Effect of Blast Loading On CFRP-Retrofitted RC Columns - A Numerical Study
Effect of Blast Loading On CFRP-Retrofitted RC Columns - A Numerical Study
Effect of Blast Loading On CFRP-Retrofitted RC Columns - A Numerical Study
,
Y.A. Al-Salloum and
S.H. Alsayed
Specialty Units for Safety and Preservation of
Structures, College of Engineering, King Saud
University, Riyadh 11421 Saudi Arabia
Received 1 Oct 2010;
In revised form 9 Feb 2011
2
d
c
clear cover to longitudinal reinforcing bars,
t
s
thickness of equivalent smeared ring for longitudinal steel,
t
s
=
n
2
8r
s
m
s
modular ratio of reinforcing steel,
m
s
=
E
s
E
c
m
f
modular ratio of CFRP,
m
f
=
E
f
E
c
The depth of neutral axis (NA) of the section may be found by taking the moment of
eective transformed areas about NA, thus giving:
A
c
y
c
+ A
esc
y
sc
+ A
efc
y
fc
A
est
y
st
A
eft
y
ft
= 0 (3)
where,
A
c
area of concrete in compression,
A
c
= r
2
{ sin () cos ()}
y
c
distance of centroid of the area of concrete in compression from NA,
y
c
=
2r
3
_
sin
3
()
sin () cos ()
_
r + y, = cos
1
_
ry
r
_
A
esc
area of smeared ring of steel in compression,
A
esc
= 2 (m
s
1) r
s
s
t
s
y
sc
distance of centroid of the smeared ring of steel in compression from NA,
y
sc
= 2r
2
s
sin (
s
) r + y
A
est
area of smeared ring of steel in tension,
A
est
= 2m
s
r
s
(
s
) t
s
y
st
distance of centroid of the smeared ring of steel in tension from NA,
y
st
= 2r
2
s
sin (
s
) + r y
A
efc
transformed area of CFRP in compression,
A
efc
= 2m
f
rt
f
Latin American Journal of Solids and Structures 8(2011) 55 81
H.M. Elsanadedy et al / Eect of blast loading on CFRP-Retrotted RC columns a numerical study 61
y
fc
distance of centroid of the CFRP in compression from NA,
y
fc
= 2r
2
sin () r + y
A
eft
transformed area of CFRP in tension,
A
eft
= 2m
f
r ( ) t
f
y
ft
distance of centroid of the CFRP in tension from NA,
y
ft
= 2r
2
sin ( ) + r y
The cracked moment of inertia of the section is given by:
I
cr
=
r
4
4
_
sin () cos () + 2 sin
3
() cos ()
_
A
c
y
2
c
+ (m
s
1) r
3
s
t
s
{
s
+ sin (
s
) cos (
s
)}
A
esc
y
2
sc
+ m
s
r
3
s
t
s
{
s
sin (
s
) cos (
s
)} A
est
y
2
st
+ m
f
r
3
t
f
{ + sin () cos ()}
A
efc
y
2
fc
+ m
f
r
3
t
f
{ sin () cos ()} A
eft
y
2
ft
(4)
where,
s
= cos
1
_
ry
r
s
_
.
2.2 Dynamic response
Considering a reinforced concrete column of length L and mass per unit length m, as shown in
Fig. 2. The mode shapes of vibration of an Euler column may be expressed as [8]:
(x) = A
1
cos(ax) + A
2
sin(ax) + A
3
cosh(ax) + A
4
sinh(ax) (5)
with the rst and second spatial derivatives given by:
(x) =
1
a
{A
1
sin(ax) + A
2
cos(ax) + A
3
sinh(ax) + A
4
cosh(ax)} (6)
(x) =
1
a
2
{A
1
cos(ax) A
2
sin(ax) + A
3
cosh(ax) + A
4
sinh(ax)} (7)
where,
x distance measured from one end of column,
a eigen value parameter (unit: Length
1
) such that,
a
4
=
4
2
f
2
m
EI
f natural frequency of column,
m mass per unit length of column
m = r
2
c
r
c
density of reinforced concrete,
E
c
modulus of elasticity of reinforced concrete,
E
c
= 4700
f
c
[3],
f
c
compressive cylinder strength of concrete,
I moment of inertia of column.
