Effect of bond degradation on fire resistance of FRP-strengthened reinforced concrete beams

Composites: Part B 42 (2011) 226–237

Contents lists available at ScienceDirect

Composites: Part B journal homepage: www.elsevier.com/locate/compositesb

Effect of bond degradation on fire resistance of FRP-strengthened reinforced concrete beams A. Ahmed, V.K.R. Kodur ⇑ Department of Civil and Environmental Engineering, 3546 Engineering Building, Michigan State University, East Lansing, MI 48824-1226, United States

a r t i c l e

i n f o

Article history: Received 7 May 2010 Received in revised form 27 September 2010 Accepted 3 November 2010 Available online 17 November 2010 Keywords: B. Debonding B. Delamination B. Fiber/matrix bond C. Numerical analysis FRP-strengthened concrete beams

a b s t r a c t FRP-strengthened reinforced concrete (RC) members experience significant loss of strength and stiffness properties when exposed to fire. At elevated temperatures, the rate of loss of such properties is influenced considerably by the bond degradation at the FRP–concrete interface. This paper presents a numerical approach for modeling the bond degradation in fire exposed FRP-strengthened RC beams. The numerical procedure is incorporated into a macroscopic finite element model which is capable of accounting high temperature material properties, different fire scenarios, and failure limit states in evaluating fire response of FRP-strengthened RC beams. The validity of the model is established by comparing predictions from the program with data from full scale fire resistance tests on FRP-strengthened RC beams. The validated model is applied to evaluate the effect of bond degradation on fire response of FRPstrengthened beams. Results from the analysis indicate that significant bond degradation occurs close to glass transition temperature of the adhesive leading to initiation of FRP delamination. The time at which bond degradation occurs depend on the fire insulation thickness and glass transition temperature of the adhesive. However, variation of adhesive thickness does not significantly influence fire resistance of FRP-strengthened RC beams. Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction In recent years, fiber reinforced polymers (FRP) are finding increasing applications in repair, rehabilitation and strengthening of reinforced concrete (RC) structural members. The increased load carrying capacity of such FRP-strengthened structures primarily depends on the effectiveness of bond between FRP and concrete substrate that is highly influenced by the properties of adhesive [1]. At room temperature, the effect of bond characteristics on debonding failure of FRP-strengthened RC beams has been well studied in the literature. Based on these studies, bond–slip models have been proposed to account for load–slip at ambient conditions [2–9]. When FRP-strengthened members are exposed to fire, bond at FRP–concrete interface degrades due to increasing temperature. The term ‘‘FRP–concrete interface’’ in this paper, refers to adhesive adjacent to concrete substrate that bonds FRP with concrete and any deterioration in adhesive properties resulting from temperature, leads to relative slip between FRP and concrete substrate. Closer to glass transition temperature (Tg), the thermo-mechanical properties (strength and stiffness) of adhesive degrade considerably and this initiates debonding of FRP. Previous studies clearly show that the bonding material (adhesive) soften at temperatures ⇑ Corresponding author. Tel.: +1 517 353 9813; fax: +1 517 432 1827. E-mail address: [email protected] (V.K.R. Kodur). 1359-8368/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.compositesb.2010.11.004

close to Tg and this leads to significant reduction in tensile strength and elastic modulus as illustrated in Fig. 1 [10]. Gluguru reported a 20% reduction in shear strength for high temperature epoxy and a 70% for general purpose epoxy, when the temperature reaches about 80 °C [11]. The bond at FRP–concrete interface is influenced by a number of factors such as type of FRP reinforcement, type of adhesive and its thickness, compressive strength of concrete, moisture, surface preparation (roughness), workmanship, and temperature level [12]. While the effect of many of these factors have been studied in the literature [4,13–15], there is little information on the influence of temperature on bond degradation. This bond at interface is critical for transfer of forces from concrete to FRP. When the temperatures of FRP/adhesive reaches Tg, bond properties (shear and bond strength) of adhesive deteriorate considerably and this introduces slip at the interface. This slip reduces the ability of adhesive to transfer forces efficiently, ultimately leading to debonding of FRP composite as illustrated schematically for an FRPstrengthened RC beam in Fig. 2. Previous research has indicated that Tg is likely to be a critical factor in the event of fire [12]. Therefore, load carrying capacity of FRP-strengthened structural members under fire conditions is primarily influenced by thermo-mechanical properties (mainly Tg) of the adhesive. Guidelines for design of FRP-strengthened RC structures at room temperature are available in codes and standards. However,

A. Ahmed, V.K.R. Kodur / Composites: Part B 42 (2011) 226–237

227

This model is applied to carry out parametric studies to quantify effect of bond on the overall fire performance of FRP-strengthened RC beams. 2. State-of-the-art review

Fig. 1. Schematic sketch for variation of elastic modulus and tensile strength of adhesive [10].

Fig. 2. Schematic sketch showing debonding of FRP at elevated temperature (a) beam elevation (b) development of shear stresses at interface of FRP–concrete (c) slip at the interface (d) debonding at the end of FRP reinforcement.

there are no specific guidelines for design under fire conditions. As an example, ACI 440.2R-08 [16] guidelines assumes no contribution from FRP to the load carrying capacity of strengthened members in the event of fire. Most of the previous research studies on fire resistance of FRP-strengthened member assumed a perfect bond up to glass transition temperature and a complete debonding, thereafter. Currently available numerical models does not account for temperature induced bond degradation in the fire resistance analysis. This paper presents a numerical procedure to model the temperature induced bond degradation at FRP–concrete interface. This procedure is implemented to a macroscopic finite element model that is capable of tracing the fire response of FRP-RC beams.

