Engineering Structures 32 (2010) 1769–1778

Contents lists available at ScienceDirect

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

Effects of bolt–plate arrangements on steel plate strengthened reinforced

concrete beams
R.K.L. Su ∗,1 , W.H. Siu 2 , S.T. Smith 3
Department of Civil Engineering, The University of Hong Kong, Pokfulam Road, Hong Kong, China

article info abstract

Article history: A structure that can behave in a ductile fashion under extreme events is of paramount importance due
Received 23 April 2009 to safety considerations. Because of such importance, ductility performance of bolted side-plated (BSP)
Received in revised form concrete beams under different bolt–plate arrangements is investigated experimentally through four
26 February 2010
large-scale tests on BSP beams. It is observed that bolt–plate arrangements have a dominant effect on
Accepted 27 February 2010
Available online 30 March 2010
the ductility performance of beams in terms of both the post-elastic strength enhancement (defined as
the ratio between the peak strength and the yield strength of the beam), and the displacement ductility
Keywords:
(defined as the ratio between displacements at the ultimate and notional yield stages). In order to ensure
Bolts ductile beam behaviour, the concept of maximum allowable plate–force demand is introduced, of which
Bi-linear the strength of additional plates should be kept below the ‘balanced’ failure point, while sufficient shear
Concrete beams strength of the bolt connections should be provided so that the strengthened beam will possess both
Curvature sufficient strength enhancement and ductility.
Ductility © 2010 Elsevier Ltd. All rights reserved.
Plate
Strengthening

1. Introduction allowable plate–force demand in the bolt–plate system has not

Due to deterioration of materials and demand for additional
strength, retrofitting of existing beam structures is often needed.
There are basically two methods for strengthening or stiffening ex-
isted reinforced concrete (RC) beams. The first method is to attach
advanced composites, such as carbon fibre reinforced polymers or
glass fibre reinforced polymers [1], onto the tension surface or the
side faces of the members. These composites are generally capable
of increasing the ultimate load resistance but are prone to peeling
and edge debonding failure [2]. The second method for strength-
ening or stiffening existed RC beams is to attach steel plates to
the external surfaces of the structural components by means of
bolting. This method can enhance both the flexural and shear per-
formances of RC beams. A number of studies were carried out to
investigate the flexural and shear behaviour of bolted side-plated
(BSP) beams [3–8] and coupling beams [9] as well as the be-
haviour of the connecting bolt groups [10,11]. However, as the
∗ Corresponding author. Tel.: +852 2859 2648; fax: +852 2559 5337.
E-mail address: klsu@hkucc.hku.hk (R.K.L. Su).
1 Associate Professor.
2 PhD graduate.
3 Assistant Professor.
been specified in these studies, the designed BSP beams may be
over-reinforced when thick plates and strong bolts are used. It
is well recognized that ductility is an important structural prop-
erty which provides noticeable warning at the beginning of fail-
ure of the structure to users for evacuation. Although substantial
research has shown that BSP beams are effective in providing
additional strength, over-reinforced BSP beams could lead to un-
desirable non-ductile failure [4]. It is therefore imperative to de-
velop retrofitting techniques that can enhance flexural strength
and maintain sufficient ductility. With this consideration, four-
point bending tests of five simply-supported RC and BSP spec-
imens were conducted. All these specimens have the same RC
geometries, but are strengthened with different bolt–plate ar-
rangements. The bolt–plate arrangements of specimens were
properly selected so that the present experimental study would
cover both over-reinforced and under-reinforced BSP beams. The
experimental data obtained was analyzed and the most appro-
priate strengthening system that can enhance flexural strength
and maintain sufficient ductility has been identified. The proposed
bolt–plate arrangement can help designers to establish a ductile
retrofitting scheme quickly and conveniently. It is noted that the
investigation of the effect of bolt–plate arrangements on the par-
tial interaction behaviour of BSP beams is beyond the scope of this
paper. The related experimental study can be found in the com-
panion paper [12].

