Construction and Building Materials 40 (2013) 145–161

Contents lists available at SciVerse ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Experimental and numerical investigation for the flexural strengthening of RC

beams using near-surface mounted steel or GFRP bars
Tarek H. Almusallam, Hussein M. Elsanadedy ⇑,1, Yousef A. Al-Salloum, Saleh H. Alsayed
Specialty Units for Safety & Preservation of Structures, Department of Civil Engineering, King Saud University, P.O. Box 800, Riyadh 11421, Saudi Arabia

h i g h l i g h t s

" The worth of NSM bars for upgrading the flexural capacity of RC beams is investigated.
" A total of eight groups of 16 beams were tested under four-point bending till failure.
" The ACI 318-11 code and ACI 440.1R-06 guidelines were used for peak load prediction.
" Nonlinear finite element analysis was conducted using LS-DYNA software.
" NSM bars were found to be successful at upgrading the flexural capacity of RC beams.

a r t i c l e i n f o a b s t r a c t

Article history: In this paper, the effectiveness of NSM bars as a means of restoring or upgrading the flexural capacity of
Received 9 June 2012 RC beams is experimentally and numerically investigated. The studied parameters included type of NSM
Received in revised form 9 September 2012 bars: steel versus GFRP, and NSM reinforcement ratio (number and diameter of inserted NSM bars). A
Accepted 25 September 2012
total of eight groups of 16 beams were tested under four-point bending. The two beams of the first group
Available online 5 December 2012
were reinforced with three main steel bars and were used as control specimens. The two beams of the
second group were reinforced with three main GFRP bars and were utilized for comparison with control
Keywords:
specimens. Assuming that one of the three main steel bars in the control specimen had corroded, three
Near surface mounting
FRP bars
groups of six beams were designed in which one NSM steel or GFRP bar was inserted in the tension side.
Concrete beams Yet, with the assumption of the corrosion of two main steel bars in the control beam, the last three groups
Flexural strengthening of six beams were planned in which two NSM steel or GFRP bars were planted in the tension side. For
Finite element modeling NSM-upgraded beams, special type of epoxy paste was used as bonding agent. Test results showed that
by using NSM steel or GFRP bars to compensate the difference in the main reinforcement, the original
load capacity of the control beam was successfully restored. The ultimate capacity of the beams was pre-
dicted using the ACI 318-11 code and ACI 440.1R-06 guidelines. A numerical investigation utilizing non-
linear finite element (FE) analysis was also carried out using LS-DYNA software. A comparison was made
between the experimental and numerical results and good agreement was obtained. Based on the valida-
tion of FE results, the numerical analysis was extended to include additional cases to study the effect of
FRP reinforcement ratio on the flexural performance of NSM-upgraded beams.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction
Reinforced concrete structures are suffering from different
types of deteriorations. The paramount of all is the one due to steel
corrosion. The damage due to corrosion or cracks due to combined
mechanical and chemical factors implies the need for appropriate
methodology of structural rehabilitation. Several techniques are
⇑ Corresponding author. Tel.: +966 597938718.
E-mail address: elsanadedy@yahoo.com (H.M. Elsanadedy).
1
On leave from the Department of Civil Engineering, Helwan University, Cairo,
Egypt.
now available to repair beams and columns. In the last few years,
near surface mounting (NSM) technique has received more atten-
tion as an alternative for externally bonded FRP laminates in the
flexural strengthening of concrete elements [1–3]. Due to the avail-
ability of high quality resins, NSM technique can offer an easy and
efficient way of rehabilitation for many parts of the structures. Ma-
jor innovative applications have been witnessed in this field to
overcome the problems arising when externally bonded FRP com-
posites are subjected to severe environmental conditions or
mechanical damage [4–6]. Currently, prestressing is employed
with the NSM-FRP strengthening technique to enhance the flexural
performance of strengthened beams [7–9].

0950-0618/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.conbuildmat.2012.09.107
146 T.H. Almusallam et al. / Construction and Building Materials 40 (2013) 145–161

In NSM strengthening technique, longitudinal grooves are first
cut into the concrete cover of beams or slabs, then the reinforcing
bars are inserted into these grooves and bonded with an appropri-
ate binding agent; typically an epoxy paste or a cement grout. The
first application of this technique was the implementation of steel
bars into slots to strengthen a bridge deck slab in Lapland, Finland
in 1940s [10]. The initial research work on NSM technique was re-
ported by Blaschko and Zilch [11] using CFRP strips inserted into
grooves cut at the surface of concrete specimens. The specimens
were tested in a double shear configuration. Test results showed
that strengthening using NSM CFRP strips has a greater anchoring
capacity compared to externally bonded CFRP strips.
Based on the results of several applications [12], it has been
concluded that the increase in the flexural capacity of strengthened
beams applying NSM technique is greater than that using
externally bonded systems for the same axial stiffness of the FRP

Fig. 1. Details of test beams.

 T.H. Almusallam et al. / Construction and Building Materials 40 (2013) 145–161 147

Table 1 Table 2
Test matrix. Material properties used in the FE modeling.

Group ID Main NSM Type of Description No. of Concrete

tension bars NSM specimens
Material model MAT_CSCM_CONCRETE
bars inserted bars
Density (kg/m3) 2320
CS 3Ø10 – – Control beam 2 Uni-axial compressive strength 36.6
(steel) (steel bars) (MPa)
CF 3Ø10 – – Beam reinforced 2 Max aggregate size (mm) 10
(GFRP) with GFRP bars
Ø10 main tension steel bars & Ø10 NSM steel bars
RW1S 2Ø10 1Ø10 Steel One steel bar 2
Material model MAT_PIECEWISE_LINEAR_PLASTICITY
(steel) added on the
Density (kg/m3) 7850
tension side
Young’s modulus (GPa) 200
RW1Ø14S 2Ø10 1Ø14 Steel One steel bar 2
Poisson’s ratio 0.3
(steel) added on the
Yield stress (MPa) 408
tension side
Tangent modulus (MPa) 1458
RW1F 2Ø10 1Ø10 GFRP One GFRP bar 2
Plastic strain to failure 0.1171
(steel) added on the
tension side Ø6 main compression steel bars
RW2S 1Ø10 2Ø10 Steel Two steel bars 2 Material model MAT_PIECEWISE_LINEAR_PLASTICITY
(steel) added on the Yield stress (MPa) 334
tension side Tangent modulus (MPa) 989
RW2Ø14S 1Ø10 2Ø14 Steel Two steel bars 2 Plastic strain to failure 0.117745
(steel) added on the Ø8 transverse reinforcement
tension side Material model MAT_PIECEWISE_LINEAR_PLASTICITY
RW2F 1Ø10 2Ø10 GFRP Two GFRP bars 2 Yield stress (MPa) 374
(steel) added on the Tangent modulus (MPa) 1177
tension side Plastic strain to failure 0.117435
Total no. of beams 16 Ø14 NSM steel bars
Material model MAT_PIECEWISE_LINEAR_PLASTICITY
Yield stress (MPa) 550
Tangent modulus (MPa) 1023
laminates. This is attributed to the fact that the NSM method de- Plastic strain to failure 0.076855
lays the debonding of the FRP laminate, and thus increases the load Epoxy
carrying capacity and the FRP strength utilization ratio; the ratio of Material model MAT_PIECEWISE_LINEAR_PLASTICITY
Density (kg/m3) 1650
the strain in the FRP at failure to its ultimate strain. Unlike the un-
Young’s modulus (GPa) 4.3
strengthened concrete beams, the failure of an externally bonded Poisson’s ratio 0.37
FRP-strengthened beam is generally governed by separation of Yield stress (MPa) 15
the concrete cover either at the end of the FRP laminates or be- Tangent modulus (MPa) 0
tween two intermediate cracks. These premature failures, caused Plastic strain to failure 0.496511