A large number of studies, both experimental and theoretical, have been undertaken to study bond degradation at between FRP–concrete interface at ambient temperature. Results from these studies have generated bond strength models and also quantified influence of various factors on bond strength degradation. Some of the proposed models are based on empirical relationships, while others take into consideration fracture mechanics principles to evaluate bond strength [17–23]. Limited studies have been conducted in the literature on the effect of temperature on bond between FRP–concrete interface. Tadeu and Branco [24] studied the influence of temperature on bond between externally bonded steel plates and concrete by testing concrete specimens of 150  100  100 mm with a gluing area of 100 mm long and 80 mm wide. The specimens were tested in double-lap shear at five selected temperatures (20, 30, 60, 90 and 120 °C). Based on these tests, authors reported significant reduction in bond strength with temperature; a 90% reduction at 120 °C. Blontrock et al. [25] conducted double-lap shear test on CFRPstrengthened concrete prisms 150  150  800 mm separated by a thin metal plate. The specimens were subjected to direct tensile load at four temperature levels of 20, 40, 55 and 70 °C. The tests conducted at 40 and 55 °C showed an increase in failure load by 41% and 24% respectively, however, the failure load decreased by about 19% (compared to maximum load at 20 °C) at 70 °C (close to Tg). Klamer et al. [26] investigated the influence of temperature on debonding behavior of externally bonded CFRP through two different test setups namely: double-lap shear test and small scale three point bending test at five temperatures (10, 20, 50, 60 and 75 °C). Results indicated an increase in failure load by about 10% for specimens tested at 50 °C. However, a further increase in temperature to 75 °C resulted in 27% decrease in failure load. A similar trend was not observed in three point bending test. Klamer et al. [27] also tested four full scale FRP-strengthened RC beams at 20, 50 and 70 °C to investigate influence of temperature on FRP debonding mechanism. Test results indicated that the type of failure and the failure load at room temperature was similar to that at 50 °C, however, at 70 °C failure loads reduced considerably. Therefore, the authors concluded that the contribution of FRP to strength capacity can be ignored when temperatures at FRP– concrete interface reach Tg. The effect of service temperature (50, 65 and 80 °C) on bond strength was studied by Leone et al. [28]. The test specimens (150  150  800 mm) were strengthened with CFRP and GFRP hand layup sheets and tested under double-lap shear test procedure. The experimental investigation showed a decrease in maximum bond strength for temperatures above Tg of adhesive. At 80 °C, the bond strength in CFRP and GFRP sheets, and CFRP laminate dropped by about 54%, 72% and 25% respectively. Data from these tests indicate that strain level along FRP reinforcement and initial transfer length for a given load increases with rise in temperature. Wu et al. [29] studied the effect of temperature on bond behavior between FRP sheet and concrete. The specimens (100  100  450 mm) were tested at temperature levels ranging from 26 to 60 °C using ordinary and thermo-resisting epoxies. Based on tests data, the study concluded that close to Tg, debonding fracture energy (Gf) decreases while length (Le) required to achieve effective bond increases. It was also observed that failure load and elastic modulus decrease with temperature. Gamage et al. [30] investigated bond characteristics of CFRP plated concrete blocks (130  130  300 mm) at elevated

228

A. Ahmed, V.K.R. Kodur / Composites: Part B 42 (2011) 226–237

temperatures. The authors conducted two series of shear tests; first series of eleven specimens without any insulation, and second series of two specimens with 50 mm thick insulation. The test data showed that bond strength is independent of bonded length of FRP when exposed to elevated temperatures. The un-insulated test specimens experienced loss of bond after 5–6 min of fire exposure time that indicated the necessity for fire protection (insulation) to maintain effective bond between FRP and concrete at elevated temperatures. The experimental study conducted by Camata et al. [12] and Di Tomasso et al. [31] on bond behavior at elevated temperatures showed degradation of bond properties at temperatures close to glass transition temperature (Tg). A closed form solution to determine interfacial shear stresses and normal stresses for a prismatic section due to thermal expansion while assuming elastic behavior, was presented by Denton [32]. The results indicated peak shear stress values near to the end of the FRP plate reducing non-linearly towards the mid-span of the beam. Numerical results also showed that for FRP plates, use of tapered end configuration significantly reduces the peak interfacial shear stress. The above review clearly illustrates that bond degradation occurs in FRP-strengthened concrete members at high temperatures. Most of these studies were conducted on small scale test specimens and data on full scale FRP-strengthened members is limited. The state-of-the-art review also indicates that bond at FRP–concrete interface is a weak link at higher temperatures since concrete and steel properties do not degrade much up to 400 °C [33]. 3. Evaluating strain (eslip) due to bond–slip In FRP-strengthened RC members, the binding material (adhesive) provides load path for transfer of stresses from concrete substrate to FRP reinforcement. At temperatures beyond Tg, bond properties (shear and bond strength) deteriorate considerably and this introduces a slip at bond interface. Due to this bond–slip, adhesive loses its ability to effectively transfer forces between concrete and FRP and this result in FRP developing only partial tensile stresses as compared to a perfect bond case where full stresses in FRP can effectively be utilized. With increasing slip, the bond dete-

riorates considerably and ultimately leads to debonding of FRP. Thus, bond degradation with temperature is to be properly accounted for reliable assessment of fire resistance in FRPstrengthened RC members. In FRP-strengthened members, FRP terminates at a distance from the support. Concentration of shear stresses, which mainly contribute in transfer of forces from concrete to FRP, is substantial near edges of FRP reinforcement. Previous studies have shown that this high shear stress concentration is a major cause of FRP debonding. Fig. 3 schematically shows development of shear stresses in a beam segment of FRP-strengthened RC beam and related bond–slip at FRP–concrete interface. Thus, variation of these shear stresses with temperature can be used to derive an expression for computing bond–slip at FRP–concrete interface. This expression will account for changing material properties of adhesive (shear stiffness) with temperature. The assumptions made in deriving this expression are; shear stresses are invariant across adhesive thickness, stress distribution is independent of flexural cracks in concrete, and curvature at beam soffit and FRP is to be the same. The beam is idealized into a number of segments along its length. For a small elemental length ‘‘dx’’ of the adhesive (see Fig. 3c), displacement (du) due to slip is given by:

du ¼

s G

tg

ð1Þ

where s is the shear stress, G is the shear modulus and tg is adhesive thickness. For each beam segment i, average shear stress si at the FRP– concrete interface can be expressed as (refer Fig. 3):

si ¼

Pfrpðiþ1Þ  PfrpðiÞ Li  b

ð2Þ

where Pfrp(i) is force in FRP reinforcement for segment i, Li is length of segment i, and b is the width of the beam. With increasing temperature due to fire, the adhesive softens and experiences a significant reduction in its shear modulus (G). This softening effect results in a relative slip (dslip) between FRP composite and concrete. Slip in a segment i can be calculated as (refer to Fig. 3c):

dslipðiÞ ¼ ci  tg ð3Þ

Fig. 3. Development of shear stresses and bond–slip in a beam segment.

