1 INTRODUCTION
Several approaches have been developed to strengthen existing reinforced concrete structure. Compared to the traditional method of steel platebonding reinforcement, the externally bonded carbon fiberreinforced polymer (CFRP) reinforcement has many advantages. ^{Kondratenko and Ding (2018}) investigated the destruction types of the CFRP reinforced concrete structures, and a new calculation method for preventing the delamination of the CFRP from concrete structures was proposed. ^{Mashrei et al. (2019}) employed grooving technique to improve the flexural capacity of CFRP strengthened beams, and it was found that CFRP with grooves could effectively delay the debonding failure of CFRP from the concrete surface. A variety of studies have focused on the mechanical properties and performances of existing structural members reinforced by CFRP. ^{Abbasloo and Maheri (2017}) studied the modal characteristics of CFRP composite and designed a method to easily analyze the modal properties of CFRPHoneycom sandwich panels. ^{Shukri et al. (2018}) investigated the mechanism of the interfacial crack debonding and introduced the momentrotation method to simulate the debonding of hybrid CFRP strengthen method. It was found that CFRP with higher elastic modulus had adverse effect on the interface of CFRP and concrete. ^{Jiang et al. (2015}) studied the prestressed CFRP reinforced RC beams, and proposed a formula for calculating the reliability of deflection and crack width by considering the prestressed conditions. In recent years, the CFRP reinforcement method have been increasingly used in bridge structures. The analysis of girder bridge reinforced by CFRP has been carried out by ^{Kohnehpooshi and Jaafar (2017}) and ^{Yang et al. (2017}). Flexural tests were conducted on the prestressed concrete bridge deck with CFRP tendons. The stress states and failure modes of the CFRP tendons were given. ^{Chen et al. (2018}) applied the CFRP in the bridge pier reinforcement and conducted the quasistatic tests. It was found that the loadcarrying capacity and antideformation ability of the bridge pier can be significantly improved by CFRP. ^{Hosseini et al. (2019}) performed the truckloading tests on an old roadway bridge by CFRP reinforcement. It was found that the mismatch of the CFRPsteel thermal expansion coefficient negative effect can be reduced at a low air temperature.
Since the CFRPconcrete interface bond strength is usually small, the interface layer is prone to debonding failure under large loads. Reasonable anchorage methods can effectively prevent or delay the occurrence of the debonding of CFRPconcrete interface (^{Ceroni and Pecce, 2010}). ^{AlMahaidi and Kalfat (2011}) made a beneficial attempt to increase the CFRPtoconcrete bond beyond the provision of effective bond length, which involved the use of additional CFRP material to expand the CFRP bonding area. ^{Galal and Sekar (2008}) adopted the mechanical fasteners to increase the CFRPtoconcrete bond strength, resulting in the improvement of CFRPtoconcrete bond strength. ^{Qazi et al. (2013}) analyzed specimens with anchorages of CFRP strip techniques from their failure modes and ultimate capacity. It was found that the CFRP anchorage measures had a positive effect on the overall performance of CFRP strengthened structures. ^{Wu et al. (2010}) investigated the intermediate crackinduced debonding of CFRP reinforced RC beams with CFRP enhancement, which demonstrated the effectiveness of CFRP in preventing intermediate crack debonding. ^{Mertoğlu et al. (2016}) conducted 14 concrete specimens by externally bonded CFRP. A modified bond slip model was proposed by comparing with the model proposed for system without anchoring. It can be concluded that the increasing application of dowel anchorages could significantly increase the bond stress of CFRPconcrete interface. Since the CFRPconcrete interface of the bridge roof is usually subjected to large bending moments and stresses, more emulative anchoring method is required to prevent the interface debonding, thereby enhancing the structural bearing capacity.