0141-0296/$ – see front matter © 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2010.02.028
1770 R.K.L. Su et al. / Engineering Structures 32 (2010) 1769–1778
Table 1
Summary of bolt–plate arrangements of BSP specimens.
No. of bolts on shear span
Strength of bolts on shear span Pb (kN)
Plate sizea (mm × mm)
Plate force at ultimate state in full interaction analysis Fp,fi (kN)
Degree of shear connection (Pb /Fp,fi )
a
Value refers to the plate on each face of the beam, dp. = deep.
Fig. 1. Test setup.
2. Experimental program
2.1. Test setup
A four-point bending test setup was adopted, as shown in Fig. 1.
The beam specimens were simply-supported with a clear span of
3600 mm. Load was applied by a 1000 kN hydraulic jack in the ver-
tical direction, and divided into two equal point loads and exerted
symmetrically onto the specimen through a transfer beam. With
this arrangement, a pure bending zone with constant moment was
created so that the flexural behaviour under pure bending could be
studied.
2.2. Specimen details
Five specimens were fabricated and tested. The RC details of
all specimens are identical. As shown in Fig. 2, the size of the
RC sections were 225 mm × 350 mm deep. The specimens were
under-reinforced by 3T16 tension bars, contributing to a steel per-
centage of 0.76%. Transverse reinforcement of T10-150 was ap-
plied throughout the span of the beam. This amount of longitudinal
and transverse reinforcement would be sufficient to ensure that
the specimens would fail in flexure in both un-strengthened and
strengthened cases. This is desirable since the objective of this ex-
perimental study is to investigate the flexural responses of beams.
Specimen NBNP is the control specimen without any strength-
ening measure. It serves as a control specimen and is used to
demonstrate the structural performance of a RC beam prior to
strengthening.
Four strengthened specimens were fabricated. Being identi-
fied as the major parameters affecting the performance of beams,
the bolt–plate arrangements of these specimens were properly
selected so that the behaviour of beams under partial shear in-
teraction in all cases could be captured in this study. Four differ-
ent bolt–plate combinations including ‘Strong Bolt Strong Plate’
(SBSP), ‘Weak Bolt Strong Plate’ (WBSP), ‘Weak Bolt Weak Plate’
(WBWP) and ‘Strong Bolt Weak Plate’ (SBWP) are adopted and the
Specimen SBSP Specimen WBSP Specimen WBWP Specimen SBWP
8 3 3 5
608 228 228 380
6 × 150 dp. 6 × 150 dp. 6 × 75 dp. 6 × 75 dp.
605 605 302 302
1.00 0.38 0.75 1.26
detail arrangements of these specimens are summarized in Table 1
and illustrated in Fig. 2.
Two different plate sizes, 6 mm × 150 mm deep and 6 mm ×
75 mm deep mild steel plates on both side faces of specimens were
chosen to be the ‘Strong’ and ‘Weak’ plate arrangements respec-
tively. This is equivalent to 2.3% and 1.1% of the gross sectional area
of concrete. The centroidal axis of the plates was set at 250 mm
from the top of the beam and in the tension region of the beam.
This arrangement can prevent buckling of steel plates which is out
of the scope of the present study. Buckling of steel plates occurs
when relatively deep plates are used of which the top parts would
be subjected to compression [13,14]. Adopting the current plate ar-
rangement, the entire steel plate was below the mid-depth of the
section and therefore the plate compression can be minimized and
hence plate buckling can be avoided.
In this study, the degree of shear connection is used to consider
if a bolt arrangement is ‘Strong’ or ‘Weak’. The degree of shear
connection is defined mathematically as
Pb /Fp,fi , (1)
where Pb and Fp,fi are the total strength of bolts on a shear span and
the plate force at ultimate state under the case of full interaction
respectively.
To determine the degree of shear connection of all strengthened
specimens, a non-linear full interaction section analysis was car-
ried out. A non-linear stress–strain relationship was adopted for
concrete and an elastic–plastic relationship was used to model the
stress–strain relationship of steel plates and reinforcement. Incre-
mental curvatures were applied to the section until the peak mo-
ment was solved. The plate force at ultimate state and the degree
of shear connection of the strengthened specimens are listed in Ta-
ble 1. The degree of shear connection of Specimens SBSP and SBWP
are 1.00 and 1.26 respectively, implying that sufficient bolt forces
have been provided. Thus the bolt arrangement is ‘Strong’. Con-
versely, the degree of shear connection of Specimens WBSP and
WBWP are, respectively, 0.38 and 0.75, meaning that these speci-
mens have ‘Weak’ bolt arrangements.
Specimens SBSP and WBSP were strengthened by the same size
of plates, i.e. 6 mm × 150 mm deep steel plate strengthened on each
side face of the beams, but with ‘Strong’ and ‘Weak’ bolt arrange-
ments respectively. Specimens SBWP and WBWP have the same
plate arrangement but with different degree of shear connections.
By comparing these two pairs of specimens, the effect of strong
and weak bolt arrangements on the behaviour of specimens could
be studied. Meanwhile, the bolt arrangements of Specimens WBSP
and WBWP are identical but with different plate arrangements. By
comparing the results of these two specimens, the effect of plate
geometry on the performance of the beams could be investigated.
The concrete beams without plate strengthening were first fab-
ricated. Holes for allowing future installation of a bolt–plate system
were reserved in the concrete core using aluminium tubes with an
internal diameter of 14 mm. The aluminium tube was permanently
cast in the concrete core as it is expected that the change in stiff-
ness due to the tube is very small compared with the shear stiffness
of the bolt and would not significantly affect the experimental re-
sults. The bolt–plate system was installed at least 14 days after cur-
ing of concrete. By using this arrangement, the concrete core and
the steel plate are interacted purely through the bolt anchors.
 R.K.L. Su et al. / Engineering Structures 32 (2010) 1769–1778 1771