A. Ahmed, V.K.R. Kodur / Composites: Part B 42 (2011) 226–237

where tg is adhesive thickness and ci is the shear strain in segment i given by:

ci ¼

si

ð4Þ

G

Substituting ci in Eq. (3), relative slip (dslip) in a beam segment can be expressed as:

dslipðiÞ ¼

Pfrpðiþ1Þ  PfrpðiÞ 1   tg G Li  b

ð5Þ

Knowing dslip, the relative strain due to slip can be established as:

dslipðiÞ change in length ¼ original segment length Li Pfrpðiþ1Þ  PfrpðiÞ 1 ¼   tg G L2i  b

eslipðiÞ ¼

ð6Þ

In Eq. (6), bond–slip (eslip) depends on shear modulus of adhesive (that decreases with temperature) as well as force in FRP (Pfrp) which in turn depends on temperature, curvature and strain in FRP. The bond–slip (eslip) in each beam segment can be calculated at any fire exposure time using Eq. (6). The variation of bond–slip is a function of distance from FRP plate ends. As schematically shown in Fig. 4, peak bond–slip occurs near FRP plate end and varies exponentially towards center of the beam. The beam segment with peak bond–slip represents critical segment of the FRP-strengthened beam since delamination of FRP initiates at this segment. For simplification, bond–slip evaluated in critical segment can be assumed consistent in all beam segments, for a given time step. Under fire conditions, FRP only develops partial tensile strength due to bond–slip. Therefore, in computing effective mechanical strain in FRP, strain due to bond–slip (eslip) is to be subtracted from the total strain. This effective mechanical strain, which takes into consideration bond degradation, can be used to calculate stress and tensile force in FRP. The effect of temperature induced bond– slip is significant when the temperature at FRP–concrete interface exceeds Tg.

229

4. Numerical model The above approach to account for bond–slip has been incorporated into a fire resistance model, which was initially developed to evaluate fire response of FRP-strengthened RC beams by assuming a perfect between FRP and concrete [34]. This fire resistance model takes into account high temperature properties of constitutive materials to generate moment–curvature (M  j) relationships for different beam segments at various time steps. These time dependant M  j relationships are utilized to trace the response of FRP-strengthened RC beam in the entire range up to collapse under fire conditions. In the analysis, the total fire exposure time is divided into number of time steps and at each time step, response of the beam is traced through the following steps:  Establishing temperatures due to fire.  Conducting heat transfer analysis to determine temperature distribution in segmental cross section.  Calculating the slip (eslip) at the interface of FRP and concrete.  Generating moment curvature (M  j) relationships for each beam segment at various time steps and performing beam analysis to compute internal forces and deflections in the FRPstrengthened RC beam. The beam is idealized by dividing it into a number of segments along its length (refer to Fig. 3b) and the mid-section of each segment is assumed to represent the overall behavior of the segment. This mid-section is further discretized into a number of elements (see Fig. 4b). A finer mesh is used for insulation (3  3 mm) and FRP (6  3 mm) since these elements are in close proximity to fire zone and are also highly sensitive to temperature rise. In the model, the beam is assumed to be exposed to fire from three sides while ambient conditions are assumed to prevail on the top side to represent the presence of slab. Fire temperatures are established from known time–temperature curves for standard or any other

Fig. 4. Schematic interfacial shear stress distribution.

230

A. Ahmed, V.K.R. Kodur / Composites: Part B 42 (2011) 226–237

specified design fire scenario. Then, at each time step, temperature distribution in the beam cross section is established through thermal analysis utilizing high temperature thermal properties of constitutive materials. The procedure for undertaking thermal analysis, including governing heat transfer equations, is given in Reference [35]. The computed cross sectional temperatures form input to strength analysis wherein time dependant M  j relationships are generated for each beam segment. Under fire conditions, each element of concrete, steel and FRP are subjected to different strains and all these strains have to be included to assess effective mechanical strain. At a given fire exposure time, the mechanical strains are computed as:

ecmec ¼ ect  ecth  eccr  ectr ðfor concreteÞ

ð7Þ

esmec ¼ est  esth  escr ðfor steelÞ

ð8Þ

frp frp frp efrp mec ¼ et  eth  ecr þ ebi þ eslip ðfor FRPÞ

ð9Þ

where et = total strain, eth = thermal strain, emec = mechanical strain, ecr = creep strain, etr = transient strain, ebi = initial strain at the soffit of the beam at the time of retrofitting with FRP and eslip = slip at the interface of FRP and concrete. The superscripts ’c’, ’s’ and ’frp’ represents concrete, steel rebar and FRP respectively. In Eqs. (7)–(9), at a time step, a total strain (et) in each element of concrete, FRP and rebar can be computed for an assumed value

of strain at top most fiber in concrete (ec) and curvature (j) by following expression (see Fig. 5a):

et ¼ ec þ jy

ð10Þ

where et = total strain, ec = strain in top most fiber in concrete, j = curvature and y = distance from uppermost fiber in concrete to center of element In Eqs. (7)–(9), creep strains for both concrete and steel, and transient strain in concrete, which depend on time, temperature and stress levels, are computed based on the models proposed by Harmathy [36,37], and Anderberg and Thelandersson [38], respectively. For FRP, creep strain is negligible and is not accounted for in the analysis since fiber direction coincides with loading direction of the strengthened beam [39]. Initial strain (ebi) in FRP is evaluated based on dead loads at the time of retrofitting, while bond–slip (eslip) at the interface of FRP–concrete is computed using proposed approach in Eq. (6). With this approach all strains components in Eqs. (7)–(9), ex frp cept mechanical strain efrp mec are known and thus emec can be evaluated. Knowing mechanical strain, stresses in each of the concrete, steel rebar and FRP elements can be obtained through temperature dependent stress–strain relationships for these materials. Knowing the stresses, force in concrete, steel and FRP can be calculated. At each time step, the computed forces are used to check force equilibrium. As schematically shown in Fig. 5a, for an assumed total strain at the top layer of concrete ðect Þ, curvature (j) is iterated

Fig. 5. Discretization of beam for analysis and relationship for idealized segment.

A. Ahmed, V.K.R. Kodur / Composites: Part B 42 (2011) 226–237

until force equilibrium is satisfied. This iterative procedure is repeated till equilibrium, compatibility and convergence criterion are satisfied. Once these conditions are satisfied, moment and curvature corresponding to that strain is computed. Through this approach, various points on the moment–curvature curve are generated for each time step. Following the generation of M  j relationships, an iterative procedure described by Cambell and Kodur [40] is employed to evaluate deflections of the beam at each time step. The beam analysis starts under a unit applied load using initial rigidity (EIo) and the moment and corresponding curvature in each beam segment is determined. The segment that has the maximum moment is selected as the critical (key) segment of the beam. Then, a target curvature in the key beam segment is selected on pre-generated M  j curve. Utilizing unit load analysis, a scaling factor is evaluated by dividing the target curvature with unit load curvature in the key segment. The unit load curvatures in all beam segments are scaled by multiplying them with the by this scaling factor. An iterative procedure, illustrated in Fig. 5, is employed till convergence of secant rigidity within a certain tolerance is achieved. Once tolerance is achieved, the above procedure is repeated for next assumed target curvature [41]. After each iteration procedure, load required to attain target curvature (key segment) is computed and stored. To compute the actual curvatures and deflections in the beam, applied load is interpolated between these stored values. In the above procedure, stiffness matrix and the loading vector are computed for each longitudinal segment and assembled in the form of a nonlinear global stiffness equation, and solved to compute deflections at that time step:

