose, side are the shelter, gas cylinder and column support unit. The shelter
and the results show a good coincidence between the damage assess­ was designed and fabricated to prevent the field-testing participants
ment through modal-based method and the damage degree based on from blast waves and was located over 100 m away from the column
axial bearing capacity within allowable errors. To this end, the proposed support unit. The column support unit was made of high-strength rein­
modal-based damage assessment method is proved to be capable of forced concrete, which was designed to sustain the blast loads and the
capturing blast-induced damage of any RC columns. reaction force from the tested column. To avoid blast-induced concrete
crushing, a high-strength steel plate was attached to the head-on façade
1.0
of the column support unit. The gas cylinder was manufactured to store
high-pressure nitrogen. Once the tested column was installed in the
designated position, the initial axial loading was provided by the cyl­
RCC1 inder piston below the tested column and was controlled by the gas
0.8 RCC2 cylinder. When the initial axial loading was exerted, the column head
RCC3 and foot were constrained laterally using steel components to ensure no
slippage occurring at both column ends, as shown in Fig. 18. In the end,
Blast damage degree

the cylinder piston was covered by sandbags and a steel plate to protect
0.6 the cylinder piston from direct interaction with blast waves.
Rectangular assembly of trinitrotoluene (TNT) blocks was adopted to
generate blast waves. Each TNT block had a net weight of 200 g and an
COV=12.8% outer dimension of 100 mm × 50 mm × 25 mm, thereby a density of
0.4
1600 kg/m3. In the rectangular TNT assembly, each layer consisted of
15 TNT blocks. The explosive charge was placed in front of the tested
column with a standoff distance of 1.5 m and a detonation height of
0.2 1.5 m above the top base surface, as presented in Fig. 19.
Initial axial loading was implemented to simulate the dead weight of
the superstructures and remains constant during the subsequent pro­
cesses. Two kinds of axial loading ratio ALR, 0.18 and 0.36, denoted as
0.0 A2 and A4 respectively, were adopted in this paper. To investigate the
0.0 0.2 0.4 0.6 0.8 1.0 blast load effect on RC columns’ performance, three sets of charge mass,
Damage assessment through modal-based method i.e., 18 kg, 21 kg, and 24 kg, were considered. Three sets scaled dis­
tance, i.e., 0.57 m/kg1/3, 0.54 m/kg1/3, and 0.52 m/kg1/3, were calcu­
Fig. 15. Extended validation of the modal-based method. lated and denoted as Z1, Z2 and Z3. The combined load cases of initial

9
Y. Shi et al. Engineering Structures 300 (2024) 117166

Fig. 16. Image and sketch of tested RC columns (in millimetre).

