Engineering Structures 257 (2022) 114007

Contents lists available at ScienceDirect

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

Finite element analyses of FRP-strengthened concrete beams with corroded

reinforcement
Milan Gotame a , Carl Lindqvist Franklin a , Mattias Blomfors b , Jincheng Yang a , Karin Lundgren a ,∗
a
Department of Architecture and Civil Engineering, Chalmers University of Technology, SE 412 96, Gothenburg, Sweden
b
Bridge and Analysis group, Norconsult AB, SE 417 55, Gothenburg, Sweden

ARTICLE INFO ABSTRACT

MSC: Existing deteriorated reinforced concrete (RC) structures need strengthening to extend service life. Fibre
00-01 reinforced polymer (FRP) has been widely used to strengthen sound structures, but its application on
99-00 damaged concrete structures still needs to be investigated. This paper presents non-linear finite element
Keywords: analyses conducted to assess the flexural behaviour of corrosion-damaged RC beams strengthened with
Reinforced concrete structures externally bonded FRP. Beams in four different categories were analysed: a reference beam, a corroded
Finite element analysis but non-strengthened beam, and corroded beams strengthened with glass FRP (GFRP) and carbon FRP
Corrosion
(CFRP) respectively. Furthermore, the strengthened beams were modelled with different modelling choices
Fibre-reinforced polymer
to investigate the effectiveness of FRP applied to the beam soffit and as U-jackets. Pre-loading and corrosion-
CFRP
GFRP induced cracks were incorporated by reducing the tensile strength of concrete elements at crack locations.
Strengthening Average and pitting corrosion were incorporated by reducing the cross-sectional area of the reinforcement
Cracks corresponding to the measured corrosion levels. Interface elements were used to simulate the bond between
Interface FRP and concrete. The modelling methods were validated against experimental results. It was found that
U-jacket modelling of pitting corrosion, especially the location of pits, lengths and number of pits considered, were
influential in predicting the load and deformation capacity of beams. A CFRP plate at the beam soffit,
combined with inclined U-jackets at its ends of the CFRP plate provided sufficient flexural strengthening. Thus,
intermediate U-jackets did not further increase the load-bearing capacity for the studied beam geometry and
corrosion damages. However, with a GFRP sheet at the beam soffit, both inclined and intermediate U-jackets
were needed to provide full utilisation of the GFRP sheet for the studied beam geometry. In further studies of
the effectiveness of the strengthening methods, it is recommended to investigate beams of varying dimensions,
corrosion patterns and levels, and FRP spacing and dimensions.

1. Introduction load-carrying capacity. Furthermore, variables such as higher volumes

of commuter and cargo activity, and heavier vehicles have increased
Concrete has been a widely utilised material in structures over the load on existing structures throughout the world [1], creating a
150 years, due to its low cost, low need for maintenance, longevity, need for increased capacity. To reduce the cost and environmental im-
fire performance and stiffness [1]. Existing concrete structures built
pacts of new constructions, it is vital (for society and the environment)
in the mid-to-late 1900s display increasing signs of deterioration due
to be able to strengthen structures such as bridges and buildings.
to adverse environmental conditions resulting in reduced capacity [1].
The most common cause of degradation of reinforced concrete (RC) Strengthening with fibre-reinforced polymers (FRP) is a well-
structures is chloride-induced corrosion [2]. Corrosion of reinforcement accepted technique for increasing the capacity of concrete structures
degrades concrete structures in several ways. Pitting corrosion causes [6]. There has been much research into strengthening of sound struc-
local reduction of the cross-sectional area of the reinforcement, reduc- tures with FRPs [1,7]. However, the need to be able to strengthen
ing its yielding and ultimate capacities as well as its ductility [3,4]. The deteriorated structures has increased, and research is needed to know
corrosion products occupy a larger volume than the original steel. This how to apply this to the strengthening of corrosion-damaged structures.
introduces splitting stresses, and, ultimately, results in spalling of the Due to the high strength-to-weight ratio and high modulus of elasticity,
concrete cover and weakening of the bond between the concrete and
externally bonded FRPs in the beam soffit are easy to handle and
reinforcement [5]. The latter may cause anchorage failure and reduced

∗ Corresponding author.
E-mail address: [email protected] (K. Lundgren).

https://doi.org/10.1016/j.engstruct.2022.114007
Received 1 October 2021; Received in revised form 1 February 2022; Accepted 7 February 2022
Available online 9 March 2022
0141-0296/© 2022 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
M. Gotame et al. Engineering Structures 257 (2022) 114007

