Engineering Structures 33 (2011) 2201–2211

Contents lists available at ScienceDirect

Engineering Structures

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

The experimental behavior of FRP-strengthened RC beams subjected to designfire exposureAqeel Ahmed, Venkatesh Kodur ∗

Department of Civil and Environmental Engineering, Michigan State University, United States

a r t i c l e i n f o

Article history:Received 27 July 2010Received in revised form16 March 2011Accepted 18 March 2011Available online 21 April 2011

Keywords:FRP-strengthened RC beamsFire testsFire resistanceDesign fireNumerical modelRestrained RC beams

a b s t r a c t

The results from fire resistance experiments on five rectangular reinforced concrete (RC) beams arepresented in this paper. Four of these RC beams were tested after being strengthened with carbonfiber reinforced polymer (CFRP), while the remaining one was tested as a control RC beam specimen.The beams were tested by exposing them to fire and service load, computed based on the nominalcapacity of an unstrengthened/strengthened beam in accordancewith ACI 318/ACI 440.2R provisions. Thetest variables included type of fire exposure, anchorage zone, insulation type, and restraint conditions.The data from the fire tests is used to evaluate the thermal and structural response, as well as failurepatterns in FRP-strengthened RC beams. The test results indicate that the anchorage configuration playsa critical role in limiting the deflections of the strengthened beam after debonding of the FRP occurs atTg ± 10 °C, where Tg is the glass transition temperature. Also, FRP-strengthened RC beams supplementedwith 25 mm thick spray-applied insulation can survive failure under ASTM E119 standard fire or adesign fire. Further, the fire-induced axial restraint force significantly increases the fire resistance ofFRP-strengthened/unstrengthened RC beams, provided that the location of restraining force is below thegeometric centroid of the beam and that the beam’s deflection is sufficiently small along the entire span.

© 2011 Elsevier Ltd. All rights reserved.

1. Introduction

In recent years, fiber reinforced polymers (FRPs) have foundwide use in civil infrastructure applications, and they are oftenused for the strengthening and retrofitting of RC beams, due totheir high strength, durability, and ease of application. Currently,most applications of FRPs are in bridges and other infrastructurewhere fire safety is not a critical issue. However, when usedin buildings, FRP-strengthened RC beams have to satisfy fireresistance requirements specified in building codes and standards.

In the last decade, a few fire tests have been conducted on FRP-strengthened RC beams to generate fire endurance ratings. Most ofthese tests were under standard fire exposure aimed at obtainingfire resistance ratings, rather than studying the response of FRP-strengthened members under fire conditions. Thus, there is a lackof understanding of the response of FRP-strengthened RC beamsunder realistic fire, loading, and failure limit states. This lack of firetest data and fire design methods is posing a major obstacle forwider use of FRP in buildings and parking structures [1].

To overcome the current knowledge gaps, a research project iscurrently underway atMichigan StateUniversity (MSU) to quantifythe influence of various parameters on the fire performance of

∗ Corresponding author. Tel.: +1 517 353 9813.E-mail addresses: kodur@msu.edu, kodur@egr.msu.edu (V. Kodur).

0141-0296/$ – see front matter© 2011 Elsevier Ltd. All rights reserved.doi:10.1016/j.engstruct.2011.03.010

FRP-strengthened RC beams. This paper presents the results fromfire tests on five beams exposed to standard and design fire(non-standard fire) conditions. The effectiveness of new insulationsystems, anchorage patterns, axial restraint effects, and designfire exposure on fire resistance of FRP-strengthened RC beams isspecifically evaluated.

2. Research significance

The main impediment for the use of FRP in buildings is due tolack of information on its behavior under elevated temperatures.Most studies reported in the literature have focused on evaluatingthe fire resistance of FRP-strengthened RC members understandard fire conditions, and this led to developing proprietary fireresistance ratings. Currently, specifications in codes and standards(including ACI 440-R2) are based on standard fire conditions, andthey often assume FRP to be completely ineffective in the event offire. The research presented herein aims to quantify performanceof FRP-strengthened RC beams under realistic fire scenarios. Thedata from the fire tests is utilized to study the influence of criticalparameters such as the effects of the anchorage zone, non-standardfire exposure, and fire-induced axial restraint force on fire responseof FRP-strengthened RC beams. The results presented in this paperoffer an insight into the fire response of FRP-strengthened RCbeams under non-standard fire exposure and also provide valuabledata for validating numerical models. The validated models can be

2202 A. Ahmed, V. Kodur / Engineering Structures 33 (2011) 2201–2211

applied to undertake parametric studies, data from which can beused to develop guidelines for fire design of FRP-strengthened RCbeams. Such guidelines will be an advancement over the currentlyavailable prescriptive fire provisions (as in ACI 440-R2) which arebased on standard fire exposure only.