Fig. 2. RC and bolt–plate detail of BSP specimens.

a Table 2
Concrete mix adopted for producing a cubic meter of concrete.
Water Cement Fine aggregate 10 mm aggregate

kg/m3 199.6 278.9 1024.5 837.6

compressive strength of cube was around 35 MPa, as shown in the

b of 460 MPa was used in this study. φ 16 bars were used as tension
Fig. 3. Dynamic set washer (a) diagrammatic illustration and (b) actual
arrangement.
2.3. Material properties
A concrete mix containing maximum coarse aggregates of
10 mm and with target mean cube strength of 30 MPa at 28 days
after casting was designed for the specimens. The constituents and
the corresponding proportions of the concrete mix are shown in
Table 2. For each specimen, three concrete cubes with dimensions
150 mm × 150 mm × 150 mm were cast and compressive tests
were carried out on the test day to obtain the compressive strength
of cubes. There were slight variations in the average compressive
strength of cube of specimens, but in general the average concrete
Table 3.
High yield steel reinforcement with characteristic yield strength
reinforcement while φ 10 reinforcing bars were used as hangers.
Three samples were taken from each type of reinforcement. Tensile
tests were carried out and the yield strength and Young’s modulus
of these samples are summarized in Tables 4 and 5.
Three strips of dimension 400 mm × 6 mm thick were taken
from each of the 75 mm deep and 150 mm deep steel plates for
tensile tests. The material properties of the steel plates are listed
in Table 6.
HAS-E anchor rods of 12 mm diameter [15] were used as
the mechanical connectors of all specimens. The HAS-E rod is in
Grade 5.8, with a minimum 5 µm galvanized surface. Dynamic
set, instead of ordinary washer and nut was used in this study.
A spherical washer, an injection washer and an ordinary nut
were included in the dynamic set, as illustrated in Fig. 3. With
this system, epoxy grouts can be easily injected into the gaps
between connecting components and the bolt shaft, so that any
undesirable slips due to the presence of gaps between components
can be avoided. Bolt shear tests were conducted and an idealized
load–deformation response of all anchor bolts (with appropriate
key parameters) is presented in Fig. 4.
2.4. Instrumentation
Linear Variable Direct Transducers (LVDTs) and Linear Displace-
ment Transducers (LDTs) were installed, as illustrated in Fig. 5. The
1772 R.K.L. Su et al. / Engineering Structures 32 (2010) 1769–1778

Table 3
Cube strength of concrete.
Concrete cube strength (MPa)
Specimen NBNP Specimen SBSP Specimen WBSP Specimen WBWP Specimen SBWP

1 35.5 34.5 34 34.3 35.5

2 36.2 35 37 34.1 35.5
3 34 34.2 31.9 37 35.0
Average 35.2 34.6 34.3 35.1 35.3

Fig. 4. Load–slip relationship of anchor bolts.

Fig. 5. LVDT and LDT arrangements for the specimens.

Table 4 Table 5
Material properties of T-16 reinforcement. Material properties of T-10 reinforcement.
Yield strength (MPa) Young’s modulus (GPa) Yield strength (MPa) Young’s modulus (GPa)