hammer excitation was adopted owing to its convenient operation,

Table 4
inexpensiveness and portability [43,44]. Single input and multiple
Blast load cases definition.
output (SIMO) modal testing method was used and the configuration of
No. B1 B2 B3 B4 B5 B6 B7 B8 B9 experimental modal testing in this paper is presented in Fig. 20. In­
Scaled distance Z 1.6 1.4 1.3 1.2 1.1 1.0 0.9 0.8 0.7 struments for pre-blast and post-blast modal tests included accelerom­
(m/kg1/3) eters, hand-held hammer with a force sensor tip, charge amplifier, data
Equivalent TNT 0.8 1.2 1.5 2.0 2.5 3.4 4.6 6.6 9.8
acquisition system, and personal computer. As shown in Fig. 20, seven
charge mass (kg)
acceleration transducers are equidistantly placed on the rear side of the
RC column. The impact hammer is set to hit on the lower quarter point,
axial loads and blast loads for the tested RC columns were listed in generating a pulse exciting load and the subsequent dynamic responses
Table 5. It should be noted that RC column under combined load case of the tested RC column. The charges produced by the acceleration
Z1A2 suffered from a tentative explosion with a TNT mass of 9 kg at a transducers and the force sensor are transported to the charge amplifier
detonation distance of 1.5 m (Z = 0.72 m/kg1/3) before being subjected through low-noise cables, and then converted to voltage signals, which
to blast loading at scaled distance of 0.57 m/kg1/3, and no residual are collected and recorded by the data acquisition system. The data from
deflection were observed on the RC column under the blast loads at the impact hammer and seven accelerometer channels are digited and
scaled distance of 0.72 m/kg1/3. further processed in personal computer using spectrum analysis soft­
ware [24]. SDOF FRF curve-fitting technique was employed to pinpoint
4.1.3. Experimental modal tests the frequencies corresponding to the vibration modes of interest [24].
To investigate the influences of blast-induced damage on modal
properties of RC columns, experimental modal tests were performed on 4.2. Result of experimental modal tests
each RC column before and after explosion test, which were separately
termed as pre-blast modal testing and post-blast modal testing. The Table 6 shows the first two vibration frequencies and frequencies
modal testing in civil infrastructures often involves exciting the struc­ decreasing percentages of the RC columns under different scaled dis­
ture with a force and measuring the vibration responses, which are tances when the axial loading ratio remains 0.18. As illustrated in the
combined to generate the frequency response functions (FRFs) [42]. numerical analysis, the vibration frequencies of RC columns will
Three common non-destructive force excitation means are the ambient decrease after suffering from the blast loads, and the decreasing per­
vibration, displacement-release, and the employment of an impact centages of the first-order and second-order frequencies gradually de­
hammer or an electromagnetic shaker [43]. In this experiment, impact creases with the increase of scaled distances.

10
Y. Shi et al. Engineering Structures 300 (2024) 117166

Fig. 17. Test site and location of live explosion testing apparatus.

Fig. 18. Column installation and constraints.

Fig. 19. Placement and appearance of TNT assembly.

11
Y. Shi et al. Engineering Structures 300 (2024) 117166

Table 5
Dimensions and reinforcing details of three RC columns.
No. Depth (mm) Width (mm) Height (m) Unconfined strength (MPa) Concrete cover (mm) Reinforcement ratio (%) Stirrup ratio (%)

RCC1 300 300 3.0 36.6 25 1.13 0.58

RCC2 400 400 3.6 40 25 1.01 0.49
RCC3 400 600 4.6 40 25 1.05 0.58

damage assessment.

