Shear strengthening of precast prestressed I-girders using carbon fiber reinforced polymers and in-fill concrete blocks

Due to the rapid increase in traffic volumes, existing infrastructure demands proper assessment and in some cases rehabilitation is needed. Prestressed concrete (PC) I-shaped girders are widely used in prefabricated girder bridges. With sequel generations of shear design provisions their shear capacity might need to be increased in the framework of assessing safety of aging bridges. Earlier attempts in shear strengthening this specific type of girder with fiber reinforced polymers (FRP) has shown early debonding along the concrete surface at the web-flange corner so that the increase in shear strength was very limited. This study presents the results of full scale PC I-girders strengthened with CFRP through epoxy bonded concrete in-fill blocks that locally transforms the I-section to a rectangular section at the position of CFRP strips. This transformation eliminated the early debonding of CFRP shear reinforcement and for the studied configurations a 30% – 40% increase in shear capacity was obtained with respect to the unstrengthened control specimen, without and with additional CFRP spike anchors, respectively. In comparison, similar configurations without in-fill blocks reached 5% and 12% increase in shear capacity, respectively. As such, the use of in-fill blocks appeared more effective than local anchorage of the CFRP at the web-flange corner. Analytical verification of the experimental results confirmed that existing formulations lack ability to predict the shear capacity of strengthened


| INTRODUCTION
Traffic demand in terms of volume and loading is proportionally increasing with time and requires upgrading of existing transportation infrastructure, which is often Discussion on this paper must be submitted within two months of the print publication.The discussion will then be published in print, along with the authors' closure, if any, approximately nine months after the print publication.decades old. 1 A significant part of this infrastructure relates to precast I-girders.In addition to the increased traffic, degradation of such bridges can be related to damages due to the impact of vehicles and cyclic loading, poor design/detailing, and aging of the bridge structure.Subject to all mentioned reasons, concrete bridges require proper assessment and repair/rehabilitation to ensure the long run serviceability.As part of the assessment, it might also be required to account for sequel generations of code provisions, among which more reliable/improved shear provisions.
The American Road and Transportation Builders Association estimated that one out of every three bridges needs repair or replacement which comprises almost 36% of bridges in the United States.Of those aging bridges, more than 43,500 are rated in poor condition and can take up to 30 years to repair. 2 Similarly, in the local context of Europe, for instance in the Netherlands there are approximately 70 existing girder bridges, comprising of post tensioned girders and transversely prestressed thin deck slabs, that require assessment. 3This number is specific only for the mentioned bridge type.
Shear failure is never a favorable design failure mode.Therefore, shear strengthening is often needed if the internal shear reinforcement is not sufficiently provided due to improper design or if it is corroded, the member is strengthened for flexure or in order to fulfill the requirements of latest shear design provisions that are becoming more reliable over the years.
There are various techniques and materials used for the repair and rehabilitation of structural elements among which the most popular include externally bonded reinforcement (EBR), 4,5 near surface mounted (NSM) 6 reinforcement, textile reinforced mortar (TRM, also known as carbon concrete). 7The choice of a repair technique for a specific project considers performance, cost, and duration of service disruption during the repair process.The use of externally bonded fiber reinforced polymer (FRP) materials qualifies this demand and has gained popularity especially during the last two decades.A few of the merits include high strength to weight ratio, easy application and chemical/corrosion resistance. 8In case of bridge structures its use, mostly by means of carbon fiber reinforced polymers (CFRP) has been defined for compensation of increased traffic volumes, dampening of vibrations, mitigation of corrosion damage, and correction of deficient designs. 9hen FRP sheets or strips are applied over the web of an I-girder, either vertical to the length of the girder or perpendicular to the expected shear crack, the bonded sheets or strips are susceptible to early debonding at the web-flange corner.1][12][13][14] The straightening of the CFRP strip acting in tension results in high-stress concentrations inducing early debonding.This results in strong underutilization of CFRP strength and the increase in shear capacity is limited.One of the ways to eliminate this early debonding is to add concrete in-fill blocks, as used in the present study, to give a rectangular shape to the I-girder.This solution has been previously used in 4 for secondary shear strengthening when the girder was tested for flexure.
Only a limited number of studies are available on shear strengthening of prestressed concrete (PC) Igirders.A state-of-the-art overview is provided in. 15Earlier studies, to postpone early debonding of I-shaped shear strengthening, have used different types of anchorages such as horizontal CFRP sheets over the vertical/ diagonal sheets, 10,12,14,16,17 mechanical anchors, 12 CFRP spike anchors, 13,18 and CFRP strips embedded in a groove cut at the web-flange corner. 16][21][22] The reported increase in shear strength of PC Igirders using horizontal CFRP strips as anchorage over the vertical/diagonal CFRP strips is not consistent and controversial.The benchmark study on scaled PC Igirders of existing bridges in Canada 10 has utilized different CFRP materials along with different strengthening configurations with horizontal sheets as anchorage.The efficiency in terms of increase in shear capacity for different configurations is reported in the range of 9%-36%.Another study 16 on decommissioned bridge girders employed similar configurations with horizontal CFRP sheets as anchors and reported 15%-27% increase in shear strength.However, no increase in shear strength is reported in 12 on very similar AASHTO I-girders.The variable effectiveness of vertical CFRP strips along with horizontal CFRP strips as anchorage is indicated in 14 in terms of crack width control and stiffness, yet, no increase in the shear capacity was reported.On the other hand, a proof load test (not up to failure) depict the efficiency of CFRP strengthening using vertical and horizontal CFRP sheets on a decommissioned bridge I-girder damaged by vehicle impact in Canada. 17hree different mechanical anchoring schemes were used in, 12 however, none of them was reported to significantly increase the shear capacity of the member, given local failure modes of anchorages and CFRP sheets.The combination of vertical and horizontal CFRP strips along with so-called spike anchors is reported in 13 as the best solution, with a shear strength increase up to 38% for the studied configuration.However, this configuration requires a lot of drilling for fixing the spike anchors and its suitability in the case of thin web I-girders is an open question.Similarly, the fixation of mechanical anchorages as in 12 required a lot of drilling in the thin webs of PC I-girders, creating weak discontinuity regions from where the shear crack can easily be triggered due to local stress concentrations.Similar observations are made in, 16 where, vertical/diagonal CFRP strips were anchored inside a groove cut along the web-flange corner.
A 38% increase in shear strength of a PC I-girder has been reported in 11 with a CFRP strip spacing less than half the effective depth of the studied PC I-girder, yet, no increase was observed when the spacing was greater than half the effective depth.The effect of CFRP configuration as continuous sheet versus strips is reported in, 13 where, a very similar shear strength increase has been reported in both cases.No debonding has been reported in 23 for a PC I-girder whose web height was small compared to the width of the flanges, with 11% increase in the shear capacity for the considered configuration.As there is no standard shape/size for PC I-girders, results for a certain cross-section might not hold true for another one with relatively different shape/size.However, in USA or Canada I-shaped cross sections for use in the bridge construction are highly standardized.
Nonlinear finite element analysis (FEA) was carried out in 24 to supplement the experimentation given in 12 focused on three dimensional interface element to capture the debonding failure and to analyze the interface behavior.Another study 25 reported nonlinear FEA based on the experimental validation of 11 discussing the effect of various parameters related to the shear strengthening of I-girders using CFRP.Additionally, several analytical or (semi-) empirical models [26][27][28][29] are given in various researches and FRP design guidelines for predicting CFRP shear contribution, nevertheless, their predictions show a lot of variations. 30,31o obtain further understanding with respect to the above points, in this work a large scale experimental campaign is reported on 700 mm high PC I-girders strengthened using four different configurations of CFRP shear reinforcement.The effect of strengthening along the I-shape is considered in comparison to strengthening along a traditional shape by using epoxy bonded concrete in-fill blocks on the web of the I-girders.Additionally, the effect of CFRP spike anchors in controlling the debonding on the mentioned strengthening configurations is also investigated.Lastly, the suitability of analytical formulations such as in FRP design guidelines are explored to predict the contribution of CFRP shear reinforcement.