Latin American Journal of Solids and Structures 8(2011) 55 81
62 H.M. Elsanadedy et al / Eect of blast loading on CFRP-Retrotted RC columns a numerical study
In the beginning, when both ends of the columns are xed i.e. (c) =
(x) = 0 at x = 0
and at x = L, thus giving:
_
_
1 0 1 0
0 1 0 1
cos(aL) sin(aL) cosh(aL) sinh(aL)
sin(aL) cos(aL) sinh(aL) cosh(aL)
_
_
_
_
A
1
A
2
A
3
A
4
_
_
=
_
_
0
0
0
0
_
_
(8)
Thus giving the characteristic equation as,
cos(aL) cosh(aL) = 1 (9)
The mass of the column being distributed along the length, there are innite sets of frequen-
cies and associated modes that satisfy the above equation. The rst few values are: aL = 4.7300,
7.8532, 10.9956, 14.1372, 17.2788 . . .
When the column is subjected to blast, plastic hinges may be formed at the ends thus the
boundary conditions get transformed to: (x) =
_
1 0 1 0
1 0 1 0
cos(aL) sin(aL) cosh(aL) sinh(aL)
cos(aL) sin(aL) cosh(aL) sinh(aL)
_
_
_
_
A
1
A
2
A
3
A
4
_
_
=
_
_
0
0
0
0
_
_
(10)
The characteristic equation obtained for the above is:
sin(aL) sinh(aL) = 0 (11)
The roots of the above equation are: aL = n, where, n = 1, 2, 3, . . ..
It may be noted here that the above analysis is based on the assumption of prismatic column
section and perfectly straight column axis. The exposure of column to blast may result in non-
uniform material erosion, permanent deformation in the column axis and the hinge action as
a result of damage may not occur at both ends or may be partial. It is due to these reasons
that the actual response of the column may dier from the analysis presented above.
2.3 Discussion
The section properties of the column taken in the study are given in Table 2. The fundamental
frequency and time period calculated based on these section properties and for the two end
conditions (i.e. both ends xed and both ends hinged) are given in Table 3. It may be noted
here that the section is considered to be prismatic. The variation in the natural frequency and
time period of the column has been plotted in Figs. 3 and 4 respectively.
It is observed from Table 2 that the spalling of cover, which is going to occur due to the
vibrations during the blast loading, reduces the cracked moment of inertia of the section by
25.2%. It is assumed that the concrete cover remains attached even after its spalling thus the
mass per unit length, which is taken to be uniform, remains unaected. The advantages of
Latin American Journal of Solids and Structures 8(2011) 55 81
H.M. Elsanadedy et al / Eect of blast loading on CFRP-Retrotted RC columns a numerical study 63
providing CFRP (longitudinal and circumferential) are two folds one of increasing the section
parameters (area and moment of inertia) and the other of preventing the spalling of concrete
cover by providing connement. The concrete cover which is otherwise brittle because of non-
connement becomes conned and adds to the ductility of the section. The provision of CFRP
enhances the cracked moment of inertia of the section by 34.6%. Further, if a comparison is
made with the moment of inertia of the section without cover then the enhancement is 79.9%.
Table 2 Section properties of column section.
S. No. Property Value
1. Radius, r (m) 0.3
2. Concrete grade, f
c
(MPa) 30
3. Reinforcement:
Longitudinal 1616 (A
st
= 32.17cm
2
, 1.14%)
Ties 10@200
4. Retrotting using CFRP
Longitudinal 2 layers of 1 mm each
Hoop 2 layers of 1 mm each
5. Gross transformed area, A
g
(cm
2
)
no retrotting 3077.4
with retrotting 3197.4
6. Gross moment of inertia, I
g
(m
4
)
no retrotting with cover 0.007053
with retrotting 0.007594
7. Depth of NA of cracked section, y (mm)
no retrotting
with cover 155.1
without cover 146.2
with retrotting 172.8
8. Cracked moment of inertia, I
cr
(m
4
)
no retrotting
with cover 0.002373
without cover 0.001776
with retrotting 0.003195
It is observed from Table 3 and Figs. 3 and 4 that the change in the values of moment of
inertia of the section results in signicant change in the natural frequency of dierent modes
of the column.