effective in improving structural behaviour [1]. However, for heavily levels, material properties of concrete, reinforcement and FRPs were
corroded RC structures, externally bonded FRPs might not improve used according to measurements made by Yang. Based on the validated
the load capacity, stiffness and ductility capacity effectively due to FE models, different examples of FRP composite applications were
damaged concrete cover [8–10]. Triantafyllou [11] and Al-Saidy [8] analysed to investigate their effects on flexural responses and failure
suggested doing a patch repair of the damaged concrete cover before modes.
applying FRP plates, to improve the structural performance of the
corrosion-damaged RC structures. In practice, a patch repair might not
be feasible for RC structures due to its vulnerability to the removal 2. Overview of analysed experiments
of concrete cover and labour-intensive uneconomical characteristics.
Thus, there is a need for alternative strengthening techniques. It has
been shown that the combination of transverse FRP wraps/U-jackets The RC beams analysed in this FEA study were tested in four-
wrapped around the beam with FRP plates, can improve the flexural point bending until failure by Yang [14]. The first test group, RN, in-
performance without needing a patch repair [11]. cluded two sound beams, non-strengthened to act as reference. Among
Nowadays, the finite element (FE) method (FEM) is commonly used the deteriorated beams, two were non-strengthened (DN), three were
in the structural analysis of RC structures. To model corrosion-damaged strengthened with GFRP (DG) and three were strengthened with CFRP
concrete beams, the damage (both in the form of corrosion-induced (DC). This section presents an overview of the experiments. For a more
cracks and reductions in reinforcement area) need to be included. detailed information about the tests, see Yang [14,32].
According to Blomfors [12,13], the weakening of concrete elements
The tested RC beam specimens are of 2.1 m length, 225 mm height
at the location of cracks in an FEA model gives reasonable results
and 150 mm width, as shown in Fig. 1(a). The concrete mix has
when compared with test results, in terms of load vs deflection and
proportions of cement:sand:gravel:water = 1:2.46:1.90:0.43 by weight.
ultimate capacity. Considering that pitting corrosion has a major impact
on the reduction of ultimate and deformation capacity of a beam [14], Section 3.3 presents the properties of concrete and reinforcement at the
modelling local corrosion is highly important in accurately predicting time of structural failure tests. The damaged beams were pre-cracked
the capacity of corrosion-damaged RC structures. The local pitting and by three-point bending, followed by 75 days of accelerated corrosion of
general corrosion of steel reinforcement can be modelled by reducing the tensile reinforcement bars at the mid-600 mm zone. The three-point
its cross-sectional area [15]. bending and accelerated corrosion resulted in transverse bending cracks
Previously, two-dimensional (2D) [16–18] and three-dimensional and longitudinal corrosion-induced cracks. At the mid-span, the crack
(3D) [13,19–22] FE modelling have been used to analyse sound, cor- width was in the range of 0.7 to 1.9 mm. The corrosion-induced cracks
roded and strengthened concrete beams. Even though 2D symmetrical were propagated within the mid-900 mm region. Fig. 1(b) shows how
FE modelling can lead to good results [16,17], 3D FE modelling may the FRPs were bonded externally to the concrete surface. The 1500 mm
be needed for reasons such as: (i) the potential corrosion effects are long GFRP sheet and CFRP plate with cross-sections of 150x3.03 mm2
typically 3D in nature (ii) for a particular beam size, the assumption and 100x1.45 mm2 respectively, were attached to the beam soffit. After
of plane stress or plane strain may not be completely true and (iii)
48 h of curing, CFRP U-jackets were installed on the DG and DC beams,
the FRP strengthening may not be applied to the entire width of
see Fig. 1(b). The vertical U-jackets have one layer of CFRP fabric,
the beam. Normally, FRP strengthening is modelled as shell elements
whereas the 45◦ inclined U-jackets have three layers of CFRP fabric.
with an orthotropic material, or as brick elements defined with an
After bonding of the FRPs, the beams were cured for four weeks prior
isotropic 3D material representing the adhesive and unidirectional
and bi-directional smeared reinforcement of fibres [23]. The bond be- to the four-point bending tests.
haviour between the concrete and FRP has an influence on the debond- The beams with an effective span of 1.8 m were tested in four-
ing of FRP and stiffness of strengthened RC structures. Researchers point bending. Variable differential transformers (LVDT) were used to
have used different methods to model the bond behaviour between con- measure the net deflection at the middle of the beam and at the two
crete and FRP in non-linear finite element analyses (NLFEA). According supports. In the FRP composites, the axial strains at critical positions
to current literature, two different methods have been found to be more were measured by strain gauges (SG5, SG6), see Fig. 1(b). After the
common: (i) full bond without failure/damage [24]; (ii) behaviour with four-point bending tests, the tensile reinforcement bars were extracted
limited strength and damage evolution [25,26]. In terms of how to from the beams and the corrosion levels were measured using 3D
assign the bond behaviour, we can (i) define constitutive law directly optical scanning [14].
between two interfacial surfaces or (ii) use elements to model the bond
line and assign the law to interfacial elements [27,28]. Between these
two methods, the interface elements have most commonly been used to 3. Numerical modelling
simulate the interaction between concrete and FRP. In this method, the
interfacial properties can be assigned by defining the constitutive bond
stress versus slip relationship or shear traction–separation law [27,28]. The flexural behaviour of reference, corrosion-damaged and FRP-
Chen et al. [29] concluded that the bond behaviour between the strengthened RC beams tested by Yang [14] was studied using 3D
concrete and FRP is important to accurately predict the failure and NLFEA in DIANA 10.4 [33]. This section presents the FE modelling
crack pattern of strengthened RC structures. Several other researchers approaches are presented.
have also applied NLFEA to FRP-strengthened RC beams, and shown
that the failure mode and behaviour can be accurately predicted when
the bond between concrete and FRP is properly included [21,22,30,31]. 3.1. Summary of analyses carried out
FE analyses of FRP-strengthened beams with corroded reinforce-
ment thus includes several major challenges. In addition to modelling
of cracking reinforced concrete, special consideration needs to be taken For ten tested specimens, one beam from each of the four different
on how to include the corrosion damages and how to model interaction categories was analysed. A reference beam (RN1), a corroded but non-
between the concrete and the FRP strengthening. Thus, modelling strengthened specimen (DN1) and corroded beams strengthened with
methods need to be developed and validated to experimental results. GFRP and CFRP (DG1 and DC1 respectively) were modelled using
In this research work, FE models were developed and the analyses different modelling choices. An overview of these analyses and the
were validated to experiments carried out by Yang [14]. The corrosion different modelling choices is presented in Table 1.

2
M. Gotame et al. Engineering Structures 257 (2022) 114007

Fig. 1. Overview of experiments.

3.2. Beam idealisation and meshing centres of the loading plates was 600 mm. The load was applied as
deformation at the midpoint of the loading beam, see Fig. 2.
Since the corrosion damage for the damaged RC beams was not
symmetrical, they were modelled with their complete geometry. The 3.3. Models for concrete, reinforcement and their interaction
reference beam was used as a basis, and was further developed for
modelling of the corroded and strengthened beams. The idealised ref- Table 2 shows the material properties of the concrete used in the
erence beam for the 3D FE model is shown in Fig. 2. The concrete FEA. These are based on compression and wedge split tests conducted
was modelled using 3D solid continuum elements of brick, tetrahedron, by Yang [32]. Concrete cracking was modelled using a smeared ro-
tating crack model based on the total strain,
√ as described in [33].
pyramid, and wedges (HX24L, TE12L, PY15L, TP18L, respectively), all 3
The crack band width was assumed as 𝑉 , where V is the volume
with linear interpolation and normal gauss integration. The average
of a 3D concrete element, as proposed by Rots [35]. This crack band
size of element lengths was set to 20 mm, fulfilling the recommendation
𝑙 ℎ 𝑏 width was later verified against approximate lengths of zones with
of using a minimum of ( 50 , 6 , 6 ), as suggested by Hendriks [34].
localised strains in the analysis. For the reference beam, the tensile
Longitudinal reinforcement bars were modelled as beam elements,
strength of the concrete elements at the mid-section, see Fig. 2, was
which were connected to surrounding concrete by interface properties
reduced by 20% to promote strain localisation; this largely improved
as described in Section 3.3. However, the top rebars and stirrups were
the convergence. The tensile behaviour of concrete was characterised
assigned as embedded reinforcement, assuming full interaction with the
by the tensile strength–strain relationship, introduced by Hordijk [36].
concrete. In the FE model, the loading and support plates were mod- The behaviour of concrete in compression was taken into account using
elled as 25 mm thick plates with solid elements. Thin wooden plates, a parabolic compression curve, according to Feenstra [37], with a
modelled using solid elements, were applied under them to avoid stress softening branch. The softening branch of the compression curve is
concentration on the concrete near the plates. The interface between based on the compressive fracture energy. A reduction factor of 0.6 due
the concrete beam and steel loading and support plates was modelled to lateral cracking was used, according to Vecchio & Collins [38]. To
using Q24IF elements. The centre nodes along the support plates were simulate the behaviour of reinforcement in the FE analyses, the tensile
restricted to displace vertically at both left and right support plates, and stress–strain curve was given as an input. The stress–strain properties of
also restricted horizontally at the right support. The load distribution the steel reinforcement bars were obtained from standard tensile tests
beam, used in the tests, was idealised with beam elements (L12BEA). conducted by Yang [32]. The curves fitting the average of six tensile
The end nodes of the loading beam were tied to the centre lines of the tests of both the top and bottom rebars were used as input in the FE
loading plates in their vertical displacement. Furthermore, the centre analyses, for detailed information, see [39].
node of the loading beam was restricted to displacing horizontally to The interaction between concrete and bottom reinforcement was
ensure the stability of the loading beam. The distance between the modelled using a bond stress–slip relationship according to fib Model