½K g ½d ¼ ½P

ð11Þ

231

ture. In literature, there is reliable data on high temperature properties of concrete and steel. However, knowledge is limited on high temperature properties of FRP, adhesive and insulation. For concrete and steel, the properties suggested by ASCE Manual [42] and for FRP and insulation, semi-empirical relationships suggested by Bisby [43], have been incorporated into the model. These high temperature relationships for thermal and mechanical are presented in Appendix A. Accordingly, temperature dependant stress–strain curves for FRP are linear elastic till ultimate rupture strain. Figs. 6 and 7 show normalized thermal conductivity and thermal capacity for FRP and insulation (vermiculite–gypsum) as a function of temperature. It can be seen that there is a considerable reduction in thermal conductivity of FRP with increasing temperature. The plateau in thermal capacity of FRP in the range of 340– 510 °C is due to additional heat absorbed for decomposition of the resin [44]. For insulation the thermal conductivity decreases up to 200 °C, remains nearly constant till 500 °C and then increases with temperature. The peak for thermal capacity of insulation is at about 100 °C (shown in Fig. 7) and is due to evaporation of trapped water that consumes most of the heat energy. In externally bonded FRP-strengthened RC members, mechanical properties of adhesive, which provides load path for transfer of stresses between FRP and concrete substrate, degrade significantly with temperature. The reduced stiffness of adhesive at elevated temperature (close to Tg) has softening effect that initiates bond–slip. Therefore, to account for bond–slip, variation of shear modulus (G) of adhesive with temperature is needed to compute bond–slip (eslip) at FRP–concrete interface utilizing Eq. (6). Bond stress–slip curves presented by Leone et al. [28] have been included in the model. These curves provide data for variation of

where Kg = global stiffness matrix, d = nodal displacements, P = Pf + Ps where Pf = equivalent load vector due to applied loading and Ps = equivalent nodal vector due to P  d effect. The model generates various output parameters, such as cross sectional temperatures, stresses, strains, deflections and moment capacity for each time increment. These parameters are checked against pre-designated failure criterion, which include thermal and structural considerations. The time increment continues until one of the limiting criteria is reached. At this time step, the beam is said to have failed. The time duration to reach this failure point is the fire resistance of the beam. In the model, any or all of the following limiting criteria can be applied to evaluate failure of the FRP-strengthened RC beam:  The moment due to applied load exceeds the strength capacity of the beam.  The temperature in reinforcing steel (tension reinforcement) exceeds 593 °C.  The deflection of the beam exceeds L/20, where L is the length of the beam, at any fire exposure time.  The rate of deflection exceeds the limit L2/9000d (mm/min) where L is the length of the beam (mm); and d, effective depth of the beam (mm).  The temperature in FRP layer exceeds glass transition temperature (Tg) of FRP.

Fig. 6. Normalized thermal conductivity of FRP and insulation as a function of temperature.

It should be noted that the user has the option to specify any (or all) of the five limit states to define failure. 4.1. High temperature material properties For modeling the response of FRP-strengthened beams, high temperature properties of concrete, steel, FRP, adhesive and insulation are required. These properties include thermal, mechanical and deformation properties which vary as a function of tempera-

Fig. 7. Normalized thermal capacity of FRP and insulation with varying temperature.

232

A. Ahmed, V.K.R. Kodur / Composites: Part B 42 (2011) 226–237

surface extending 100 mm on the two sides of the beam (refer to Ref. [46]). Both the beams were analyzed with the above described numerical model and results from analysis are compared with measured test data in Figs. 9–12. For the analysis, the high temperature material properties as discussed above were used. Fig. 9 shows the comparison of predicted and measured temperature for Beam I in steel reinforcement and at the interface of FRP–concrete, respectively. A good agreement between the predicted and

Fig. 8. Variation of adhesive shear modulus versus temperature.

shear modulus (G) with temperature. Fig. 8 shows the variation of shear modulus as function of temperature. It can be noticed that shear modulus reduces consistently with increasing temperature and thus significantly influence the bond performance. 5. Model validation The above described fire resistance model is verified by comparing predictions from the model with measured test data on FRPstrengthened RC beams. As highlighted in state-of-the-art review section, there is limited test data that accounted for fire induced bond degradation in FRP-strengthened RC members. The two beams selected for validation, designated as Beams I and II, were tested by Blontrock et al. [45] and Ahmed and Kodur [46]. The geometric properties of the beams as well as material properties of concrete, steel, CFRP and insulation are given in Table 1. Beam I is 200  300 mm in cross section and 3 m in span length. The beams had two 10 mm and 16 mm rebars as compression and tensile reinforcement respectively. The beam was strengthened with 1.2 mm thick and 100 mm wide CFRP laminates, and was insulated with Promatect-H type fire protection. The beam had 25 mm of insulation thickness at the beam soffit, and an additional 12 mm insulation thickness extending on both sides of the beam up to a height of 105 mm (measured from the soffit insulation thickness). Beam II is of 254 mm width, 406 mm depth and 3.96 m span length. The flexural strength of Beam II was enhanced by installing two CFRP sheets of 2 mm thick and 203 mm width and spray-applied with TyfoÒ WR Advance Fire Protection (AFP) system. The insulation layout comprised of 25 mm at the bottom

Fig. 9. Measured and predicted temperatures at the interface of FRP–concrete and corner rebar for Beam I.

Fig. 10. Measured and predicted deflection as a function of fire exposure time for Beam I.