3. Previous research

The few notable experimental studies on FRP-strengthened RCbeams were conducted by Deuring [2], Blontrock et al. [3], Barnesand Fidell [4], and Williams et al. [5]. Deuring conducted firetests on six RC beams, strengthened with external carbon fiberreinforced polymer (CFRP) strips and steel plates, under ISO 834standard fire exposure to assess the post-fire residual strength ofthe beams. Through these fire tests, Deuring demonstrated that theinteraction between FRP and concrete is lost in the early stages offire exposure. Thus, the author recommended the application ofexternal thermal insulation to maintain an effective bond underfire conditions.

Blontrock et al. [3] conducted fire tests on FRP-strengthenedRC beams using a variety of supplemental fire insulation schemes.Based on the results from the fire tests, the authors concluded thatsupplemental fire insulation is needed for externally bonded FRPto limit the temperature rise at the FRP/concrete interface to bebelow the glass transition temperature (Tg). The glass transitiontemperature is the temperature at which the polymer goes from ahard, glass-like state to a rubber-like state. The two primary testprocedures to determine Tg are differential scanning calorimetry(DSC) and thermo-mechanical analysis (TMA). Typically, the glasstransition temperature for commonly used polymers (epoxy)varies in the range 60–82 °C [1].

Williams et al. [5,6] conducted full-scale fire tests on four FRP-strengthened beam–slab assemblies (T-beams) protected withvermiculite-gypsum (VG) insulation and a top layer intumescentcoating (EI-R), under standard fire exposure. 100mmwide layers ofCFRPwere applied for flexural strengthening. The study concludedthat FRP-strengthened RC beams with an appropriate thicknessof 38 mm VG-EI-R insulation are capable of achieving 4 h of fireendurance rating, although Tg of the adhesivewas exceededwithin1 h into the test.

The above review illustrates that there have been only limitedexperiments reported, and these have been conducted to evaluatethe fire resistance ratings of FRP-strengthened RC beams. Mostof the previous tests were on RC beams strengthened withboth flexural and shear strengthening. However, in practicalapplications the beams are often strengthened for flexure alonein buildings. Furthermore, the previous fire tests were conductedunder standard fire exposure without any due consideration torealistic fire, loading, and restraint scenarios. In addition, there isa lack of data on FRP-strengthened beams protected with the newand improved (Tyfo r⃝ WR AFP) fire insulation system.

4. Experimental studies

To generate test data, fire resistance experiments on four CFRP-strengthenedRCbeamswere conducted. For comparison purposes,an additional RC beam was also tested under fire conditions. Thetest variables included fire scenario, axial restraint, anchoragezone, and insulation type.

4.1. Fabrication of test specimens

Five rectangular RC beams were fabricated for undertakingthe fire resistance tests.The beams were designed to be as close

to typical building geometries as possible, in order to maximizethe usefulness of the test results. The beams were of 254 mmwidth and 406 mm depth and had 3.96 m span length. The RCbeams were designed as per ACI 318 [7] specifications and werefabricated at the Civil Infrastructure Laboratory at Michigan StateUniversity. The RCbeamshad three#6 rebars (Grade 60) as flexuralreinforcement and two #4 rebars as compression reinforcement.The stirrups used as shear reinforcement were #2 rebars (6 mmdiameter), which are typically used in RC beams where the sheardemand is not significant, and these were spaced at 150 mmintervals. The elevation and cross-sectional details of the beams areshown in Fig. 1.

The RC beams were cast with concrete that had a designcompressive strength of 42 MPa. Type I Portland cement andcarbonate-based coarse aggregate was used in the batch mix. Themeasured compressive strength of the concrete at 28 days was52 MPa, while on the day of the test (2 years or more), it was55 MPa. The mix proportion of concrete comprised the following.