| EXPERIMENTAL PROGRAM
The experimental program consisted of five large scale shear tests on PC I-girders provided by a precast company located in Lier, Belgium.Four tests employed different shear strengthening configurations using externally bonded CFRP strips and one test was carried out as a reference (control) without external strengthening in order to quantify the relative improvement in shear strength.After application of the strengthening configurations, PC I-girders were monotonically loaded until failure.The aim of the study was to develop and test the novel practical/efficient solution to shear strengthen PC I-girders using CFRP strips, epoxy bonded concrete in-fill blocks and CFRP spike anchors, thus to eliminate the early debonding of CFRP shear reinforcement around the webflange corner of the I-section.

| Test specimen
The PC I-girder test specimen was specially designed in order to have enough margin for shear strengthening and to ensure shear failure mode even after the application of the shear strengthening system.A typical I-shaped crosssection as shown in Figure 1 was selected and the reinforcement and prestressing was designed in collaboration with the precast company.The I-shape cross-section has a height of 700 mm and width of 240 mm at the flanges and 70 mm at the web.The longitudinal reinforcement consisted of 8 Ø 15.2 mm strands in the tension zone and two similar strands in the compression zone, where, all the strands were prestressed to 50% (930 MPa, 129 kN each strand) of their tensile strength (as defined by the supplier = 1860 MPa).This atypical prestressing level was needed to introduce extra flexural reinforcement for ultimate limit state, while maintaining acceptable strain levels at service load and to allow for the specific test configuration as shown later in Figure 5. Single leg stirrups having Ø 6 mm were provided (alternating in both directions) at a spacing of 180 mm.The total length of each PC I-girder test specimen was 10 m.After curing for a month at the precast company's yard, the specimens were transported to Ghent University, Magnel-Vandepitte Laboratory for strengthening and testing.