1 545 189 1 550 185

2 525 184 2 540 194
3 540 188 Average 545 190
Average 537 187

moment zone, installation at the middle of the top flange is not

LVDTs other than those in the constant moment zone were in- possible as a loading beam was present. In this case, pairs of LDTs
stalled and pointed at the middle of the top flange. At the constant were installed and the average of the readings was used to repre-
 R.K.L. Su et al. / Engineering Structures 32 (2010) 1769–1778
Fig. 6. Strain gauges located on the steel plates of the specimens.
sent the displacement along the centroidal longitudinal axis of the
beam, as illustrated in Fig. 5.
Electric resistance strain gauges were attached on plates to
measure the curvatures in various sections of the specimens. These
strain gauges were installed along the direction of beam span
at five different sections. Three sections were within the peak
moment zone and two at the middle of the left and right shear
spans, as illustrated in Fig. 6.
2.5. Loading history
In each test, the specimen was first incrementally loaded to
50% of the ultimate capacity of the beam as estimated by a full
interaction sectional analysis. After that, the loading process was
changed to a displacement-controlled mode in which incremental
displacements were applied. The loading process was terminated
at the instant when the post-peak loading in the beam was 85% of
the peak load, or any of the two bolts in the specimen had fractured.
3. Experimental results
3.1. Moment–deformation response
The moment–deformation responses of all specimens are
shown in Fig. 7. For Specimen SBSP, which is strengthened with
‘strong’ plate and ‘strong’ bolt arrangements, the initial deforma-
tion response of the beam was elastic. The bottom reinforcement
yielded when the applied moment reached 154 kN m, equal to 95%
of the peak moment (Mu ) and the tangential stiffness of the beam
dropped abruptly to 15% of the stiffness prior to yielding of the bot-
tom reinforcement. The remaining stiffness was provided by both
the axial interaction between concrete in compression and steel
plate in tension and the flexure of steel plates. The stiffness kept
constant up to the ultimate state (M = 161 kN m). After that, con-
crete crushed and the beam degraded gradually when further de-
formation was applied and the beam failed in concrete crushing
when the mid-span deflection reached 55 mm.
For Specimen WBSP, which has the same plate arrangement as
Specimen SBSP, but with lesser bolts installed, the behaviour of the
beam was governed by bolts. The elastic stage terminated when
the applied moment reached 131 kN m (0.88Mu ) due to yielding of
the bottom reinforcement. At that instant, the tangential stiffness
abruptly dropped to 26% of the stiffness just prior to yielding. The
stiffness dropped gradually when the beam was deformed further
and reached zero when the moment reached 147 kN m (0.99Mu ).
The gradual drop in stiffness was due to the stiffness drop in bolts
as bolt slips increased. Two bolts on the same shear span and at the
same side fractured consecutively beginning from the one closer to
1773
Fig. 7. Moment–deformation responses of all specimens.
the support, when the mid-span deflection was 51 mm and 54 mm
respectively.
For Specimen WBWP, which was strengthened by the same
‘Weak’ bolt arrangement with that of Specimen WBSP, but the
‘Weak’ plate arrangement, the bottom reinforcement yielded at
the end of the elastic stage when the moment reached 124 kN m
(0.93Mu ). The remaining stiffness was about 10% of the stiffness
prior to yielding. The remaining stiffness sustained up to the failure
point when the moment and deflection were 133 kN m and 50 mm
respectively. Failure of the beam was due to consecutive bolt
fractures, which was similar to Specimen WBSP.
For Specimen SBWP, which had the same ‘Weak’ plate
arrangement as that of Specimen WBWP but 60% more bolts
to anchor the plates, the cracked elastic stiffness of Specimen
SBWP was 14% larger than that of Specimen WBWP. The bottom
reinforcement yielded when the moment was equal to 124 kN m
(0.86Mu ). After yielding, a constant tangential stiffness, equal
to 17% of its stiffness just prior to yielding, remained to resist
further loading. When the applied load reached 145 kN, the
load–deflection curve changed gradually from an ascending trend
to a descending trend and crushing of concrete began. The
descending branch of Specimen SBWP was milder than Specimen
SBSP and the residual strength in beam reached 85% of Mu at the
point when the beam was further deformed by 22 mm.
3.2. Strength and ductility
Compared with the control specimen (Specimen NBNP), the
strengthened specimens show various percentage of strengthen-
ing ranging from 32% to 59%. A summary of the strengthening per-
formance of the specimens is given in Table 7. Among the four
1774 R.K.L. Su et al. / Engineering Structures 32 (2010) 1769–1778
Table 6
Material properties of steel plates.
75 mm plate
Young’s modulus (GPa) Yield strength (MPa)
1 202 325
2 215 355
3 191 335
Average 203 338
Table 7
Summary of failure modes and increase in strength of BSP beams.
Failure mode
Specimen NBNP Concrete crushing
Specimen SBSP Concrete crushing
Specimen WBSP Bolt fracture
Specimen WBWP Bolt fracture
