Numerical evaluation of the retrot effectiveness for GFRP retrotted concrete

slab subjected to blast pressure

Jin-Won Nam
a,c
, Ho-Jin Kim
b
, Sung-Bae Kim
c
, Na-Hyun Yi
c
, Jang-Ho Jay Kim
c,
*
a
Department of Civil and Environmental Engineering, Southern University at Baton Rouge, Pinchback Building #327, Baton Rouge, LA 70813, USA
b
Institute of Technology Research Planning, ATMACS CO.LDT, Sangdaewon-dong, Jungwon-gu, Seongnam-si, Gyunggi-do 462-120, South Korea
c
School of Civil and Environmental Engineering, Yonsei University, 134 Shinchon-dong, Seodaemun-gu, Seoul 120-794, South Korea
a r t i c l e i n f o
Article history:
Available online 21 October 2009
Keywords:
Blast load
Fiber Reinforced Polymer (FRP)
Retrotting effectiveness
High-strain rate dependent material model
Debonding failure model
Reinforced concrete (RC)
Rened FEM analysis
a b s t r a c t
For retrotting structures against blast loads, sufcient ductility and strength should be provided by
using high-performance materials such as ber reinforced polymer (FRP) composites. The effectiveness
of retrot materials needs to be precisely evaluated for the retrotting design based on the dynamic
material responses under blast loads. In this study, rened FEM analysis with high-strain rate dependent
material model and debonding failure model is conducted for evaluating the FRP retrotting effective-
ness. The structural behavior of reinforced concrete (RC) slab retrotted with glass ber reinforced poly-
mer (GFRP) under blast pressure is simulated and the analysis results are veried with the previous
experimental results.
2009 Elsevier Ltd. All rights reserved.
1. Introduction
Generally, concrete can be considered as a highly effective con-
struction material in resisting against blast loading compared to
other materials. However, concrete structures designed for the ser-
vice loads of normal strain-rate require special retrotting to in-
crease the structural resistance against blast loading. Retrotting
method of attaching extra structural members or supports to in-
crease the blast resistance is undesirable from the perspectives of
construction cost increase and useable space elimination. Also, this
method generally does not greatly improve the overall structural
resistance against blast load [13]. Therefore, less expensive and
more convenient ber reinforced polymers (FRP) sheets or plates
are being used as surface attachments to retrot specied areas
of structural members [3,4]. The FRP surface attachments signi-
cantly improve the blast resistance of structures without forfeiting
usable space and requiring long construction time, thereby saving
money. For the retrotting of concrete structures for blast resis-
tance, the selection of the type of FRP is important. The selected
FRP has to improve stiffness, strength, and ductility of a retrotted
structure to satisfy required blast safety resistance and absorb
blast energy whereby transforming structural failure mode from
brittle to ductile.
To analyze and design FRP retrotted structures under blast
loads, both experimental and numerical studies are necessary [3
12]. Simplied lumped mass models for blast resistant structure
design and analysis allow rudimentary ways of designing and ana-
lyzing global concrete structure displacement behavior in terms of
applied load, mass, and resistance factors [13]. In such methods,
the applied load is calculated based on the application of simple
blast wave function and conversion of overall structural stiffness
to a single stiffness value. Due to its simplicity, this method is still
commonly used for blast resistant structure design and analysis.
However, recently, in order to improve the simplied analysis
methods, studies on the precise blast analysis methods with accu-
rate material models and rened nite element models for the
simulation of retrotted concrete structure behavior have been ac-
tively pursued for the accuracy and reliability of analysis results
[1218].
If properly validated, the rened FEM analyses can be used as a
replacement for costly structure blast experiments. Furthermore,
even when specialized testing facilities and related resources are
available, some conditions and data are more readily obtained
through such virtual experiments using the FEM. For these reasons,
it is vital to establish effective analysis tools for new and retrotted
concrete structures under blast loading to predict structural behav-
iors, select optimum retrotting materials, and ensure desired fail-
ure mechanisms.
In this study, the blast resistance of glass ber reinforced poly-
mer (GFRP) retrotted reinforced concrete (RC) slabs is analyzed
0263-8223/$ - see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compstruct.2009.10.031
* Corresponding author. Tel.: +82 2 2123 5802; fax: +82 2 364 1001.
E-mail address: jjhkim@yonsei.ac.kr (Jang-Ho Jay Kim).
Composite Structures 92 (2010) 12121222
Contents lists available at ScienceDirect
Composite Structures
j our nal homepage: www. el sevi er . com/ l ocat e/ compst r uct
using the explicit analysis code LS-DYNA, which accommodates
the modeling of high-strain rate dependency and debonding fail-
ure. Also, the retrot effectiveness of GFRP fabrics is evaluated by
comparing the analysis results for non-retrotted and retrotted
slabs. The verication of the analyses is performed through com-
parisons with experimental results [10,19].
2. Constitutive material models considering blast load
2.1. Concrete damage model
The concrete damage model used in this study is based on Mal-
vars model [16,18], which modies the WillamWarnke failure
surface with three parameter surface denition and pressure cutoff
[20]. The concrete damage model of this study represents blast dy-
namic hardeningsoftening non-linear behavior by accumulated
effective plastic strain in the regime of continuum mechanics. In
the model, three independent failure surfaces are dened as
Dr
m
a
0