3
M. Gotame et al. Engineering Structures 257 (2022) 114007

Table 1
Overview of FE analyses.
Beam FEA model Description 𝐹𝑢 a 𝛥𝑢 b Failure mode
[kN] [mm]
RN1 RN1_FEA Reference RC beam 88.8 53.0 Concrete crushing
DN1_GC_FEA Reference beam modified to 76.0 64.1 Concrete crushing
damaged beam with only general
corrosion corresponding to DN1 beam
DN1_FEA Three major corrosion pits in each rebar 50.8 17.9 Rebar rupture
DN1 DN1P_FEA One major corrosion pit in each rebar 51.4 9.9 Rebar rupture
DN1PE_FEA One major corrosion pit in each rebar 51.4 9.1 Rebar rupture
with extended yield penetration length
DG1N_FEA Corroded but non-strengthened 50.9 10.4 Rebar rupture
DG1S_FEA Corroded, strengthened with GFRP 94.5 14.7 Concrete separation
sheet only
DG1IU_FEA Corroded, strengthened with GFRP 111.0 18.6 Anchorage failure
sheet and inclined U-jackets
DG1NLI_FEA Corroded, strengthened with GFRP sheet, 112.0 18.0 Anchorage failure
vertical and inclined U-jackets,
DG1
non-linear interface for U-jackets
DG1_FEA Corroded, strengthened with GFRP sheet, 131.0 22.8 GFRP rupture
vertical and inclined U-jackets,
linear interface U-jackets
DC1N_FEA Corroded, but non-strengthened 49.6 10.1 Rebar rupture
DC1P_FEA Corroded, strengthened with 119 11.8 Anchorage failure
CFRP plate only
DC1IU_FEA Corroded, strengthened with 160 14.5 Anchorage failure
DC1 CFRP plate and inclined U-jackets
DC1_FEA Corroded, strengthened with 168 14.6 Anchorage failure
CFRP plate, vertical and inclined U-jackets
a
𝐹𝑢 – total ultimate load applied.
b
𝛥𝑢 – deflection at mid-span of beam corresponding to the ultimate load.

Fig. 2. 3D FE mesh of the reference beam (FE model RN1).

Code 2010 [40], assuming ‘‘Good’’ bond condition and ‘‘Pull-Out’’ Table 2
Material properties of concrete used in the non-linear analyses.
failure condition. The effect of corrosion on the bond–slip relationship
Concrete properties Value Reference Test
was not considered, as it is of minor importance for the analysed
beams. Note that this effect is important for other cases, such as when Mean compressive strength 𝑓𝑐𝑚 = 0.8𝑓𝑐𝑚.𝑐𝑢𝑏𝑒 63 [MPa] [32,40] 𝑓𝑐𝑚.𝑐𝑢𝑏𝑒 a
Mean tensile strength 𝑓𝑐𝑡 = 2.12𝑙𝑛(1 + 0.1𝑓𝑐𝑚 ) 4.21 [MPa] [40] –
anchorage or shear failure takes place (note also that corrosion may
Young’s modulus 𝐸𝑐𝑚 33.3 [GPa] [32] 𝐸𝑐𝑚.𝑐𝑦𝑙𝑖𝑛𝑑𝑒𝑟 b
change the failure mode to these). This effect can then be included Tensile fracture energy 𝐺𝐹 134 [N/m] [32] Wedgec
either by modifying the bond–slip behaviour, or by use of a frictional Compressive fracture energy 𝐺𝐶 = 250𝐺𝐹 33500 [N/m] [40] –
model between the steel and the concrete, see [13]. The normal stiff- a
𝑓𝑐𝑚.𝑐𝑢𝑏𝑒 – from standard compressive test of concrete cubes (side length 150 mm).
ness was set to 1013 N/m3 . Full interaction was assumed between top b
𝐸𝑐𝑚.𝑐𝑦𝑙𝑖𝑛𝑑𝑒𝑟 – from standard test of cylindrical specimens of 100 mm diameter and
reinforcement and concrete, as well as between stirrups and concrete. 200 mm height.
c
Wedge – from wedge splitting test.
3.4. Implementation of pre-existing cracks

For beams in the categories DN1, DG1 and DC1, the cracks in- fracture energy for the corresponding concrete elements, as suggested
duced due to pre-loading (transverse cracks) and corrosion (longi- by Blomfors et al. [12,13]. The reduction was based on the measured
tudinal cracks) were incorporated by reducing tensile strength and crack widths, 𝑤𝑐 . The reduced tensile properties of the weakened

4
M. Gotame et al. Engineering Structures 257 (2022) 114007

Fig. 3. Principle of deriving reduced stress versus strain curve for weakened concrete using the crack widths (𝑤𝑐 ), proposed by Blomfors [12]: (a) Bi-linear stress-crack opening
relationship [41] (b) Stress–strain relationship, using the crack band width h, (c) The resulting stress–strain relationship for the weakened elements. Here, 𝑓𝑐𝑡,𝑐 is the reduced
𝑤 −𝑤
tensile stress, 𝐺𝐹 ,𝑤𝑐 is the reduced residual fracture energy and 𝜀𝑤𝑐,𝑢𝑙𝑡 = 𝑢𝑙𝑡ℎ 𝑐 for 𝑤𝑐 > 𝑤𝑠 .

concrete elements were determined from the corresponding crack width

𝑤𝑐 in a bilinear mode-I stress-to-crack width relationship with a break
point, according to Wittmann et al. [41], see Fig. 3(a), which was
characterised by 𝜎𝑠 = 1.05 MPa, 𝑤𝑢𝑙𝑡 = 0.159 mm [40], and 𝑤𝑠 =
0.0239 mm. This is a simplified way to include existing damage, and
is specially well suited when inspection results of existing structures
are available. At prediction of future damages, several models for
development of corrosion-induced cracking exist, e.g. [42,43].
For each of the transverse cracks, the crack width was assumed to
be constant along the width of the beam. Moreover, the longitudinal
cracks were assumed to have the same crack width throughout their
crack length. For those crack widths larger than the breaking point in
the bilinear mode-I stress-to-crack width curve, the widest crack width
was chosen. Furthermore, for cracks with a width close to or larger
than the ultimate crack width (𝑤𝑢𝑙𝑡 ), a tensile strength corresponding
to 0.99𝑤𝑢𝑙𝑡 was assumed. Fig. 3 shows how the reduced tensile stress
versus strain was calculated. In FE modelling, this was given as input
for the weakened concrete element rows. For the weakened elements,
the strain localisation was assumed over an element, so that ℎ = 20 mm.
Fig. 4 shows a 3D view of weakened concrete elements corre-
sponding to pre-existing cracks, in DN1 and DC1 beams. In general,
the corrosion-induced cracks, on the bottom face of the beams, were
assumed to extend to as much as twice the level of the tensile rebars
from the beam soffit. Moreover, for corrosion-induced cracks at the side
face, elements at the level of the tensile rebars were weakened. These
cracks were assumed to extend to twice the effective cover of the tensile
rebar along the width of beam.

3.5. Modelling of corrosion damage in rebars Fig. 4. 3D view of weakened elements for cracks induced by pre-loading and corrosion.