Latin American Journal of Solids and Structures 8(2011) 55 81
64 H.M. Elsanadedy et al / Eect of blast loading on CFRP-Retrotted RC columns a numerical study
Table 3 Fundamental time period and frequency of dierent modes.
State of section
Mode-1 Mode-2 Mode-3 Mode-4
T f T f T f T f
(ms) (Hz) (ms) (Hz) (ms) (Hz) (ms) (Hz)
Both ends xed
No cracking
no retrotting 8.7 115 3.2 317 1.6 622 1.0 1028
with retrotting 8.4 119 3.0 329 1.5 645 0.9 1067
Cracked section
with cover 15.0 67 5.4 184 2.8 361 1.7 596
no retrotting
without cover 17.0 59 6.2 163 3.1 319 1.9 527
with retrotting 12.9 77 4.7 214 2.4 419 1.4 692
Both ends hinged
No cracking
no retrotting 19.7 51 4.9 203 2.2 457 1.2 813
with retrotting 19.0 52 4.7 211 2.1 474 1.2 843
Cracked section
no retrotting
with cover 34.0 29 8.5 118 3.8 265 2.1 471
without cover 38.4 26 9.6 104 4.3 234 2.4 416
with retrotting 29.3 34 7.3 137 3.3 308 1.8 547
Figure 3 Variation in frequency of dierent modes for moment of inertia varying from cracked to uncracked
section.
Latin American Journal of Solids and Structures 8(2011) 55 81
H.M. Elsanadedy et al / Eect of blast loading on CFRP-Retrotted RC columns a numerical study 65
Figure 4 Variation in time period of dierent modes for moment of inertia varying from cracked to uncracked
section.
3 NUMERICAL MODELING
LS-DYNA [15], a general purpose nite element program was used to develop the 3-D model
of the column. Two cases were considered in the modeling of the column. The rst case
involved the column to be modeled without any strengthening and the second case involved
strengthening of the column with Carbon Fiber Reinforced Polymer (CFRP) sheets. Damping
has been ignored, as it has a negligible eect for short duration, impulsive loads.
3.1 Finite element mesh
Modeling of the column was rst completed using ANSYS-Version 11 as it has a very strong
graphical user interface and the le was then imported to FEMB (which is a preprocessor for
LS-DYNA) database for incorporating the dierent parts as well as the blast interface and
contact segments. A combination of eight and six node solid elements was used to model the
concrete volume. The longitudinal reinforcing bars and ties were modeled using 2-node Hughes
Lui beam elements. For the modeling of CFRP sheets, 4-node shell elements were employed.
Perfect bond was assumed between rebar elements and the surrounding concrete volume and
also between the FRP and the concrete substrate. Figure 5 details the mesh discretization for
the concrete elements, the CFRP elements and the reinforcing cage used in the study. In order
to study the eect of rening the mesh on the numerical results, another ne mesh was created
as shown in Table 4. A comparison of the two meshes used in the study is also detailed in the
table. The main dierence between the two meshes is in the number of concrete elements per
column section. The total numbers of elements in the model are 13472 and 18592 for Mesh 1
and Mesh 2 respectively.
Latin American Journal of Solids and Structures 8(2011) 55 81
66 H.M. Elsanadedy et al / Eect of blast loading on CFRP-Retrotted RC columns a numerical study
(a) Concrete solid
elements
(b) CFRP shell elements (c) Reinforcing beam
elements
Figure 5 Mesh discretization of column in LS-DYNA.
3.2 Material modeling
The Karagozian & Case (K&C) model [17], designated as Material type 072R3 in LSDYNA, was
employed to represent concrete for the column. The model is specially designed for predicting
the response of concrete under blast loads. It is a three-invariant model which uses three
shear failure surfaces and includes damage and strain rate eects. It also incorporates many
important features of concrete behavior such as tensile fracture energy, shear dilation and
eects of connement. The reinforcement was modeled using material type 024 to model the
elasto-plastic response with strain rate dependency. In order to model the CFRP material,
type 054-055 was utilized, which is capable of dening orthotropic material characteristics.