by debonding, often limit the effectiveness of this strengthening
technique and prevent these beams from attaining their ultimate
flexural capacities [13]. However, these debonding failures are less
likely to occur with NSM-FRP rods compared to those with exter-
nally bonded FRP sheets.
The effectiveness of strengthening RC beams with prestressed
near-surface mounted (NSM) carbon fiber reinforced polymer
(CFRP) rods was experimentally and numerically investigated by
Badawi and Soudki [9]. Four RC beams (254 mm deep by
152 mm wide by 3500 mm long) were tested under monotonic
loading. One beam was kept un-strengthened as a control beam.
One beam was strengthened with a non-prestressed CFRP NSM
rod. Two beams were strengthened with prestressed CFRP NSM
rods stressed to 40% and 60% of the rod’s ultimate strength. The
test results showed that strengthening with non-prestressed CFRP
NSM rod enhanced the flexural response of the beam compared to
that of the control specimen. A remarkable improvement in the re-
sponse was obtained when the RC beams were strengthened with
prestressed (40% and 60%) CFRP NSM rods. An increase up to 90% in
the yield load and 79% in the ultimate load compared to the control
beam was obtained. An analytical model was developed using sec-
tional analysis method to predict the flexural response of RC beams
strengthened with prestressed CFRP NSM rods. The proposed mod-
el showed excellent agreement with the experimental results.
Soliman et al. [14] investigated experimentally and numerically
the behavior of RC beams strengthened in flexure with NSM-FRP
bars. A total of twenty reinforced concrete beams (3100-mm
long  200-mm wide  300-mm deep) were constructed and tested
till failure. Different parameters including internal steel reinforce-
ment ratio, type of NSM-FRP bars, FRP bar diameter, bonded length
and groove size were investigated in this research. Test results
Ø10 NSM GFRP bars
Material model MAT_PIECEWISE_LINEAR_PLASTICITY
Density (kg/m3) 1990
Young’s modulus (GPa) 40
Poisson’s ratio 0.26
Yield stress (MPa) 743
Tangent modulus (MPa) 0
Plastic strain to failure 1.00E05
showed that the use of NSM-FRP bars is effective in increasing the
flexural capacity of concrete beams. In addition, an incremental
nonlinear displacement-controlled 3D finite element (FE) analysis
was used to numerically simulate the behavior of the test beams
with the inclusion of bond–slip relationship. Comparisons between
the FE predictions and experimental results showed very good
agreement in terms of the load–deflection and load–strain relation-
ships, ultimate capacities, and modes of failure for the tested beams.
Hawileh [15] recently presented the development of a detailed
3D nonlinear FE model that can accurately predict the load carrying
capacity and response of RC beams strengthened with NSM-FRP rods
subjected to four-point bending loading. The developed FE model
considers the nonlinear constitutive material properties of concrete,
yielding of steel reinforcement, cracking of the filler bonding mate-
rials, bond slip of the steel and NSM reinforcements with the adja-
cent concrete surfaces, and bond at the interface between the
filling materials and concrete. The numerical FE simulations were
compared with experimental measurement tested by other investi-
gators comprising of seven specimens strengthened with NSM CFRP
rods in addition to one un-strengthened control specimen. Overall,
the predicted FE mid-span deflection responses agreed very well
with the corresponding measured experimental tested data at all
148 T.H. Almusallam et al. / Construction and Building Materials 40 (2013) 145–161

stages of flexural loading. Furthermore, the developed models were
also capable of predicting the failure mode of the strengthened
tested specimen such as NSM rod debonding (peeling off) and con-
crete cover separation. The validated FE models were then used to
study the effect of different NSM material bar types and sizes.
In the present study, the effectiveness of NSM bars as a means
of maintaining or upgrading the flexural capacity of RC beams is
investigated. The studied parameters include type of NSM bars:
steel versus GFRP, and NSM reinforcement ratio (number and
diameter of inserted NSM bars). A total of eight groups of 16 beams
were tested under four-point bending till failure. The ultimate
capacity of the beams was predicted using the ACI 318-11 code
[16] and ACI 440.1R-06 guidelines [17]. A numerical investigation
utilizing nonlinear finite element (FE) analysis was also carried
out and a comparison was made between the experimental and
numerical results. Based on the validation of results, the numerical
analysis was extended to include additional cases to study the ef-
fect of GFRP reinforcement ratio on the flexural performance of
NSM-upgraded beams.
2. Experimental program
2.1. Test matrix
The test matrix was designed to investigate the viability of using NSM tech-
nique in restoring (or upgrading) the flexural load capacity of RC beams with cor-
roded steel reinforcement. The test program consisted of 16 beams divided into
eight groups consisting of two beams each. All beams were rectangular in cross-
section (150  200 mm) and 2200 mm long with an effective span of 2000 mm.

0.5P 0.5P
Strain gage(s) at
middle of main steel bar(s)

Roller
Strain gage(s) at LVDT
middle of NSM bar(s)

(a) Instrumentation layout

Steel beam

Load cell
AMSLER
testing machine Test beam

Roller

(b) Test setup

Fig. 2. Test setup and instrumentation.

Table 3
Ultimate load comparison for all beams.

Beam ID Ultimate load (Pu) (kN) % Gain in flexural capacity

EXP FEM ACI EXP/FEM EXP/ACI FEM/ACI EXP FEM ACI
Control beams
CS 40.54 38.65 31.32 1.05 1.29 1.23 – – –
CF 52.83 51.20 36.80 1.03 1.44 1.39 – – –
Beams upgraded with single NSM bar
RW1S 41.93 42.35 33.32 0.99 1.26 1.27 3.4 9.6 6.4
RW1Ø14S 57.54 59.88 51.90 0.96 1.11 1.15 41.9 54.9 65.7
RW1F 48.53 47.97 39.05 1.01 1.24 1.23 19.7 24.1 24.7
RW1Ø12F – 56.16 43.54 – – 1.29 – 45.3 39.0
RW1Ø14F – 62.61 47.95 – – 1.31 – 62.0 53.1
RW1Ø16F – 69.67 52.21 – – 1.33 – 80.3 66.7
Beams upgraded with two NSM bars
RW2S 45.75 48.66 35.31 0.94 1.30 1.38 12.9 25.9 12.7
RW2Ø14S 79.03 83.42 71.16 0.95 1.11 1.17 94.9 115.9 127.2
RW2F 51.17 56.49 43.92 0.91 1.16 1.29 26.2 46.2 40.2
RW2Ø12F – 71.58 50.33 – – 1.42 – 85.2 60.7
RW2Ø14F – 81.53 56.41 – – 1.45 – 111.0 80.1
RW2Ø16F – 90.53 62.10 – – 1.46 – 134.3 98.2
 T.H. Almusallam et al. / Construction and Building Materials 40 (2013) 145–161 149