The bridge roofs are always subjected to the coupling of a variety of loads and environment conditions, such as the temperature effects, during their service life (^{Luo, 2018}). The experimental fatigue performance of CFRP plate reinforced under hotseawater environment and cyclic loads coupled action was investigated. Recent researches have investigated the mechanical properties of CFRP reinforced concrete structures under high temperature. ^{Hong et al. (2016}) and Wang et al. (2017) analyzed the longterm performance of concreteCFRP interface under the combined impacts of temperature and sustained loads. The basis of the durability design was proposed for the high strength concrete structures reinforced by CFRP. Several studies (^{Biosca et al., 2019}; ^{Ferrier and Agbossou, 2017}; ^{Lin et al., 2019}; ^{Lopez et al., 2003}) focused on the flexural and fatigue properties of concrete beams strengthened with CFRP plates under temperature field. It can be concluded that the high temperature condition has an adverse effect on the bearing capacity of interfacial debonding. However, most previous studies have not quantitatively investigated the crack development and mechanical properties of the CFRPreinforced bridge roof under high temperature, and little attention had been given to the reasonable CFRP anchoring schemes in a hot environment.
The interfacial debonding of CFRP plateconcrete is an inevitable problem in the reinforcement structures, which would lead to premature failure of CFRP plate at anchorage point. Recent years, several researches have been conducted on the interfacial debonding of CFRP plateconcrete. However, few studies focus on the effective anchoring method of CFRP plate reinforced bridge roof at high temperature. Aiming at closing the gap, we studied the interfacial debonding problem especially for the bridge roof with different CFRP anchoring measures, under high temperature effect of paving asphalt. A theoretical model was proposed for calculating the intermediate crack debonding bearing capacity of the CFRP reinforced bridge roofs. Based on the comparison with the test results, it can be found that the calculating model proposed in this paper is more suitable for the CFRP reinforced bridge roofs, compared to the models used in the previous literature.
This paper investigates the crack and mechanical behavior of CFRP platereinforced bridge roof under high temperature with different anchoring measures. Taking Bangabandhu bridge roof as the research background, the bridge roof reinforcement schemes with different CFRP anchoring measures were designed firstly. Six CFRPreinforced test beams were then tested under fourpoint bending loads. The crack propagation, midspan strain, deflection, and flexural capacity of the anchoring beams were comparatively analyzed. In addition, typical failure modes of the CFRP platereinforced concrete specimens were carefully studied, and the debonding bearing capacity and crack development features were then obtained. Finally, the calculation model of the intermediate crack debonding bearing capacity was proposed to quantify the debonding capacity of the CFRP reinforced bridge roofs.
2 CRACKS OF A BRIDGE ROOF AND THE CFRP REINFORCEMENT
The Bangabandhu Bridge serves as the connection between Central Asia and Northwestern Europe. Cracks appeared in the early stages of bridge construction, seventeen different types of longitudinal cracks were found on the bridge roof, which were mainly caused by the temperature and shrinkage stress. The crack morphology is shown in Figure 1.
A CFRP plate repair and reinforcement scheme was proposed to reduce and control the development of longitudinal cracks in the bridge roof deck. The main procedure included: 1) the epoxy resin was grouted into the cracks on the bridge deck; 2) the CFRP plates with a width of 100 mm and a thickness of 1.4 mm were applied horizontally in four lanes of the bridge to constrain the development of longitudinal cracks; 3) leveled the bridge deck and set the insulating layer and waterproof layer; and 4) paved the SMA wear resistant layer with a thickness of 5 cm. Figure 2 illustrates the deck paving scheme of the bridge roof.
3 EXPERIMENTAL PROGRAM
In order to study the debonding failure of the bridge reinforced by CFRP plates under high temperature asphalt paving, the high temperature and tests of CFRP platereinforced specimens were conducted. The debonding failure modes of CFRP plates with different anchoring were compared in detail. In the test, the high temperature in the asphalt pavement construction of Bangabandhu Bridge was simulated and measured to evaluate the influence of high temperature on the reinforcement effect of CFRP plates.
3.1 Temperature simulation and measurement
In order to simulate the high temperature effect on the CFRP plates during the asphalt paving process and the vehicle load effect, the specific fabrication sequence of the specimen was as follows: 1) the CFRP platereinforced specimens were cured for 3 to 7 days; 2) the surface of the specimens were coated with latex paint as a protective layer; 3) asphalt was paved on site for high temperature simulation and rolling (shown in Figure 3). Figure 4 shows the test site of the asphalt paving and rolling. It can be seen from Figure 4 that the maximum temperature of the asphalt surface layer reaches 172 °C. After 60 minutes, and the bottom of the asphalt surface returned to about 67 °C.