Table 1 Summary of properties for FRP-strengthened RC beams used in the fire resistance analysis. Property

Beam I

Beam II

Beam III

Description Cross Section (mm) Length (m) Reinforcement

Tested by Blontrock et al. [45] 200  300 3.15 2 / 10 mm 2 / 16 mm 57.5 591 2  40.6 (kN) 25 Siliceous Sika carbodur S1012 1.2 2.8 165 1.7% 25 Promatect – H

Tested by Ahmed and Kodur [46] 254  406 3.96 2 / 13 mm 3 / 19 mm 52 450 2  70 (kN) 54 Carbonate CFRP (TyfoÒ SCH-41) 2 986 95.8 1.0% 25 TyfoÒ WR AFP system

Typical FRP-strengthened RC beam [48] 380  610 6.7 2 / 15.8 mm 4 / 25 mm 38 414 60 (kN/m) 40 Carbonate CFRP 3 2450 176 1.41% 20 Vermiculite–gypsum (VG)

0

2

Top bars Bottom bars

f c (N/mm ) fy (N/mm2) Applied total load Concrete cover thickness (mm) Aggregate type FRP type FRP thickness (mm) FRP ultimate tensile strength (kN/mm2) Modulus of elasticity FRP (kN/mm2) Rupture strain of FRP (mm/mm) Insulation thickness (mm) Insulation type

A. Ahmed, V.K.R. Kodur / Composites: Part B 42 (2011) 226–237

233

Fig. 11. Measured and predicted temperatures for FRP-RC Beam II.

Fig. 12. Measured and predicted deflections for FRP-RC Beam II.

measured values can be seen in entire range of fire exposure. Predicted and measured deflections at mid-span of FRP-strengthened RC beam (Beam I) are compared in Fig. 10. Results from analysis indicate that debonding of FRP occurs at 30 min of fire exposure time that is slightly higher than that reported by Blontrock et al. (at 26 min when the temperature at the interface was 52.1 °C i.e., less than measured Tg = 62 °C). This variation in predictions may be due to slight differences in bond properties used in the model as compared to the actual properties of tested beam. In early stages of fire exposure, the response of the beam is relatively stiff (lower deflections) which can be attributed to high strength and stiffness properties of bonded FRP. When the temperature at FRP–concrete interface exceeds Tg, a slight jump in the time–deflection curve can be observed that indicates debonding of FRP. After FRP debonds, the beam experience high magnitude of stresses as compared to an un-strengthened RC beam and this result in an increase in deflection rate with early strength failure (reduced fire resistance) as shown in Fig. 10. The deflection predicted by the model before and after composite action between FRP and concrete was lost, matches closely with measured test data. Fig. 11 provides a comparison of temperatures at FRP/concrete (FRP/C) and FRP/insulation (FRP/insulation) interfaces, and at three different locations (TC5, TC6 and TC9) in the beam cross section for Beam II [46]. TC5 represent temperature in compression reinforcement, TC6 represent corner rebar temperature (flexural reinforcement) while TC9 is middepth of beam cross section (203 mm), as shown in Fig. 11a. The model predicts temperature fairly well in compression and flexural reinforcement as well as at the middepth of the beam cross section (refer to Fig. 11a). For first 35 min of fire exposure time, the predicted temperature at FRP/insulation and FRP/concrete interfaces matches well with the measured data, as

shown in Fig. 11b. Beyond this point, the model predictions does not match well with measured temperatures since a portion of insulation fell off when FRP delaminated around 38 min and the model could not account for falling-off of insulation [46]. The predicted and measured mid-span deflection of FRP-RC beam (Beam II) is compared in Fig. 12. There is a good agreement between measured and predicted deflections for the entire duration of the test. Compared to measured time of FRP debonding which is around 20 min, the model predicts it to be about 25 min. This variation can be attributed to the discrepancy between measured and predicted temperatures at interface of FRP as discussed above. Overall, the model provides reasonable estimates of temperature at different locations of beam cross section and deflections match fairly well with measured data. The above comparison indicate that the model is capable of tracing overall thermal and structural response of FRP-strengthened beams, including the effect of fire induced bond degradation on fire resistance.

6. Effect of bond degradation on fire resistance To illustrate the effect of bond degradation at FRP–concrete interface on fire response, a case study has been carried out on a FRP-strengthened RC beam, designated as Beam III. The analysis was carried out for three cases namely; with perfect bond, with temperature induced bond degradation and with a plain RC beam (with and without externally applied insulation). The properties of the beam used in the analysis are summarized in Table 1. The RC beam is strengthened with 2 mm thick CFRP laminate and has 20 mm thick insulation at the beam soffit extending on two sides of the beam up to 105 mm height (measured from the soffit insulation thickness). The beam is analyzed under ASTM E119 standard fire exposure. For structural analysis, the beam is loaded with 60 kN/mm (52% load ratio, which is defined as the ratio of applied loading at the time of fire to capacity at room temperature). The analysis is carried out using the above described model and fire resistance is evaluated based on four failure criteria as described above. The results are presented in Figs. 13–19. Fig. 13 shows a comparison of time–deflection response for two cases of FRP-strengthened RC beams, namely; with a perfect bond, with temperature induced bond–slip, and two cases of un-strengthened RC beams, namely with and without externally applied fire protection. In early stages of fire exposure, the response of both strengthened beams (with and without accounting for bond degradation) is stiffer as compared to un-strengthened RC beam due to high strength and stiffness properties provided by FRP composite. For the un-strengthened RC beam, the rate of deflection is much higher since mechanical properties of concrete and steel

234

A. Ahmed, V.K.R. Kodur / Composites: Part B 42 (2011) 226–237

Fig. 13. Deflection of beams as function of fire exposure time.

Fig. 14. Ultimate tensile strength of CFRP as a function of temperature.

Fig. 15. Temperature variation at the interface of FRP–concrete interface as a function of fire exposure time.

degrade faster in the absence of any external fire protection. For strengthened beam with a perfect bond, the response of the beam is stiffer for entire duration of fire exposure as compared to the FRP-RC beam that accounts for bond degradation. This is because overall behavior of the beam (deflection) primarily depends on high temperature properties of FRP. From Fig. 14, which illustrates the variation of tensile strength of FRP with temperature, it can be seen that FRP strength reduces significantly when the temperature range is beyond 300–400 °C. Therefore, the strength and stiffness properties of FRP are not much affected for the duration of fire exposure time since insulation works efficiently in keeping FRP temperatures

Fig. 16. Moment capacity of FRP-strengthened and RC beam as function of fire exposure time.

Fig. 17. Variation of interfacial shear stress as a function of fire exposure time.