All beam sections were designed to be under-reinforced to ensure tension failure
during testing. Two beams (the first group ‘‘CS’’) were reinforced with 3£10 mm
steel bars in the tension side and were used as control specimens. Two beams
(the second group ‘‘CF’’) were reinforced with 3£10 mm GFRP bars in the tension
side and were prepared for comparison with group ‘‘CS’’. Assuming that one of
the three main steel bars in the control specimen ‘‘CS’’ had corroded, a set of beam
groups was designed in which one NSM steel or GFRP bar was inserted in the ten-
sion side. In this regard, six beams (groups ‘‘RW1S’’, ‘‘RW1£14S’’ and ‘‘RW1F’’) were
originally reinforced with 2£10 mm steel bars in the tension side and were up-
graded with 1£10 mm NSM steel bar, 1£14 mm NSM steel bar and 1£10 mm
NSM GFRP bar in the tension side, respectively. With the assumption of the corro-
sion of two main steel bars in the control beam ‘‘CS’’, the last set of beam groups
was planned in which two NSM steel or GFRP bars were planted in the tension side.
The last six beams (groups ‘‘RW2S’’, ‘‘RW2£14S’’ and ‘‘RW2F’’) were originally rein-
forced with 1£10 mm steel bar in the tension side and were upgraded with
2£10 mm NSM steel bars, 2£14 mm NSM steel bars and 2£10 mm NSM GFRP
bars in the tension side, respectively. Stirrups used for all specimens were
£8 mm @ 100 mm spacing. Test beams were designed to have singly reinforced
sections with tension reinforcement only and hence, compression steel should
not be used; however, all beams had 1£6 mm steel bar in the compression side just
for holding the stirrups in position during concreting. A piece of polystyrene of
30  30  2200 mm was fixed at the base of the formwork to maintain the grooves
for the inserted NSM bars. After 28 days of curing, the desired additional bars were
placed in the grooves and fixed using epoxy. Detailing and dimensions of test beams
are shown in Fig. 1. Test matrix is summarized in Table 1.
580 and 629 MPa for £6 mm, £8 mm, £10 mm and £14 steel bars, respectively.
For £10 mm GFRP bars, six samples were collected and then tested as per the ACI
440.3R-04 guidelines [20]. The average tensile strength, modulus of elasticity and
strain at rupture of the six samples were 743 MPa, 40 GPa and 2%, respectively.
Mechanical properties of different materials are summarized in Table 2.
2.3. Installation of NSM bars
All specimens with one and two grooves (30  30 mm each) along the beam
were sand blasted and cleaned with acetone before the application of the repair
epoxy-bonding agent (Sikadur-31) which was applied in two layers. The second
layer was applied after insertion of the required steel or GFRP bars. After application
of epoxy, it was leveled with the adjacent concrete level and left for curing.
2.4. Test setup
The beam specimens were subjected to 4-point bending at a total span of 2.0 m
and a shear span of 0.9 m. The load was applied using a stiff steel beam connected to
2000-kN AMSLER testing machine. A load cell was mounted between the machine
and the rigid beam as shown in Fig. 2 in order to record load during the experiment.
All specimens were monotonically loaded at a displacement rate of 1 mm/min till
failure. A Linear Variable Displacement Transducer (LVDT) was affixed underneath
the mid-span of beams to measure their deflection during the test. Moreover, strain
gages were used to record strains at the level of main tension steel and NSM rein-
forcement during the experiment. The locations of sensors are shown in Fig. 2.

2.2. Material properties

Ready-mix concrete was used to cast the beam specimens. Six standard cylinders
(150  300 mm) were also prepared for compression testing at 28 days as per the
ASTM C39/39M standard [18]. The average compressive strength of the six cylindri-
cal samples of the concrete mix was 36.6 MPa. In order to determine the actual char-
acteristics of steel reinforcement, three samples of steel bars from each diameter
were tested under tension as per ASTM E8/E8M test standard [19]. The average val-
ues for yield strength are 334, 374, 408 and 550 MPa for £6 mm, £8 mm, £10 mm
and £14 steel bars, respectively. The average values for tensile strength are 451, 513,
3. Test results and discussion
Tables 3 and 4 show a summary of the flexural behavior of all
test groups in terms of: (i) flexural loading capacity, (ii) load and
mid-span deflection at yielding of main steel, (iii) mid-span deflec-
tion at ultimate, (iv) effective pre-yield stiffness, (v) deflection
ductility, (vi) energy ductility, (vii) reinforcement strains at peak
load, and (viii) failure mode. It should be noted that that the values

Table 4
Experimental with FE results with key points of load–deflection curves.a

Beam ID Results Py (kN) Dy (mm) Du (mm) Ke (kN/m) % DKe lD lE esu (le) eNSM, u (le) Failure mode
Control beams
CS EXP 31.82 6.02 47.10 5288 – 7.83 8.64 24,409 – SY–CC
FE 32.07 5.73 50.43 5596 – 8.80 9.93 26,489 – SY–CC
EXP/FE 0.99 1.05 0.93 0.95 – 0.89 0.87 0.92 –
CF EXP – – 50.68 1042 – – – 19,649 – FR–CC
FE – – 54.71 936 – – – 18,790 – FR–CC
EXP/FE – – 0.93 1.11 – – – 1.05 –
Beams upgraded with single NSM bar
RW1S EXP 30.60 5.11 37.14 5991 13 7.27 7.06 40,640 45,308 SY–CC
FE 37.96 6.40 41.54 5931 6.0 6.49 6.56 48,772 48,581 SY–CC
EXP/FE 0.81 0.80 0.89 1.01 – 1.12 1.08 0.83 0.93
RW1Ø14S EXP 43.93 6.26 39.51 7019 32.7 6.31 6.41 26,411 33,160 SY–CC
FE 48.25 6.90 40.17 6988 24.9 5.82 5.96 28,080 35,841 SY–CC
EXP/FE 0.91 0.91 0.98 1.00 – 1.09 1.08 0.94 0.93
RW1F EXP 24.82 5.12 38.75 4852 8.2 7.57 7.59 33,456 20,163 SY–FR–CC
FE 27.83 6.07 37.15 4584 18.1 6.12 6.99 32,664 18,575 SY–FR–CC
EXP/FE 0.89 0.84 1.04 1.06 – 1.24 1.09 1.02 1.09
RW1Ø12F FE 29.97 6.27 41.78 4778 14.6 6.66 7.80 33,750 18,553 SY–FR–CC
RW1Ø14F FE 31.21 6.47 44.42 4821 13.8 6.86 8.48 35,059 18,575 SY–FR–CC
RW1Ø16F FE 32.52 6.48 45.91 5022 10.3 7.09 8.86 35,621 17,478 SY–CC
Beams upgraded with two NSM bars
RW2S EXP 36.30 5.60 41.66 6483 22.6 7.44 7.95 42,375 44,166 SY–CC
FE 44.31 7.11 43.71 6227 11.3 6.14 6.16 49,275 49,954 SY–CC
EXP/FE 0.82 0.79 0.95 1.04 – 1.21 1.29 0.86 0.88
RW2Ø14S EXP 66.73 7.88 32.18 8466 60.1 4.08 4.04 25,209 33,675 SY–CC
FE 71.96 8.91 29.17 8076 44.3 3.27 3.52 26,832 28,047 SY–CC
EXP/FE 0.93 0.88 1.10 1.05 – 1.25 1.15 0.94 1.20
RW2F EXP 20.85 5.14 43.91 4052 23.4 8.53 8.84 20,756 19,300 SY–FR–CC
FE 25.54 6.29 44.23 4063 27.4 7.04 9.29 23,780 18,535 SY–FR–CC
EXP/FE 0.82 0.82 0.99 1.00 – 1.21 0.95 0.87 1.04
RW2Ø12F FE 30.04 6.67 46.23 4504 19.5 6.93 9.48 25,075 16,922 SY–CC
RW2Ø14F FE 32.28 6.70 45.39 4820 13.9 6.78 9.79 23,654 14,285 SY–CC
RW2Ø16F FE 36.12 7.08 43.05 5102 8.8 6.08 8.86 25,039 12,678 SY–CC
a
Py and Dy = load and mid-span deflection at yielding of main steel; Du = mid-span deflection at ultimate; Ke = effective pre-yield stiffness; % DKe = percent gain in effective
pre-yield stiffness over the control group CS; lD = deflection ductility ratio = Du/Dy; lE = energy ductility index; esu = main steel (or GFRP) strain at peak load; eNSM, u = steel or
GFRP NSM strain at peak load; SY = steel yielding; CC = concrete crushing; FR = FRP rupture.
150 T.H. Almusallam et al. / Construction and Building Materials 40 (2013) 145–161
enlisted in Tables 3 and 4 are the average of the two test beams in
each group. It should be also revealed that the energy ductility in-
dex (lE) shown in Table 4 is calculated as per Emadi and Hashemi
[21] from the following formula:
 