| Test configurations and preparations
Four CFRP shear strengthening configurations were designed for this experimental study.Figure 2 shows the schematic details of each configuration along with the specimen designations.The first configuration 'I-NA' is the one having CFRP sheets bonded along the geometry of the PC I-girder with no anchors (NA).Second configuration 'I-SA' is similar to 'I-NA' with the only difference that continuous CFRP spike anchors (SA) were provided passing through both the top and bottom web-flange corners of the I-girder.The third configuration, designated as 'U-NA', has concrete in-fill blocks epoxy-bonded to the web on both sides to render a rectangular shape to the I-girder locally at the positions where CFRP strips have to be bonded.The last configuration 'U-SA' is similar to 'U-NA' with the only difference that spike anchors were provided at the mid height of the top flange (in the compression zone).
The exploded view of the anchored configuration I-SA and U-SA is shown in Figure 3. Similar details apply to unanchored configurations I-NA and U-NA except for the CFRP spike anchors.
The first step was the grinding of the concrete substrate of the test specimens, in order to remove the cement laitance and to roughen the concrete.The bottom corners of each specimen were also ground to give a radius of approximately 30 mm.Concrete in-fill blocks having a width of 100 mm were cast in a specifically designed mold and cured for a month before bonding.Afterwards, the bottom surface of the concrete in-fill blocks (to be bonded to web) and the top surface (where CFRP strips have to be bonded) were ground in a similar way as for the PC I-girders.Concrete in-fill blocks were bonded using a thixotropic two component structural epoxy adhesive. 32The blocks were pressed against the web of the I-girder using temporary clamps, so that the excessive epoxy rolls out at the sides, resulting in a very thin bond interface thickness of 1-2 mm.The aim was to align the in-fill concrete blocks as close as possible with the girder flanges.Indentations due to too thin concrete blocks would lead to early debonding, while, projection of too thick in-fill blocks would lead to early rupture of CFRP strips due to stress concentrations around the bulged edges of the in-fill blocks.The clamps were F I G U R E 2 Carbon fiber reinforced polymers (CFRP) shear strengthening configurations.
removed after 12 h of bonding and the epoxy adhesive was cured for minimum of 7 days.During the curing period of the epoxy, other operations such as drilling of the holes for spike anchors, bonding of the CFRP strips and fixing of the spike anchors were carried out.For the anchored configurations I-SA and U-SA holes of Ø14 mm were drilled as the CFRP spike anchor has a Ø10 mm.In case of I-SA, the depth of the anchor hole was all through the width of the thin part of the web (70 mm).Similarly, in case of U-SA the depth of the anchor hole was equal to 85 mm as shown in Figure 4.
Unidirectional CFRP sheets-400 g/m 233 was cut into strips of width 75 mm.To apply the CFRP strips to the specimen two types of epoxy were used.First the CFRP strips were adhered to the concrete applying the same epoxy as for the concrete in-fill blocks, 32 next the sheets were fully impregnated with high-performance epoxy 34 from the top.After filling the anchor holes with injectable epoxy, 35 thoroughly saturated CFRP spike anchors with high-performance epoxy 34 were fixed into the holes.For I-SA, the spike anchors were spread in opposite directions referred as two-way spike anchors, where, the density of the material is equally divided into two halves.For U-NA, the anchors were only spread in one direction thus referred as one-way spike anchors.The details of anchors are given in Figure 4.

| Material properties
The PC I-girders were cast at the precast factory using high-strength self-compacting concrete having maximum aggregate size of 16 mm (crushed aggregate).The concrete in-fill blocks were cast in the lab using a more traditional concrete mix having maximum aggregate size of 16 mm (rounded aggregate).The results of the compressive strength (average of three cubes with side length 150 mm, tested according to NBN EN 12390-3), flexural and splitting tensile strength (average of three prisms of 100 Â 100 Â 400 mm 3 for in-fill blocks and 150 Â 150 Â 600 mm 3 for PC I-girder, tested according to NBN EN 12390-5 and NBN EN 12390-6, respectively) are given in Table 1.The age of the I-girders and the reference test specimens (cubes and prisms) was approximately 7 months when tested.
The mechanical properties of the steel reinforcement (average of three samples, tested according to NBN EN  15630-19) and CFRP strips (values as provided by the manufacturer) are given in Table 2.