The measured general and pitting corrosion characteristics were

implemented in the FE model by reducing the cross-sectional area of the where 𝐴𝑖𝑛𝑖 is the cross-sectional area of a rebar before corrosion and
corroded bottom tensile rebars. The general corrosion was considered 𝐴𝑐𝑜𝑟𝑟.𝑚𝑖𝑛 is the minimum cross-sectional area of the rebar at pit after cor-
by reducing the cross-sectional areas of the rebars to correspond with rosion, extracted from 3D-scanning data. Thereafter, the cross-sectional
the average corrosion level in the mid 900 mm. The pitting corro- area of the rebar just outside the yield length, 𝐴𝑝 , was calculated by
sion was considered by further reducing the cross-sectional areas of using equation:
the bottom rebars at the locations of major pits, according to the
3D scanned values. At loading, localised strains are generated at the 𝐴𝑝 = (1 − 𝜇𝑝 )𝐴𝑖𝑛𝑖 (3)
location of a severe corrosion pit [4]. At yielding, the localised strains The expected yield length (𝑙𝑦 ) was estimated by measuring the
will spread within a certain region, here denoted as the yield length, length of the pit with corrosion levels greater than 𝜇𝑝 using the 3D
𝑙𝑦 , but limited by the fact that rupturing of the rebar will occur at the scanned data of the rebars, see Fig. 5. Moreover, Fig. 5 presents how
pit before yielding takes place outside the yield length. By looking at the average and local corrosion were implemented in the FE model for
the equilibrium of a uniaxially loaded bar with varying corrosion levels bar I of the DN1 beam.
and a corrosion pit with maximum corrosion level (𝜇𝑚𝑎𝑥 ), the corrosion As indicated in Fig. 5 (bottom), the cross-sectional area of the
level at the end of yield length, 𝜇𝑝 , can be calculated as: rebar elements were reduced along length 𝑙𝑦 at the location of pits
in the FE model. However, the yield length was not always equal to
𝜇𝑝 = 1 − (1 − 𝜇𝑚𝑎𝑥 )𝑓𝑢 ∕𝑓𝑦 (1)
an integer multiple of the element length, 𝑙𝑒 . Accordingly, the closest
where 𝑓𝑢 and 𝑓𝑦 are the ultimate and yield strength of the rebars and integer multiple (n) was chosen, and the ultimate and rupture strain
𝜇𝑚𝑎𝑥 is the maximum corrosion level of the pit: was modified to:

𝜇𝑚𝑎𝑥 = 1 − 𝐴𝑐𝑜𝑟𝑟.𝑚𝑖𝑛 ∕𝐴𝑖𝑛𝑖 (2) 𝜀𝑚𝑜𝑑 = 𝜀𝑖𝑛𝑖 𝑙𝑦 ∕𝑛𝑙𝑒 . (4)

5
M. Gotame et al. Engineering Structures 257 (2022) 114007

Fig. 5. Example of how the yield length (𝑙𝑦 ) was estimated using the 3D scanning data of bar I of DN1 beam, plus how this was implemented in the FE model. 3D scanned data
shows the cross-sectional area of the rebar before and after corrosion. 𝐴𝑖𝑛𝑖𝑡𝑖𝑎𝑙 and 𝐴𝑐𝑜𝑟𝑟 are cross-sectional areas of rebar before and after corrosion, 𝐴𝑐𝑜𝑟𝑟.𝑚𝑖𝑛.𝑎 , 𝐴𝑐𝑜𝑟𝑟.𝑚𝑖𝑛.𝑏 and 𝐴𝑐𝑜𝑟𝑟.𝑚𝑖𝑛.𝑐
are the minimum cross-sectional areas and 𝐴𝑝.𝑎 , 𝐴𝑝.𝑏 and 𝐴𝑝.𝑐 are cross-sectional areas of a rebar at the end of yield penetration, corresponding to Pit a, b and c respectively.

where 𝜀𝑖𝑛𝑖 is the initial strain of the rebars. 3.7. Iterative scheme
A number of models were analysed to study the effects of the num-
ber of pits and how close those pits were to each other in the different The NLFEA of RC beams were done by applying deformation-
rebars. Furthermore, the authors investigated how adding ‘‘neighbour- controlled loading, as shown in Fig. 2. For the reference beam, the
ing’’ elements to the pit (thus increasing the yield length) affects the deformation was applied in 500 steps of 0.05 mm followed by 400 steps
deformation capacity of the beam. These neighbouring elements were of 0.1 mm. For the analyses of deteriorated (DN1) and strengthened
assigned the cross-sectional area of 𝐴𝑝 . A detailed explanation of the (DG1 and DC1) beams, the deformation was applied in 250 steps of
incorporation of the corrosion damages may be found in [39]. 0.02 mm followed by 500 steps of 0.05 mm. The BFGS (Broyden,
Fletcher, Goldfarb, and Shanno) method was used to perform equilib-
3.6. Modelling of FRP
rium iteration [33]. The maximum number of iterations was set to 400
The FRPs were modelled as 3D plane stress (3D membrane) quadri- to achieve convergence. As suggested by Hendriks, an energy norm of
lateral elements (Q12GME elements). Isotropic material properties 0.001 and a force norm of 0.01 were implemented as tolerance limits
were assigned to the GFRP sheet, CFRP plate and inclined U-jackets, for convergence in analysis of RN1 and DN1 beams [34]. However, for
while orthotropic material properties were assigned for the unidi- the analysis of beams under the DC1 category, the energy norm limit
rectional vertical U-jackets, see Table 3. The stiffness in the weak was set to 0.01 to achieve the convergence.
direction, x-direction in Fig. 6, of the vertical U-jackets was assumed
to be the same as the epoxy which was used between the CFRP plate 4. Results and discussion
and the concrete. The linear stress–strain relationship of the FRP was
implemented in FE analyses as a linear elastic material with a brittle This chapter presents the results of the FE analyses with different
failure when reaching the tensile strength. The tensile strength for the modelling choices. The flexural behaviour of four beams (RN1, DN1,
FRP material was calculated by multiplying the elastic modulus (𝐸𝑓 ) DG1 and DC1) are presented and compared with the test results. The
by the rupture strain (𝜀𝑓 𝑢 ). effects of pitting corrosion on the deformation and ultimate capacity
Fig. 6 presents the meshing of the FRPs for the DG1 beam, with of the RC beams are also presented, in terms of load versus mid-
the different geometries indicated by arrows. The average element size span deflection curves. In addition to the tested beams, the results of
of the FRP was set to 20 mm to align with the concrete. However, a analyses with different modelling choices (see Table 1), are presented to
mesh size of 5 mm was chosen for CFRP plate in beam category DC1,
study the effectiveness of the FRP strengthening methods in improving
as recommended by Kalfat [23].
the ultimate load and deformation capacity.
The interface properties between FRPs and concrete were mod-
elled as 3D structural plane interface (Q24IF elements). The non-linear
4.1. Behaviour in bending
interface properties were modelled by the bi-linear bond stress–slip
relationship proposed by Lu et al. [28], characterised by 𝑠0 = 0.0611 mm
and 0.07 mm, 𝑠𝑓 = 0.149 mm and 0.182 mm, 𝜏𝑚𝑎𝑥 = 4.71 MPa and The flexural behaviour of four beams from experiments and FEA
5.74 MPa for GFRP-concrete and CFRP-concrete interfaces respectively. are shown in Fig. 7 and Table 4. The FE analyses were able to capture
The same bond stress–slip relationships were considered for the con- the same failure mode as in the experiments, see Table 4. Failure due
nection between U-jackets to concrete and U-jackets to CFRP plate. to concrete crushing after yielding of tensile reinforcement caused the
However, for the DG1 beam, one additional analysis was carried out; flexural failure of the reference beam, see Fig. 8. In the FE analysis
assuming linear properties corresponding to the stiffness in the first of the reference beam, crushing of the concrete was localised to one
segment of the bi-linear relationship. element row, which agreed with the assumed localisation zone, see

6
M. Gotame et al. Engineering Structures 257 (2022) 114007

Fig. 6. Mesh for FRPs in the model of the DG1 beam.