1 Eu
lE ¼ þ1
2 Ey
where Eu is the energy of the beam at ultimate state (area under
load–deflection curve up to ultimate deflection) and Ey is the energy
of the beam at first yield of main steel (area under load–deflection
curve up to yield displacement). Both peak load and effective pre-
yield stiffness of each group of upgraded beams are compared with
the control group CS, and the corresponding percent gains are given
in Tables 3 and 4, respectively. Load versus mid-span deflection
curves are presented as shown in Fig. 3 for all test groups. The
curves shown in Fig. 3 are for one test beam of each group. In addi-
tion, final modes of failure are illustrated in Fig. 4 for representative
samples of test beams.
As demonstrated in Fig. 3, the control group CS in addition to
steel-NSM-upgraded groups (RW1S, RW1£14S, RW2S and
RW2£14S) displayed the standard nearly-bilinear response charac-
teristics of under-reinforced beams. They failed in flexure through
(a) Beams upgraded with single NSM bar
(b) Beams upgraded with two NSM bars
Fig. 3. Load–deflection curves for test beams.
the formation of wide flexural crack at the mid-span with the final
mode of failure being concrete crushing at the critical section as pre-
sented in Fig. 4a, c and d. Tables 3 and 4 identify that insertion of
steel NSM bars on the tension side was successful at maintaining
the original flexural capacity and stiffness of the beams. Having
ð1Þ the same reinforcement ratio as the control group CS, groups
RW1S and RW2S achieved 3.4% and 12.9% (Table 3) increase in the
flexural capacity. This may be attributed to the increase in the effec-
tive depth for planted NSM bars. In addition, groups RW1S and
RW2S had 13% and 22.6% (Table 4) gain in their effective stiffness,
respectively. Deflection and energy ductility ratios of the two
groups RW1S and RW2S were very close to the control group CS.
Groups RW1£14S and RW2£14S (with 32% and 64%, respectively,
increase in the total steel ratio over the control group CS) attained
41.9% and 94.9% gain in their flexural capacity, respectively. The
two groups RW1£14S and RW2£14S had 32.7% and 60.1% gain
in their effective stiffness, respectively. Yet, due to the use of higher
reinforcement ratios for groups RW1£14S and RW2£14S, their
deflection and energy ductility ratios were significantly reduced.
The control group CF and the two GFRP-NSM-upgraded groups
RW1F and RW2F failed by FRP rupture in the constant moment re-
gion followed by concrete crushing as demonstrated in Fig. 4b for
group RW1F. This type of failure may be attributed to the sufficient
end anchorage of the GFRP NSM bars in addition to the good bond-
ing of the epoxy adhesive used in this study. Even though group CF
had higher flexural capacity than group CS, its effective stiffness
was about 20% of that of group CS. This may be attributed to the
low tensile modulus of the GFRP bars which was about 20% of that
of the reinforcing steel. The two groups RW1F and RW2F were suc-
cessful at maintaining the original flexural capacity and ductility of
the control group CS. The gain in flexural capacity was about 19.7%
and 26.2% for groups RW1F and RW2F, respectively. However, the
effective stiffness was reduced by about 8.2% and 23.4% for groups
RW1F and RW2F, respectively. This indicates that the insertion of
GFRP bars with enough end anchorage could be used as a success-
ful technique at restoring the flexural load capacity of RC beams
with corroded steel reinforcement. However, in case FRP bars with
low modulus are used, special attention should be given to the re-
duced stiffness that may result in increased beam deflection at ser-
vice load level.
4. Calculation of beam capacity using ACI 318-11 and ACI
440.1R-06
Since slippage of NSM bars was not observed throughout the
testing, computation of the strength of the beam cross-section
was based on perfect bond assumption between bars and sur-
rounding concrete. For beams strengthened with steel NSM bars,
the ACI 318-11 code [16] was followed in the calculation of their
flexural capacity. Yet, for beams upgraded with GFRP NSM bars,
flexural capacity calculations were in line with both the ACI 318-
11 code [16] and ACI 440.1R-06 guidelines [17]. The flexural capac-
ity is dependent on whether failure is governed by concrete crush-
ing or GFRP rupture. The failure mode is determined by comparing
the GFRP NSM reinforcement ratio (qNSM) to the balanced rein-
forcement ratio of the GFRP NSM bars calculated from
 0     
fc Ef ecu d fy
qb;NSM ¼ 0:85b1  qs ð2Þ
ffu Ef ecu þ ffu dNSM ffu
where b1 is the ratio of depth of equivalent rectangular compressive
stress block to neutral axis depth as given by the ACI 318-11; fc0 is the
specified compressive strength of concrete; ffu is the tensile strength
of GFRP reinforcement; Ef is the tensile modulus of GFRP reinforce-
ment; ecu is the ultimate concrete strain taken as 0.003; qs1 is the ra-
tio of main steel reinforcement; d is the distance from extreme
 T.H. Almusallam et al. / Construction and Building Materials 40 (2013) 145–161 151

compression fiber to the centroid of main tension steel bars; dNSM is When qNSM > qb,NSM, the failure of the beam is initiated by
the distance from extreme compression fiber to centroid of GFRP crushing of the concrete and the neutral axis depth is calculated
NSM bars; and fy is the yield strength of main steel reinforcement. from

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2
ðqNSM bdNSM Ef ecu  qs bdfy Þ2 þ 3:4b1 fc qNSM b dNSM Ef ecu
=
qs bdfy  qNSM bdNSM Ef ecu þ
c¼ ð3Þ
1:7b1 fc= b

Fig. 4. Mode of failure for representative samples of test beams.