| Test setup and instrumentation
A schematic of the test setup used is shown in Figure 5. Two shear tests were performed on each girder, considering that 2.5 m of the girder length was overhanging on the support away from the load.Thus, the second test zone was out of the influence of the applied loading during the first test.After completing the first test, the girder was rotated and the already tested region was placed overhanging on the support away from the load.The adjacent longer span was externally strengthened using Ø16 mm high-strength threaded rebars and steel channel sections at a spacing of 500 mm.This was done to ensure the failure within the CFRP strengthened test region and to restrict the opening of the secondary shear cracks in the adjacent longer span.The load was applied using two hydraulic jacks of capacity 500 kN coupled together over a steel profile (HEM-160).The tests were executed in a stroke control way at constant rate of 0.4 mm/min.The shear span to effective depth ratio (a/d = 1800/615) for all the tests was taken nearly equal to three.As such the angle of an imaginary shear crack between the edge of the support plate and the edge of the loading profile equals 23 .
The detail of various sensors used to capture the response of test specimen is shown in Figure 6.Displacement measurement under the load and over the supports was carried out using linear variable displacement transducers (LVDTs).For local strain measurements over the CFRP strips, nine strain gauges (designated as SG1 to SG9) were used at different locations (at mid height and at web-flange corner) in the region of the anticipated shear crack.For global strain measurements, three dimensional digital image correlation (3D DIC) was used on the side opposite to the one equipped with the other sensors.Further, strain gauge based strain sensors (designated as SS1 to SS4) of length 200 mm were used to monitor the shear deformation and crack openings in between the CFRP strips.

| TEST RESULTS AND OBSERVATIONS
Test results and observations in terms of load-deflection behavior, ultimate load, failure aspect, development of shear cracks, and strain variation in the CFRP strips are discussed in the following.

| Load-deflection behavior and ultimate load
Point load versus deflection under the load is given in Figure 7a.In order to have a more closer look in Figure 7b, each curve is also plotted at an offset of 5 mm.The loads at first shear crack and at ultimate, along with the corresponding deflections are listed in Table 3 and the relative comparison is shown in Figure 8.
It can be seen that the specimen I-NA exhibited almost the same behavior as the control specimen, with only a small (5%) increase in shear capacity.This is due to the initiation of debonding of CFRP strips at the web-flange corner (just after the initiation of shear cracking) that continued further with increase in crack widths and further cracking at higher loads.A similar behavior was shown by specimen I-SA in terms of early debonding, however, the presence of anchors controlled/delayed it slightly resulting in limited (12%) increase of the shear capacity.The efficiency of the spike anchors remained limited as discussed later in the section on debonding of CFRP strips.
The specimens with concrete in-fill blocks exhibited better performance of the CFRP shear reinforcement, where, an increase of 31% and 40% in shear capacity was noticed for specimens U-NA and U-SA, respectively.For the same CFRP shear strengthening reinforcement ratio (and 8% less material in length because of the U-shape), the extra gain in the shear capacity is pronounced.This demonstrates the feasibility and positive effect of bonded concrete in-fill blocks in changing the shape of the CFRP shear reinforcement to a more conventional U-shape thus resolving the problem of early debonding associated with the complex shape of PC I-girder.
The increase in deflection under the point load for each strengthening configuration is also listed in Table 3. Results are basically in line with observed increases in peak loads, yet increase in deflection is more pronounced than increase in peak load.

| Failure aspect
Figure 9 shows the crack pattern at peak load from DIC analysis as major principle strains (e 1 ), as well as the photos of the failure aspect after the end of the tests.The failure of the control PC I-girder was typically a web shear failure, where, the crack appeared at the mid height that later extended further towards the support and point load.Before the failure, the crack was fully developed over the shear span exhibiting the yielding of the internal steel stirrups.The failure was brittle involving rupture of internal steel stirrups in the shear span as well as disintegration of concrete in the compression zone as shown in Figure 9a.
The failure mode of strengthened PC I-girders varied with respect to the strengthening configuration used.For specimen I-NA and I-SA, the loading was stopped once the beam showed multiple load drops with excessive deflections and crack openings (subject to safety of the instrumentation).This can be clearly seen in the load deflection curves of these specimens.At that instant, CFRP strips being debonded become redundant and further increase in the load is not expected.For specimen I-SA, local failure in the CFRP strips was observed near the peak load around multiple spike anchors due to the stress concentration.The crack pattern for both of these specimens is similar as shown in Figure 9b,c.
The failure of specimen U-NA occurred due to the sudden debonding of the CFRP strips followed by rupture of the steel stirrups as shown in Figure 9d.In case of specimen U-SA, where the CFRP strips were anchored using the CFRP spike anchors in the compression zone, the failure triggered due to the rupture of the middle CFRP strip close to the bottom corner as shown in Figure 9e.After the rupture of the primary CFRP tie, the surrounding ties and the internal steel stirrups could not redistribute the applied load resulting in a brittle failure.The more prominent strain values (red zones in Figure 9c-e) represent stress concentrations at the location of the shear cracks and to a lesser extent where the CFRP bridges the shear crack.These prominent values are however also located at the edges of the CFRP strips where local debonding is occurring.Note that for larger crack widths or debonding the DIC stops correlating, so that in the proximity of some of the wider cracks strain values are missing as can be seen in Figure 9. Therefore, the actual strain in the direction of the fibers of the CFRP strips are given later in Section 3.5 (Figure 15).