Fig. 18. Slip distribution for mid-span of the beam as a function of fire exposure time.

sufficiently low. This results in relatively low deflections in FRP-RC beam with a perfect bond. This analysis also illustrates that behavior of the FRP-strengthened RC beam predicted with a perfect bond is not realistic. For the case of FRP-strengthened beam, where slip is accounted for in the analysis, it can be noticed from Fig. 13 that debonding of FRP occurred at around 40 min. This can be attributed to the loss of bond when the temperature reaches glass transition temperature of the adhesive (Tg of adhesive is 81 °C). An examining of results from thermal analysis indicated that temperature at the interface

A. Ahmed, V.K.R. Kodur / Composites: Part B 42 (2011) 226–237

Fig. 19. Effect of adhesive thickness on slip at FRP–concrete interface as function of fire exposure time.

of FRP–concrete (Tfrp–concrete) reached 83 °C at about 40 min (see Fig. 15). It can be seen from the results presented in Fig. 13 that the effect of slip starts around 25 min into fire exposure ðT frp—concrete was around 46  CÞ and full debonding occurred around 40 min when temperature at the bond interface exceeds Tg. At this stage, stiffness of the beam decreases significantly, leading to increase in deflections at a faster rate. However, rate of increase in deflection in this FRP-strengthened RC beam and un-strengthened insulated RC beam is much slower than that in RC beam (no insulation) which is mainly due to beneficial effect of insulation that slows the temperature rise in steel reinforcement leading to a slower stiffness degradation. For initial 20–30 min of fire exposure, the behavior of unstrengthened insulated RC beam is similar to that of RC beam with no external fire protection. This is because in absence of any strengthening in these beams, flexural steel reinforcement that mainly contribute to moment capacity, maintains its full strength due to slower rebar temperature increase resulting from effective protection provided by the concrete cover. In later stages with increasing temperatures, fire insulation continues to protect the un-strengthened insulated RC beam, while non-insulated RC beam loses much of its strength and stiffness in the absence of externally applied fire protection. This leads to rapid increase in deflections in un-insulated RC beam, as shown in Fig. 13. It can also be noticed that after FRP fully debonds, the deflections in FRP-strengthened RC beam and insulated un-strengthened RC beam closely match. This is because, after debonding, FRP does not contribute towards capacity of the beam and the steel rebars are the one that carry tensile forces. Thus, FRP-strengthened RC beam behaves similar to un-strengthened insulated RC beam. Effect of slip on strength of FRP-strengthened RC beams is illustrated in Fig. 16 where the variation of moment capacity is plotted as a function of time for two cases of the FRP-strengthened RC beam. In both cases, the behavior of the beams is identical for early stages of fire exposure (up to 20 min) where there is slight increase in moment capacity of the beams followed by reduction. At room temperature, the behavior of an externally bonded FRPstrengthened beam is stiffer due to applied FRP reinforcement and this result in a non uniform distribution of curvature along the beam length for a given applied loading (minimal towards the supports). Therefore, full capacity of FRP reinforcement at beam soffit is not utilized. With increasing fire temperatures, concrete on two sides of the beam (upper zone) which is directly exposed to heat flux in absence of any fire protection, starts to lose its strength and stiffness and this introduces more ductility in the beam (curvature increases with time). At this stage, the strain distribution along FRP reinforcement is more uniform. This uniform strain distribution results in an increase in tensile force in FRP

235

which leads to a slight increase in moment capacity. However, with further increasing temperature at FRP–concrete interface, properties of FRP and the adhesive starts to degrade and this reduces the moment capacity of the FRP-strengthened beam. The adhesive loses its strength and stiffness with temperature, and composite action between FRP and concrete is lost when the temperature approaches glass transition temperature. At this point, results show an abrupt decrease in the moment capacity of FRP-strengthened beam that accounts for bond degradation. However, for FRP-strengthened beam with a perfect bond, the moment capacity of the beam decreases almost linearly since strength and stiffness of FRP degrade with temperature. The variation of interfacial shear stress and strain (bond–slip) distribution over half the span of the beam for various fire exposure times is presented in Figs. 17 and 18. It can be noticed that the peak interfacial shear stress and strain (eslip) occurs in the vicinity of FRP termination zones (edges) and are relatively more uniformly distributed in portions away from plate ends. As expected, shear stresses predicted at the FRP end are zero and matches the free surface stress condition. As the temperature at interface increases with fire exposure time, the maximum interfacial shear stress and corresponding strain (eslip) also increase towards plate end. These increasing shear stresses contribute to debonding of FRP when its magnitude exceeds decreasing shear capacity of the adhesive with temperature. Also, with increase in temperature, strain distribution along bonded length of FRP becomes more uniform which can be attributed to reduction in stiffness of the adhesive with temperature. As the temperature reaches close to glass transition temperature, magnitude of peak strains increases significantly in FRP end zones just prior to FRP delamination as shown in the Fig. 18. These trends in shear stress and strains (eslip) distribution predicted by the model closely follow the trends from test results conducted by Denton, and Klamer et al. [32,47]. From the above analysis, it can be concluded that the strength and stiffness of the adhesive reduces considerably at temperatures close to glass transition temperature. This results in peak shear stress and strain (eslip) concentration near the end of FRP composite that ultimately contribute to initiation of FRP debonding. 7. Effect of adhesive thickness on bond degradation To study the effect of adhesive thickness on temperature induced debonding of FRP, an FRP-strengthened RC beam was analyzed for varying thickness of adhesive from 1 to 4 mm. The beam had same characteristics as that of Beam III. Fig. 19 shows the deflection–time curves for the four cases of insulation thicknesses and for the case of fully bonded FRP beam. Results from the analysis indicate that up to first 20 min of fire exposure, there is no noticeable effect of adhesive thickness on time–deflection response. Beyond 20 min when debonding starts to occur, insulation thickness has minor influence on the deflections. For an increased adhesive thickness, bond–slip starts to occur in an earlier fire exposure time and as a consequence, the beam deforms slightly more. However, irrespective of adhesive thickness, beam experiences similar deflection after debonding of FRP. Therefore, adhesive layer thickness does not have significant effect on bond degradation and fire resistance of FRP-strengthened RC beam. 8. Effect of insulation thickness on bond degradation From above study, it clear that the bond degradation is a function of temperature at the FRP–concrete interface. The interface temperature depends on the insulation thickness. To investigate effect of insulation schemes on bond degradation of FRP-strengthened RC beams, beam designated as Beam III, has been analyzed with

236

A. Ahmed, V.K.R. Kodur / Composites: Part B 42 (2011) 226–237

 In computing moment capacity of a fire exposed insulated FRPstrengthened RC beam, a perfect bond can be assumed till temperature at FRP–concrete interface (adhesive) reaches Tg. Beyond Tg, FRP contribution can be taken as zero towards the moment capacity. Acknowledgments The research presented in this paper is supported by the National Science Foundation (Grant No. CMMI 0855820) and Michigan State University through Strategic Partnership Grant (Award No. SPG 71-4434). Any opinions, findings, and conclusions or recommendations expressed in this paper are those of the authors and do not necessarily reflect the views of the sponsors. Appendix A Fig. 20. Effect of insulation thickness on time to reach Tg.