Figure 5 illustrates the temperaturetime curves of the CFRP plate in the test. It can be seen from the figure that the temperature on the upper and the lower sides of the CFRP plate changes significantly in the ascending section of the temperature curve, while the temperature curve in the descending section tends to be flat. The highest temperature of the upper and lower sides of the CFRP plate are respectively 72.7 °C and 64.8 °C, the temperature variations of the CFRP plates are relatively large.
According to the storage modulus curves proposed by ^{Cheng (2016}), the CFRP plateconcrete interfacial adhesion properties may change under high temperature. The surfacebonded CFRP reinforced beams could experience premature debonding failure in the high temperature environment. Thus, in this paper, the additional anchoring measures are taken on the CFRPreinforced specimens after the high temperature action, to ensure the good effect of the CFRP reinforcement.
3.2 Test scheme of bridge roof reinforcement
According to the Bangabandhu bridge roof beam, the CFRP platereinforced specimens was designed as 4.00 m in length, 0.65 m in width and 0.28 m in height. Figure 6 shows the summary of test specimens for investigation.
The specimens were divided into two types: the contrast specimens and the reinforcement specimens(Group A and Group B). Different reinforcement schemes were tested and analyzed for further study of the reinforcement measures with different CFRP plates. In order to compare the reinforcement effect of CFRP plates under different anchorage measures, six different anchorage schemes were adopted as additional anchorage measures for the reinforcement of specimens. The classification and specific parameters of specimens are shown in Table 1. The different schemes of additional anchoring measures of the specimens are shown in Figure 7. Figure 7(a) shows specimens in Group A without additional anchorage measures. Specimens additionally anchored by sticking strips at different positions are presented in Figures 7(b) to (d), and specimens with bolt anchorage measures at different positions are illustrated as Figures 7(e) to (g).
Grouping of specimens  Specimen number  Reinforcement measures with CFRP  Additional measure taken by anchoring the CFRP plate  Purpose  Figure number 

Contrast Specimen  FDBL  None  No additional anchorage measure  Beam for Comparison  / 
Group A  FX1  Three CFRP plates in the roof of the bridge transversely  No additional anchorage measure  For comparing the CFRP reinforced effect  Figure 7(a) 
FX2  
FX3  
Group B  TM1  Three CFRP plates in the roof of the bridge transversely  End sticking anchorage  For comparing the CFRP reinforced effect with sticking anchoring measures  Figure 7(b) 
TM2  End sticking anchorage & middle sticking anchorage  Figure 7(c)  
TM3  End sticking anchorage & bearing sticking anchorage  Figure 7(d)  
LM1  Three CFRP plates in the roof of the bridge transversely  End bolt anchorage  For comparing the CFRP reinforced effect with bolt anchoring measures  Figure 7(e)  
LM2  End bolt anchorage & middle bolt anchorage  Figure 7(f)  
LM3  End bolt anchorage & bearing bolt anchorage  Figure 7(g) 
The performance parameters of the CFRP plate, including tensile strength, tensile modulus and elongation at break, are shown in Table 2. The measurement performance parameters of the adhesive, including tensile strength, tensile modulus, flexural strength and compressive strength are shown in Table 3. Concrete with a compressive strength of 21.1 MPa and steel with a tensile strength of 360 MPa were adopted in the tests.
Thickness (mm)  Tensile strength (MPa)  Tensile modulus (GPa)  Elongation at break (%) 

1.4  23002400  150155  1.41.42 
Model  Tensile strength (MPa)  Tensile modulus of elasticity (MPa)  Bending strength (MPa)  Compressive strength (MPa) 

K801  2526  25002600  3032  7075 
A 30ton hydraulic jack new bridge was applied in the test specimens. Firstly, a load of 5 kN was applied to the test piece for preloading, and then the load per stage was increased by 20kN. Figure 8 presents the diagram of the test system. Figure 9 shows a representative specimen in the test rig. Figure 9 shows a representative specimen in the test rig. Figure 9 (a) presents the specimen before the bending test. Asphalt layer was paved on the specimens, and the high temperature effect of the asphalt acted on the CFRPreinforced specimen. After the bending test was completed, the asphalt surface layer was removed to facilitate the observation of the debonding of the interface between the CFRP plate and concrete. Figure 9(b) shows the specimen with asphalt layer partially removed after the bending test. The failure of the specimen during the test is characterized by the yielding of the steel, the fracture of the concrete and the peeling of the CFRP plates.