| Debonding of CFRP strips
All the strengthened specimens demonstrated different degrees of debonding of CFRP shear reinforcement before actual failure.An insight of visual observations on debonding made during the different tests is given in Figure 10.In case of specimen I-NA, the CFRP strip in the middle debonded heavily between the two cracks as can be seen in Figure 10a.The debonding started at the web-flange corner after occurrence of the shear crack and extended further with the occurrence of the second shear crack at the mid height.The other strips (2nd and 4th) debonded also at the web-flange corner to a relatively less degree.In specimen I-SA, with the anchors at the webflange corner, the debonding mainly occurred in between the anchored length as can be seen in Figure 10b.At and after the ultimate load, the shear crack continued to widen with successive load drops corresponding to the loosening of the CFRP strips around the spike anchors.
For the specimens with epoxy bonded concrete in-fill blocks U-NA and U-SA, shear cracks extended in the concrete in-fill blocks and only slight debonding was observed at the periphery of a few CFRP strips close to failure, as can be seen in Figure 10c,d.The debonding displacement (w) just before failure, expressed as out of plane displacement captured using 3D DIC is shown in Figure 11.It can be seen that in both the specimens I-NA and I-SA the middle strip is heavily debonded with 15 mm out of plane displacement.The adjacent strips were also considerably debonded at the web-flange corner as the shear crack passed through this zone.This corresponds to the observations mentioned above in Figure 10.For specimens U-NA and U-SA only slight out of plane debonding displacement was noticed even at higher loads corresponding to the better performance of CFRP shear reinforcement over the adapted U-shape.

| Development of shear cracks
The load values at which the first shear crack appeared are listed in Table 3 and comparatively shown in Figure 8.It can be seen that for all the specimens the shear cracking started in the similar range of load, that is, 337-376 kN.This shows that the applied strengthening configurations could not delay the (first) shear cracking showing the passive nature of CFRP shear reinforcement before concrete cracking.
The crack pattern illustrated as a plot of major principal strains at peak load from DIC is shown in Figure 9.It can be seen that the control specimen failed with the growth of only one major shear crack, however, the strengthened configurations showed two major parallel shear cracks.All the major shear cracks are highlighted in Figure 9 with dotted black lines and the corresponding inclination angles are given in Table 3.The range of shear crack angles was between 20 and 29 degrees for all of the test specimens, that are in line with Model Code 2010 36 shear provisions for PC beams.With such shallow crack angels, almost all of the CFRP strips contributed to the shear resistance, however, the three in the middle contributed more compared to the two at the ends that were only engaged once the web shear crack is fully developed into the flanges of the beam.Minor flexural cracks were also observed in all the tests as can be seen in the plots of major principal strains (e 1 ) in Figure 9.
The crack width and crack slip for each specimen was measured using DIC as well as strain sensors (designated as SS1 to SS4 in Figure 6).The record of strain sensors was used only for validation of DIC virtual sensor's F I G U R E 1 0 Debonding of carbon fiber reinforced polymers (CFRP) strips at the given percentage of peak load (PL).location and magnitude.DIC virtual sensors perpendicular to the crack were used for measuring crack width and similar sensors across (but in the direction of) the crack were used for measuring crack slip as shown in Figure 12.The plot of crack width and crack slip with respect to the load are shown in Figure 13.It can be seen that the crack width (Figure 13a) of specimen I-NA and I-SA is very similar to that of the control specimen, where, the crack width continued to increase without considerable increase in the load.The specimens U-NA and U-SA showed much smaller crack widths compared to the other specimens even at higher loads, corresponding to the more effective shear crack bridging effect of the CFRP strips bonded through the concrete in-fill blocks.A similar trend can also be seen in Figure 13b, in case of crack slip.