Table 3
Dimensions and material properties of FRP composites [32].
FRP composites width × thickness 𝐸𝑓 a 𝐸𝑓 𝑥 b 𝐸𝑓 𝑦 c 𝐺𝑥𝑦 d 𝜀𝑓 𝑢 e 𝜈f
[mmxmm] [GPa] [GPa] [GPa] [GPa] [%] [mm]
GFRP laminate (isotropic) 150 × 3.03 20.2 – – – 1.82 0.2
CFRP plate (isotropic) 100 × 1.45 214 – – – 1.27 0.2
Inclined U-jacket (isotropic) 100 × 2.39 57.6 – – – 1.24 0.2
vertical U-jackets (orthotropic) 100 × 0.797 – 7.1 57.6 5 1.24 0.2
a 𝐸 – isotropic modulus of elasticity.
𝑓
b𝐸
𝑓 𝑥 – modulus of elasticity in x-direction.
c𝐸
𝑓 𝑦 – modulus of elasticity in y-direction.
d
G𝑥𝑦 – shear modulus.
e
𝜀𝑓 𝑢 – ultimate tensile strain at rupture.
f
𝜈 – Poisson’s ratio.

Fig. 8. As shown in Fig. 9, in the non-strengthened deteriorated beam, Table 4

Ultimate load capacity and failure modes of all beam specimens in experiments (EXP)
rupture of the bar at the left pit in the second (II) bottom rebar caused
and FEA.
a brittle failure (FE model DN1_FEA). A brittle failure due to rupture
𝐹𝑢 a [kN] 𝛥𝑢 b [mm] Failure mode
of the GFRP sheet was found in DG1 beam, see Fig. 10 (FE model Specimens
EXP FEA Ratio EXP FEA Ratio EXP and FEA
DG1_FEA). Furthermore, the DC1 beam failed due to a brittle anchorage
failure at the end of the CFRP plate, see Fig. 11 (FE model DC1_FEA). RN1 77.7 88.8 1.14 46.7 53.0 1.13 CCc
DN1 41.5 50.8 1.22 16.3 17.9 1.10 RSd
It is worth noting that the ultimate load capacity of the reference
DG1 135.8 131.0 0.96 20.1 22.8 1.13 RGe
beam (RN1) was 88.8 kN in the FE analysis, which is 14% larger than DC1 170.8 168.0 0.98 16.4 14.6 0.89 ACf
the test result. This difference was carefully investigated. Analytical
a
𝐹𝑢 - total ultimate load applied.
calculations were done to calculate the cracking, yielding, and the b
𝛥𝑢 – corresponding deflection at mid-span of beam.
ultimate load of the reference beam. These agreed well with the FE c
CC – failure due to concrete crushing.
results, see Fig. 7. A number of possible reasons for the discrepancies d
RS – rupture of steel rebar.
to the test results were investigated by confirming: e RG – rupture of GFRP sheet.

f AC – anchorage failure of CFRP plate.

• beam height and placement of the reinforcement bars against test
photos.
• force capacity for the bottom reinforcement bars with additional
material tests. predict the same failure mode as in the test; in other words, the rupture
• calibration of load cells used in beam and reinforcement testing. of the pitted rebars. The analysis including only general corrosion was
developed from the RN1 beam by incorporating only general corrosion
However, no explanation was found for the discrepancies. The analyses levels corresponding to DN1 beam measured by Yang [14] at mid-
of the strengthened beams agreed better with the tests results. Thus, 900 mm region (model DN1_GC_FEA). In this analysis, the load capacity
when the FRPs were the main component providing stiffness for the was decreased compared to the reference beam, but the deformation
beams, the results matched the tests better. capacity was not influenced, see Fig. 13.
The crack patterns at a mid-span deflection of 9.39 mm were By including the pitting corrosion, the load capacity was further
compared, based on DIC measurements during experiments and from decreased. Similar load capacity was found in all analyses, including
FE analyses, see Fig. 12. As the cracks in the FE analyses were localised pitting corrosion. However, the deformation capacity was found to be
into one element row, the assumed crack bandwidth was validated. highly sensitive to the modelling of pitting corrosion, see Fig. 13. This
In RN1 beam, the cracks were well distributed over the mid-900 mm was studied by trying different modelling choices for the corrosion
region, both in the experiment and the FE analysis. In the damaged pits. It was considered important to find a balance between having
beam, the crack width increased and cracks were localised to the mid- a simple and practical modelling approach for the corrosion pits and
section in-between the loading plates. Both in FE analyses and tests, the to have a response that was close to the test results. Three different
application of FRPs suppressed the crack opening effectively, thereby modelling approaches were examined for modelling the pitting corro-
reducing both the number and width of cracks. sion. It was found that the analysis with three major pits in each tensile
rebar agreed best with the test results (model DN1_FEA). This analysis
4.2. Effects of corrosion displayed a higher deformation capacity compared to the other ones.
This was because it avoided premature failure of the pits in the rebars
The effects of general and pitting corrosion were studied by compared to the model with only one major pit at a critical section
analysing the beam of category DN1. As described, the analysis of the (models DN1P_FEA and DN1PE_FEA). Model DN1PE_FEA was similar
damaged beam (including the general and pitting corrosion) was able to to model DN1P_FEA but the yield length was extended by reducing

7
M. Gotame et al. Engineering Structures 257 (2022) 114007

Fig. 7. Load versus mid-span deflection and FE analyses results compared with the experimental results..

Fig. 8. Concrete compressive stress vs strain at one of the critical integration points in the analysis of RN1 beam. All elements in a row along the 𝑦-direction failed in the crushed
state: (a) Stress vs strain at the critical integration point associated with the concrete element indicated in figure (b), (b) Failure of the RN1 beam due to concrete crushing and
localisation of concrete crushing into one element row. The test photo shows crushing of concrete at failure.

Fig. 9. Stress versus strain for tensile bar II at the critical pit of DN1 beam from FE analyses: (a) Stress versus strain at left pit, indicating failure of the beam, (b) Reinforcement
element with high strain which initiated failure of the beam due to rupture. The experimental photo shows failure due to rupture of corroded tensile rebar.

8
M. Gotame et al. Engineering Structures 257 (2022) 114007

Fig. 10. Strain development for DG1 beam: (a) The strain versus mid-span deflection for GFRP sheet at an integration point corresponding to rupture at mid-section plus FE results
compared to the test results, (b) Rupture of GFRP sheet from experiment and FE analysis.

Fig. 11. (a) End slip versus mid-span deflection of CFRP plate in DC1 beam from FE analyses, (b) High slip values at the end of CFRP plate, indicating anchorage failure as in
experiment. The test photo shows slip of the CFRP plate plus inclined U-jackets towards the centre of the beam.