152 T.H. Almusallam et al. / Construction and Building Materials 40 (2013) 145–161

where b = width of beam section. The tensile strain in the GFRP procedure were followed to estimate the beam flexural capacity
NSM bars is then given by and the results are as given in Table 3. For control beam CS, the
  ACI 318-11 code was conservative by about 29%; however, the
dNSM  c
eNSM ¼ ecu ð4Þ ACI 440.1R-06 guidelines were conservative by about 44% in pre-
c dicting the flexural capacity of control beam CF. The ACI 318-11
The moment capacity of the beam section is then estimated from code underestimated the peak load for steel-NSM-upgraded beams
  by 11% to 30% as evidenced from Table 3. Yet, for GFRP-NSM-up-
b c b c graded beams, the above detailed procedure was conservative in
M u ¼ qs bdfy d  1 þ qNSM bdNSM Ef eNSM ðdNSM  1 Þ ð5Þ
2 2 estimating the flexural capacity by about 16–24%. It should be
mentioned that contrary to the experimental observations, FRP
When qNSM < qb,NSM, the failure of the strengthened beam is initi-
rupture was not anticipated for beams CF, RW1F and RW2F. The
ated by rupture of GFRP bars, and the ACI stress block is not appli-
calculated FRP strains at ultimate condition were 12,150, 15,200
cable because the maximum concrete strain (ecu) may not be
and 14,000 le for the three beams, respectively. These values are
attained. The analysis incorporates several unknowns: the concrete
65%, 82% and 75% of the theoretical FRP rupture strain. Conse-
compressive strain at failure, the depth to the neutral axis and the
quently, a more accurate prediction procedure that matches the
rectangular stress block factors. The analysis involving all these un-
test results is highly needed. In this regard, 3-D nonlinear FE mod-
knowns becomes very complex. Yet, similar to the ACI 440.1R-06, a
eling was utilized as a numerical simulation tool.
simplified and conservative calculation of the flexural strength of
the beam section can be estimated from
    5. Finite element modeling
b cb b cb
M u ¼ qs bdfy d  1 þ qNSM bdNSM ffu dNSM  1 ð6Þ
2 2
LS-DYNA [22], a general-purpose finite element program, was
where employed for the numerical simulation of the test beams. The 3-
  D finite element model was developed using a general-purpose
ecu pre-processor FEMB. Only half of the beam was modeled account-
cb ¼ dNSM ð7Þ
ecu þ efu ing for its symmetry.
and efu = rupture strain of GFRP reinforcement.
Since we are not dealing with a design-related problem and the 5.1. Model geometry
goal is to predict the flexural capacity of test beams, a strength
reduction factor of one (u = 1) was utilized. The ACI 318-11 code, In order to model the real behavior of tested RC beams, it is
the ACI 440.1R-06 guidelines and the above-mentioned detailed imperative that the concrete volume be modeled using solid

Fig. 5. Finite element mesh in FEMB software showing the modeled components.
 T.H. Almusallam et al. / Construction and Building Materials 40 (2013) 145–161 153

elements. For this reason, 8-node reduced integration solid hexa-
hedron elements were used to model both concrete and epoxy
adhesive. These elements have three degrees of freedom at each
node. Single point volume integration is carried out by Gaussian
quadrature. The biggest advantage of using solid elements with
one-point integration is the substantial savings in computer time;
yet, the biggest disadvantage is the need to control the zero energy
modes, which arise, called hourglassing modes. Undesirable hour-
glass modes tend to have periods that are typically much shorter
than the periods of the structural response, and they are often ob-
served to be oscillatory. One way of resisting undesirable hourglas-
sing is with a viscous damping or small elastic stiffness capable of
stopping the formation of the anomalous modes but having a neg-
ligible effect on the stable global modes. In LS-DYNA software,
three-dimensional algorithms for controlling the hourglass modes
fs
fsu
fy
are typically used. The longitudinal steel and GFRP bars and the
transverse ties were modeled using 2-node Hughes–Liu beam ele-
ments. The Hughes–Liu beam element formulation is based on the
shell elements developed by Hughes and Liu [23,24]. The Hughes–
Liu beam element is compatible with the solid elements, because
the element is based on a degenerated solid element formulation,
and hence, it was used in this study. Even though shear deforma-
tion is included in the Hughes–Liu beam element formulation,
shear deformations in the steel (or GFRP) bars are usually negligi-
ble and will not affect the global behavior of the concrete beams.
The 8-node solid elements were 15  15  20 mm in size.
Numerical convergence study showed that further decrease in
the mesh size has little effect on the numerical results but leads
to the risk of computer memory overflow and substantially in-
creases the computing time. Fig. 5 shows the typical mesh of an
NSM-upgraded beam, which consists of 7480 solid concrete ele-
ments, 220 solid epoxy elements, 576 beam elements for the rein-
forcing bars to give a total of 8276 elements. The FE analysis was
based on perfect bond assumption between steel bars and sur-
rounding concrete, between NSM bars and epoxy adhesive and in
between epoxy adhesive and concrete volume.
5.2. Material modeling

The material model type 159, MAT_CSCM_CONCRETE was em-

ployed to model the concrete volume. This is a smooth or continu-
Es ous cap model available in LS-DYNA for solid elements, with a
1 smooth intersection between the shear yield surface and the hard-
ening cap. In this model, the initial damage surface coincides with
εy εsu εs the yield surface. Rate effects are modeled with viscoplasticity.
(a) Concrete cracking is considered using the traditional smeared
crack approach. More details of this material model can be found
in Refs. [25,26].
ff The material model type 24, MAT_PIECEWISE_ LINEAR_PLAS-
TICITY was utilized to model steel bars, GFRP bars and epoxy adhe-
ffu sive. This material is suited to model elasto-plastic materials with
an arbitrary stress versus strain curve and an arbitrary strain rate
dependency. The schematic multi-linear stress–strain curves as-
signed to the steel bars, GFRP bars and epoxy adhesive are pre-
sented in Fig. 6. A summary of the material properties used in
the FE analysis is presented in Table 2. It should be noted that since
the GFRP bars are linear elastic up to failure, a very small value of
1.00E05 was assigned for the plastic strain to failure as identified
Ef
in Table 2.
1
εf
εfu 5.3. Erosion

(b) 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 phe-
fa nomena and produce graphical plots which are more realistic rep-
resentations of the actual events. By activating this feature, the
fau eroded solid element is physically separated from the rest of the
mesh. Material failure was simulated by element erosion at a spe-
cific plastic strain; thus, whenever an element reaches this critical
value, it is removed from the computation. This erosion model rep-
resents a numerical remedy to distortion, which can cause exces-
sive and unrealistic deformation of the mesh. In this study,
elements of concrete and epoxy adhesive were allowed to erode
Ea when the maximum principal strain reached 0.05 [26].

1 5.4. Boundary conditions

εau εa
(c) Only half of the beam was modeled in LS-DYNA taking into ac-
count the symmetry of the beam specimens. A node set was
Fig. 6. Stress–strain curves: (a) steel bars; (b) GFRP bars and (c) epoxy adhesive. created which consisted of nodes at support location of the beam
154 T.H. Almusallam et al. / Construction and Building Materials 40 (2013) 145–161

which had to be restricted for the displacement in the global Z
direction thus representing a roller support near the beam end.
Symmetric boundary conditions were applied for the nodes in
elements for the plane representing the continuation of the beam
in reality. This included restriction of displacement in the global
X direction and the rotation about the global Y and Z directions
for those nodes. Since the loading was displacement controlled,
another node set was created which comprised of nodes
along the loading plane which were controlled to have the same
Z-displacement throughout the test.