| Strain variation in CFRP shear reinforcement
For all the strengthening configurations, the development of strains in CFRP strips is plotted at different load levels until failure in Figure 14.The location of strain gauges in the zone of anticipated shear crack is shown in Figure 6.However, the plotted strain values were taken on the opposite side of I-girder using DIC, so as to correlate with the other DIC observations given in this study (note that the strain gauges showed a very similar trend compared to DIC at the strain gauge locations).The strain plots in Figure 14 show the maximum values observed along each CFRP strip.Before the shear cracking (for loads <300 kN) no change in strains were observed, hence not plotted.
For specimen I-NA, it can be seen that small strains were developed after shear cracking in the CFRP strips, and underwent only limited variation after 400 kN due to the debonding of strips.For specimen I-SA, thanks to the spike anchors, almost double CFRP strains are observed in two strips at 400 kN compared to I-NA.Beyond 400 kN these strains tend to reduce with further load increase and corresponding debonding of CFRP strips.Close to the failure, the strains in strip 3 again started to increase showing the local contribution of the spike anchors.For specimens U-NA and U-SA, significantly higher strains are observed, equal to 7122 and 9574 με, respectively.The strain plots show regular trend and proportional increase with respect to the loads.This confirms the effectiveness of CFRP strips bonded to the I-section through concrete in-fill blocks.
Figure 15 shows the strain contours along the CFRP strips at peak load.These figures give clear idea of the strip regions actively contributing in shear capacity enhancement.In specimen I-NA strains could not be developed subject to early debonding of the CFRP strips.In specimen I-SA, only strip 3 in the middle showed some contribution even after debonding due to the spike anchors, while strips 2 and 4 could not contribute as the major shear crack passed through the anchor holes.However, for specimen U-NA and U-SA the three strips in the middle are fully active in contributing to the shear capacity.The strips on the sides are also active only in the region of shear cracking.In case of specimen U-SA, it can be seen that the regions of spike anchors show less strains subject to increase in the density of CFRP material due to thickness of the spike anchor.The bottom of the strip 3 showed maximum strains where it actually ruptured.

| ANALYTICAL PREDICTION OF CFRP SHEAR CONTRIBUTION
The analytical prediction for the shear resistance of the concrete and the internal stirrups is not given here, however, a detailed overview of standard models is given in 37 in the context of PC I-girders.For verification of the CFRP shear reinforcement contribution, the models given in FRP design guidelines such as fib Bulletin 90, 27 fib Bulletin 14, 26 ACI 440, 28 and TR 55 29 have been used only for configurations with concrete in-fill blocks, consistent with a U-shape configuration.To the knowledge of the authors, there is only one model available 38 for calculating the shear contribution when CFRP is bonded along an I-shape as in configurations I-SA and I-NA.The basic expression in all these guidelines is given by Equation 1.
where, A fw is the sectional area of two legs of CFRP strips bonded at a spacing of s f over the effective depth of girder equal to d.The elastic modulus of CFRP is E f , while the effective strain in the strips is given as ε fe .The inclination of the shear crack is represented by θ and the angle of the FRP fibers with respect to the length of the beam is represented by α (90 in this study).
The two unknown parameters in this expression, that is, the shear crack angle and the effective stress/strain are needed to evaluate the shear contribution of CFRP.The recommended shear crack angle is 45 according to fib Bulletin 14, 26 ACI 440, 28 TR 55, 29 and Hutchison. 38fib Bulletin 90 allows the choice of shallow shear crack angle, thus, 25 was selected as suggested by MC2010 36 for prestressed beams.28][29]38 The possible contribution of concrete in-fill blocks were not considered, given the tensile strength of concrete is not comparable to that of CFRP.The anchored configuration U-SA was considered equivalent to the fully wrapped system, where CFRP fracture governs.The only exception was fib Bulletin 90, where, for the anchored systems, the effective stress is defined as maximum 90% of the fully wrapped system.The given partial safety factors and the strain limits were not considered for calculations so as to estimate the physical CFRP shear contribution.
F I G U R E 1 5 Contours for strain (in the direction of the fibers, e yy ) on fiber reinforced polymers (FRP) shear reinforcement at peak load.
The predicted results and their comparison with the experimental values are listed in Table 4. Here, the given experimental strain values (ε f, exp ) represent the average of the maximum strain values observed over each strip at peak load (as shown in Figure 14).For anchored configuration U-SA, the most close conservative prediction of shear contribution is made by fib Bulletin 14, while, fib Bulletin 90 and ACI 440 overestimated the CFRP shear contribution.This overestimation resulted from the use of shallow shear crack angle in case of fib Bulletin 90 and very high CFRP strains (equal to 0.75 ε fu , considering anchored configuration as fully wrapped) for ACI 440.For the configuration without anchors U-NA, ACI 440 predicted the shear contribution accurately, while, it is over and underestimated by fib Bulletin 90 and fib Bulletin 14, respectively.A more or less reasonable prediction was obtained by TR55 for the anchored configuration (U-SA), however, in case of the unanchored configuration (U-NA) the margin of underestimation is very high.The semi-empirical model of Hutchinson is based on the strain observations of configurations with horizontal strips as anchorage, therefore, only the shear contribution of anchored configuration along I-shape (I-SA) is given for comparison.The predicted value is quite reasonable based on the small magnitude of CFRP shear contribution.Note that trends in predicting the shear capacity (V f, pre /V f, exp ) are generally not well in line with predictions of the effective strain (ε f, pre /ε f, exp ).