Specimen SBWP Concrete crushing
strengthened beams, Specimen SBSP, being strengthened with a
‘Strong’ bolt and ‘Strong’ plate arrangement, was the strongest with
59% additional strength over the control beam. Specimen WBSP,
with the same ‘Strong’ plate arrangement as Specimen SBSP, but
with ‘Weak’ bolts, achieved a 47% increase in strength. Even with
half of the plate size as the ‘Strong’ plate case, Specimen SBWP
achieved a 43% increase in strength, which is comparable to Spec-
imen SBSP. Specimen WBWP was the weakest among the four
strengthened beams and achieved a 32% strength increase. This is
much lower than that of Specimens WBSP and SBWP, which have
the same bolt and plate arrangements respectively.
The present experimental results indicate that the load–defo-
rmation responses of the specimens, as shown in Fig. 7, can
be idealized by a bi-linear curve (see Fig. 8). The displacement
ductility factor µ∆ , which is defined as the ratio between the
displacement at peak load ∆u and the notional yield displacement
∆y , was adopted to measure the ductility performance of the plated
beams. As the strengthened beams considered are predominantly
subjected to gravity loads, the strengthened beam would collapse
when the applied gravity load is higher than the load-carrying
capacity of the beam. The post-peak descending branch of the
load–deflection curve would alter the rate of failure but could not
avoid collapse of the beam. Because of that, the displacement at
peak load ∆u was chosen to represent the ultimate deformation of
the BSP beams. As shown in Fig. 8, the notional yield displacement
∆y is defined as the intersection of the two straight lines
associated with the load–deflection curves at the elastic and post-
elastic stages respectively. The displacement ductility factors of
specimens were calculated using the above definitions and the
results are given in Table 8. The four strengthened specimens have
displacement ductility, ranging from 1.70 (Specimen SBSP) to 2.57
(Specimen WBWP). The displacement ductility of all specimens
were lower than the un-strengthened specimen as expected since
the bolt–plate system acts as an additional reinforcement to the
specimen and hence would reduce the displacement ductility of
the BSP specimens.
Substantial post-elastic strength enhancement could be found
in Fig. 7. The degree of post-elastic strength enhancement is
paramount important for a gravity load-resisting structure as the
excessive deformation under additional loading in the post-elastic
stage can provides noticeable warning of impending failure of the
structure to occupants for evacuation. The post-elastic strength
enhancement factor υ is defined as the ratio between the peak
strength Pu and the yield strength Py , see Fig. 8. The post-elastic
strength enhancement factors of the specimens are tabulated in
Table 8. It can be observed that the post-elastic strength enhance-
ment factor of Specimen SBWP was 1.17 which is substantially
higher than the corresponding values of other specimens, ranging
150 mm plate
Young’s modulus (GPa) Yield strength (MPa)
214 335
216 350
206 320
212 335
Mu (kN m) Increase in strength (%)
101.4 –
161.5 59
149.2 47
133.4 32
144.6 43
Fig. 8. Definitions of displacement ductility factor and post-elastic strength
enhancement factor.
from 1.05 to 1.08. From the displacement ductility factor and post-
elastic strengthening factor, it is found that combining the use of
‘Strong bolts’ with limited amount of steel plates can effectively
enhance the post-elastic strength of beam with a satisfactory level
of ductility, meaning that Specimen SBWP is the safer strengthen-
ing system as the post-elastic branch provides a warning signal to
occupants if the structure is overloaded prior to collapse.
3.3. Failure modes
Both Specimens SBSP and SBWP failed by concrete crushing
similar to the control specimen. The failure of Specimen SBSP
is associated with concrete crushing prior to yielding of the
steel plates whereas that of Specimen SBWP is due to excessive
deformation of concrete after yielding of both steel plates and
reinforcement in the beam.
Bolt-fracture was observed in Specimen WBWP and Specimen
WBSP during the tests. The failure processes of Specimen WBWP
and Specimen WBSP were similar. As the applied load increased,
significant tilting of bolts towards the centre of beam was observed
(see Fig. 9(a)). This is due to an extra moment being induced due
to eccentricity of shears exerted by the steel plates and RC beam,
as illustrated in Fig. 9(b). The extra moment induced increases
the demand of bolts, which may become critical if partial shear
connection is adopted. Finally, when the combined moment and
shear on bolts reached the capacity of the bolt, sudden fracture
occurred which caused the failure of the specimen.
Outward deflections of plates were also observed at the plate
ends as the beam approached failure as shown in Fig. 10. The
deflection was due to an uneven distribution of plate forces along
the thickness of the plate, which occurred simultaneously with
the tilting of the bolts. The outward deflections of plates were
restrained by the anchoring bolts located nearest to the plate ends.
 R.K.L. Su et al. / Engineering Structures 32 (2010) 1769–1778 1775