(a) Beam CS

(b) Beam RW1F

(c) Beam RW2S

(d) Beam RW1∅14S

Fig. 7. FE mode of failure for representative samples of test beams.
 T.H. Almusallam et al. / Construction and Building Materials 40 (2013) 145–161
5.5. Loading strategy
LS-DYNA uses explicit time integration algorithms for solving
the problems, which are less sensitive to machine precision than
other finite element solution methods. The load application pro-
cess in LS-DYNA is time-history dependent. Since the testing pro-
cedure involved displacement controlled static loading, the
inertia effects were removed from the dynamic equation by assign-
ing a constant velocity to the displacement controlled node set.
This will lead to zero acceleration and hence zero inertia force.
The rate of change of displacement was defined as 1 mm/min to
match with the experimental loading.
6. Finite element results and discussion
6.1. Modes of failure
Fig. 7 depicts the modes of failure for representative samples of
test beams as observed from the FE analysis post-processing soft-
ware. The failure modes are based on contours of mid-surface max-
imum principal strains. It is noted that the failure modes predicted
(a) Beam CS
(b) Beam CF
Fig. 8. Load–deflection comparison for control beams.
155
from the FE analysis matches very well with the experimental
observations. From the analysis it was found that, like the experi-
mental observations, the control beam CS in addition to steel-
NSM-upgraded beams (RW1S, RW1Ø14S, RW2S and RW2Ø14S)
failed due to concrete crushing after the formation of wide flexural
cracks in the maximum moment region. However, similar to the
experimental testing, the FE analysis revealed that the control
beam CF and the two GFRP-NSM-upgraded beams RW1F and
RW2F failed by FRP rupture in the constant moment region fol-
lowed by concrete crushing as demonstrated in Fig. 7b for beam
RW1F.
6.2. Load–deflection curves
A comparison was made between the load–deflection curves
obtained from the experimental and the FE studies for all the beam
specimens. Figs. 8 and 9 depict this comparison. As seen from the
figures, the experimental load–deflection curves showed good
agreement especially for the ultimate load carrying capacity, com-
pared with the FE analysis of the control beams as well as NSM-
upgraded beams. Tables 3 and 4 enlist the comparison details. As
seen in Table 3, maximum deviation of 9% was found for the
numerical results for ultimate load. Yet, as evidenced from Table
4, compared with the experimental results, maximum deviation
of 11% was observed for mid-span deflection at ultimate load. Both
the stiffness and ductility ratios were also predicted efficiently by
the FE models in comparison with the experimental results. Figs.
8 and 9 also show that the FE models were successful at imitating
the softening behavior which demonstrates the accuracy and
applicability of the material model.
The FE analysis evidenced that due to the use of higher rein-
forcement ratios for beams RW1Ø14S and RW2Ø14S, their deflec-
tion and energy ductility ratios were significantly reduced
compared with the control beam CS. The FE analysis also supported
the conclusion that use of steel or GFRP NSM bars with enough end
anchorage could be a successful technique at restoring the flexural
load capacity of RC beams with corroded steel reinforcement. Nev-
ertheless, in case FRP NSM bars with low modulus are used, the
effective stiffness of the beam will be reduced that may bring about
increased deflection at service load level.
6.3. Strain gage results
The maximum tensile strain in the main reinforcement obtained
from the post-processing software of LS-DYNA for the tested beam
specimens was compared with the experimental strains obtained
using the strain gages. Figs. 10 and 11 depict this comparison for
control and NSM-upgraded beams, respectively. Table 4 enlists
measured and predicted main steel (or GFRP for beam CF) strains
for test specimens. Good agreement was achieved between the
experimental and predicted main steel strains with maximum devi-
ation of 17%. It is clear that all test beams (except for CF) had a high
ductile behavior with a strain ductility (ratio of main steel strain at
ultimate load to steel yield strain) of more than 10, which is typical
for such under-reinforced beams failing in flexure. Having the same
reinforcement ratio as the control beam CS, beams RW1S and RW2S
achieved strain ductility increase of 66% and 74%, respectively.
However, for beams RW1Ø14S and RW2Ø14S with higher rein-
forcement ratios, the strain ductility increase was only limited to
1.1% and 1.0%, respectively. For GFRP-NSM-upgraded beams
RW1F and RW2F, the achieved strain ductility was low compared
with their steel NSM counterparts RW1S and RW2S.
In addition to main steel, strain gages were attached to steel
and GFRP NSM bars to record their strain throughout the test.
Fig. 12 and Table 4 show the NSM strain comparison for up-
graded
156 T.H. Almusallam et al. / Construction and Building Materials 40 (2013) 145–161

(a) Beam RW1S (b) Beam RW1F

(c) Beam RW2S (d) Beam RW2F

(e) Beam RW1∅14S (f) Beam RW2∅14S

Fig. 9. Load–deflection comparison for NSM-upgraded beams.

beams where good agreement was accomplished between the
experimental and predicted values. From Table 4, it is clear that
the two GFRP-NSM-upgraded beams RW1F and RW2F failed due
to GFRP fracture as both measured and predicted NSM strains at
peak load were very close to the values at rupture (20,000 le as
measured experimentally and 18,575 le as assumed in the FE
analysis).
7. Effect of FRP reinforcement ratio
The FE analysis was extended to study the effect of FRP rein-
forcement ratio on the flexural performance of NSM-upgraded
beams. In addition to the two FRP NSM-upgraded beams used for
FE validation (RW1F and RW2F), another six specimens were
numerically investigated. Table 5 enlists the details of the
 T.H. Almusallam et al. / Construction and Building Materials 40 (2013) 145–161 157

additional six beams considered for the numerical study. It should qNSM =qb;NSM
ar ¼ ð8Þ
be noted that the six specimens RW1Ø12F to RW2Ø16F have the qs =qsb
same dimensions and material properties as those used for beams
RW1F and RW2F. It should be also noted that in the analysis of the where qNSM = ratio of NSM reinforcement, qb,NSM = balanced rein-
six additional beams, perfect bond was assumed between NSM forcement ratio of the NSM bars calculated from Eq. (2), qs = ratio
bars and epoxy adhesive and in between epoxy adhesive and con- of main steel reinforcement and qsb = balanced reinforcement ratio
crete volume. As per the manufacturer’s datasheet, the epoxy of the main steel bars estimated from
! 
adhesive employed in this study has bond strength to concrete of fc= ecu
about 3.5 MPa. Also, the GFRP NSM bars have enough end anchor- qsb ¼ 0:85b1 ð9Þ
fy ecu þ ey
age beyond the section of maximum moment (bars extended to the
end of the beam). Hence, the calculated bond stresses along the bar where ey = yield strain of main steel bars. The relationship between
perimeter (even for Ø16 mm bars) will be significantly less than the FRP reinforcement coefficient (ar) and the response parameters
3.5 MPa. Consequently, the perfect bond assumption is valid. Sum- is shown in Fig. 13a and b for peak flexural capacity and NSM
mary of FE analysis results for the additional six beams is listed in strain at ultimate load level, respectively. It is evident that as the
Tables 3 and 4. It is evident that for the same main steel, as the FRP reinforcement coefficient increases the flexural capacity of
NSM reinforcement ratio increases both the flexural capacity and the upgraded beam increases. As demonstrated from Fig. 13b, rup-
effective stiffness increase. ture of FRP bars is predicted up to a reinforcement coefficient of
In order to study the effect of FRP reinforcement, a new param- about 20 after which NSM strain decreased indicating that the
eter called ‘‘FRP reinforcement coefficient (ar)’’ was first introduced beams would fail by steel yielding followed by concrete crushing.
in this study. This coefficient is given from the following formula: Consequently, in order to eliminate the unfavorable brittle FRP
fracture in NSM-upgraded beams, it is recommended for practicing
engineers not to design an NSM strengthening scheme with FRP
reinforcement coefficient (ar) less than 30, which corresponds to
a strain level of approximately 90% of the FRP rupture strain as
seen in Fig. 13b. However, in order to guarantee for main steel
yielding and avoid the other extreme brittle concrete compression
failure, the FRP reinforcement coefficient (ar) should be limited.
Using strain compatibility analysis, the maximum value of ar is
estimated as 50% of that corresponding to balanced strain condi-
tion between extreme concrete compressive fiber and main steel.
In conclusion, the recommended limits of the FRP reinforcement
coefficient are:
2   3 !
0:425b1 fc= ecuecuþey d  0:5qs dfy qsb
30 6 ar 6 4 h    i 5 ð10Þ
e þe
Ef ecu cuecu y dNSM  1 dNSM qs  qb;NSM
d