4.3. Blast-induced damage assessment

To determine the damage severity of RC columns subjected to close-

in blast loadings, the residual bearing capacity test of blast-damaged RC
columns was carried out, as presented in Fig. 22 [45]. It is recorded that
the RC column under load case Z2A2 (Z = 0.54 m/kg1/3 and ALR =
0.18) exhibits the highest residual bearing capacity, thereby the lowest
blast-induced damage. In contrast, the RC column under load case Z3A2
(Z = 0.52 m/kg1/3 and ALR = 0.18) presents the lowest residual bearing
capacity and the highest blast-induced damage. It is reasonable since the
smaller scaled distance leads to more serious damage of RC column. The
residual bearing capacity of blast-damaged RC column under the load
case Z1A2 (Z = 0.57 m/kg1/3 and ALR = 0.18) falls in between. The
reason is that RC column Z1A2 is initially exposed to a tentative ex­
plosion with a TNT mass of 9 kg at a standoff distance of 1.5 m
(Z = 0.72 m/kg1/3) before being subjected to blast loading at scaled
distance of 0.57 m/kg1/3. It is concluded from Fig. 22 that the tentative
explosion has a negative influence on the residual bearing capacity of RC
column although no plastic deformation was observed after the tentative
explosion. In addition, the comparison of RC columns under load case
Fig. 20. Configuration of the experimental modal tests.
Z2A2 and Z2A4 shows the increase of axial load ratio reduces the re­
sidual bearing capacity of blast-damaged RC columns at the same scaled
Table 6 distance of 0.54 kg/m1/3. The residual bearing capacity of RC column
Mechanical properties of HRB400 steels. under Z2A4 decreases by 9.5% compared to that of RC column under
Z2A2. The observation of negative effect of axial loading ratio on re­
Item 8 mm Stirrups 22 mm longitudinal rebar
sidual bearing capacity is in accordance with previous studies [7,8], in
Density (kg/m3) 7800 7800
which the severity of concrete crack and spall in the prestressed column
Young’s modulus (GPa) 196.4 163.1
Yield strength (MPa) 482.5 426.0
(ALR = 0.32) is higher relative to the non-stressed columns at the same
Ultimate strength (MPa) 644.4 589.5 scaled distance Z = 0.54 kg/m1/3.
Failure strain 0.17 0.24 The peak axial bearing capacity of the intact RC column which is
calculated in accordance with Eq. (27) [46,47], where fc denotes the
compressive strength of concrete; fy is the tensile strength of the longi­
Table 7 depicts the first two vibration frequencies and decreasing
tudinal reinforcement; Ag is the gross section area of RC column; As is the
percentages of RC columns under different axial loading ratios when the
area of the longitudinal reinforcement.
scaled distance keeps at 0.54 m/kg1/3. When the scaled distance remains
( )
unchanged, the decreasing percentage of the first-order vibration fre­ P0 = fc ⋅ Ag − As + fy ⋅As (27)
quencies increases from 14.6% to 22.3% with the axial loading ratio
increasing from 0.18 to 0.36. However, the decreasing percentage of the To validate the proposed modal-based damage assessment method,
second-order vibration frequency varies slightly with the axial loading the blast-induced damage was calculated based on the experimental
ratio increasing from 0.18 to 0.36. modal data through the proposed modal-based method and the axial
The variation of the first two mass-normalized displacement mode bearing capacity, respectively. The comparison is presented in Fig. 23, in
shapes of RC columns under different blast load cases is presented in which the abscissa is damage degree through the modal-based damage
Fig. 21. It can be seen from the figures that the amplitude changes in the assessment method while the ordinate is blast-induced damage degree
first mode shapes are more sensitive to blast-induced damage than those based on axial bearing capacity. The damage degree based on axial
of the second mode shapes, especially for load cases Z1A2 and Z2A4. The bearing capacity falls between 0.2 and 0.6, and according to the clas­
distinct variation of mode shapes should not be ignored in blast-induced sification of damage level proposed by Shi et al. [13], the tested columns
are categorized as medium-damaged and severe-damaged. Given the
experimental data in this paper, the coefficient of variation (COV) of
Table 7 using the proposed modal-based method to estimate the blast-induced
Definition of combined load cases. damage of RC columns is 11.7%. Particularly, for load cases Z2A4, the
No. Charge Detonation Scaled Initial axial Axial blast damage degree based on axial bearing capacity and the damage
mass (kg) distance (m) distance (m/ loading ratio, load assessment through modal-based method basically conform to corre­
kg1/3) ALR (kN) sponding relationship. For load case Z1A2, Z2A4 and Z3A2, the relative
Z1A2 18 1.5 0.57 0.18 260 errors between the modal-based damage degree and the damage degree
Z2A2 21 1.5 0.54 0.18 260 based on axial bearing capacity are within 15%. Such slight errors of
Z2A4 21 1.5 0.54 0.36 520 using modal-based method clearly demonstrate that the assumption of
Z3A2 24 1.5 0.52 0.18 260
damage mapping during the derivation of modal-based damage method

12
Y. Shi et al. Engineering Structures 300 (2024) 117166

Table 8
First two vibration frequencies and decreasing percentage at different scaled distances.
NO. ALR Scaled distance (m/kg1/3) First-order frequency Second-order frequency

Undamaged (Hz) Damaged (Hz) Relative reduction (%) Undamaged (Hz) Damaged (Hz) Relative reduction (%)

Z1A2 0.18 0.57 27.9 24.1 13.9 159.7 109.2 31.6

Z2A2 0.54 29.7 25.4 14.6 155.1 97.4 37.2
Z3A2 0.52 40.5 25.7 36.5 165.8 99.2 40.2

Table 9
First two vibration frequencies and decreasing percentage at different axial loading ratios.
NO. ALR Scaled distance (m/kg1/3) First-order frequency Second-order frequency