4 EXPERIMENTAL RESULTS AND ANALYSIS
4.1 Fracture morphology and crack propagation
Figure 10(a) presents the fracture morphological distribution of unreinforced Specimen FDBL. Figures 10(b) to (d) show the crack distribution of three specimens reinforced by CFRP plates. Figures 11(a) to (f) illustrate the fracture morphological distribution of the CFRP reinforced specimens with different additional anchoring measures. It can be seen from the figures 10 and 11 that the flexural main crack appeared first and was almost perpendicular to the tensile side of the member at the beginning of the formation and development of the crack. The root cracks were located near the position of the steel bars, mainly appeared at the moment when the member was close to failure.
As can be seen from Figures 10(a), the crack pitch and width in the purely curved portion of the comparative specimen FDBL were large. The crack shape of the comparative specimen FDBL was roughly triangular because it did not have a CFRP plate attached. The crack width of the concrete surface is the largest. As the crack height increases, the crack width gradually decreases.
For the specimens in Group A, the CFRP plate was adhered to the surface of specimen. Therefore, the interfacial bond shear stress between concrete and CFRP plate prevented the shrinking of the concrete under tension near the cracking section. The cracks present the shape of the jujube, as shown in Figures 10(b) to (d). The crack distribution of the specimen in Group B is characterized by a small crack pitch and a crack width in the purely curved region, as shown in Figure 11. It should be pointed out that the specimens TM3 and LM3 present flexural shear cracks in the vicinity of the bearing, as illustrated in Figures 11(c) and (f). As the load increases, the crack develops toward the compression zone of the concrete in the direction of the support.
Table 4 lists the crack distribution of the pure flexural region of the specimen. It can be seen from Table 4 that the ratio of the maximum crack spacing to the average crack spacing is between 1.33 and 2.92, and the ratio of the maximum crack spacing to the minimum crack spacing is between 1.09 and 1.55. The rising height of the main flexural crack in the pure flexural region of the specimen FDBL was 260.1 mm. Due to the enhanced constraint of the CFRP plate, the maximum height of the bending main crack of the Group A and Group B specimens is 245.8 mm, which is smaller than the maximum height of the bending main crack of the specimen FDBL.
Specimen number  Number of cracks  Maximum crack spacing

Minimum crack spacing

Mean crack spacing



Maximum rise height of cracks (mm) 

FDBL  5  183.6  137.7  168.2  1.09  1.33  260.1 
FX1  7  175.5  81.5  129.2  1.36  2.15  215.2 
FX2  8  146.0  78.6  102.2  1.43  1.86  232.2 
FX3  8  155.3  73.2  101.1  1.54  2.12  234.2 
TM1  6  174.6  95.2  127.0  1.37  1.83  235.2 
TM2  8  141.5  73.2  105.2  1.35  1.93  234.8 
TM3  7  144.8  65.5  115.0  1.26  2.21  245.8 
LM1  6  151.9  77.8  123.0  1.23  1.95  241.5 
LM2  6  150.0  77.8  115.6  1.30  1.93  215.0 
LM3  8  158.3  54.2  102.4  1.55  2.92  237.2 
Figure 12 shows the variation of the loadcrack curves of the specimens. The tensile stress unreinforced specimen (Specimen FDBL) is transmitted only by the bond between steel and concrete. The tensile stress of CFRP Platereinforced specimens can also be shared by steel bars and CFRP plates. The additional anchoring reinforcement of the CFRP plate effectively reduces the tensile stress and strain values of the steel bar. The stress gradient value and crack width of the specimens gradually decrease with the reinforced effect of the CFRP plate.
In this paper, the specimens in Group B are affixed with transverse strips at different locations as an additional anchoring measure. Figure 13 shows the average crack spacing of the specimens with different additional anchoring measures. The average crack spacing and the maximum crack width of the specimens with additional anchoring measures at the purely curved section and the support position have the most significant reduction effect. It can be seen from Fig. 14 that the maximum crack width of the specimens LM3 and TM1 is similar to the specimen FDBL before the load reaches 0.8M _{ y } , and when loaded to 0.9M _{ y } . The maximum crack width of the specimen LM1 and the specimen TM1 are significantly reduced compared to the maximum crack width of the specimen FDBL. There are two reasons for the above phenomenon: 1) the number of curved main cracks gradually decreases as the root crack propagation increases; 2) the strain gradient of concrete gradually decreases with the effect of additional anchoring.