In the above equation, the lower limit was set to avoid FRP rupture
with a 10% safety margin as demonstrated in Fig. 13b. Yet, the upper
limit corresponds to the condition of having the concrete reaching
its ultimate compression strain (ecu) synchronized with yield strain
of the main steel (ey) with a 50% safety margin. Derivation of the
(a) Beam CS upper limit of Eq. (10) is given in Appendix A. For GFRP NSM-
upgraded beams used in this study, the upper limits of the FRP
reinforcement coefficient (ar) were estimated to be 155 and 169
for beams with 2Ø10 and 1Ø10 main steel bars, respectively. The
lower and upper limits of ar are presented in Fig. 13b.

8. Conclusions

On the basis of experimental and numerical results presented in

this paper the following conclusions can be drawn:

1. The bond behavior of all planted steel and GFRP bars was excel-
lent, in which no debonding or bond failure was observed in any
of the tested beams. This is attributed to the sufficient end
anchorage of the NSM bars in addition to the good bonding of
the epoxy adhesive used in this study.
2. The use of steel NSM bars with enough end anchorage can be
considered as a successful technique at restoring the flexural
load capacity of RC beams with corroded steel reinforcement.
As the NSM reinforcement ratio increases, both the flexural
capacity and effective pre-yield stiffness of the beam increase.
(b) Beam CF
However, both deflection and energy ductility ratios get signif-
Fig. 10. Load versus tensile steel (or FRP) strain comparison for control beams. icantly reduced with higher ratios of NSM reinforcement. In all
158 T.H. Almusallam et al. / Construction and Building Materials 40 (2013) 145–161

(a) Beam RW1S (b) Beam RW1F

(c) Beam RW2S (d) Beam RW2F

(e) Beam RW1∅14S (f) Beam RW2∅14S

Fig. 11. Load versus main tensile steel strain comparison for NSM-upgraded beams.

Table 5
Additional beams considered for the numerical study.

Beam ID Main tension bars NSM bars inserted Type of NSM bars Description
RW1Ø12F 2Ø10 (steel) 1Ø12 GFRP One GFRP bar added on the tension side
RW1Ø14F 2Ø10 (steel) 1Ø14 GFRP One GFRP bar added on the tension side
RW1Ø16F 2Ø10 (steel) 1Ø16 GFRP One GFRP bar added on the tension side
RW2Ø12F 1Ø10 (steel) 2Ø12 GFRP Two GFRP bars added on the tension side
RW2Ø14F 1Ø10 (steel) 2Ø14 GFRP Two GFRP bars added on the tension side
RW2Ø16F 1Ø10 (steel) 2Ø16 GFRP Two GFRP bars added on the tension side
 T.H. Almusallam et al. / Construction and Building Materials 40 (2013) 145–161 159

(a) Beam RW1S (b) Beam RW1F

(c) Beam RW2S (d) Beam RW2F

(e) Beam RW1∅14S (f) Beam RW2∅14S

Fig. 12. Load versus NSM strain comparison for upgraded beams.

cases, the total tension steel ratio of the beam section should
not exceed the maximum limits given by codes of practice in
order to alleviate the unwanted sudden compression failure.
3. FRP NSM bars with enough end anchorage can be used efficiently to
maintain the flexural load capacity of RC beams with corroded steel
reinforcement. Nevertheless, in case FRP NSM bars with low mod-
ulus are used, the effective stiffness of the beam will be reduced
that may bring about increased deflection at service load level.
4. Both the ACI 318-11 code and ACI 440.1R-06 guidelines with
strength reduction factor of one (u = 1) are conservative in esti-
mating the flexural capacity of NSM-upgraded beams. The ACI
318-11 code underestimated the peak load for steel-NSM-
upgraded beams by 11–30%. Yet, for GFRP-NSM-upgraded
beams, procedure based on both ACI 318-11 and ACI 440.1R-
06 was conservative in estimating the flexural capacity by
about 16–24%.
160 T.H. Almusallam et al. / Construction and Building Materials 40 (2013) 145–161

ratio) to main steel ratio (with respect to its balanced ratio).

Based on the FE analysis, it was found out that rupture of FRP
bars is predicted up to a reinforcement coefficient of about 20
after which NSM strain decreased. Accordingly, in order to
avoid the unfavorable brittle FRP fracture in NSM-upgraded
beams, it is recommended for practicing engineers not to design
an NSM strengthening scheme with FRP reinforcement coeffi-
cient (ar) less than 30, which corresponds to a strain level of
approximately 90% of the FRP rupture strain. Yet, in order to
ensure yielding of main steel and avoid the sudden concrete
compression failure, a maximum value is proposed for ar as
given in Eq. (10).
6. It is worth mentioning that some drawbacks may arise when
using GFRP NSM bars in strengthening of RC beams. Special
attention should be given to the issue of creep rupture which
is mostly associated with GFRP bars. This may reduce the
maximum permissible stresses resisted by the GFRP bars. In
addition, GFRP bars may degrade in the alkaline ambient
provided by the surrounding concrete, especially in the case
of beams with GFRP bars used as original main
reinforcement.

Acknowledgments

The Authors express their appreciation to the Deanship of Sci-

entific Research at King Saud University for funding the work
through the research group Project No. RGP-VPP-104. Thanks are
also extended to the MMB Chair for Research and Studies in
Strengthening and Rehabilitation of Structures, at the Department
of Civil Engineering, King Saud University for providing technical
support.