Undamaged (Hz) Damaged (Hz) Relative reduction (%) Undamaged (Hz) Damaged (Hz) Relative reduction (%)

Z2A2 0.18 0.54 29.7 25.4 14.6 155.1 97.4 37.2

Z2A4 0.36 35.0 27.2 22.3 162.0 103.9 35.9

1st mode shape, initial 1st mode shape, initial

1st mode shape, damaged 1st mode shape, damaged
2nd mode shape, initial 2nd mode shape, initial
3.0 2nd mode shape, damaged 3.0 2nd mode shape, damaged

2.5 2.5
Column Height (m)

2.0 Column Height (m) 2.0

1.5 1.5

1.0 1.0

0.5 0.5
Z1A2 Z2A2
0.0 0.0

-0.05 -0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 -0.05 -0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05

mass-normalized displacement (kg-1/2) mass-normalized displacement (kg-1/2)

1st mode shape, initial 1st mode shape, initial

1st mode shape, damaged 1st mode shape, damaged
2nd mode shape, initial 2nd mode shape, initial
3.0 2nd mode shape, damaged 3.0 2nd mode shape, damaged

2.5 2.5
Column Height (m)

Column Height (m)

2.0 2.0

1.5 1.5

1.0 1.0

0.5
Z2A4 0.5
Z3A2
0.0 0.0

-0.05 -0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 -0.05 -0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05

mass-normalized displacement (kg-1/2) mass-normalized displacement (kg-1/2)

Fig. 21. Variation of mass-normalized displacement mode shapes.

is applicable. Hence, from the perspective of engineering application, through numerical simulations and experimental tests. The blast-
the estimating error was allowable, and the validity of modal-based induced damage of RC columns was defined as the lost proportion of
damage assessment method was proved experimentally. axial bearing capacities. To estimate the blast-induced damage of RC
columns, a modal-based damage assessment method was derived using
5. Conclusions the concept of homogenous damage of concrete material and further
expressed as the explicit function of vibration frequencies and mass-
In this paper, a modal-based method for the post-blast damage normalized mode shapes. Finite element analysis and experimental
assessment of RC columns was proposed and then separately validated modal test showed that the vibration frequencies of RC column would

13
Y. Shi et al. Engineering Structures 300 (2024) 117166

analysis, Writing – original draft. Ye Hu: Data curation. Zhong-Xian Li:

Writing – review & editing, Supervision, Project administration. Yang
Ding: Writing – review & editing, Supervision, Funding acquisition.

Declaration of Competing Interest

The authors declare that they have no known competing financial

interests or personal relationships that could have appeared to influence
the work reported in this paper.

Data availability

Data will be made available on request.

Acknowledgement

The authors gratefully acknowledge the support from the National

Natural Science Foundation of China under grant numbers 51938011,
Fig. 22. Residual loading histories of blast-damaged RC columns. 52178498 and 52325807.