• Cement 390 kg/m3.• Coarse aggregate 1038 kg/m3.• Fine aggregate 830 kg/m3.• Water 156 kg/m3.• Water reducing agent 2 kg/m3.

The cast beamswere sealedwithin the forms for the first 7 days.Thereafter, the beams were lifted out from the forms and storedin the laboratory at 25 °C (40% relative humidity) for 18 monthsbefore applying CFRP for flexural strengthening. The details of thespecimens, test parameters, and properties of CFRP are provided inTables 1 and 2.

Four of the RC beams (B1–B4) were strengthened with CFRPsheets to enhance their flexural strength capacity by 50%. Theconcrete surface, which was fairly smooth due to even form work,was roughened through sand blasting to partially expose theaggregate at the beam soffit. Then two CFRP sheets, each of 1 mmthickness, 203 mm width and saturated in S-epoxy, were roller-applied at the beam soffit for flexural strengthening. The resin usedto bond fibers was two-component epoxy material with a glasstransition temperature (Tg) of 82 °C. Tg of the resin was evaluatedby the manufacturer as per ASTM-D4065 specifications [8,9].Table 3 provides relevant details of the S-epoxy properties. Forbeams B1 and B2, CFRP was applied to the entire unsupportedlength of the beam (3.66 m) terminating at a distance d from thesupports. For beams B3 and B4, only the center 2.44 m of the beamwhich was exposed to fire was retrofitted with CFRP. This varyingconfiguration of FRP layout was adopted to study the influence ofcooler anchorage zones on the fire response of FRP-strengthenedRC beams. Unlike in most of previous tests, no external shearstrengthening was provided, in order to study the failure patternsunder flexural strengthening only.

The strengthened beams were cured for 72 h and thenspray-applied with a Tyfo r⃝ weather resistant (WR) advance fireprotection (AFP) system, an improved version as compared tothe previously developed Tyfo r⃝ AFP system [5]. The Tyfo r⃝ WRAFP system is comprised of vermiculite-gypsum (VG) insulationand a top layer intumescent coating (EI-R). It is available in twoforms, Tyfo r⃝ WR AFP-Type A and Tyfo r⃝ WR AFP-Type B. Thisinsulation is a non-combustible and a non-flammable lightweightmaterial available in powder form. Beams B1, B3 and B4 werespray-applied with Type A insulation, while Type B insulation wasused for beam B2. On the top surface of the insulation, a spray-on EI-R coating was applied. This EI-R is a crack-resistant surfacecoating with excellent adhesion and fire resistance properties thatprovides additional stability to insulation. The insulation layoutwas comprised of 25mm of insulation thickness at the beam soffit,extending 100 mm on the two sides (refer to Fig. 1(c)).

A. Ahmed, V. Kodur / Engineering Structures 33 (2011) 2201–2211 2203

(a) Elevation.

(b) Cross-sectional details and location of thermocouples and strain gauges.

(c) Mid-span cross section showing location of thermocouples.

Fig. 1. Elevation and cross-sectional details of tested FRP-strengthened RC beam.

Table 1Summary of test parameters and results.

Beamdesignation

CFRP strengthening Insulation type Insulation thickness(mm)

Fire scenario Support condition Load (kN) Failure time(min)

VG# EI-R Test Model

B01 – – – – ASTM E119 SSb 50 180 182B02 – – – – Design fire ARc 50 NFa NFB1 2 layers of 203 mmwide Tyfo r⃝ WR AFP%-Type A 25 0.1 Design fire SS 70 NF NFB2 Tyfo r⃝ WR AFP-Type B 25 0.1 Design fire SS 70 NF NFB3 Tyfo r⃝ WR AFP-Type A 25 0.1 ASTM E119 SS 70 NF NFB4 Tyfo r⃝ WR AFP-Type A 25 0.1 ASTM E119 AR 70 NF NF

% Advance fire protection system; # Vermiculite-gypsum.a NF — no failure.b Simply supported.c Axially restrained.

The insulation application comprised spraying a thin coat of VGprimer, followed by a dash coat, on the FRP-strengthened beamsoffit to enhance the FRP/insulation bond. Thereafter, insulationmaterial, which is available in powdered form, was mixed withan appropriate amount of clear water and spray-applied on thebeams using a hopper gun. Special attentionwas given tomaintainuniform insulation thickness throughout the beam length. This

sprayed insulation was cured for 24 h before a final coat of EI-Rwas applied.