Table 8
Summary of displacement ductility factors and post-elastic strength enhancement factors.
∆y (mm) ∆u (mm) µ Py (kN) Pu (kN) υ
Specimen NBNP 17.4 63.0 3.62 101.4 101.4 1.00
Specimen SBSP 20.8 35.2 1.70 154.4 161.5 1.05
Specimen WBSP 21.0 48.2 2.30 140.2 149.2 1.06
Specimen WBWP 19.4 49.9 2.57 124.0 133.4 1.08
Specimen SBWP 18.0 44.0 2.45 123.6 144.6 1.17
µ = displacement ductility, υ = post-elastic strength enhancement factor.

Fig. 9. Tilting of anchor bolts (a) deformation mode and (b) shear transfer mechanism through the action of bolts.

plates was estimated by the difference of strain gauge readings at

Fig. 10. Outward deflection of the plate of Specimen WBSP.
3.4. Moment–curvature responses
The moment–curvature relationship of the constant moment
zones was captured in order to study the flexural responses of BSP
beams. As partial shear interaction existed [3], the curvatures of
the RC and steel plates were different and two different measuring
systems were used. The average curvature of the RC was measured
by three sets of LVDTs installed at the two loading points and
mid-span (see Fig. 5) so that the abrupt changes in curvatures
due to localized cracks would be minimized. As the theoretical
curvature profile of the constant moment zone is constant, the
average curvature could be estimated from the deformations of the
three LVDT readings. On the other hand, the curvature of the steel
different levels of the steel plate divided by the vertical separation
between them (see Fig. 6 for the arrangements of strain gauges).
The moment–curvature relationships of all four BSP specimens are
plotted in Fig. 11(a)–(d).
For Specimen SBSP, the curvature of the RC beam and the steel
plate were the same up to the point when the applied moment
reached 135 kN m. Then the curvature of the steel plate increased
much less per specimen moment added to the beam than in the
initial elastic stage. Plastic hinges began to form in the RC beam
when the moment reached 154 kN m as the moment–curvature
relationship of the RC became flat. Finally, the beam failed when
the curvatures of the beam and steel plate were 39 × 10−3 m−1
and 15 × 10−3 m−1 respectively.
For Specimen WBSP, the moment–curvature relationship of the
RC beam was elastic until the moment and curvature of the RC
beam reached 131 kN m and 14 × 10−3 m−1 respectively. Plastic
hinges gradually formed in the RC beam until the beam failed when
its curvature reached 33 × 10−3 m−1 . Curvature of the steel plates
was almost the same as the RC beam prior to the formation of
plastic hinges, and was slightly larger than that of the beam as
further loads were added. The curvature in the steel plates stopped
to increase with the curvature of the beam since reaching 23 ×
10−3 m−1 and it remained constant until failure.
For Specimen WBWP, the moment–curvature relationship of
both the RC beam and steel plates in Specimen WBWP were
similar to Specimen WBSP. Both curvatures of the RC beam and
the steel plates responded identically and linearly up to the point
when the moment and curvatures were 124 kN m and 16 ×
10−3 m−1 respectively. Plastic hinges formed in the RC beam and
the corresponding curvature reached 31 × 10−3 m−1 when failure
occurred. The curvature in the steel plate stopped to increase with
that of the RC beam when it reached a maximum of 23 × 10−3 m−1 .
For Specimen SBWP, the moment–curvature relationship of the
RC beam was elastic until the moment reached 129 kN m. Plastic
1776 R.K.L. Su et al. / Engineering Structures 32 (2010) 1769–1778
Fig. 11. Moment–curve responses of RC and steel plates (a) Specimen SBSP, (b) Specimen WBSP, (c) Specimen WBWP, and (d) Specimen SBWP.
hinges formed afterwards and the beam failed when the curvature
of the RC beam was 61×10−3 m−1 . This value was much larger than
the other specimens. The curvature responses of steel plates were
about 10% smaller than that of the RC beam up to the elastic limit,
when the curvature in the steel plates was 14×10−3 m−1 . Then, the
curvature in the steel plates decreased gradually to 6 × 10−3 m−1
at the peak moment (M = 145 kN m). At the post-peak stage the
curvature in steel plates remained unchanged.
It can be seen that the curvatures of steel plates were consis-
tently lower than the RC beam in the inelastic stage. In some cases,
the curvature even reduced when further loads were applied. In
the inelastic range of response, the additional moment was mainly
resisted by the axial force couple between the RC beam in compres-
sion and the steel plate in tension. The axial tension force extended
and straightened the plate and hence reduced the curvature of steel
sections. Also, as the stiffness of bolts drop, their ability to mobilize
the transverse deflections of the plates also drop and that also con-
tributes to the separation of curvatures between the RC and plates.
4. Various design controlling factors of BSP beams
From the results mentioned in the previous sections, it is ob-
served that the behaviour of BSP beams could be significantly af-
fected by using different strengthening schemes. The strength and
ductility of BSP beams could be controlled by concrete compo-