p
a
1
a
2
p
Maximum failure surface 1
Dr
r

p
a
1f
a
2f
p
Residual failure surface 2
Dr
y
a
0y

p
a
1y
a
2y
p
Yield failure surface 3
Dr
e
r
f
Drp=r
f
Enhanced failure surface 4
where Dr is stress difference (on the deviatoric stress failure sur-
face) and p is conning pressure for each stage of behavior. The vari-
ables a
0
, a
1
, and a
2
are constants obtained by the unconned
compression test and conventional triaxial compression tests at
various degrees of conning pressure. r
f
is the strength enhance-
ment factor, which represents strain-rate effect of concrete. The en-
hanced concrete strength is obtained by multiplication of the
enhancement factor to static concrete strength. The strength
enhancement factors used in this study are shown in Fig. 1.
2.2. Steel reinforcement model
For the material model of steel reinforcement in concrete, dy-
namic and strength increasing factors are considered [21]. The
yield stress function of steel reinforcement based on the von Mises
criterion is
r
y
br
0
f
h
e
p
eff
5
where b is the variable for strain-rate effect and is calculated using
Eq. (6), which is based on Cowper and Symonds model [23]. r
0
is
initial yield stress and f
h
e
p
eff
is hardening function, which can be
expressed as Eq. (7) using plastic stiffness E
p
and effective plastic
strain e
p
eff
.
b 1
_
e
C

1=r
6
f
h
e
p
eff
E
p
e
p
eff
7
where C and r are strain-rate parameters. In this study, the strain-
rate effect is considered by dening the relationship between effec-
tive plastic strain and yield strength based on the experimental
data. The relations between effective plastic strain and yield
strength are shown in Fig. 2.
2.3. Rate dependent FRP failure model
The failure model of FRP in this study is based on progressive
failure criteria of Hashin and Chang and Changs model [24,25].
The strain-rate effect on the material strengths is incorporated into
failure model based on Parks model [26].
1
10
100
1.E-06 1.E-04 1.E-02 1.E+00 1.E+02
Strain rate (1/s)
S
t
r
e
n
g
t
h

e
n
h
a
n
c
e
m
e
n
t

f
a
c
t
o
r
Tension(This study)
Compression(This study)
Tension(CEB)
Compression(CEB)
Fig. 1. Concrete strength enhancement due to high-strain rates.
0
200
400
600
800
1000
1200
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Strain rate = 0
Strain rate = 10
-4
Strain rate = 10
0
Strain rate = 10
2
Strain rate = 10
3
Strain rate = 0
Strain rate = 10
-4
Strain rate = 10
0
Strain rate = 10
2
Strain rate = 10
3
Effective plastic strain
Y
i
e
l
d

s
t
r
e
s
s

(
M
P
a
)
Fig. 2. Rate dependent relationship of effective plastic strain and yield stress for
steel reinforcement.
Enhanced failure criteria on dynamic condition
1
2
Y
t
static
Y
t
dynamic
Failure criteria on static condition
X
t
static
X
t
dynamic
Y
c
static
Y
c
dynamic
X
c
static
X
c
dynamic
Fig. 3. Enhanced failure criteria of FRP.
J.-W. Nam et al. / Composite Structures 92 (2010) 12121222 1213
2.3.1. Fiber breakage
For tension failure, the ber breakage criterion can be expressed
as
e
r
11
X
t

2

s
12
SC
12

2
and
r
11
X
t

2
P
s
12
SC
12

2
; for r
11
>0 8
where e is a failure index representing the combined effect of the
normal and shear stresses; r
11
is normal stress; s
12
is shear stress;
X
t
is the tensile strength; SC is the in-plane shear strength. The cri-
terion states that ber breakage occurs when the failure index is
equal to or greater than unity, for cases where the effect of normal
stress is greater than that of shear stress. For the compressive fail-
ure, the failure criterion can be expressed as
e
r
11
X
c

2
; for r
11
< 0 9
where X
c
is the compressive strength. The criterion states that ber
failure in compression is mainly due to the normal stress, because
ber buckling dominates the failure in compression.
2.3.2. Enhanced failure criteria
The strength values of failure criteria are dependent on strain-
rate and can be expressed as follows.
X
t
f
1