The material angles for the longitudinal and circumferential layers were specied as 0
and
90
respectively. The manufacturers data sheet for the CFRP material was used for dening
dierent material parameters. The laminated shell theory was used for the purpose of correcting
the assumption of a uniform constant shear strain throughout the thickness of the composite
shell, thus avoiding very sti results. The failure criteria of composite material used in the
analysis is the one proposed by Chang and Chang [7] with special features of compression
failure governed by the criteria of Matzenmiller and Schweizerhof [18]. A summary of material
properties used in the analysis are presented in Table 1.
3.3 Erosion
The erosion option provides a way of including failure to the material models. This is not a
material or physics-based property; however, it lends a great means to imitate concrete spalling
phenomena and produce graphical plots which are more realistic representations of the actual
events. By activating this feature, the eroded solid element is physically separated from the rest
of the mesh. This erosion model represents a numerical remedy to distortion, which can cause
Latin American Journal of Solids and Structures 8(2011) 55 81
H.M. Elsanadedy et al / Eect of blast loading on CFRP-Retrotted RC columns a numerical study 67
Table 4 Comparison between nite element meshes used in this study.
Parameter Mesh 1 Mesh 2
Finite element mesh of
column section
No. of concrete elements
per section
112 176
Total No. of concrete ele-
ments
8960 14080
Size of concrete element ranges from 30 to 90 mm ranges from 30 to 50 mm
No. of beam elements for
longitudinal bars
1280 1280
Length of longitudinal bar
element
50 mm 50 mm
No. of beam elements for
transverse ties
672 672
Length of transverse tie el-
ement
50 mm 50 mm
No. of FRP shell elements 2560 2560
Size of FRP shell element 50 59 mm 50 59 mm
Total No. of elements per
model
13472 18592
excessive and unrealistic deformation of the mesh. The application of erosion to the simulated
model requires calibration with experimental results; however in the absence of experimental
validation, the consequence of possible discrepancy in the erosion specied is limited. This is
because the damage level of the concrete material is basically governed by the material model
itself. In this study, the concrete elements in the RC column were allowed to erode when the
principle tensile strain reached 50% [14]. Column failure is characterized by the volume of
eroded concrete elements within a particular section with respect to the total elements in the
section, which will have an index about the axial load resistance of the column.
Latin American Journal of Solids and Structures 8(2011) 55 81
68 H.M. Elsanadedy et al / Eect of blast loading on CFRP-Retrotted RC columns a numerical study
3.4 Loading and boundary conditions
Fixed boundary conditions were assigned for the top and bottom nodes of the column. The
axial load acting upon the column due to dead plus live loads from upper stories was applied
as nodal loads at the column top. This axial load was applied as a ramp function over a period
of 0.5 s as shown in Fig. 6.
Dierent charge weights of 100, 200, 500 and 1000 kg equivalent weight of TNT at stand-
o distances of 1, 4 and 15 m were considered in the study. Both the un-strengthened and
CFRP- strengthened columns were subjected to these blast loads. The blast loads impinging
on the contact segments of the column were calculated by the software using ConWep [25].
The contact segments of the blast were the solid elements of the front face of the column which
were taken to be in contact with the blast. The vertical height of the charge was taken as 1.0
m above the base of the column because the explosive is assumed to be carried in a vehicle.
Thus, the shock transmitted to the column through ground gets diminished due to which it
has been ignored in the analysis. The blast loading was set to trigger at 0.5 seconds as shown
in Fig. 6.
Figure 6 Loading procedure for dead, live and blast loads on column.
3.5 Solution strategy
LS-DYNA uses explicit time integration algorithm for solving the problems, which is less sen-
sitive to machine precision than other nite element solution methods. The benets of this
are greatly improved utilization of memory and disk. An explicit FE analysis solves the in-
cremental procedure and updates the stiness matrix at the end of each increment of load (or
displacement) based on changes in geometry and material. The termination time of 1.5 s was
set in order to realize the complete blast related response of the column.