The beams were instrumented with Type-K thermocouples atthree different sections along the span of the beam to measurethe temperatures at various depths in the concrete, reinforcement,and FRP/concrete and FRP/insulation interfaces. A total of 27 ther-mocouples were installed for each beam. In addition, normal- and

2204 A. Ahmed, V. Kodur / Engineering Structures 33 (2011) 2201–2211

Fig. 2. Structural fire test furnace and loading setup at MSU Civil and Infrastructure Laboratory.

Table 2Properties of composite laminate.

Property ASTMmethod Typical test value

Ultimate tensile strength inprimary fiber direction

D-3039 986 MPa

Elongation at break D-3039 1.0%Tensile modulus D-3039 95.8 GPaNominal laminate thickness D-1777 1.0 mm

Table 3Properties of epoxy used in FRP strengthening.

Property ASTMmethod Typical test value

Tg D-4065 82 °CTensile strength D-638 72.4 MPaTensile modulus D-638 3.18 GPaElongation percent D-638 5.0%Flexural strength D-790 123.4 MPaFlexural modulus D-790 3.12 GPa

high-temperature strain gauges of type ZFLA-6-11 (specified by themanufacturer to be reliable up to 300 °C)were installed tomeasurethe strains in the compression and tension rebars, respectively. Thelocations and numbering of thermocouples and strain gauges areshown in Fig. 1(b). Also, three linear variable differential transduc-ers (LVDTs) were installed on each beam, one at mid-span and twounder point loads, to measure the deflections.

4.2. Test setup and procedure

The fire resistance tests on FRP-strengthened RC beams wereconducted at MSU’s structural fire testing furnace. The test furnaceis specially designed to produce conditions, such as temperature,structural loads, and heat transfer, to which a member mightbe exposed during a fire. The furnace, shown in Fig. 2, consistsof a steel framework supported by four steel columns, with afire chamber that is 2.44 m wide, 3.05 m long, and 1.68 mhigh. Six propane burners located within the furnace providethermal energy, while six thermocouples, distributed throughoutthe test chamber, monitor the furnace temperature during a firetest. The furnace temperature can be maintained along a desiredtime–temperature curve as in a standard or design fire. Two smallview ports on either side of the furnace wall facilitate visualmonitoring of the fire-exposed test specimens during fire tests.The furnace accommodates two beams at a time, and differentload levels and restraint conditions can be simulated for eachbeam. One of the two beams can be tested under axial restraint

support conditions, while the other beam has to be tested undersimply supported end conditions. The axial restraint stiffness of theloading frame was found to be 13 kN/mm. To minimize damage tothe loading frame, the stiffness of the axial restraint system is setto adjust automatically to zero (using a relief valve) when the axialrestraint force exceeds a value of 120 kN.

In each test, two beams were tested simultaneously underloading and fire conditions. The beams were simply supported atthe ends with an unsupported length of 3.66 m, of which 2.44 mwas exposed to fire in the furnace. To investigate the effect of firescenario on the response, the beams were tested under standardand design (non-standard) fire scenarios. Beams B1 and B2 weretested under a design fire exposure, while beams B01, B3, andB4 were tested under ASTM E119 standard fire. All beams weretested under four-point bending. The unstrengthened RC beam(B01) was loaded with two point loads, each of 50 kN, whilethe strengthened FRP-RC beams had applied loading of 70 kN.This loading represents 55% of the unstrengthened and 50% ofthe strengthened beam nominal capacity computed in accordancewith ACI 318 [7] and ACI 4402R [1] specifications. The point loadswere applied at a distance of 1.4m from the end supports, as shownin Fig. 1(a). The loading was applied approximately 30min prior tothe start of fire until a steady condition (no increase in deflectionwith time) was reached. This was selected as the initial conditionfor measuring deflections in the beam.