nents, anchor bolts or steel plates. A new generic classification
of BSP beams is herein established according to the controlling
factors and three groups of beams, namely concrete-controlled,
bolt-controlled and plate-controlled are identified. The structural
behaviour of these three groups of BSP beams are discussed herein.
Case 1: Concrete-controlled BSP beams
In a concrete-controlled case, the structural behaviour of the
strengthened beam is controlled by concrete in compression. The
typical load–deflection response of a concrete-controlled case,
such as that of Specimen SBSP, is shown in Fig. 12(a). The tension
reinforcement yields at the end of the elastic stage (Segment OA).
The stiffness of the beam drops abruptly after yielding of the
tension reinforcement (Point A) but the beam is still able to take up
further loads as resisted by the axial force couple between the RC
beam and the steel plate (Segment AC). At the ultimate state (Point
C), although the steel plate is still able to provide additional tensile
force, concrete is unable to provide the required compression to
give additional moment resistance. So concrete crushes and leads
to brittle failure of the beam (Segment CE).
Case 2: Bolt-controlled BSP beams
Fig. 12(b) illustrates the idealized load–deflection curve of a
bolt-controlled beam like that of Specimen WBSP or Specimen
WBWP. The beam responds elastically (Segment OA) until the
yielding of the tension reinforcement in the RC beam (Point A).
When further loads are applied, the loads are taken up by the
axial force couple similar to that in the concrete-controlled case
(Segment AC). However, as the stiffness of the bolts significantly
drops, the interaction between the RC beam and steel plates
decreases and results in a gradual drop in the stiffness of the beam
(Segment AC). At point C, the average stiffness of the bolts reaches
zero so that the force induced in the plates goes to its maximum
and the beam becomes plastic. Further deformation applied to the
bolt–plate system is taken up by the bolts alone and the bolts fail
when the ultimate deformation is reached (Point D). The failure
path of bolt-controlled beams depends on the post-peak responses
of the bolts, which can be gradual or brittle, as indicated by the path
DE1 or DE2 respectively.
Case 3: Plate-controlled BSP beams
The load–deflection response of plate-controlled beams, similar
to that of Specimen SBWP, is illustrated in Fig. 12(c). Similar to the
concrete-controlled case, an elastic stage (Segment OA), followed
 R.K.L. Su et al. / Engineering Structures 32 (2010) 1769–1778
Fig. 12. Idealized load–deflection curves of (a) concrete-controlled, (b) bolt-
controlled and (c) plate-controlled BSP beams.
by an abrupt drop in stiffness (Point A) and a further linear stage
(Segment AB) is present. Instead of concrete crushing, the steel
plate yielded first. The yielding process of the steel plates occurs
gradually as the strain within the plate is non-uniform, and begins
from the bottom fibre to the top fibre when subjected to a sagging
moment. This results in a gradual drop in stiffness (Segment BC).
The beam becomes fully plastic afterwards (Segment CD) and the
concrete starts crushing due to excessive deformation at point D.
Concrete crushing then continues until failure occurred at point E.
Among the three cases, the plate-controlled type is recom-
mended as the elastic and post-elastic strengths can be enhanced,
and sufficient ductility can be maintained. To achieve this, the
amount of plate added should be properly selected and bolts with
adequate strength are required. The bolt-controlled case can also
be used in practice if the anchor bolts have a very high plastic de-
formation capacity. However, most of the anchor bolts can only
1777
Fig. 13. Generic classification of BSP beams according to the strength of bolts and
plates.
provide limited ductility. The concrete-controlled case results in
brittle failures and should be avoided under any circumstance.
Whether the BSP beam is controlled by concrete, bolt or plate
failure, it is actually dependent on the strength of the bolts and
the steel plates. The relationship between the strength of bolts,
strength of steel plates and the corresponding beam classification
is illustrated in Fig. 13. If the strength provided by bolts and plates
is higher than the compressive capacity of the concrete stress
block in the RC beam, the beam is controlled by the concrete. The
failure behaviour of the beam is brittle and is similar to an over-
reinforced concrete beam. If either the bolts or steel plates added
are of insufficient strength, the strengthened beam is controlled by
the corresponding weaker element (see the corresponding regions
in Fig. 13). In such cases, the beams would fail in a more ductile
manner which would provide noticeable warning of impending
failure of the structure.
The intersection of the three broken lines in Fig. 13, indicat-
ing bolts, steel plates and concrete having a similar strength cor-
responds to the ‘balanced’ failure case. In this ideal case, crushing
of concrete, fracture of bolts and full yielding of steel plates will
occur simultaneously at the ultimate loading stage.