Latin American Journal of Solids and Structures 8(2011) 55 81
H.M. Elsanadedy et al / Eect of blast loading on CFRP-Retrotted RC columns a numerical study 69
3.6 Blast load
The reected pressure and positive phase duration found from ConWep for dierent charge
weights considered in the study are plotted in Fig. 7. ConWep calculates air blast parameters
using the equations found in Ref. [13] which is based on the data from explosive tests using
weights from less than 1kg to over 400,000 kg. It is found from the air blast parameters
calculation that ConWep may not be used for the explosion of 500 and 1000 kg charge weights
at 1m range. The minimum range for the applicability of ConWep for 500 and 1000 kg charge
weights is found as 1.42 and 1.79 m respectively. It is due to this reason that the column has
been analyzed for these charge weights (i.e. 500 and 1000 kg) at 2m range and found to have
completely destroyed. Thus obviously these charge weights at 1 m range would also destroy
the column. Though no analysis for these charge weights at 1 m range has been carried out
but the results reported latter for these cases are based on the results of analysis for 2 m range.
Figure 7 Reected pressure and positive phase duration for dierent charge weights.
4 ANALYSIS RESULTS
4.1 Eect of mesh size
Two cases, one each from the un-retrotted and the retrotted column cases of blast scenarios
were used to compare the results of the two meshes considered above for the purpose of mesh
sensitivity analysis. Table 5 shows the results of the numerical convergence study. The numer-
ical convergence study showed that further decrease in the mesh size has little eect on the
numerical results but leads to the risk of computer memory overow and substantially increases
the computing time. In order to achieve maximum computing eciency and thereby reduce
the run-time, it was decided to use Mesh-1 for all parametric cases of blast loading simulation.
Latin American Journal of Solids and Structures 8(2011) 55 81
70 H.M. Elsanadedy et al / Eect of blast loading on CFRP-Retrotted RC columns a numerical study
Table 5 Results of mesh sensitivity analysis.
Parameter
Column without FRP
(200 kg charge at 4m
stando distance)
Column with FRP (1000
kg charge at 4m stando
distance)
Mesh 1 Mesh 2 Mesh 1 Mesh 2
Peak lateral displace-
ment (mm)
41.05 43.23 124.98 123.40
Permanent lateral dis-
placement (mm)
20.39 23.21 103.14 101.78
Max stresses in longitu-
dinal bars (MPa)
512 508 623 606
Maximum stresses in
transverse bars (MPa)
510 545 701 708
4.2 Displacements and time period of vibration
The time history of maximum lateral displacement of column for two typical combinations
of charge weights and stand-o distances are shown in Figs. 8 and 9. The peak lateral and
permanent displacement of column for the blast scenarios considered in the analysis are given
in Table 6.
Figure 8 Lateral displacement of column subjected to blast due to 100kg charge at 15m stand-o distance.
The observations made from the displacement record are summarized below:
i) As stated previously, the charge weights of 500 and 1000 kg at a stand-o distance of
2.0 m completely destroyed both the columns with and without CFRP strengthening.
Accordingly, these charge weights at 1 m range would also destroy the columns. So it can
Latin American Journal of Solids and Structures 8(2011) 55 81
H.M. Elsanadedy et al / Eect of blast loading on CFRP-Retrotted RC columns a numerical study 71
Figure 9 Lateral displacement of column subjected to blast due to 100kg charge at 1m stand-o distance.
be assumed that the columns within the focus of a blast of this magnitude would be totally
destroyed and may not be protected by retrotting.
ii) The retrotted as well as un-retrotted columns subjected to the blast of 100 kg charge
weight at 15 m stand-o distance do not undergo any damage, thus the time period of
vibration for this load case (Fig. 8) at the close of the period of analysis when the vibration
is almost free should be the initial period of vibration of the column in the undamaged
state. Thus the time period of the column initially in the undamaged state obtained from
the free vibration response of column subjected to the least intensity blast (100 kg charge
weight at 15 m stand-o distance) obtained from the end of the period of analysis is 10.2
and 9.0 s respectively for the un-retrotted and retrotted columns (Fig. 8). A comparison
of these values with the analytically obtained time period (Table 3 and Fig. 4) shows that
the values lie between the time period calculated for the rst mode of the cracked and
uncracked state with both ends xed. This is due to the cracking of the column section
due to vibration. The predominant mode of vibration initially is thus mode number 1 for
both ends xed condition.
iii) The retrotting of column considered in the study results in 11.7% reduction in the initial
natural time period of vibration of the column as obtained from the free vibration response
of column subjected to the least intensity of blast (100 kg charge weight at 15 m stand-o
distance) obtained from the end of the period of analysis.
iv) The retrotting of column reduces the peak lateral displacement considerably. The retrotting
of column reduces the peak displacement by 21% when the damage to the column is almost
negligible i.e. when the intensity of blast is least severe (100 kg charge weight at 15 m
stand-o distance). A study of all blast cases considered indicates that the reduction of
peak displacement varies from 8% for 100 kg charge weight at stand-o distance of 4 m to
79% for 500 kg charge weight at a stand-o distance of 4 m.