After 30 min of loading, the beams were exposed to firefrom three sides. Beams B1 and B2 were exposed to a design(non-standard) fire that comprised a rising temperature (growth)phase followed by a cooling phase. The provisions of Eurocode1 [10] were utilized to arrive at the design fire (non-standardfire) used in the test. The heating phase of the fire is influencedby the compartment characteristics including fuel load (amountof combustible materials), ventilation openings, and wall liningmaterials [11]. The selected design fire for the test corresponds toEurocode 1 parametric fire for a room with dimensions of 6 m ×

4 m × 3 m having a fuel load of 600 MJ/m2. Further, it is assumedthat the ventilation factor increases to 0.2 after attaining peaktemperature due to breakage of windows (glass). These fuel loadand ventilation characteristics are representative of a typical officecompartment. The values of fuel load and opening dimensionsare given in Table 4. The decay phase of the fire is generated byproviding a constant cooling rate of 10 °C/min. Thus, the growthand decay phase of the fire used in tests represent one possibledesign (parametric) fire scenario as per Eurocode 1 [10] provisions.

Beams B01, B3, and B4 were tested under ASTM E119 standardfire exposure. For the duration of the fire test, the applied loading

A. Ahmed, V. Kodur / Engineering Structures 33 (2011) 2201–2211 2205

Fig. 3. Measured temperatures as a function of fire exposure time for beams B1–B4.

a b

Fig. 4. Measured temperatures at the FRP/insulation and FRP/concrete interfaces.

Table 4Compartment and material characteristics assumed for developing design firescenarios.

Liningmaterial

Thermal capacity of liningmaterial (W s0.5/m2 K)

Openingdimension (m)

Fire load (MJ/m2

floor area)

Concrete 1900 2.25 × 1.5 600

was kept constant. During the tests, the temperatures at variouslocations of the beam cross section, strains, and deflections wererecorded at 5 s intervals. Also, visual observations were madethrough view ports in the furnace to record the progressionof cracks in the insulation, localized burning of the FRP, anddelamination of the insulation and FRP.

5. Test results and discussion

The data generated from the above fire tests can be usedto study the overall performance of FRP-strengthened RC beamsunder fire conditions. The thermal and structural responses ofbeams are compared to evaluate the effect of fire scenario,

insulation type, anchorage zones, and axial restraint force. The fireconditions represented a typical compartment fire in a buildingwhere a portion (center) of the beam is exposed to fire.

5.1. Thermal response

The thermal response of FRP-strengthened RC beams undertwo different fire scenarios can be studied by comparing thetemperature progression as illustrated in Fig. 3. Although onlytemperature measurements at Section B (mid-span) are presentedin Figs. 3, 4 and 7, an analysis of the test data showed similartemperature trends for Sections A and C. Fig. 3 shows that beamsB3 and B4, exposed to ASTM E119 standard fire, experiencedan increase in temperature at various cross section locationsuntil the test terminated after 180 min. For beams B1 and B2,exposed to design fire, the measured temperature increases toa maximum value and then starts to decrease. This decreasein temperature can be attributed to the decay (cooling) phasein the time–temperature curve of the design fire. Initially, thetemperature increases slowly, with a plateau around 100 °C, whichcan be attributed to evaporation of free and chemically bonded

2206 A. Ahmed, V. Kodur / Engineering Structures 33 (2011) 2201–2211

Fig. 5. Development of cracks in insulation — middle portion of FRP-strengthened RC beam.

Fig. 6. Left quarter span of beam B3 after falling off of insulation.

water in the insulation that consumes a significant amount ofenergy. The VG insulation, which is spray-applied in the formof slurry, contains both free and chemically bonded water. Mostof the free water in the insulation, close to the exposed surface,dries out at room temperature; however, the remaining free andchemically bonded water consumes significant energy from thefire to evaporate.

Fig. 4 shows temperature progression at the FRP/insulationand FRP/concrete interfaces for the tested FRP-strengthened RCbeams. The data in Fig. 4(a) shows a rapid increase in measuredtemperatures at the FRP/insulation interface immediately after themoisture evaporation induced plateau. This sudden temperatureincrease is attributed to the localized burning of the FRP matrixas a result of wide crack openings in the insulation (refer toFig. 5). Visual observation during the test indicated early crackpropagation in the insulation for beam B2, which was spray-applied with Type B insulation. Also, analysis of test data showsthat the temperature increase at both interfaces is slightly higher(in comparison to beams insulated with Type A insulation) withrelatively short duration in the temperature plateau. This can beattributed to the direct flow of heat flux due to early formationof cracks in the insulation that resulted in higher measuredtemperatures, as well as quick evaporation of water.