_
e; X
c
f
2

_
e; Y
t
f
3

_
e; Y
c
f
4

_
e; SC
12
f
5

_
e 10
where X is the longitudinal strength; Y is the transverse strength;
SC
12
is the in-plane shear strength;
_
e is strain-rate. Subscripts t
and c represent tensile and compressive state. Specic functional
forms f
_
e can be determined from experiment such as uniaxial
and eccentric tension tests. Therefore, the uniaxial and eccentric
tension tests should be carried out to predict the appropriate
strain-rate effect on the longitudinal and shear strength. Linear
functions of material strength according to strain rates can be de-
ned by Eqs. (11)(13) [27].
X X
static
f
_
e
11
MPa 11
Y Y
static
n
_
e
22
MPa 12
SC
12
SC
12

static
j
_
e
12
MPa 13
where f, n, j are material constants. From Eqs. (11)(13), the strain-
rate effect is simply related to the in-plane failure criteria of FRP
failure model. The enhanced failure criterion is shown in Fig. 3.
0.0
0.2
0.4
0.6
0.8
1.0
0 5 10 15 20
L_e/s _min=2
L_e/s _min=5
L_e/s _min=10
L_e/s _min=50
L_e/s _min=100
t
a
/ t
e
P
i
m
p
/

P
s
t
L
e
/ s
min
= 2
L
e
/ s
min
= 5
L
e
/ s
min
= 10
L
e
/ s
min
= 50
L
e
/ s
min
= 100
Fig. 4. Ratio of the bond failure loads of blast loading and the static cases.
Target FRP retrofitted structures
Blast load generation
Boundary condition check Explosion condition check
Establishment of numerical model
Material modeling Finite element modeling
Solid element
Beam element
Shell element
Contact interface modeling FRP-concrete interface
Input boundary and blast loading conditions
Blast analysis based on FEM
Structural responses check
Displacement, stress, and strain Failure strain, debondingfailure
Evaluation of the blast retrofit
performance
Global behavior Local behavior of retrofit material
D
y
n
a
m
i
c

m
a
t
e
r
i
a
l

p
r
o
p
e
r
t
i
e
s
Concrete
Reinforcing steel
FRP (Rate dependent failure)
Fig. 5. Evaluation procedure of the blast retrot effectiveness for FRP retrotted concrete structures.
1214 J.-W. Nam et al. / Composite Structures 92 (2010) 12121222
2.4. Debonding failure modeling under blast loading
The debonding failure model under blast pressure is presented
based on the Lorenzis and Tegolas model [28]. The dynamic bond
strength is calculated from the debonding failure load ratio of dy-
namic to static for the FRP retrotted concrete and is used in deb-
onding failure criteria under arbitrary blast loading. The static
debonding failure load of FRP retrotted concrete is derived from
the Yuans model [29].
The loaddisplacement (PD) of bonded plate can be expressed
as Eq. (14) [29]
P
s
f
b
p
k
1
D
d
f
tanhk
1
L: 14
When the interfacial softening begins, the displacement of the
edge becomes equal to the slip (D = d) and the Eq. (14) becomes
P
s
f
b
p
k
1
for infinite bond length: 15
The length of interface to resist the applied load is referred as
the effective bond length. This effective bond length can be dened
as the bond length over which the shear stresses resist more than
97% of the applied load with an innite bond length. Based on this
denition and considering that tanh(2) ; 0.97, the effective bond
length is given by
L
e

2
k
1
: 16
From the Eqs. (15) and (16), the bond shear stress of unit width at
the loaded end under a static force P can be determined as
s
st
max
P
2
L
e
17
where L
e
is the effective bond length between FRP and concrete in
which the bond load increases linearly to a constant value. The lin-
ear relationship between s
max
and P, and between s
st
max
and P is ex-
pressed as a function of t
a
/t
e
as
s
st
max
s
max

P
imp
P
st
f t
a
=t
e
18
where P
st
and P
imp
are the tensile forces which induce the local bond
strength s
f
at the loaded end under the static and impulsive blast
loading cases, respectively. t
a
is the duration of blast loading and
t
e
is the duration for the stress wave propagation to the effective
bond length. If the local bond slip relationship is assumed to be lin-
ear elastic and s
f
is assumed to be the same in the static and in the
dynamic cases, then Eq. (18) gives the ratio of the debonding failure
loads in the blast loading and static loading cases. A ratio less than 1
indicates early debonding of the FRP as a result of the impulsive
application of the load.
In this study, the bond strength under blast load is evaluated by
using the blast impulsetime history, which is obtained by integra-
tion of blast pressuretime history. For the blast pressuretime
history function, the Friedlander decay function can be used [1].
Pt P
m
1
t
t
p