Fig. 12. Comparison of crack pattern from tests (DIC results) and FE analyses. E1 is the main tensile strain. Shown here are the RN1 - reference beam; DN1 - damaged but
non-strengthened beam; DG1 - strengthened with GFRP sheet and U-wraps; and DC1 - strengthened with CFRP sheet and U-wraps..

the cross-sectional area of one neighbouring element on both sides happening simultaneously, see Fig. 13. It may be noted that in these
of major pits on rebar I and II of DN1 beam. The FE analyses of analyses, pitting corrosion was present only where the bending moment
the DN1PE_FEA model resulted in that the rupture of the rebars not was constant, and thus, the weakest section determined the capacity.

9
M. Gotame et al. Engineering Structures 257 (2022) 114007

the CFRP plate was plotted against the mid-span deflection to see
the utilisation of CFRP plate in each strengthening configuration, see
Fig. 15(b). Moreover, the ultimate load-capacity and corresponding
mid-span deflection are presented in Table 1. The application of CFRP
plate at the beam soffit increased the load capacity of the damaged
beam from 50.8 kN to 119 kN. This further increased to 160 kN after
applying inclined U-jackets at the end of CFRP plate on the soffit, see
Table 1.
For the beam in category DC1, the CFRP plate had relatively high
stiffness compared to the U-jackets and acted as the main load-carrying
member. Therefore, the behaviour of the DC1 beam was less influenced
by the connection between the CFRP plate and U-jackets, compared to
the beams in category DG1. Thus, non-linear interface properties were
successfully implemented. The stiff CFRP plate had brittle anchorage
failure, limiting the utilisation of its high tensile strength. The inclined
U-jackets delayed the anchorage failure of the CFRP plate by providing
extra anchorage to its end. With a mesh size of 5 mm, it was possible to
capture the stress concentrations in the interface at the end of the CFRP
Fig. 13. Load deflection curves for the DN1 beam with different corrosion modelling, plates. In an additional analysis of DC1, equivalent to DC1_FEA but with
compared to test results. Shown here are the RN1_FEA - analysis of reference beam;
DN1_GC_FEA - reference beam modified to damaged beam with general corrosion
a coarse mesh for the CFRP, high stress concentrations resulted in a
corresponding to DN1 beam; DN1PE_FEA - analysis with extended yield length; premature anchorage failure of the CFRP plate. The utilisation of the
DN1P_FEA - analysis with only one major pit in each bar at a critical section; DN1_FEA - CFRP plate was 28% for the beam with CFRP plate only. This increased
the analysis with three major pits in each tensile rebar; DN1_EXP - experimental results to 41% after applying inclined U-jackets at the end of CFRP plate. It is
of DN1 beam.
interesting to note that applying intermediate U-jackets had negligible
influence on load and deformation capacity for the beam with CFRP
plate at soffit. This might be because the CFRP plate has high stiffness
For load cases with varying bending moment, it may be necessary to compared to U-jackets, and corrosion cracks were localised only in mid
include more corrosion pits to ensure that the behaviour is described zone. In other words, corrosion cracks did not extend all the way in the
correctly. studied beam. This may also be taken to mean that the inclined and
vertical U-jackets were not mobilised effectively in the experiments. For
4.3. Flexural strengthening with GFRP sheet
cases with corrosion-induced cracks present in a larger part of the span,
the vertical U-jackets may be of larger importance. It can be noted that
When analysing the beams strengthened with GFRP (category DG1),
in analyses of such cases, it will be important to properly describe the
it was found that applying GFRP plate at the beam soffit increased the
effect of corrosion-induced cracks on bond and anchorage [13]. This
load capacity of the damaged beam to 94.5 kN. This further increased to
requires further research.
111 kN after inclined U-jackets were applied at the end of GFRP sheet
on the soffit, see Table 1. The failure modes changed depending on
4.5. Discussion
how many FRP components were applied. When only the GFRP sheet
was applied, the failure was caused by concrete separation due to high
The structural behaviour of strengthened beams with corroded re-
tensile forces at the ends of the sheet (model DG1S_FEA). Meanwhile,
applying the inclined U-jackets changed the failure mode to anchorage inforcement is complex, and thus, analyses must include a sufficient
failure (model DG1I_FEA). The utilisation of the GFRP sheet was 54% level of complexity. In the analyses presented, modelling of pitting
for model DG1S_FEA; this increased to 71% after applying inclined U- corrosion, was important. Further, the interface properties, between
jackets at the end of GFRP sheet, with non-linear interface properties FRP-to-concrete and between the FRPs, influenced to some extent.
between FRP connections (model DG1I_FEA). Modelling of corrosion-induced cracking did not influence the results
The stiffness of the beams under the DG1 category was affected to any major extent in most of the analyses; this was concluded as
significantly by the stiffness of the U-jackets and also the interface the presence of vertical U-jackets did not make any major difference.
properties between the GFRP sheet and the U-jackets. When the in- However, modelling of the corrosion-induced cracks most likely af-
termediate vertical U-jackets were modelled to the GFRP sheet with fected the results in the analysed case exhibiting concrete separation as
non-linear interface properties (see Section 3.6), premature failure of failure mode (analysis DG1S_FEA). Further, as shown experimentally,
the beam occurred due to anchorage failure of the GFRP sheet (model intermediate U-jackets contribute for other configurations [8]. Thus,
DG1NLI_FEA). This did not correspond with the experimental results. it is considered important to include the effect of corrosion-induced
Thus, non-linear interface properties between the inclined U-jackets cracks in analyses of strengthened beams with corroded reinforcement.
and the GFRP sheet might be conservative. However, when linear Further, the effect of corrosion-induced cracks on bond and anchorage
interface properties were used, the failure mode changed to GFRP was not of importance in the analysed cases but should be included
sheet rupture. In other words its strength was fully utilised (model when that is critical [13].
DG1_FEA), consistent with the experimental results. The assumption
of linear interface properties between the inclined U-jackets and the 5. Conclusions
GFRP sheet was accepted due to; (i) the interface properties from Lu
et al. [28] was considered to be too conservative, as it is meant for a In this work, finite element analyses were carried out to study
connection between concrete and FRP, not between FRPs, (ii) relatively corrosion-damaged RC beams strengthened with externally bonded FRP
low slip values at the ends of the GFRP sheet (see Fig. 14). composites. From this work, the following conclusions were drawn:

4.4. Flexural strengthening with CFRP plate • Non-linear finite element analysis were able to predict the struc-
tural behaviour of damaged and strengthened RC beams. The FE
Fig. 15(a) shows the load versus mid-span deflection curves for analyses were useful in studying the effects of the different parts
different FRP strengthening alternatives. Furthermore, the strain in of the applied FRP composites.

10
M. Gotame et al. Engineering Structures 257 (2022) 114007

Fig. 14. (a) Load versus mid-span deflection curves for DG1 beam with different strengthening alternatives. (b) Strain at an integration point corresponding to the critical section
of GFRP sheet vs mid-span deflection. Shown here are the DG1_EXP - experimental result; DG1_FEA - strengthened with GFRP sheet, inclined U - jackets and vertical U - jackets
with linear interface properties between vertical U - jackets and GFRP sheet; DG1NLI_FEA - strengthened with GFRP sheet, inclined U-jackets and vertical U-jackets with non-linear
interface properties between vertical U-jackets and GFRP sheet; DG1I_FEA - strengthened with GFRP sheet and inclined U-jackets; DG1S_FEA - strengthened with GFRP sheet; and
DG1N_FEA - analysis of non-strengthened beam.