Fig. 4(b) shows that the temperature increase at the FRP/con-crete interface is slower, possibly due to the formation of aprotective char layer as a result of the pyrolysis process of thematrix [12]. This char layer acts as a thermal barrier and insulatesthe interior interface between the FRP and concrete, which iscritical in maintaining a bond at elevated temperatures. Forbeam B3, an abrupt increase in temperature was recorded at theFRP/concrete interface after 30 min. This rapid increase probablyoccurred due to the direct exposure of the thermocouple after aportion of the insulation and FRP was delaminated, as shown inFig. 6.

The fire resistance of typical FRP-strengthened flexural mem-bers is mainly influenced by the strength and stiffness propertiesof the FRP and steel reinforcement (since the temperatures in con-crete remain below 500 °C for most of the fire duration), and theseproperties degradewith temperature. In a FRP, the rate of degrada-tion of strength and stiffness properties ismuch faster as compared

Fig. 7. Comparison of corner rebar temperature as a function of fire exposure time.

to concrete and steel due to the low tolerance of the polymer ma-trix to high temperatures. Therefore, after the FRP loses its strength(bond) at high temperatures (mostly after reaching Tg ), the rebartemperatures become an important indicator of the fire resistanceof FRP-strengthened RC beams. Fig. 7 shows the temperature pro-gression in a corner rebar as a function of time for the five testedbeams, including the RC control beam. As expected, the test datashows that supplemental insulation notably influences the tem-perature increase in the rebars. The rebars in FRP-strengthenedRC beams are well insulated; therefore, they experience a steadyrise in temperature for the entire test duration, unlike the controlbeam, which was not insulated. This temperature trend can be at-tributed to the low thermal conductivity of the insulation material(Tyfo r⃝ WRAFP system) that helps to keep rebar temperatures low.

The measured rebar temperatures in beam B3 are higher ascompared to the other three insulated beams since 1/4 of thespan of beam soffit was directly exposed to fire after the FRPbecame delaminated (refer to Fig. 8). In beam B2, the steelreinforcement displayed slightly higher temperatures due to theearly development of cracks in the insulation. It can be seen thatthe rebar in the control RC beam reaches a high temperature valueof 593 °C in 180 min. However, in all FRP-strengthened RC beams,the rebar temperature ismuch below 400 °C at 3 h of fire exposure.This is due to the presence of insulation, which plays a vital role inkeeping the temperature low in rebars for the entire duration of

A. Ahmed, V. Kodur / Engineering Structures 33 (2011) 2201–2211 2207

Fig. 8. Comparison of mid-span deflection in FRP-strengthened RC beams.

the fire test. Since the rebars do not lose any significant strengthup to 400 °C, the steel reinforcementmaintains nearly full strength(flexural capacity) for the test duration [13].

5.2. Structural response

The structural response of the beams can be gauged throughdeflection progression with fire exposure time. Fig. 8 shows thevariation of mid-span deflection as a function of time for all thetested beams, including the control RC beam (B01). All beams,except beam B01, were loaded with two point loads (70 kNeach), which represent 50% of the ultimate capacity of the beamscalculated as per ACI 440.2R-08 provisions [1]. Beam B01 wassubjected to two point loads of 50 kN each, which represents 55%of the beam capacity according to ACI 318 [7].

Under fire, the mid-span deflection increased gradually untilaround 20 min. In the case of FRP-strengthened RC beams, themeasured deflection showed a sudden increase at 20 min. Thiscan be attributed to bond loss at the FRP/concrete interface withtemperature. A closer examination indicated that the temperaturein the FRP had exceeded the glass transition temperature of epoxy(Tg = 82 °C) at 20 min of fire exposure. When the temperature ofthe polymer matrix approaches the glass transition temperature,the matrix becomes soft and viscous. This results in a reducedcapability of the polymer matrix to transfer forces between thefibers, and interaction between the FRP and concrete substrate islost due to the degradation of the bond. This leads to a suddenincrease in deflections. This sudden jump in deflections at 24 minwas reconfirmed through deflection calculations that were carriedout using the crackedmoment of inertia of the transformed section.