| DISCUSSION
The test results show a clear advantage of combining CFRP shear strengthening with epoxy bonded concrete in-fill blocks as well as spike anchors in enhancing the shear capacity of PC I-girders by eliminating the problem of early debonding (or straightening) of CFRP strips at the web-flange corner.The change in shape of the I-section to a rectangular section as for specimens U-NA and U-SA, increased the shear as well as the deflection capacity of the PC I-girder.
The development of shear cracking in case of specimens strengthened along the I-section (I-NA and I-SA) resulted in debonding at the web-flange corner, particularly for the three strips in the middle.At the web-flange corner (also called the re-entrant corner) the CFRP strip has a reverse curvature along the geometry.Once the shear crack passes this zone, this reverse curvature is pulled out of the substrate leading to early debonding.Similarly, if the crack passes through the center of the strip away from the re-entrant corner (as in case of middle strip), the CFRP is strained initially but later the strain transmission takes place towards the reverse curvature region where the debonding occurs.During this phenomenon of strain transmission, if another shear crack appears (as in case of specimen I-NA shown in Figure 11), the debonding is more easily triggered due to the availability of less bond length.Thus, shear strengthening along the I-shape is always inefficient and does not suit the characteristic tensile properties of the CFRP material.
The performance of CFRP anchors is greatly dependent on the location over the PC I-girder.In case of specimen I-SA, the spike anchors at the web-flange corner restricted the degree of debonding of the CFRP strips.Figure 16a shows the shear crack passing through the anchor hole, because of which the anchorage losses its stiffness and the CFRP strip is pulled out leading to stress concentration at the anchor.A similar configuration tested in a study 13 also showed no significant increase in the shear strength.Drilling in the relatively thin concrete web of I-girders, for anchoring externally bonded FRP reinforcement, indeed endangers web weakening and can attract the crack to run through the drilled hole.It might be considered to avoid such configurations, e.g. by using an in-fill concrete block and anchoring in the compression flange.
A comparison between specimen U-NA and U-SA shows the combined positive effect of the rectangular geometry (because of concrete in-fill blocks) as well as the CFRP spike anchors in the compression zone (top flange).For specimen U-SA, compared to U-NA an extra increase of 9% is associated to the spike anchors.Important to note, is not only a 9% increase but also the change of failure mode from CFRP debonding to CFRP tensile rupture (at least for one of the CFRP strips).
Another important aspect to consider is the contribution of the epoxy bonded concrete in-fill blocks to the shear capacity of the specimens U-NA and U-SA.The appearance of the first shear crack U-NA and U-SA in comparison to the control and other specimens without the in-fill blocks shows no significant difference.For all the specimens, the first shear crack appeared within the similar load range between 337 and 376 kN. Figure 16b shows two parallel web shear cracks passing through the bonded in-fill block.This shows that the in-fill blocks do not have a considerable contribution to the shear capacity of the I-girders.Though, the in-fill blocks could potentially act as bonded shear ties, the shear cracking load was too high in comparison to the capacity of these ties.As a result they started to crack in tension once the shear crack appeared.This happened without any noticeable debonding of the in-fill blocks from the concrete web.
It is observed that all the applied CFRP strengthening configurations did not provide any beneficial effect in enhancing the shear crack initiation load.This indicates, as generally observed in externally bonded FRP applications, that the first component which resists the applied shear is only the concrete cross-section itself.After the shear cracking of the concrete cross-section, a part of the applied shear is retained by aggregate interlocking, dowel action of longitudinal reinforcement and the uncracked concrete in the compression zone.The rest of the applied shear is transferred to the internal steel stirrups and the externally bonded CFRP strips.This demonstrates the passive nature of internal steel stirrups as well as the externally bonded CFRP strips before first shear cracking.In contradiction to this observation, another study 13 reported 20%-50% increase in the shear cracking load in case of vertical and horizontal CFRP sheets and spike anchors.Though, there were no strains observed in the vertical CFRP strips before shear cracking, the reported increase may probably be due to the intrinsic variability of the different test specimens.
In shear strengthening applications, the spacing of the externally bonded CFRP strips has a direct influence on the shear contribution.Specifically for PC I-girders, a study 11 reported that the shear strength increase is not possible when the CFRP spacing is greater than half the effective depth of the specimen.This observation (and tentative conclusion) came from experiments with strengthening along the I-shape.The question can be raised if this is due to the low efficiency of I-shaped shear strengthening, rather than the spacing as such.Indeed, for all the configurations used in the current study, the spacing was somewhat more than half the effective depth of the specimen (0.6d) and significant differences in effectiveness were found in I-NA and I-SA versus U-NA and U-SA.
The models used for analytical predictions do not show a consistent trend for different configurations (anchored and unanchored).This shows that the choice of shear crack angle and the effective stress/strain value is not so obvious.Until more specific shear design provisions become available, specific to CFRP strengthened PC I-girders, a design by testing approach should be recommended.