Appendix A. Derivation of the upper limit of Eq. (10)

Fig. A.1 illustrates section analysis of FRP NSM-strengthened RC

beam at balanced strain condition between extreme concrete com-
pressive fiber and main steel. The concrete compression force is gi-
ven by

C c ¼ 0:85fc= babs ¼ 0:85fc= bb1 cbs ðA:1Þ

where cbs = neutral axis depth corresponding to balanced condition
between concrete and main steel, and is estimated from
Fig. 13. Relationship between FRP reinforcement coefficient (ar) and response  
ecu
parameters for FRP-NSM-upgraded beams (based on FE analysis). cbs ¼ d ðA:2Þ
ecu þ ey
5. A new parameter called ‘‘FRP reinforcement coefficient (ar)’’ The tensile force in main steel bars is given by
was first introduced in this study. This coefficient is the propor-
T s ¼ As fy ¼ qs bdfy ðA:3Þ
tion of NSM reinforcement ratio (with respect to its balanced

Fig. A.1. Strain and stress distribution at balanced condition between concrete and main steel.
 T.H. Almusallam et al. / Construction and Building Materials 40 (2013) 145–161 161

The tensile force in FRP NSM bars is calculated from [6] Soliman SM, El-Salakawy E, Benmokrane B. Flexural behaviour of concrete
beams strengthened with near surface mounted FRP bars. In: Proc. on CD-
T NSM ¼ ANSM eNSM Ef ¼ qNSM bdNSM eNSM Ef ðA:4Þ ROM, 4th int. conf. on FRP composites in civil engineering (CICE2008), Zurich,
Switzerland; 22–24 July, 2008.
where eNSM = strain at level of NSM bars, and is estimated from [7] Nordin H, Täljsten B. Concrete beams strengthened with prestressed near
      surface mounted CFRP. ASCE J Compos Constr 2006;10(1):60–8.
dNSM ecu þ ey dNSM [8] El-Hacha R, Gaafar M. Flexural strengthening of reinforced concrete beams
eNSM ¼ ecu  1 ¼ ecu 1 ðA:5Þ using prestressed near-surface mounted CFRP bars. PCI J Fall 2011:134–51.
cbs ecu d
[9] Badawi M, Soudki K. Flexural strengthening of RC beams with prestressed NSM
CFRP rods – experimental and analytical investigation. Constr Build Mater
From the equilibrium of internal forces, we get
Elsevier Publ 2009;23(10):3292–300.
C c ¼ T s þ T NSM ðA:6Þ [10] Asplund S. Strengthening bridge slabs with grouted reinforcement. ACI Struct J
Am Concr Inst 1949;20(6):397–406.
      [11] Blaschko M, Zilch K. Rehabilitation of concrete structures with strips glued
ecu ecu þ ey dNSM into slits. In: Proc. of 12th int. conf. on composite materials, Paris, CD-ROM;
or; 0:85fc= bb1 d ¼ qs bdfy þ qNSM bdNSM ecu  1 Ef
ecu þ ey ecu d 1999.
[12] El-Hacha R, Rizkalla S. Near surface mounted fiber reinforced polymer
ðA:7Þ
reinforcements for flexural strengthening of concrete structures. ACI Struct J
Am Concr Inst 2004;101(5):716–7.
From which we can get
[13] De Lorenzis L, Teng JG. Near-surface mounted FRP reinforcement: an emerging
  technique for strengthening structures. Compos Part B: Eng Elsevier Publ
0:85b1 fc ecuecuþey d  qs dfy
=
2007;38:119–43.
qNSM ¼ h
e þe
  i ðA:8Þ [14] Soliman SM, El-Salakawy E, Benmokrane B. Flexural behaviour of concrete
Ef ecu cuecu y dNSM d
 1 dNSM beams strengthened with near surface mounted fibre reinforced polymer bars.
Can J Civil Eng 2010;37(10):1371–82.
From Eq. (8), the FRP reinforcement coefficient (ar) is given by [15] Hawileh RA. Nonlinear finite element modeling of RC beams strengthened
  with NSM FRP rods. Constr Build Mater Elsevier Publ 2012;27:461–71.
! [16] ACI Committee 318. Building code requirements for structural concrete and
0:85b1 fc= ecuecuþey d  qs dfy qsb commentary. ACI 318-11. American Concrete Institute, Detroit, MI, USA; 2011.
ar ¼ h   i ðA:9Þ [17] ACI Committee 440. Guide for the design and construction of structural
e þe
Ef ecu cuecu y dNSM  1 dNSM qs  qb;NSM concrete reinforced with FRP bars. ACI 440.1R-06, American Concrete Institute,
d
Detroit, MI, USA; 2006.
Assuming a 50% safety margin, the upper limit of ar is then given by [18] ASTM. Standard test method for compressive strength of cylindrical concrete
  specimens. ASTM C39/C39M. American Society for Testing and Materials, West
2 3 ! Conshohocken, PA, USA; 2010. http://dx.doi.org/10.1520/C0039_C0039M-10.
0:425b1 fc= ecuecuþey d  0:5qs dfy qsb [19] ASTM. Standard test methods for tension testing of metallic materials. ASTM
ar;max ¼4 h   i 5 : ðA:10Þ
e þe
Ef ecu cuecu y dNSM  1 dNSM qs  qb;NSM E8/E8M. American Society for Testing and Materials, West Conshohocken, PA,
d USA; 2009. http://dx.doi.org/10.1520/E0008_E0008M-09.
[20] ACI Committee 440. Guide test methods for fiber-reinforced polymers (FRPs)
for reinforcing or strengthening concrete structures. ACI 440.3R-04. American
References Concrete Institute, Detroit, MI, USA; 2004.
[21] Emadi J, Hashemi SH. Flexural study of high strength RC beams strengthened
with CFRP plates. World Acad Sci Eng Technol 2011;78:380–4.
[1] Cruz JMS, Barros JAO. Bond between near surface mounted carbon fiber
[22] Livermore Software Technology Corporation (LSTC). LS-DYNA user’s keyword
reinforced polymer laminate strips and concrete. ASCE J Compos Constr
manual (nonlinear dynamic analysis of structures in three dimensions), vol. 1.
2006;8(6):519–27.
Version 971, LSTC, Livermore, California; 2007.
[2] Soliman SM, El-Salakawy E, Benmokrane B. Bond properties of near surface
[23] Hughes TJR, Liu WK. Nonlinear finite element analysis of shells: Part I. Two-
mounted (NSM) carbon FRP bars in concrete. In: Proc. on CD-ROM, annual
dimensional shells. Comput Methods Appl Mech 1981;27:167–81.
general meeting & conference Canadian society of civil engineering, June 6–9,
[24] Hughes TJR, Liu WK. Nonlinear finite element analysis of shells: Part II. Three-
2007, Yellowknife, Northwest Territories, Canada.
dimensional shells. Comput Methods Appl Mech 1981;27:331–62.
[3] Nanni A, Ludovico M, Parretti R. Shear strengthening of a PC bridge girder with
[25] Murray YD. User’s manual for LS-DYNA concrete material model 159. Report
NSM CFRP rectangular bars. Adv Struct Eng 2004;7(4):97–109.
No. FHWA-HRT-05-062. US Department of Transportation, Federal Highway
[4] Hassan T, Rizkalla S. Bond mechanism of NSM FRP bars for flexural
Administration National Transportation Systems Center, USA; 2007.
strengthening of concrete structures. ACI Struct J 2004;101(6):830–9.
[26] Murray YD, Abu-Odeh A, Bligh R. Evaluation of concrete material model 159.
[5] Kotynia R. Analysis of the flexural response of NSM FRP-strengthened concrete
Report No. FHWA-HRT-05-063. US Department of Transportation, Federal
beams. In: Proc. of the 8th int. symp. of FRP reinforcement for concrete
Highway Administration National Transportation Systems Center, USA; 2007.
structures (FRPRCS-8). Greece: University of Patras; 2007.