High temperature properties of carbon/epoxy (CFRP) [43]. Thermal conductivity (kw,T). varying insulation thicknesses. The dimensions and material properties of the beams are tabulated in Table 1. The beam (Beam III) is provided with supplement insulation of varying thicknesses and configuration schemes. On the sides of the beam, the insulation thickness (20 mm) and application depth (105 mm) is kept consistent. However, insulation thickness at the beam soffit was varied to be 15, 25, 40, and 50 mm, respectively. The analysis was carried out by exposing the beams to the standard ASTM E119 fire from three sides. The applied load ratio on the beam was kept constant at 52% for all the cases. Time to reach glass transition temperature (Tg) of the adhesive was determined. Fig. 20 shows effect of insulation thickness on time to reach glass transition temperature of the adhesive. As expected, time to reach Tg increases with increasing insulation thickness. An increase in insulation thickness from 15 to 40 mm enhances time to reach Tg by about 70 min. This can be attributed to the low thermal conductivity of the fire insulation that helps to keep temperatures low at the FRP–concrete interface. This study shows that for FRPstrengthened structural members, where reaching Tg is critical for structural performance, provision of external fire protection of appropriate thickness is necessary. Numerical models, like the one discussed in this paper, will help practitioners to arrive at optimum fire insulation scheme for a given fire resistance application. 9. Conclusions Based on the information presented, the following conclusions can be drawn:  FRP-strengthened beams experience significant degradation in moment capacity and stiffness when the temperature at FRP– concrete interface exceeds glass transition temperature.  FRP-strengthened RC beam protected with insulation, attains lower deflection under fire conditions as compared to an unstrengthened RC beam, due to beneficial effect of external insulation that slows the temperature rise and strength loss in steel reinforcement.  Adhesive thickness does not have significant influence on bond properties and thus the fire resistance of FRP-strengthened RC beam is not influenced by adhesive thickness.  In fire design of FRP-strengthened RC beams, it is conservative to assume that debonding of FRP occurs at Tg. An insulated FRP-strengthened RC beam even after debonding exhibits better fire performance (higher strength capacity, fire resistance and lower deflections) than an un-strengthened RC beam.

Temperature range (Tw) is °C

Thermal conductivity (kw,T) in W/m°C

0 6 Tw 6 500

kw;T ¼ 1:4 þ 1:1 500  T w

500 6 Tw 6 650

kw;T ¼ 1:4 þ 0:1 150  ðT w  500Þ kw,T = 0.2

Tw P 650

Specific heat (Cw,T). Temperature range (Tw) is °C

Specific heat (Cw,T) is kJ/kg °C

0 6 Tw 6 325

C w;T ¼ 1:25 þ 0:95 325  ðT w Þ

325 6 Tw 6 343

C w;T ¼ 2:2 þ 2:8 18  ðT w  325Þ

343 6 Tw 6 510

C w;T ¼ 5:0 þ 0:15 167  ðT w  343Þ

510 6 Tw 6 538

C w;T ¼ 4:85 þ 3:59 28  ðT w  510Þ

538 6 Tw 6 3316

C w;T ¼ 1:265 þ 1:385 2778  ðT w  538Þ Cw,T = 0

Tw P 3316

Density (qw,T). Temperature range (Tw) is °C

Density (qw,T)is g/cm3

0 6 Tw 6 510

qw,T = 1.6 qw;T ¼ 1:6 þ 0:35 28  ðT w  510Þ qw,T = 1.25

510 6 Tw 6 538

538 6 Tw 6 1200

Tensile strength (fcom,T) in MPa.

fcom;T ¼ fcom

  1  ar 1 þ ar tanh½br ðT w  cr Þ þ 2 2

Elastic Modulus, ðEcom;T Þ in MPa.

 Ecom;T ¼ Ecom

1  aE 1 þ aE tanh½bE ðT w  cE Þ þ 2 2

where ar ¼ 0:1; br ¼ 5:83e  3; bE ¼ 8:68e  3; cE ¼ 367:41



cr ¼ 339:54; aE ¼ 0:05;

References [1] Triantafillou TC, Plevris N. Strengthening of RC beams with epoxy-bonded fibre-composite materials. Mater Struct 1992;25(4):201–11.

A. Ahmed, V.K.R. Kodur / Composites: Part B 42 (2011) 226–237 [2] Dai J, Ueda T, Sato Y. Development of the nonlinear bond stress–slip model of fiber reinforced plastics sheet–concrete interfaces with a simple method. J Compos Construct 2005;9:52. [3] Dai JG, Ueda T. Local bond stress slip relations for FRP sheets–concrete interfaces. In: 6th International symposium on FRP reinforcement for concrete structures. Singapore: World Scientific; 2003. p. 143–52. [4] 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. [5] Monti G, Renzelli M, Luciani P. FRP adhesion in uncracked and cracked concrete zones. In: 6th International symposium on FRP reinforcement for concrete structures. Singapore: World Scientific; 2003. p. 183–92. [6] Nakaba K, Kanakubo T, Furuta T, Yoshizawa H. Bond behavior between fiberreinforced polymer laminates and concrete. ACI Struct J 2001;98(3):359–67. [7] Neubauer U, Rostasy FS. Bond failure of concrete fiber reinforced polymer plates at inclined cracks – experiments and fracture mechanics model. American Concrete Institute Special Publ 1999 [SP-188:369-82]. [8] Savoia M, Ferracuti B, Mazzotti C. Non linear bond–slip law for FRP–concrete interface. In: 6th International symposium on FRP reinforcement for concrete structures. Singapore; 2003. p. 163–72. [9] Ueda T, Dai J, Sato Y. A nonlinear bond stress–slip relationship for FRP sheet– concrete interface. In: International symposium on latest achievements of technology and research on retrofitting concrete structures; 2003. p. 113–20. [10] Borchert K, Zilch K. Time depending thermo mechanical bond behavior of epoxy bonded pre-stressed FRP-reinforcement. In: Proceedings of the 7th international symposium on fiber reinforced polymer (FRP) reinforcement for concrete structures (FRPRCS7); 2005. p. 671–84. [11] Gluguru. Hysol epoxy/structural adhesives; 2009 (retrieved on 7.12.09). [12] Camata G, Pasquini F, Spacone E. High temperature flexural strengthening with externally bonded FRP reinforcement. In: Proceedings of 8th international symposium on fiber reinforced polymer (FRP) reinforcement for concrete structures (FRP8RCS). Patras, Greece; 2007. p. 1–10. [13] Chen JF, Teng JG. Anchorage strength models for FRP and steel plates bonded to concrete. J Struct Eng 2001;127(7):784–91. [14] Ueda T, Dai J, Sato Y. A nonlinear bond stress–slip relationship for FRP sheet– concrete interface. In: Proceedings of international symposium on latest achievement of technology and research on retrofitting concrete structures. Kyoto, Japan; 2003. p. 113–20. [15] Yuan H, Teng JG, Seracino R, Wu ZS, Yao J. Full-range behavior of FRP-toconcrete bonded joints. Eng Struct 2004;26(5):553–65. [16] American Concrete Institute (ACI). Guide for the design and construction of externally bonded FRP systems for strengthening concrete structures. Farmington Hills, MI: ACI 440.2R-02, ACI Committee 440; 2002. p. 1–45. [17] Blaschko M, Niedermeier R, Zilch K. Bond failure modes of flexural members strengthened with FRP. Fiber Compos Infrastruct 1998:315–27. [18] Hiroyuki Y, Wu Z. Analysis of debonding fracture properties of CFS strengthened member subject to tension. In: Proceedings of 3rd international symposium. Sapporo, Japan; 1997. p. 287–94. [19] Maeda T, Asano Y, Sato Y, Ueda T, Kakuta Y. A study on bond mechanism of carbon fiber sheet. In: Proceedings of 3rd international symposium. Sapporo, Japan; 1997. p. 279–86. [20] Sato Y, Kimura K, Kobatake Y. Bond behavior between CFRP sheet and concrete part 1. J Struct Construct Eng 1997;500:75–82. [21] Tanaka T. Shear resisting mechanism of reinforced concrete beams with CFS as shear reinforcement. Graduation thesis, Hokkaido University, Japan; 1996. [22] Yuan H, Wu Z. Interfacial fracture theory in structures strengthened with composite of continuous fiber. In: Proceedings of the symposium of China and Japan, science and technology of the 21st century. Tokyo, Japan; 1999. p. 142– 55. [23] Yuan H, Wu Z, Yoshizawa H. Theoretical solutions on interfacial stress transfer of externally bonded steel/composite laminates. Struct Earthquake Eng 2001;18(1):27–40. [24] Tadeu AJB, Branco FJFG. Shear test of steel plates epoxy-bonded to concrete under temperatures. J Mater Civil Eng 2000;12(1):74–80.