exp a
t
t
p

19
where a is constant and t
p
is the duration for positive pressure. The
blast impulsetime history function by integration of Friedlander
function can be determined as
1,000
1
,
0
0
0
7 bars (5.74 mm diameter)
A
A
Welded steel mesh
Section A-A
70
10
50
10
Unit : mm
Geometry and reinforcement details
GFRP retrofitted slab
(b)
(a)
Fig. 6. Test specimen of experiment [10,19].
Table 1
Material properties [10,19].
Concrete Compressive strength (MPa) 42
Tensile strength (MPa) 4.0
Youngs modulus (GPa) 29
Poissons ratio 0.167
Reinforcing
steel
Yield strength (MPa) 480
Ultimate strength (MPa) 600
GFRP Longitudinal modulus, E
x
(GPa) 27.5
Transverse modulus, E
y
(GPa) 1.3
Shear modulus, G
xy
(GPa) 0.27
Poissons ratio, v
xy
0.23
Longitudinal tensile strength, X
t
(MPa)
410:2log _ e
11
459:2
Longitudinal compressive strength,
X
c
(MPa)
318.0
Transverse tensile strength, Y
t
(MPa) 146:5log _ e
22
17:2
Transverse compressive strength, Y
c
(MPa)
10.5
In-plane shear strength, SC (MPa) 6:2loglog_ e
12
7:9
Density, q (kg/m
3
) 2,100
Thickness (mm) 1.3
Epoxy Tensile strength (MPa) 54
Elastic modulus (GPa) 3.1
Maximum elongation (%) 5
J.-W. Nam et al. / Composite Structures 92 (2010) 12121222 1215
It

t
0
Ptdt P
m
t
p
a
1 e
a
t
tp
1
t
t
p

1
a
e
a
t
tp
1


20
and Eq. (20) can be rewritten as a function of the maximum im-
pulse, I
max
, as
It I
max
1
1
a
e
a
t
tp
1
t
tp

1
a

1
1
a

1
a
e
a
: 21
From Eqs. (20) and (21), the ratio of the bond failure loads under
static and blast loads can be determined as
s
st
max
s
max

1
1
a

1
a
e
a

1
0
me
at
dt
1
a
coth
2s
min
Le

e
a
ne
a 1
s
min
te
Letp

22
where s
min
is the minimum distance of stress wave with
m a1 t coth
2tp1t
te
and n 1 a
s
min
te
Letp
. From the above ap-
proach, the ratios of the dynamic bond failure load of FRPconcrete
interface under blast loading to the static bond failure load are cal-
culated as shown in Fig. 4. As shown in Fig. 4, the blast impulse may
produce a substantial reduction of the debonding load compared to
the static load. This reduction is more signicant for larger L
e
/s
min
ratio when the minimum activated length of the joint is shorter.
3. Blast analysis for FRP retrotted RC slab
For evaluating the FRP retrotting effectiveness for concrete
structures under blast pressure, the evaluation procedure consid-
ering rate dependent FRP failure model and debonding model is
presented as shown in Fig. 5. In order to verify the procedure,
the previous eld blast test [10,19] is simulated following the pre-
sented analysis procedure using the explicit analysis code LS-DYNA
[21,22]. Finally, the analyses results are compared with the previ-
ously reported experimental results.
3.1. Descriptions of eld blast test
To verify the validity of the rened FE analysis, the eld blast
test for FRP retrotted reinforced concrete slab performed by
Razaqpur and Tolba [10,19] is simulated. As shown in Fig. 6a, the
specimen is a RC slab with the dimensions of 1.0 1.0 0.07 m.
The slabs are doubly reinforced with welded steel mesh, which
has cross-sectional area of 25.8 mm
2
and center-to-center spacing
of 152 mm in each direction. The yield and ultimate strengths of
Fig. 7. Test setup and the tripod holding the explosive charge [10,19].
-2
0
2
4
6
8
1 2 3 4 5 6
Experimental(A) (Tolba 2001)
Experimental(B) (Tolba 2001)
ConWep
Time (msec)
B
l
a
s
t

p
r
e
s
s
u
r
e

(
M
P
a
)
Fig. 8. Comparison between CONWEP [30] and measured reected pressure.
1216 J.-W. Nam et al. / Composite Structures 92 (2010) 12121222
the reinforcing bars are 480 MPa and 600 MPa, respectively, and
the compressive strength of concrete is 42 MPa. The slabs are ret-
rotted on both faces with two laminates of GFRP arranged in a
crossed form as shown in Fig. 6b. Each GFRP laminate is 500 mm
wide and covers a middle half of the slab. The laminate is basically
unidirectional composite fabric with E-glass bers in the main
reinforcing direction. The material properties of the test specimen
are tabulated in Table 1. Dynamic strengths of GFRP are calculated
using Eqs. (11)(13).
The explosive used in the test is ANFO, which contains 82% of
the energy of same amount of TNT. Specimen is subjected to the
blast pressure from 33.4 kg of ANFO. The distance from the center
of the charge to the center of the front face of the slab is 3.0 m. The
test setup is shown in Fig. 7.
3.2. Blast Analysis of GFRP retrotted RC slab
3.2.1. Blast load generation
The blast load is generated using CONWEP. CONWEP is software
for calculating blast pressure according to the weight and stand-off
distance of explosives. It has been developed stochastically [30].
The blast is assumed as a spherical burst in the free atmosphere
and 33.4 kg of ANFO is converted into 27.4 kg of TNT by conversion
factor of 0.82. The comparisons between CONWEP predicted pres-
sures and eld test measured pressures [10,19] are shown in Fig. 8.
The CONWEP prediction for the pressuretime prole agrees well
with the test measurement.
3.2.2. Finite element models
Solid and beam elements with rate dependent material models
are used for concrete and reinforcing steels, respectively. For the
GFRP sheets modeling, shell element with rate independent and
dependent failure models are adopted to incorporate GFRP rate ef-
fect on the global structural behavior. The shell elements of GFRP
are attached to the solid elements of concrete with contact interfa-
cial element [21,22] with perfect bonding and debonding models.
The debonding model considers the dynamic bond strength calcu-
lated by Eq. (22) with the failure of contact elements. The shell ele-
ments are placed at a distance of half-thickness of GFRP sheet from
the concrete surface. This virtual thickness offset is used to simu-
late the attachment of GFRP to the concrete in the analysis. For
the determination of element size, noise analyses for the different
element sizes were conducted and 2.5% of specimen length was
determined as an optimum element size considering the comput-
ing time and the reliability. The FE modeling including concrete,
reinforcing steel and GFRP sheet of the retrotted slab is schemat-
ically shown in Fig. 9.
Regarding the boundary conditions, translations and rotations
for the x, y and z directions of all of the edges of RC slab are con-
strained. According to the Razaqpurs paper [19], the specimens
are supposed to be restrained in only uplift direction. However, a
-5
0
5
10
15
0.00 0.01 0.02 0.03 0.04
k9 w/o strain effect
k10 w/ strain effect
k11 perfect bonding
Non-retrofit
Time (sec)
M
i
d
-
s
p
a
n