Figure 14 shows that the mean crack spacing of specimen LM2 and LM3 are reduced by 6.02% and 16.75%, respectively, compared to that of the specimen LM1. It indicates that the additional bolt anchoring measures in the midspan and the bearing positions were more effective. The statistics shows that the maximum crack width of the specimen LM2 and the specimen LM3 are reduced by 23.21% and 14.26%, respectively, comparing with the specimen LM1 with end bolt anchoring measure. The additional bolt anchoring measure in midspan position has most effective in reducing the maximum crack width.
4.2 Midspan loaddeflection curve
Figure 15 shows the variation of the loaddeflection curves of the specimens. It can be seen from Figure 15 that the cracking load values of the specimens reinforced by the CFRP plates are approximately the same. The tensile stress is shared by the CFRP plate and the steel bar when the concrete is cracked. With the gradual increase of CFRP tensile strength, the cracking of the specimen beam strengthened by CFRP plate is obviously delayed and the rigidity is improved.
It can be seen from Figure 16 that the strains of steel bars of CFRP platereinforced specimen increase slowly after the occurrence of the crosssection crack. Since the CFRP plate and the steel bar share the tensile stress that the loadstrain curve trend of the CFRP platereinforced specimens are steeper than that of the unreinforced specimen FDBL. The CFRP plate has a good reinforcing effect in the specimens of bridge roof.
4.3 Development of CFRP plate strain
It can be seen from Figure 17 that the variations of the strain curve of CFRP plates are similar. Taking specimen TM2 as an example, when the load reaches 240 kN, the strain curve of the CFRP plate in the flexural shear section is almost horizontal. This indicates that the reinforcement effect of the CFRP plate can be improved by the anchoring reinforcement methods, and the occurrence of debonding of the CFRP plate can be effectively limited.
5 THEORETICAL MODELS FOR DEBONDING BEARING CAPACITY
Intermediate crack debonding (IC debonding) of CFRPconcrete interface is the dominant failure modes in CFRP reinforced bridge roof structure. Several researches have been carried out to investigate the IC debonding of CFRPconcrete interface failure and its influencing factors. Some models have been proposed to calculate the CFRPconcrete interface debonding failure (^{Leung et al., 2006}; ^{Lu et al., 2007}; ^{Said and Wu, 2008}; ^{Teng et al., 2003}). The CFRPconcrete debonding failure model is introduced based on the test data and previous research results, the proposed CFRPconcrete debonding failure proposed model is compared with the previous CFRPconcrete debonding failure model. The validity and accuracy of the proposed model are presented in this paper.
5.1 Existing theoretical models
In order to assess the applicability of existing models in calculating the intermediate crack debonding bearing capacity of CFRP reinforced concrete components, the theoretical models proposed by ^{Lu et al. (2004}), ^{Huang and Ye (2004}), ^{Teng et al. (2003}), FIB 14 (2001) are analyzed in this paper. The adaptation range of these formulas is the midspan debonding failure of CFRPreinforced reinforced test beams.
The symbols commonly used in the formulas below are uniformly described in this paper.
^{Lu et al. (2004}) proposed a calculation formula for ICdebonding bearing capacity, based on finite element simulation and data regression, as follows:
where
^{Huang and Ye (2004}) used finite element analysis and experimental methods to propose the formula as follows:
^{Teng et al. (2003}) corrected the formula of the inplane sheardebonding bearing capacity, and proposed a calculation formula for the debonding bearing capacity caused by ICdebonding as follows:
where
The debonding failure predicted by using the crack width in FIB 14 (2001) was assessed based on the following equations:
where
5.2 The proposed theoretical model
When the width of the curved main crack is equal to twice the crack width of the CFRPconcrete interface bond slip, the corresponding load value is equal to the peeling capacity of the CFRPreinforced specimen. Based on the abovementioned crack debonding criterion, a new calculation method of bearing capacity is proposed, and a unified analysis model for evaluating debonding is established. The proposed model can be used to calculate the peel load capacity relatively accurately. It is derived from the theoretical and numerical investigation conducted in this paper.