237

[25] Blontrock H, Taerwe L, Vanwalleghem H. Bond testing of externally glued FRP laminates at elevated temperature. In: Proceeding of the international conference bond in concrete – from research to standards. Budapest, Hungary; 2002. p. 648–54. [26] Klamer EL, Hordijk DA, Janssen HJM. The influence of temperature on the debonding of externally bonded CFRP. In: Proceedings of 7th international symposium on fiber reinforced polymer (FRP) reinforcement for concrete structures (FRP7RCS). Kansas City; 2005. p. 1551–92. [27] Klamer EL, Hordijk DA, Hermes MCJ. The influence of temperature on RC beams strengthened with externally bonded CFRP reinforcement. Heron 2008;53(3):157–85. [28] Leone M, Matthys S, Aiello MA. Effect of elevated service temperature on bond between FRP EBR systems and concrete. Composites Part B 2009;40(1):85–93. [29] Wu ZS, Iwashita K, Yagashiro S, Ishikawa T, Hamaguchi Y. Temperature effect on bonding and debonding behaviour between FRP sheets and concrete. FRP composites in civil engineering (CICE); 2004. p. 905–12. [30] Gamage J, Al-Mahaidi R, Wong MB. Bond characteristics of CFRP plated concrete members under elevated temperatures. Compos Struct 2006;75(1– 4):199–205. [31] Di Tommaso A, Neubauer U, Pantuso A, Rostasy FS. Behaviour of adhesively bonded concrete-CFRP joints at low and high temperatures. Mech Compos Mater 2001;37(4):327–38. [32] Denton SR. Analysis of stresses developed in FRP plated beams due to thermal effects. FRP Compos Civil Eng 2001;1:527–36. [33] Eurocode 2. EN 1992-1-2: design of concrete structures. Part 1–2: general rules – structural fire design. Brussels (Belgium): European Committee for Standardization; 2004. [34] Kodur VKR, Ahmed A, Dwaikat MB. Modeling the fire performance of FRPstrengthened reinforced concrete beams. Composite & polycon, American composites manufacturers association (ACMA). Tampa, Florida; 2009. [35] Kodur VKR, Ahmed A. A numerical model for tracing the response of FRPstrengthened reinforced concrete beams exposed to fire. ASCE J Compos Construct 2010;14(6):730–42. [36] Harmathy TZ. A comprehensive creep model. J Basic Eng 1967;89(3):496–502. [37] Harmathy TZ. Fire safety design and concrete. New York (NY): John Wiley & Sons, Inc.; 1993. [38] Anderberg Y, Thelandersson S. Stress and deformation characteristics of concrete at high temperatures, 2-experimental investigation and material behaviour model. Lund, Sweden Bulletin 54, Lund Institute of Technology; 1976. [39] Rahman AH, Taylor DA, Kingsley CY. Evaluation of FRP as reinforcement for concrete bridges. ACI Special Publ 1993;138:71–86. [40] Campbell TI, Kodur VKR. Deformation controlled nonlinear analysis of prestressed concrete continuous beams. PCI J 1990;35(5):42–90. [41] Dwaikat MB. Flexural response of reinforced concrete beams exposed to fire. Doctoral thesis, Michigan State University, East Lansing, Michigan (USA); 2009. [42] Lie TT. Structural fire protection. New York (NY): ASCE Committee on Fire Protection, Structural Division, American Society of Civil Engineers; 1992. p. 225–9. [43] Bisby LA. Fire behavior of fibre-reinforced polymer (FRP) reinforced or confined concrete. Doctoral thesis, Queen’s University, Kingston (Canada); 2003. [44] Griffis CA, Masumura RA, Chang CI. Thermal response of graphite epoxy composite subjected to rapid heating. J Compos Mater 1981;15(5):427. [45] Blontrock H, Taerwe L, Vandevelde P. Fire tests on concrete beams strengthened with fibre composite laminates. Third PhD symposium. Vienna, Austria; 2000. [46] Ahmed A, Kodur VKR. Performance of FRP-strengthened reinforced concrete beams under design fire exposure. In: Proceedings of sixth international conference on structures in fire. East Lansing, Michigan; 2010. [47] Klamer EL, Hordijk DA, Janssen HJM. The influence of temperature on debonding of externally bonded CFRP. Special Publ 2005;230:1551–70. [48] GangaRao HVS, Taly N, Vijay PV. Reinforced concrete design with FRP composites. Boca Raton (FL): CRC Press; 2007.