Fig. 15. (a) Load vs mid-span deflection curves for DC1 beam with different strengthening alternatives. (b) Strain at an integration point corresponding to mid-span section
vs mid-span deflection. Shown here are the DC1_EXP - experimental result; DC1_FEA - strengthened with CFRP plate, inclined U-jackets and vertical U-jackets; DC1IU_FEA -
strengthened with CFRP plate and inclined U-jackets; DC1P_FEA - strengthened with CFRP plate; and DC1N_FEA - analysis of non-strengthened.

• The pitting corrosion majorly influenced the load and deforma- of the GFRP-strengthened beam than the interface properties
tion capacity, while the average corrosion had a lesser effect on between the CFRP plate and U-jackets.
the load capacity and did not influence deformation capacity.
This work forms the basis for FE analyses of corrosion-damaged
• Modelling of the corrosion pits was highly influential and crucial
concrete beams strengthened with FRP, and was validated with well-
in predicting the deformation capacity and failure mode of the
defined experiments. The developed modelling technique may be used
beam. However, the ultimate load remained unaffected regardless
to further evaluate the effectiveness of strengthening methods in pa-
of how the pits were modelled.
rameter studies, including beams of varying dimensions, corrosion
• Strengthening significantly increased the ultimate load capacity
patterns and levels, and spacing and dimensions of FRP. In the current
and stiffness of the corrosion-damaged RC beams. The application study, corrosion-induced cracks were located in mid-region only, and
of FRP effectively suppressed the opening of cracks. even if the cracks were up to 1.9 mm wide, no spalling was included.
• For the studied beam geometry and corrosion damage, strength- It would be interesting to analyse specimens with corrosion-induced
ening with intermediate vertical U-jackets did not further improve cracks extending up to the beam supports and investigate whether this
the flexural performance of the damaged beams with CFRP plate would increase the demand for intermediate vertical U-jackets. For
at the beam soffit. practical applications, it is also worth noting that detailed knowledge of
• The interface properties between FRP-to-concrete and between the corrosion pits is of major importance to proper modelling. However,
the FRPs in the FE model were important in accurately predicting this information is very difficult to obtain for existing structures. Thus,
the load-carrying capacity and failure mode of the strengthened a reliable and efficient damage detection technique, providing accurate
beams. The interface properties between the GFRP sheet and U- information about pitting corrosion, would be of major benefit in
jackets were more influential in predicting the flexural behaviour assessing existing structures.