References

[1] Osteraas JD. Murrah Building bombing revisited: a qualitative assessment of blast
damage and collapse patterns. J Perform Constr Facil 2006;20(4):330–5.
[2] Sivaraman S, Varadharajan S. Investigative consequence analysis: a case study
research of Beirut explosion accident. J Loss Prev Process Ind 2021;69:104387.
[3] Luccioni BM, Ambrosini RD, Danesi RF. Analysis of building collapse under blast
loads. Eng Struct 2004;26(1):63–71.
[4] Shi YC, Li ZX, Hao H. A new method for progressive collapse analysis of RC frames
under blast loading. Eng Struct 2010;32(6):1691–703.
[5] Omran ME, Mollaei S. Investigation of axial strengthened reinforced concrete
columns under lateral blast loading. Shock Vib 2017:3252543.
[6] Siba F. Near-field explosion effects on reinforced concrete columns: an
experimental investigation [Master Thesis]. Department of Civil and
Environmental Engineering. Carleton University,; 2014.
[7] Braimah A, Siba F. Near-field explosion effects on reinforced concrete columns: an
experimental investigation. Can J Civ Eng 2018;45(4):289–303.
[8] Codina R, Ambrosini D, de Borbón F. Experimental and numerical study of a RC
member under a close-in blast loading. Eng Struct 2016;127:145–58.
[9] Liu Y, Yan J, Li Z, Huang F. Improved SDOF and numerical approach to study the
dynamic response of reinforced concrete columns subjected to close-in blast
loading. Structures 2019;22:341–65.
[10] Hu Y, Chen L, Fang Q, Xiang H. Blast loading model of the RC column under close-
in explosion induced by the double-end-initiation explosive cylinder. Eng Struct
2018;175:304–21.
[11] Chen L, Hu Y, Ren H, Xiang H, Zhai C, Fang Q. Performances of the RC column
under close-in explosion induced by the double-end-initiation explosive cylinder.
Int J Impact Eng 2019;132:103326.
[12] Kyei C, Braimah A. Effects of transverse reinforcement spacing on the response of
reinforced concrete columns subjected to blast loading. Eng Struct 2017;142:
Fig. 23. Comparison between two damage degrees. 148–64.
[13] Shi YC, Hao H, Li ZX. Numerical derivation of pressure–impulse diagrams for
prediction of RC column damage to blast loads. Int J Impact Eng 2008;35(11):
decrease after suffering from blast damage and the decreasing percent of 1213–27.
vibration frequencies increased as the blast damage intensified. The [14] Thiagarajan G, Rahimzadeh R, Kundu A. Study of pressure-impulse diagrams for
effect of blast-induced damage on mode shapes of RC column is also reinforced concrete columns using finite element analysis. Int J Prot Struct 2013;4
(4):485–504.
significant, indicating that merely using vibration frequencies in blast-
[15] Parisi F. Blast fragility and performance-based pressure–impulse diagrams of
induced damage assessment is insufficient. Both numerical and experi­ European reinforced concrete columns. Eng Struct 2015;103:285–97.
mental results demonstrate that the proposed modal-based damage [16] Yu R, Chen L, Fang Q, et al. Generation of pressure–impulse diagrams for failure
modes of RC columns subjected to blast loads. Eng Fail Anal 2019;100:520–35.
assessment method has a reliable capacity of estimating the blast-
[17] Zhang CW, Abedini M, Mehrmashhadi J. Development of pressure-impulse models
induced damage of RC columns. Given the data in the present study, and residual capacity assessment of RC columns using high fidelity Arbitrary
the COVs of numerical analysis and experimental test are 17.8% and Lagrangian-Eulerian simulation. Eng Struct 2020;224:111219.
11.7% respectively. Since experimental modal test is non-destructive [18] Wu KC, Li B, Tsai KC. Residual axial compression capacity of localized blast-
damaged RC columns. Int J Impact Eng 2011;38(1):29–40.
and easy to be performed, the proposed modal-based method can be [19] Li J, Hao H. Numerical study of concrete spall damage to blast loads. Int J Impact
employed to assess the blast-induced damage of RC columns after ex­ Eng 2014;68:41–55.
plosion events without conducting residual bearing capacity test. [20] Ibrahim A, Salim H, Flood I. Damage prediction for RC slabs under near-field blasts
using artificial neural network. Int J Prot Struct 2011;2(3):315–32.
[21] Bao X, Li B. Residual strength of blast damaged reinforced concrete columns. Int J
CRediT authorship contribution statement Impact Eng 2010;37(3):295–308.
[22] Cui J, Shi YC, Li ZX, et al. Failure analysis and damage assessment of RC columns
under close-in explosions. J Perform Constr Facil 2015;29(5):B4015003.
Yanchao Shi: Conceptualization, Writing – review & editing, Su­ [23] Xu K, Deng Q, Cai L, Ho S, Song G. Damage detection of a concrete column subject
pervision, Project administration, Funding acquisition. Shaozeng Liu: to blast loads using embedded piezoceramic transducers. Sensors 2018;18(5):1377.
Conceptualization, Methodology, Investigation, Validation, Formal [24] Heylen W, Lammens S, Sas P. Modal Analysis Theory and Testing. Leuven,
Belgium: Katholieke Universiteit Leuven,; 1997.