| CONCLUSIONS
This study on shear strengthening of PC I-girders using CFRP strips, concrete in-fill blocks and CFRP anchors has shown the following aspects: • The shear strength increase in PC I-girders with CFRP shear reinforcement along the I-shape lacks effectiveness.Alternative strengthening configurations should be used to postpone or avoid the early debonding at web-flange corners.• The epoxy bonded concrete in-fill blocks proved to be efficient in eliminating the early debonding by transforming the I-section to a rectangular section.For this study, an increase of 31% has been achieved with a CFRP and concrete in-fill blocks configuration.The same configuration also with additional spike anchors resulted in a 40% increase in the shear capacity.• The presence of CFRP spike anchors at the web-flange corner could not avoid the debonding, rather limited debonding deformations, and the increase in shear capacity was limited.These web through spike anchors lacked effectiveness because of stress concentrations in the CFRP strip around the anchor hole as well as the weakening of the web which attracted the crack to run through the drilled hole.As a result, the stiffness of these anchors is reduced further.• The CFRP spike anchors in the compression flange were effective in delaying the CFRP debonding of the Ushaped configuration with concrete in-fill blocks.This forced, for the tests in this study, the failure mode from debonding to rupture of the CFRP strips.As such an additional 9% increase in shear capacity was observed.• Analytical models given in FRP design guidelines are currently unable to give accurate predictions compared to the experimental results reported in this study.

F
I G U R E 3 Exploded view of anchored configurations.F I G U R E 4 Geometric details of spike anchors in (a) I-SA and (b) U-SA (mm).

F
I G U R E 6 Detail of instrumentation used.F I G U R E 7 Load deflection curves at (a) origin and (b) an offset of 5 mm.

F I G U R E 8
Comparison of first cracking load, ultimate load, and percentage increase in ultimate load.F I G U R E 9 Crack pattern (left, maximum principal strains from digital image correlation [DIC] at peak loads) and failure aspect (right) of specimens, dotted lines indicate the inclination of major shear cracks.

F
I G U R E 1 1 Maximum out of plane displacement (debonding) before failure, captured by 3D DIC.F I G U R E 1 2 Location of digital image correlation (DIC) virtual sensors for measuring crack width and crack slip.

F
I G U R E 1 3 Progression of shear crack (a) crack width and (b) crack slip.F I G U R E 1 4 Maximum carbon fiber reinforced polymers (CFRP) strains (in the direction of the fibers, e yy ) at different load levels (with respect to digital image correlation [DIC] side).

F I G U R E 1 6
Shear crack passing through (a) anchor hole in specimen I-SA and (b) concrete in-fill block in specimen U-SA.

Christoph
Czaderski is a Senior Researcher at Empa, Swiss Federal Laboratories for Materials Science and Technology, Dübendorf, Switzerland.His research interests include strengthening of concrete structures with (prestressed) fiber reinforced polymers and with iron based shape memory alloys.He is a member of the fib Task Group5.1 'FRP (Fibre Reinforced Polymer) Reinforcement for Concrete Structures'.Email: christoph.czaderski@empa.ch.Stijn Matthys is a Professor at Magnel-Vandepitte Laboratory, Department of Structural Engineering and Building Materials, Ghent University, Ghent, Belgium.His research interests include repair and strengthening of (concrete) structures, advanced composite reinforcement, structural behaviour of reinforced and prestressed concrete, damage diagnostics, durability, repair and monitoring, accidental load cases such as fire and explosion, prefabricated concrete, technologies for durable building materials and techniques among which circular concrete.He is the chair of the fib Task Group 5.1 'FRP (Fibre Reinforced Polymer) Reinforcement for Concrete Structures'.Email: stijn.matthys@ugent.be.How to cite this article: Yaqub MA, Czaderski C, Matthys S. Shear strengthening of precast prestressed I-girders using carbon fiber reinforced polymers and in-fill concrete blocks.Structural Concrete.2023;24(2):3091-108. https://doi.org/10.1002/suco.202200439 Strength of concrete used for specimens.
T A B L E 1 Abbreviation: PC, prestressed concrete.*Test prisms size 150 Â 150 Â 600 mm 3 .**Test prisms size 100 Â 100 Â 400 mm 3 .T A B L E 2 Mechanical properties of steel reinforcement and CFRP strips (experimental determined values for strands and stirrups, CFRP values from technical data sheet).Abbreviation: CFRP, carbon fiber reinforced polymers.F I G U R E 5 Test setup.3096 YAQUB ET AL. 17517648, 2023, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/suco.202200439 by Paul Scherrer Institut PSI, Wiley Online Library on [09/05/2023].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Ultimate loads and deflections.
T A B L E 3Abbreviation: NA, unanchored configurations; SA, spike anchors.
T A B L E 4 Analytical predictions of CFRP shear contribution.