d
i
s
p
l
a
c
e
m
e
n
t

(
m
m
)
Proposed analytical model
Displacement increasing
due to rate independency
Displacement decreasing
due to perfect bonding condition
Rate dependent failure model with debonding
Rate independent failure model with debonding
Rate dependent failure model with perfect bonding
Non-retrofit (control)
Fig. 10. Analysis results of mid-span displacement history.
Beam element
for steel mesh
Solid element
for concrete
Shell element
for FRP
42 MPa
Unidirectional Glass ply
Fiber direction
x
x
y
y
y
x
Contact interface
Concrete surface
FRP surface
Contact surface to surface to tiebreak
Shell element
Solid element
Contact criteria :
imp
A
Zoom in A
Fig. 9. Finite element models of GFRP retrotted concrete slab specimen.
J.-W. Nam et al. / Composite Structures 92 (2010) 12121222 1217
Fig. 11. Displacement distribution at 0.005 s.
1218 J.-W. Nam et al. / Composite Structures 92 (2010) 12121222
steel frame on the steel box is provided with a clamping system
and the clamping system for all edges can be considered as all
directional restraint. The boundary conditions can be identied
in Fig. 7.
3.2.3. Material models
Three material models of LS-DYNA are used in the analysis. Con-
crete damage model (MAT_72), piecewise linear plasticity model
(MAT_24), and orthotropic elastic model (MAT_2) are adopted for
concrete, steel reinforcement, and FRP reinforcement, respectively
[21]. In the analysis, the modied strength enhancement function
of concrete damage model and the dynamic failure criteria of FRP
in Chapter 2 are implemented to LS-DYNA explicitly.
3.3. Results
The mid-span displacement history of slab is shown in Fig. 10.
For the rate dependent and independent GFRP failure models with
debonding, the maximum displacements are calculated as
10.56 mm and 11.14 mm, respectively. The difference of displace-
ments between rate dependent and independent models indicates
the GFRP strength enhancement effect on the global structural
behavior. In case of the rate dependent GFRP failure model with
perfect bonding, the maximum displacement is calculated as
10.23 mm. The minute difference between debonding and perfect
bonding models indicates the energy absorbing capacity of perfect
bonding between GFRP and concrete.
From the analytical displacement history, the effectiveness of
FRP can be decided as either sufcient or insufcient retrotting.
If the behavioral curve of rate dependent FRP failure model with
debonding is closer to the behavioral curve of rate independent
FRP failure model with debonding, it means that more retrotting
is needed. If the behavioral curve is closer to perfect bonding case,
it means that the retrotting is relatively sufcient. From the anal-
ysis results, the behavioral curve of rate dependent failure model
with debonding is slightly closer to the behavioral curve of perfect
bonding model and has the maximum displacement of 10.56 mm,
which corresponds to a low damage level in the ASCE criteria [1].
Comparing the displacement results with the maximum displace-
ment of non-retrotted RC slab, the maximum displacement is re-
duced by approximately 20% by GFRP retrotting.
The displacement contours of structures also can be used as an
index to indicate structural components damage states and local-
ized failures. From the analysis of the non-retrotted slab, the cen-
tral region has large out-of-plane displacements and the edges
have large shear rotation concentration as shown in Fig. 11.
The analyses results for the strains of GFRP and concrete are
shown in Fig. 12. The maximum concrete strain of 0.0039 was ob-
served at the bottom of the slab using the GFRP strain-rate depen-
dent model with debonding failure mechanism. When the GFRP
strain-rate dependent model is not used, a signicant increase in
strain is observed and the maximum strain is calculated as
0.0046 with GFRP material failures as shown in Fig. 12a.
In Fig. 12b, some localized material failures are observed when
the stresses of GFRP exceeded the material strength and debonding
failures are observed when local GFRP strains exceeded maximum
strain capacity. The local strain exceeding GFRP strain capacity
indicates local debonding failures in which the strain of GFRP is
transferred to concrete. This phenomenon is conrmed by two
opposite tendencies shown in Fig. 12a and b. After 0.02 s from
the wave pressure application, the GFRP strains of debonding mod-
-0.002
0.000
0.002
0.004
0.006
0.008
0.010
0 0.01 0.02 0.03 0.04
w/ strain-rate
w/o strain-rate
Perfect bonding
Non-retrofit
Time (sec)
C
o
n
c
r
e
t
e