After debonding of the FRP, the rate of deflection increase ishigher for beam B3 as compared to beams B1, B2, and B4. Thiscan be attributed to loss of strength and stiffness of the beamwith temperature, in addition to almost zero contribution of FRPtowards the capacity of the beam. However, it is interesting tonote that deflections in beam B3 (which acts as a reinforcedconcrete beam after the FRP is lost) are lower as compared to thecontrol beam (B01). This is because the insulation (low thermalconductivity property) contributed effectively to control the riseof temperature in the flexural reinforcement over much of thelength of the beam, and thus the loss of strength with temperatureis gradual. Therefore, the increase in deflection (in beam B3)is not significant as compared to control beam B01. Also, theeffect of the axial restraint force can be gauged by comparing thedeflections of beams B3 and B4. Debonding of the FRP occurredin both beams; however, beam B4 experienced lower deflectionssince the axial restraining force that resulted from restrainingthe thermal expansion of the beam produces a counteractingmoment by an arch mechanism. This arch mechanism reducesthe deflections in the beam after the FRP is lost. A comparison ofthe deflections in beam B4 with those in the axially restrainedRC beam (designated as B02) tested by Dwaikat and Kodur [14]

indicates that the plain RC beam experienced higher deflectionsuntil 180min and thereafter showed a slight recovery in deflectionduring the cooling phase. This can mainly be attributed to therelatively fast degradation of the strength and stiffness propertiesof the plain RC beam in the absence of any external fire protection.

The effect of the cooler anchorage zone (located outside thefire-exposed zone) is pronounced in beams B1 and B2, as canbe seen from the structural response shown in Fig. 8. In thesetwo beams, the mid-span deflection after debonding of FRP islower due to cable action (similar to tensile membrane actionin slabs) provided by anchored continuous carbon fibers at thebeam soffit (refer to Fig. 9). When FRP-strengthened RC beams areexposed to fire, the polymer (epoxy) decomposes after prolongedexposure to high temperatures. For a typical polymer resin, thedecomposition temperature is around 500 °C [15]. However, high-temperature tests reported in the literature indicate that carbonfibers have high tolerance against thermal decomposition and areless sensitive to temperatures up to 900 °C [12]. It should benoted that these high-temperature tests on carbon fibers wereconducted under inert environment conditions, whereas underfire conditions (both in furnaces and real buildings) heating (ofcarbon fibers) occurs in an oxidizing environment, causing theearly decomposition of carbon fibers at 620 °C. It can be seenin Fig. 3 that the temperatures at the concrete/FRP interface forbeams B1 and B3 is below 600 °C. Therefore, in the absence ofany delamination of FRP within anchorage zones located outsidethe fire zone, the unbonded continuous fibers at the beam soffitcontinue to contribute towards the tensile strength of the beamthrough the cable mechanism. However, if the anchorages are noteffective, delamination of the FRP occurs, which leads to negligiblestrength contribution from the FRP, as seen from test results ofbeam B3.

The measured fire-induced axial restrained force in the axiallyrestrained beam (B4) as a function of fire exposure time isplotted in Fig. 10. The axial force increases with fire exposuretime due to restraining of the beam against thermal expansion.A comparison with the unstrengthened RC beam (B02), testedby Dwaikat and Kodur [14], shows development of a highermagnitude of axial restraint force in the RC beam as comparedto the FRP-strengthened RC beam (B4). This is mainly attributedto higher thermal expansion in the RC beam due to the absenceof any external fire protection system, which produces a fastertemperature increase in the concrete and rebar. The external fireinsulation in beam B4 helps to keep the thermal expansion lowdue to the slow temperature increase, and as a result the axial forceincreases gradually in the FRP-strengthened RC beam.

No failure occurred in any of the FRP-strengthened RC beamsfor the entire duration of the test, except for the control RC beam(B01), which failed in 180 min. The rebar temperature, which isoften used to define failure under prescriptive-based approaches,was applied to evaluate failure of the beams, and the results areplotted in Fig. 7. It can be seen that the measured temperature inthe rebars for the RC beam (B01) was 580 °C just prior to failure.Thus, the control RC beam had reached its strength and criticaltemperature limit state at almost the same time, whereas the rebartemperatures in the four FRP-strengthened RC beams were below400 °C. Thus, these beams achieved fire endurance of more than3 h, based on strength and rebar temperature criteria. It is alsoimportant to discuss the glass transition temperature (Tg) failurecriteria, which are often linked to FRP-strengthened RC beams.The analysis of test data as well as visual observations duringthe experiments showed that the glass transition temperature ofepoxy was exceeded in around 20–25 min. However, no structuralfailure of the beams occurred, and the beams successfully resistedapplied loading for the full duration of fire exposure. Therefore,reaching the glass transition temperature may not represent arealistic failure limit state for FRP-strengthened RC beams from astructural point of view, and this is in agreement with the findingsof previous fire tests.