The crack width in the crack debonding criterion (FIB, 2001; ^{Lu et al., 2004}) is described as the deformation difference between the FRP and the concrete in the two adjacent cracks. The crack width of the concrete at the height of the FRP can be derived by the following equations:
where
The tensile strain of the FRP at the crack section
where
The maximum crack width is related to the average crack width, and when the maximum crack width is equal
where
Submit Eq. (14), Eq. (15), Eq. (16), Eq. (17), Eq. (19) and Eq. (20) into Eq. (18), the debonding moment of CFRP plate can be derived as follows:
where
The interface slip displacement
The experimental data derived in this paper and previous work in ^{Hu (2013}), ^{Wu (2002}), ^{Xie et al. (2009}) and ^{Ye et al. (2003}) were used to fitted the proposed model. The suggested value was substituted into the Eq. (21), and the debonding moment
where the average cracking distance
where
The debonding bearing capacities of the specimens in the tests were calculated by the suggested model given in this paper and the models suggested by ^{Lu et al. (2004}), ^{Huang and Ye (2004}), ^{Teng et al. (2003}) and FIB 14 (2001) respectively. Table 5 lists comparison of the debonding moment (M _{ db } ) derived from the tests and different theoretical models.
Specimen number  FX1  FX2  FX3  TM1  TM2  TM3  LM1  LM2  LM3  Average error* 

Test results  168.0  168.0  154.0  182.0  168.0  196.0  168.0  168.0  182.0  / 
The proposed model  149.3  149.3  149.3  149.3  149.3  149.3  149.3  149.3  149.3  23.4 
Model in Lu et al. (2004)  136.3  136.3  136.3  136.3  136.3  136.3  136.3  136.3  136.3  36.4 
Model in Huang and Ye (2004)  140.8  140.8  140.8  140.8  140.8  140.8  140.8  140.8  140.8  31.8 
Model in Teng et al. (2003)  129.2  129.2  129.2  129.2  129.2  129.2  129.2  129.2  129.2  43.5 
Model in Fib 14 (2001)  85.2  85.2  85.2  85.2  85.2  85.2  85.2  85.2  85.2  87.5 
^{*} The average error of each model is obtained from averaging the differences between the calculated values by the theoretical models and the test values of specimens.
It can be seen from Table 5 that the average error of the proposed model is the lowest among all the theoretical models. The average error of the values calculated by the proposed model is 23.4 kN·m, which is much smaller than those of other theoretical models, such as the FIB 14 model (87.5 kN·m, 273.9% larger than that of the proposed model). It indicates that the model proposed in this paper has good predictive ability for the calculation of debonding moment between CFRP plateconcrete interface. It is worth noting that the calculation results are the same for different test specimens. This is because the current models cannot take into account the effects of different anchoring measures. Thus, further research is needed to establish a theoretical calculation model that can consider the influences of the anchoring methods.
6 CONCLUSIONS
This paper investigated the crack and mechanic behavior of CFRP platereinforced bridge roof under high temperature. Six CFRPreinforced test beams with different anchoring schemes were designed and constructed. The beam specimens, after the high temperature effect, were tested under fourpoint bending loads. The crack propagation, loaddeflection curves, loadstrain curve, and failure modes of specimens of different reinforcement schemes were observed and recorded. A theoretical model was proposed for calculating the debonding bearing capacity of the CFRPconcrete interface. The following conclusions can be drawn:
The CFRP plate has a good inhibitory effect on the crack propagation. The CFRP platereinforced test beams exhibit higher load carrying capacity than the unreinforced test beams.
Additional anchoring can effectively delay the occurrence of CFRP plate debonding. The debonding ductility and bearing capacity of the test specimens are both improved with the additional anchoring measures.
The additional anchoring measures placed at the midspan and support positions are more effective in reducing the maximum crack width and the mean crack spacing of the test specimens.
The calculating results of the debonding bearing capacity by proposed theoretical model agree well with the experimental results. The model proposed in this paper could be applied to calculate the debonding bearing capacity of the CFRP platereinforced bridge roofs.