s
t
r
a
i
n

a
t

b
o
t
t
o
m

Rate dependent failure model with debonding
Rate independent failure model with debonding
Rate dependent failure model with perfect bonding
Non-retrofit (control)
Concrete
-0.02
-0.01
0
0.01
0.02
0.03
0.04
0.05
0.06
w/ strain-rate
w/o strain-rate
Perfect bonding
Time (sec)
G
F
R
P

s
t
r
a
i
n

a
t

b
o
t
t
o
m
Ultimate strain of GFRP
Rate dependent failure model with debonding
Rate independent failure model with debonding
Rate dependent failure model with perfect bonding
GFRP
0 0.01 0.02 0.03 0.04
(a)
(b)
Fig. 12. Analysis results of strains in GFRP and concrete.
Table 2
Summary of analysis results.
Non-
retrotted
Strain-rate independent FRP and
debonding
Strain-rate dependent FRP and perfect
bonding
Strain-rate dependent FRP and
debonding
Arrival time at max.
displacement (s)
0.0010 0.0090 0.0085 0.0095
Max. displacement (mm) 12.29 11.14 10.23 10.56
Duration time of positive phase
(s)
>0.040 0.034 0.027 0.030
Max. strain of concrete (mm/
mm)
0.005 0.0046 0.0039 0.0039
Max. strain of GFRP
(mm/mm) 0.021 0.025 0.019
Material failure Concrete Concrete, GFRP Concrete, GFRP Concrete, GFRP
Interfacial failure Debonding Debonding
J.-W. Nam et al. / Composite Structures 92 (2010) 12121222 1219
el are lower than that of perfect bonding model. And, the concrete
strains of debonding model are higher than that of perfect bonding
model. The analysis results using the different FRP models are
summarized in Table 2.
4. Discussions
The displacement history results of the analyses and experi-
ments are shown in Fig. 13. The analysis result of rate dependent
FRP failure model with debonding is used for the comparison.
The overall analyses results show good agreement with the exper-
imental data. However, the experimental results show some irreg-
ular tendency in the earlier part of the displacement behavioral
curves. Also, the experimental displacement behavioral curves ob-
tained from non-retrotted and GFRP retrotted slabs crossed each
other in the latter part of the displacement behavioral curves.
These unreasonable results are caused by measuring errors in the
test eld. According to Razaqpur and Tolba [10,19], there exist
numerous variability in the eld blast test, causing test results to
be inaccurate. The variability can be attributed to measuring
instrumentation inaccuracies and changing environmental condi-
tions such as atmospheric pressure, temperature, wind, etc. Since
all of the variability cannot be considered in the analysis, analysis
results should be compared to experimental results with an allow-
able error margin. In this study, the analyses and the experimental
results are within an allowable error margin as shown in Fig. 14.
Regarding the articial energy loss of the nite element analy-
sis, the internal energy is compared with the hourglass energy.
The hourglass effect can come out easily in specic elements such
as solid elements. In this study, FlanaganBelytschko viscous form
with exact volume integration for solid elements is used to control
hourglass effect in the analysis [21]. Viscous hourglass and stiff-
ness controls are accepted as suitable solutions for problems with
elements deforming with high and low velocities, respectively,
especially if the number of time steps is large. Since the blast anal-
ysis is in the regime of very high velocity problem, the viscous
hourglass control is adopted in the study. Also, for solid elements,
the exact volume integration provides some advantage for highly
distorted elements. Fig. 15 shows the comparison of internal en-
ergy and hourglass energy of target structural model. The hour-
glass effect of elements is controlled effectively as shown in Fig. 15.
For the local failure of GFRP, the analyses results simulate and
locate GFRP failures as shown in Fig. 16. The dynamic debonding
strength from the calculation of impulse curve of blast pressure
have reasonable values and trends for the blast analysis. The sum-
marized comparisons of analyses results and experimental data are
tabulated in Table 3. From the comparisons, the presented blast
analysis methodology can predict the behavior of FRP retrotted
concrete structures under blast loading and the rened FE analysis
can be used to evaluate effectiveness of FRP retrotting design of
concrete structural members subjected to blast loading.
5. Conclusions
The conclusions from this study are as follows:
1. The blast analysis methodology for the evaluation of FRP retro-
tting effectiveness is presented. Rate dependent FRP material
model and dynamic debonding failure model are adopted for
the numerical analysis technique. With the presented analysis
methodology, the evaluation procedure for the FRP retrotting
design is also presented. The blast analysis is carried out for
GFRP retrotted RC slab and analysis results are compared with
the previously reported experimental results.
2. For a rate dependent FRP model, the enhanced failure criteria is
dened. In order to evaluate the dynamic bond behavior
between concrete and FRP under blast load, the dynamic bond
strength is calculated by integration of Friedlander decay func-
tion. The calculated dynamic bond strength is implemented in
the nite element analysis using the contact interface model.
3. In the comparative study, the FRP retrotting effectiveness is
evaluated by using different FRP failure models: rate dependent
FRP failure with debonding; rate independent FRP failure with
debonding; rate dependent FRP failure with perfect bonding.
The displacement history, damage index, and strain history
are analyzed in the respects of global structural behavior, retro-
tting sufciency, and debonding failure.
-5
0
5
10
15
20
0.00 0.01 0.02 0.03 0.04
GSS4
CS2
k9 w/o strain effect
Non-retrofit
Experiment for GFRP retrofit slab (Tolba2001)
Experiment for non-retrofit slab (Tolba2001)
FE analysis for GFRP retrofit slab (this study)
FE analysis for non-retrofit slab (this study)
Time (sec)
D
i
s
p
l
a
c
e
m
e
n
t