11
M. Gotame et al. Engineering Structures 257 (2022) 114007

CRediT authorship contribution statement [14] Yang J, Haghani R, Blanksvärd T, Lundgren K. Experimental study of FRP-
strengthened concrete beams with corroded reinforcement. Constr Build Mater
2021;301:124076. http://dx.doi.org/10.1016/j.conbuildmat.2021.124076.
Milan Gotame: Conceptualization, Methodology, Software, Vali-
[15] Biondini F, Vergani M. Deteriorating beam finite element for nonlinear analysis
dation, Formal analysis, Investigation, Data curation, Writing – origi-
of concrete structures under corrosion. Struct Infrastruct Eng 2015;11(4):519–32.
nal draft, Visualization. Carl Lindqvist Franklin: Methodology, Soft- http://dx.doi.org/10.1080/15732479.2014.951863.
ware, Validation, Formal analysis, Investigation, Data curation. Mat- [16] Belletti B, Damoni C, Hendriks MA, de Boer A. Validation of the
tias Blomfors: Methodology, Writing – review & editing, Supervision. Guidelines for Nonlinear Page 2 of 129 Finite Element Analysis of Concrete
Jincheng Yang: Conceptualization, Methodology, Writing – review Structures. June, 129. 2017, https://homepage.tudelft.nl/v5p05/RTD%201016-
3C(2017)%20version%201.0%20Validation%20of%20the%20guidelines%20for%
& editing, Supervision. Karin Lundgren: Conceptualization, Method-
20NLFEA%20of%20RC%20structures%20Part%20Slabs.pdf.
ology, Validation, Writing – review & editing, Supervision, Project [17] Mathern A, Yang J. A practical finite element modeling strategy to capture
administration, Funding acquisition. cracking and crushing behavior of reinforced concrete structures. Materials
2021;14(3):1–26. http://dx.doi.org/10.3390/ma14030506.
Declaration of competing interest [18] Kodur VK, Bhatt PP. A numerical approach for modeling response of
fiber reinforced polymer strengthened concrete slabs exposed to fire. Com-
pos Struct 2018;187(November 2017):226–40. http://dx.doi.org/10.1016/j.
The authors declare that they have no known competing finan- compstruct.2017.12.051.
cial interests or personal relationships that could have appeared to [19] Blomfors M, Zandi K, Lundgren K, Coronelli D. Engineering bond model for
influence the work reported in this paper. corroded reinforcement. Eng Struct 2018;156:394–410. http://dx.doi.org/10.
1016/j.engstruct.2017.11.030.
[20] Earij A, Alfano G, Cashell K, Zhou X. Nonlinear three–dimensional finite–
Acknowledgments element modelling of reinforced–concrete beams: Computational challenges and
experimental validation. Eng Fail Anal 2017;82(September):92–115. http://dx.
This work was financially supported by the Swedish Transport doi.org/10.1016/j.engfailanal.2017.08.025.
[21] Hawileh RA, El-Maaddawy TA, Naser MZ. Nonlinear finite element modeling of
Administration [grant number BBT-2018-011]. The computations were concrete deep beams with openings strengthened with externally-bonded com-
performed on resources at the Chalmers Centre for Computational posites. Mater Des 2012;42:378–87. http://dx.doi.org/10.1016/j.matdes.2012.06.
Science and Engineering (C3SE) provided by the Swedish National 004.
Infrastructure for Computing (SNIC). [22] Hawileh RA. Nonlinear finite element modeling of RC beams strengthened with
nsm FRP rods. Constr Build Mater 2012;27(1):461–71. http://dx.doi.org/10.
1016/j.conbuildmat.2011.07.018.
References [23] Kalfat R, Al-Mahaidi R. Numerical and experimental validation of FRP patch
anchors used to improve the performance of FRP laminates bonded to concrete. J
[1] Al-Mahaidi R, Kalfat R. Introduction. In: Rehabilitation of concrete structures Composit Construct 2014;18(3):1–10. http://dx.doi.org/10.1061/(asce)cc.1943-
with fiber-reinforced Polymer, no. 2004. Matthew Deans; 2018, p. 1–5. http: 5614.0000437.
//dx.doi.org/10.1016/b978-0-12-811510-7.00001-x. [24] Cui S, Wang J, Shi L, Liu Y. Test and FEM analysis of debonding failure of
[2] Zhou Y, Gencturk B, Willam K, Attar A. Carbonation-induced and chloride- RC beam strengthened with CFRP. J Reinf Plast Compos 2009;28(17):2151–60.
induced corrosion in reinforced concrete structures. J Mater Civ Eng http://dx.doi.org/10.1177/0731684408091932.
2015;27(9):04014245. http://dx.doi.org/10.1061/(asce)mt.1943-5533.0001209. [25] Wu YF, Wang Z, Liu K, He W. Numerical analyses of hybrid-bonded FRP strength-
[3] Fernandez I, Herrador MF, Marí AR, Bairán JM. Structural effects of steel ened concrete beams. Comput-Aided Civ Infrastruct Eng 2009;24(5):371–84.
reinforcement corrosion on statically indeterminate reinforced concrete members. http://dx.doi.org/10.1111/j.1467-8667.2009.00596.x.
Mater Struct/Mater Construct 2016;49(12):4959–73. http://dx.doi.org/10.1617/ [26] Freddi F, Savoia M. Analysis of FRP-concrete debonding via boundary integral
s11527-016-0836-2. equations. Eng Fract Mech 2008;75(6):1666–83. http://dx.doi.org/10.1016/j.
[4] Chen E, Berrocal CG, Fernandez I, Löfgren I, Lundgren K. Assessment of the engfracmech.2007.05.016.
mechanical behaviour of reinforcement bars with localised pitting corrosion by [27] Lu XZ, Ye LP, Teng JG, Jiang JJ. Meso-scale finite element model for FRP
digital image correlation. Eng Struct 2020;219(May):110936. http://dx.doi.org/ sheets/plates bonded to concrete. Eng Struct 2005;27(4):564–75. http://dx.doi.
org/10.1016/j.engstruct.2004.11.015.
10.1016/j.engstruct.2020.110936.
[5] Lundgren K. Bond between ribbed bars and concrete. Part 1: Modified model. [28] Lu XZ, Teng JG, Ye LP, Jiang JJ. Bond-slip models for FRP sheets/plates
bonded to concrete. Eng Struct 2005;27(6):920–37. http://dx.doi.org/10.1016/j.
Mag Concr Res 2005;57(7):371–82. http://dx.doi.org/10.1680/macr.2005.57.7.
engstruct.2005.01.014.
371.
[29] Chen GM, Chen JF, Teng JG. On the finite element modelling of RC beams
[6] Al-Mahaidi R, Kalfat R. Methods of structural rehabilitation and strengthening.
shear-strengthened with FRP. Constr Build Mater 2012;32:13–26. http://dx.doi.
In: Rehabilitation of concrete structures with fiber-reinforced polymer, no. 2004.
org/10.1016/j.conbuildmat.2010.11.101.
Matthew Deans; 2018, p. 7–13. http://dx.doi.org/10.1016/b978-0-12-811510-
[30] Anıl Ö, Durucan C, Emin M, Özge Baeğmez K, Kara ME, Baeğmez O. Nonlinear
7.00002-1.
three-dimensional FE analyses of RC beams retrofitted using externally bonded
[7] Hawileh RA, Rasheed HA, Abdalla JA, Al-Tamimi AK. Behavior of reinforced con-
CFRP sheets with or without anchorages. Composites Part B: Engineering
crete beams strengthened with externally bonded hybrid fiber reinforced polymer 2017;31(7):770–86. http://dx.doi.org/10.1080/01694243.2016.1232040.
systems. Mater Des 2014;53:972–82. http://dx.doi.org/10.1016/j.matdes.2013.
[31] Hawileh RA, Musto HA, Abdalla JA, Naser MZ. Finite element modeling of rein-
07.087. forced concrete beams externally strengthened in flexure with side-bonded FRP
[8] Al-Saidy AH, Al-Jabri KS. Effect of damaged concrete cover on the behav- laminates. Composites B 2019;173(May):106952. http://dx.doi.org/10.1016/j.
ior of corroded concrete beams repaired with CFRP sheets. Compos Struct compositesb.2019.106952.
2011;93(7):1775–86. http://dx.doi.org/10.1016/j.compstruct.2011.01.011. [32] Yang J. Strengthening reinforced concrete structures with FRP composites [Ph.D.
[9] Al-Saidy AH, Saadatmanesh H, El-Gamal S, Al-Jabri KS, Waris BM. Structural thesis], Chalmers University of Technology; 2021, https://research.chalmers.se/
behavior of corroded RC beams with/without stirrups repaired with CFRP sheets. publication/522501.
Mater Struct/Mater Construct 2016;49(9):3733–47. http://dx.doi.org/10.1617/ [33] TNO DIANA. DIANA 10.4. users manual 2020. In: DIANA FEA. 2020, URL
s11527-015-0751-y. https://dianafea.com.
[10] Bergström M. Assessment of existing concrete bridges. Bending stiffness as a [34] Hendriks MA, de Boer A, Belletti B. Guidelines for nonlinear finite element
performance indicator [Ph.D. thesis], Luleå University of Technology; 2009, p. analysis of concrete structures: Girder members - report RTD:1016:2012. 2012.
248, http://urn.kb.se/resolve?urn=urn:nbn:se:ltu:diva-26096. [35] Rots JG. Computational modeling of concrete fracture. 1988, http://resolver.
[11] Triantafyllou GG, Rousakis TC, Karabinis AI. Effect of patch repair and strength- tudelft.nl/uuid:06985d0d-1230-4a08-924a-2553a171f08f.
ening with EBR and NSM CFRP laminates for RC beams with low, medium and [36] Cornelissen H, Hordijk D, Reinhardt H. Cornelissen, hordijk, reinhardt - 1986.pdf.
heavy corrosion. Composites B 2018;133:101–11. http://dx.doi.org/10.1016/j. 1986, p. 45–56, Heronjournal.Nl, URL http://heronjournal.nl/31-2/6.pdf.
compositesb.2017.09.029. [37] Feenstra PH. Computational aspects of biaxial stress in plain and reinforced
[12] Blomfors M, G. Berrocal C, Lundgren K, Zandi K. Incorporation of pre-existing concrete. [Ph.D. thesis], Delft University of Technology; 1993, http://resolver.
cracks in finite element analyses of reinforced concrete beams without trans- tudelft.nl/uuid:faf2fd16-1c43-4711-b783-9e8e00d10c21.
verse reinforcement. Eng Struct 2021;229:111601. http://dx.doi.org/10.1016/j. [38] Vecchio FJ, Collins MP. Compression response of cracked reinforced concrete.
engstruct.2020.111601. 1995, http://dx.doi.org/10.1061/(ASCE)0733-9445(1995)121:7(1153.x).
[13] Blomfors M, Lundgren K, Zandi K. Incorporation of pre-existing longitudinal [39] Franklin C, Gotame M. FE analyses of strengthened concrete beams with corroded
cracks in finite element analyses of corroded reinforced concrete beams fail- reinforcement [M.Sc. thesis], Chalmers University of Technology; 2021, https:
ing in anchorage. Struct Infrastruct Eng 2020;1–17. http://dx.doi.org/10.1080/ //hdl.handle.net/20.500.12380/302745.
15732479.2020.1782444. [40] FIB. Fib model code for concrete structures 2010. 2010.

12
M. Gotame et al. Engineering Structures 257 (2022) 114007

[41] Wittmann FH, Rokugo K, Brühwiler E, Mihashi H, Simonin P. Fracture en- [43] Jiradilok P, Nagai K, Matsumoto K. Meso-scale modeling of non-uniformly
ergy and strain softening of concrete as determined by means of compact corroded reinforced concrete using 3D discrete analysis. Eng Struct
tension specimens. Mater Struct 1988;21(1):21–32. http://dx.doi.org/10.1007/ 2019;197(April):109378. http://dx.doi.org/10.1016/j.engstruct.2019.109378.
BF02472525.
[42] Lundgren K. Bond between ribbed bars and concrete. Part 2: The effect of
corrosion. Mag Concr Res 2005;57(7):383–95. http://dx.doi.org/10.1680/macr.
2005.57.7.383.

13