14
Y. Shi et al. Engineering Structures 300 (2024) 117166

[25] Kim TH, Lee KM, Chung YS, Shin HM. Seismic damage assessment of reinforced [37] Malvar L.J., Crawford J.E. Dynamic increase factors for concrete. DTIC Document,
concrete bridge columns. Eng Struct 2005;27(4):576–92. 1998.
[26] Zhang CW, Gholipour G, Mousavi AA. Blast loads induced responses of RC [38] Abedini M, Zhang C, Mehrmashhadi J, Akhlaghi E. Comparison of ALE, LBE and
structural members: State-of-the-art review. Compos Part B: Eng 2020;195:108066. pressure time history methods to evaluate extreme loading effects in RC column.
[27] Mutalib AA, Hao H. Development of PI diagrams for FRP strengthened RC columns. Structures 2020;28:456–66.
Int J Impact Eng 2011;38(5):290–304. [39] Zhong S, Oyadiji SO. Analytical predictions of natural frequencies of cracked
[28] Zhang Z, Aktan AE. Application of modal flexibility and its derivatives in structural simply supported beams with a stationary roving mass. J Sound Vib 2008;311
identification. J Res Nondestruct Eval 1998;10(1):43–61. (1–2):328–52.
[29] LS-DYNA R12 Keyword user’s manual. Livermore Software Technology (LST). An [40] Sayyad FB, Kumar B. Identification of crack location and crack size in a simply
Ansys Company, 2013. supported beam by measurement of natural frequencies. J Vib Control 2012;18(2):
[30] Hallquist J.O. LS-DYNA theory manual. Livermore. CA: Livermore Software 183–90.
Technology Corporation, 2006. [41] Kindova-Petrova D. Vibration-based methods for detecting a crack in a simply
[31] Li ZX, Zhong B, Shi Y, Yan JB. Nonlocal formulation for numerical analysis of post- supported beam. J Theor Appl Mech 2014;44(4):69.
blast behavior of RC columns. Int J Concr Struct Mater 2017;11(2):403–13. [42] Ewins D.J. Modal testing: theory, practice and application (2nd ed). Taunton, UK:
[32] Zhao CF, Chen JY, Wang Y, Lu SJ. Damage mechanism and response of reinforced Research Studies Press Ltd. and John Wiley & Sons, 2000.
concrete containment structure under internal blast loading. Theor Appl Fract [43] Green M.F. Modal test methods for bridges: a review. In PROCEEDINGS-SPIE THE
Mech 2012;61:12–20. INTERNATIONAL SOCIETY FOR OPTICAL ENGINEERING, 1995.
[33] Thiagarajan G, Kadambi AV, Robert S, Johnson CF. Experimental and finite [44] Mao YM, Guo XL, Zhao Y. Experimental study of hammer impact identification on
element analysis of doubly reinforced concrete slabs subjected to blast loads. Int J a steel cantilever beam. Exp Tech 2010;34(3):82–5.
Impact Eng 2015;75:162–173.. [45] Shi Y, Hu Y, Chen L, Li ZX, Xiang H. Experimental investigation into the close-in
[34] Tavárez F. Simulation of behavior of composite grid reinforced concrete beams blast performance of RC columns with axial loading. Eng Struct 2022;268:114688.
using explicit finite element methods [Master of Science Thesis]. Madison, [46] MacGregor J.G.G. Reinforced concrete: mechanics and design. Professional
Wisconsin: University of Wisconsin-Madison, 2001. technical reference, Englewood Cliffs, NJ: Prentice-Hall, 1996.
[35] Fanning P. Nonlinear models of reinforced and post-tensioned concrete beams. [47] ACI Committee. Building code requirements for structural concrete (ACI 318–08)
Electron J of Struct Eng 2001;2(2):111–9. and commentary. American Concrete Institute, 2008.
[36] Béton CE-Id. CEB-FIP model code 1990: design code. Thomas Telford, 1993.

15