(
m
m
)
Fig. 13. Comparison of experiment and analysis results in displacements.
0
4
8
12
16
Non-retrofit GFRP retrofitted
M
a
x
i
m
u
m

d
i
s
p
l
a
c
e
m
e
n
t

(
m
m
)
Response range for non-retrofitted cases
(Razaqpuret al. 2006; Tolba2001)
Response range for GFRP retrofitted cases
(Razaqpuret al. 2006; Tolba2001)
Fig. 14. Comparison of the maximum displacement response ranges.
0.00E+00
5.00E+02
1.00E+03
1.50E+03
2.00E+03
2.50E+03
3.00E+03
3.50E+03
0 50 100 150 200
Time (msec)
E
n
e
r
g
y

(
k
J
)
Internal Engergy
Hourglass Energy
Fig. 15. Comparison of internal and hourglass energy of analysis model.
1220 J.-W. Nam et al. / Composite Structures 92 (2010) 12121222
Experimental results [10, 19]
Compression side Tension side
M
D
D
M
M, D
M, D
M, D
D
M
M
M : Material failure
D : Debonding failure
Debonding failure
Analysis results
(b)
(a)
Fig. 16. Comparison of local debonding failure in the experiment and the analysis: (a) experimental results [10,19]; (b) debonding failure in analysis.
Table 3
Comparisons of analysis results and experimental data [8,17].
Experimental data [8,17] Analysis results
Non-retrot GFRP retrot Non-retrot GFRP retrot
Arrival time at max. displacement (s) 0.0013 0.0075 0.0010 0.0095
Max. displacement (mm) 12.50 9.58 12.29 10.56
Duration time of positive phase (s) >0.030 0.024 More than 0.040 0.030
Max. strain of concrete (mm/mm) 0.0027 0.005 0.0039
Max. strain of GFRP (mm/mm) 0.013 0.019
Material failure Concrete Concrete Concrete Concrete, GFRP
Debonding failure Debonding Debonding
J.-W. Nam et al. / Composite Structures 92 (2010) 12121222 1221
4. The numerical analysis results are veried with experimental
data which have been measured by previous researchers. Even
though the experimental data varies with ranges, the analysis
results of overall displacement history and FRP failure corre-
spond with experimental results. The presented blast analysis
procedure is conrmed to be used for the evaluation of FRP
effectiveness in the blast retrotting design.
5. Further experiments on the strain-rate effect according to the
types and layers of FRP are needed to build more accurate
model and retrotting design procedure for RC structure under
blast loading. Also, further researches on the dynamic interfa-
cial behavior between FRP and concrete are required. The differ-
ences of bond behaviors under various dynamic loads could be
experimentally obtained by dynamic pull-out test.
Acknowledgments
The authors would like to thank Korea Research Foundation
(D00792) and Korea Science and Engineering Foundation (R01-
2008-000-1117601) for its nancial support of this research.
References
[1] ASCE. Structural design for physical security, state of the practice; 1999.
[2] US Department of the Army. Structures to resist the effects of accidental
explosions. Technical manual 5-1300; November 1990.
[3] Nam JW, Kim HJ, Kim SB, Kim JHJ, Byun KJ. Analytical study of nite element
models for FRP retrotted concrete structure under blast loads. Int J Damage
Mech 2009;18(5):46190.
[4] Buchan PA, Chen JF. Blast resistance of FRP composite and polymer
strengthened concrete and masonry structures: a state-of-the-art review.
Compos Part B: Eng 2007;38(5/6):50922.
[5] Ross CA, Purcell MR, Jerome EL. Blast response of concrete beams and slabs
externally reinforced with bre reinforced plastics (FRP). In: Proceedings of the
struct cong XV building to last, Portland, USA; 1997. p. 6737.
[6] Muszynski LC, Purcell MR, Sierakowski R. Strengthening concrete structures by
using externally applied composite reinforcing material. In: Proceedings of the
seventh international symposium on interaction of the effects of munitions
with structures, Germany; 1995. p. 2918.