Latin American Journal of Solids and Structures 8(2011) 55 81
72 H.M. Elsanadedy et al / Eect of blast loading on CFRP-Retrotted RC columns a numerical study
Table 6 Peak and permanent lateral displacement of column for dierent blast scenarios
.
Charge
Weight
(kg)
Peak displacement Permanent displacement
(mm) (mm)
d = 1 m d = 4 m d = 15 m d = 1 m d = 4 m d = 15 m
Column without retrotting
100 277.53 10.54 0.89 260.21 3.95 0.08
200 489.53 41.05 1.83 478.89 20.39 0.10
500 214.67 4.19 173.55 0.53
1000 474.85 8.39 465.95 0.33
Retrotted Column
100 61.32 8.87 0.70 53.67 2.04 0.13
200 314.91 16.27 1.36 294.87 6.42 0.24
500 45.00 3.50 30.56 0.47
1000 124.98 6.62 103.14 0.61
d: stand-o distance
v) There is exponential increase in peak lateral displacement as well as the permanent dis-
placement with the reduction in the stand-o distance.
vi) At 15 m stand-o distance, the blast of even 1000 kg charge weight does not cause any
signicant damage to the column even without retrotting. Considering 30 mm as the
acceptable permanent lateral displacement for the column, at 4 m stand-o distance, blast
of 100 kg charge weight may be resisted by the column even without retrotting, whereas,
200 and 500 kg charge weights may be resisted by the column after retrotting. Higher
charge weight of 1000 kg could not be resisted by the retrotting considered in the study.
At a stand-o distance of 1 m, the blast of even 100 kg may not be resisted by even the
retrotted column. The increase in the number of layers of CFRP may however help the
column to resist it.
vii) The time period of vibration gets elongated with cracking and damage to the column. The
amount of damage (concrete fracture and yielding/fracture of steel) caused to the column
increases with the increase in the intensity of blast, thus the time period of vibration also
increases with increase in the charge weight and/or reduction in the stand-o distance
(Figs. 8-9).
4.3 State of stress and consequent damage
Table 7 depicts the maximum tensile stress in the longitudinal as well as the transverse reinforce-
ment bars as a result of dierent blast scenarios for both the retrotted and the un-retrotted
columns. From the results in Table 7, it is noted that the maximum values of tensile stress for
the transverse reinforcing bars in some blast scenarios for the retrotted columns were found
Latin American Journal of Solids and Structures 8(2011) 55 81
H.M. Elsanadedy et al / Eect of blast loading on CFRP-Retrotted RC columns a numerical study 73
to be higher than the corresponding blast scenarios of un-retrotted columns. This is due to
the erosion of concrete in un-retrotted column which results in the release of stress in ties
because the stress in ties is mainly due to the presence of concrete; whereas, the retrotted
column concrete being conned concrete, erosion is less and hence the stress in ties is more.
For these cases it was also noticed that the un-retrotted column had failed as a result of direct
shear due to the proximity of the blast. However, the retrotting of the columns for these cases
improved the overall blast resisting capacity by increasing the shear capacity of the column.
Table 7 Maximum tensile stress in longitudinal and transverse reinforcement bars for dierent blast loading
scenarios
.
Charge
Weight
(kg)
Maximum Tensile Stress Maximum Tensile Stress
longitudinal reinforcement transverse reinforcement
(MPa) (MPa)
d = 1 m d = 4 m d = 15 m d = 1 m d = 4 m d = 15 m
Column without retrotting
100 591 473 119 497 510 8.4
200 665 512 193 638 510 40
500 F 577 294 F 515 222
1000 F 620 443 F 508 502
Retrotted column
100 502 417 88 667 477 15
200 726 492 138 700 502 38
500 F 516 246 F 518 127
1000 F 623 367 F 701 384