When more bolts are added, the point corresponding to this
strengthening scheme (e.g. Specimen WBSP) in Fig. 13 would be
shifted upward; likewise, when larger sectional area or higher
strength of the steel plates is used, the point (e.g. Specimen SBWP)
in Fig. 13 would be shifted along the horizontal direction. For
practical design, it is recommended that the amount of steel plates
provided should not exceed the value of the ‘balanced’ failure as
shown in Fig. 13. This is because in that case no matter how the
bolts are arranged, the strengthened beam will either be controlled
by bolts or by concrete and will fail in a relatively brittle manner.
5. Conclusions
The results of a series of tests on BSP concrete beams with
different bolt–plate arrangements have been presented. The
post-elastic strength enhancement and displacement ductility are
identified as two important structural performance criteria for
structures predominantly subjected to gravity loads. These two cri-
teria are greatly influenced by the strength of the bolts and plates
used. It is concluded that the ‘Strong bolt weak plate’ arrangement
will lead to a design in which sufficient strength enhancement and
ductility can be achieved, while the ‘Strong bolt strong plate’ ar-
rangement results in a brittle and undesirable failure. This suggests
that the amount of steel plates added should be limited, while suf-
1778 R.K.L. Su et al. / Engineering Structures 32 (2010) 1769–1778
ficient bolts should be used to ensure the beam failure is in a de-
sirable ductile mode.
According to the experimental results presented herein, the
behaviour of BSP beams dominated by flexure can be divided
into different categories based on the strength of concrete, bolts
and steel plates. The strength and ductility performance of BSP
beams is controlled by the weakest component. The concept of a
balanced failure point, corresponding to simultaneous failure of
concrete, bolt and plate, is introduced. It provides a quick tool
for designers to check if a retrofitting scheme would fail in a
ductile manner with sufficient post-elastic strength enhancement.
While the concept of the balanced failure point is introduced,
the present limited number of tests are unable to identify the
relationship for determining the balanced failure point from a
given bolt–plate arrangement. More experimental and theoretical
studies are recommended for the determination of the balanced
failure point of BSP beams.
Acknowledgements
The research described here was supported by the Research
Grants Council of Hong Kong SAR (Project No. HKU7166/08E).
The generous support of the anchor bolts by HILTI Corporation is
gratefully acknowledged.
References
[1] Correia JR, Branco FA, Ferreira JG. Flexural behaviour of multi-span GFRP-
concrete hybrid beams. Eng Struct 2009;31(7):1369–81.
[2] Carpinteri A, Cornetti P, Pugno N. Edge debonding in FRP strengthened beams:
stress versus energy failure criteria. Eng Struct 2009;31(10):2436–47.
[3] Oehlers DJ, Nguyen NT, Ahmed M, Bradford MA. Transverse and longitudinal
partial interaction in composite bolted side-plated reinforce-concrete beams.
Struct Eng Mech 1997;5(5):553–64.
[4] Ahmed M, Oehlers DJ, Bradford MA. Retrofitting reinforced concrete beams by
bolting steel plates to their sides. Part 1: behaviour and experiments. Struct
Eng Mech 2000;10(3):211–26.
[5] Oehlers DJ, Ahmed M, Bradford MA, Nguyen NT. Retrofitting reinforced
concrete beams by bolting steel plates to their sides. Part 2: transverse
interaction and rigid plastic design. Struct Eng Mech 2000;10(3):227–43.
[6] Ahmed M, Oehlers DJ, Nguyen NT, Bradford MA. Reinforced concrete beams
with steel plates bolted to their sides. In: Proceedings of Fifth International
Conference on Structural Failure, Durability and Retrofitting. 1997. p. 362–9.
[7] Nguyen NT, Oehlers DJ, Bradford MA. An analytical model for reinforced
concrete beams with bolted side plates accounting for longitudinal and
transverse partial interaction. Inter J Solids Struct 2001;38:6985–96.
[8] Subedi NK, Baglin PS. External plate reinforcement for concrete beams. J Struct
Eng ASCE 1998;124(12):1490–5.
[9] Su RKL, Zhu Y. Experimental and numerical studies of external steel plate
strengthened reinforced concrete coupling beams. Eng Struct 2005;27(10):
1537–50.
[10] Su RKL, Siu WH. Nonlinear response of bolt groups under in-plane loading. Eng
Struct 2007;29(4):626–34.
[11] Siu WH, Su RKL. Load–deformation prediction for eccentrically loaded bolt
groups by a kinematic hardening approach. J Construct Steel Res 2009;65(2):
436–42.
[12] Siu WH, Su RKL. Effects of plastic hinges on partial interaction behaviour of
bolted side-plated beams. J Construct Steel Res 2010;66(5):622–33.
[13] Smith ST, Bradford MA, Oehlers DJ. Local buckling of side plated reinforced
concrete beams. II: experimental study. J Struct Eng, ASCE 1999;125(6):
635–43.
[14] Smith ST, Bradford MA, Oehlers DJ. Buckling tests on steel plates restrained at
discrete points in the retrofit of reinforced concrete beams. Proc Institut Civil
Eng Struct Build 2001;146(2):115–27. May.
[15] Fastener Technology Manual. Issue 2005-06. Hilti Corporation.