[7] White TW, Soudki KA, Erki MA. Response of RC beams strengthened with CFRP
laminates and subjected to a high rate loading. J Compos Constr
2001;5(3):15362.
[8] Jerome DM, Ross CA. Simulation of the dynamic response of concrete beams
externally reinforced with carbonber reinforced plastic. Comput Struct
1997;64(5/6):112953.
[9] Silva PF, Lu B. Improving the blast resistance capacity of RC slabs with
innovative composite materials. Compos Part B: Eng 2007;38:52334.
[10] Tolba A. Response of FRP-retrotted reinforced concrete panels to blast
loading. PhD thesis, Ottawa, Canada, Carleton University; 2001.
[11] Muszynski LC, Purcell MR. Composite reinforcement to strengthen existing
concrete structures against air blast. J Compos Constr 2003;7(2):937.
[12] Nam JW, Kim HJ, Yi NH, Kim IS, Kim JHJ, Chot HJ. Blast analysis of concrete arch
structures for FRP retrotting design. Comput Concrete 2009;6(4):30518.
[13] Biggs JM. Introduction to structural dynamics. New York: McGraw-Hill; 1964.
[14] Crawford JE, Malvar J, Wesevich JW, Valancius J, Raynolds AD. Retrot of
reinforced concrete structures to resist blast effects. ACI Struct J
1997;94(4):3717.
[15] Crawford JE, Malvar LJ, Morrill KB, Ferritto JM. Composite retrots to increase
the blast resistance of reinforced concrete buildings. In: Proceedings of the
tenth international symposium on interaction of the effects of munitions with
structures. San Diego, USA; 2001. p. 113.
[16] Malvar LJ, Crawford JE, Wesevich JW, Simons D. A plasticity concrete material
model for DYNA3D. Int J Imp Eng 1997;19(9/10):84773.
[17] Malvar LJ, Crawford JE. Dynamic increase factor of concrete. In: The twenty-
eighth DDESB seminar. Orlando, FL; August 1998.
[18] Malvar LJ, Morrill KB, Crawford JE. Numerical modeling of concrete conned
by ber-reinforced composites. J Compos Constr 2004;8(4):31522.
[19] Razaqpur AG, Tolba A, Contestabile E. Blast loading response of reinforced
concrete panels reinforced with externally bonded GFRP laminates. Compos
Part B: Eng 2007;38(5/6):53546.
[20] Chen WF. Plasticity in reinforced concrete. New York: McGraw Hill; 1982.
[21] LSTC. LS-DYNA keyword users manual version 9.71. Livermore Software
Technology Corporation; July 2006.
[22] Nam JW, Kim JHJ, Kim SB, Yi NH, Byun KJ. A study on mesh size dependency of
nite element blast structural analysis induced by non-uniform pressure
distribution from high explosive blast wave. KSCE J Civil Eng 2008;12(4):
25965.
[23] Jones N. Structural aspects of ship collisions. In: Jones N, Wierzbicki T, editors.
Structural crashworthiness. London: Butterworths; 1983. p. 30837.
[24] Hashin Z. Failure criteria for unidirectional ber composite. ASME J Appl Mech
1980;47(2):32934.
[25] Chang FK, Chang KY. A progressive damage model for laminated composite
containing stress concentration. J Compos Mater 1987;21(9):83455.
[26] Park H, Lee K, Lee SW, Kim K. Dynamic analysis of nonlinear composite
structures under pressure wave loading. J Compos Mater 2006;40(15):
136183.
[27] Al-Hassini STS, Kaddour AS. Strain rate effect on GFRP, KFRP and CFRP
composite laminates. Key Eng Mater 1998;141143(Pt2):42752.
[28] Lorenzis LD, Tegola AL. Bond of FRP laminates to concrete under impulse
loading: a simple model. In: Proceedings of the international symposium on
bond behaviour of FRP in structures (BBFS 2005); 2005. p. 495500.
[29] Yuan H, Wu Z, Yoshizawa H. Theoretical solutions on interfacial stress transfer
of externally bonded steel/composite plates. J Struct Mech Earthq Eng, JSCE
2001;18(1):2739.
[30] Hyde DW. Fundamental of protective design for conventional weapons,
CONWEP (conventional weapons effects). TM5-8511-1, United States Army
Waterway Experiment Station, Vicksburg, Miss; 1992.
1222 J.-W. Nam et al. / Composite Structures 92 (2010) 12121222