2208 A. Ahmed, V. Kodur / Engineering Structures 33 (2011) 2201–2211

Fig. 9. Unbonded continuous carbon fibers at the beam soffit providing cable action.

Fig. 10. Comparison of axial restraint force as a function of fire exposure time inbeams B4 and B02.

6. Numerical model

The main objective of above experimental study was togenerate test data under non-standard fire conditions to validatethe model. A macroscopic finite element (FE) based computerprogram was developed to evaluate the fire response of FRP-strengthened RC beams [16]. This fire resistance model takes intoaccount high-temperature properties of the constitutive materialsto generate moment–curvature (M–κ) relationships for differentbeam segments at various time steps. These time-dependentM–κ relationships are utilized to trace the response of the FRP-strengthened RC beam in the entire range up to collapse under fireconditions.

In the analysis, the total fire exposure time is divided intonumber of time steps, and at each time step the response of thebeam is traced through the following steps.

• Establishing the temperatures due to the fire.• Conducting a heat transfer analysis to determine the tempera-

ture distribution in a segmental cross section.• Calculating the slip (εslip) at the FRP/concrete interface.• Generating moment–curvature (M–κ) relationships for each

beam segment at various time steps and performing a beamanalysis to compute the internal forces and deflections in theFRP-strengthened RC beam.

The beam is idealized by dividing it into a number of segmentsalong its length (refer to Fig. 11(b)), and the mid-section of eachsegment is assumed to represent the overall behavior of thesegment. This mid-section is further discretized into a numberof elements (see Fig. 11(c)). In the model, the beam is assumedto be exposed to fire from three sides, while ambient conditionsare assumed to prevail on the top side. The fire temperatures areestablished from known time–temperature curves for standard orany other specified design fire scenario. Then, at each time step, thetemperature distribution in the beam cross section is establishedthrough thermal analysis utilizing the high-temperature thermalproperties of constitutive materials. The convective heat transfer

coefficient is 25 and 9 W/m2 K for surfaces exposed to fire and atambient temperatures, respectively [17]. The value of emissivity(the total emissivity, which encompasses the emissivity of thefire as well as the material) is 0.12, which was arrived at basedon temperature measurements in the MSU furnace. As a partof the experimental program, a thermocouple was installed onthe surface of one of the tested FRP-strengthened RC beams torecord the surface temperatures. This temperature data was usedto back-calculate the value of the emissivity through thermalanalysis (the value of emissivity was iterated until the surfacetemperatures matched). This emissivity was then used for thermalanalysis (validation) of all the four beams. Full details of thetemperaturemeasurements in the beamare documented in a Ph.D.thesis [18]. The generated temperatures in each segment form theinput to the strength analysis, which includes the computation ofthe fire-induced bond degradation at the FRP/concrete interface,axial restraint force, and the moment–curvature relationships.It should be noted that the model does not take into accountany debonding or cracking of the insulation in its analysis.These moment–curvature relationships are utilized to set up thestiffness matrix for undertaking beam analysis to evaluate thedeflections, stresses, and strains in the beam. In FRP-strengthenedRC members, the binding material (adhesive) provides the loadpath for the transfer of stresses from the concrete substrate tothe FRP reinforcement. At temperatures beyond Tg , the bondproperties (shear and bond strength) deteriorate considerably, andthis introduces a slip at the bond interface. Due to this bond-slip,the adhesive loses its ability to effectively transfer forces betweenthe concrete and the FRP, and this results in the FRP developingonly partial tensile stresses as compared to a perfect bond case,where full stresses in the FRP can effectively be utilized. Withincreasing slip, the bond deteriorates considerably, and ultimatelythere is debonding of the FRP. Previous studies have shown thatthe high shear stress concentration at terminating edges of the FRPis a major cause of FRP debonding. The variations of these shearstresses with temperature are used to compute the bond-slip atthe FRP/concrete interface. Full details on bond-slip calculations,including the derivations of the equations, are given in [19].

All strain components that result during fire exposure suchas mechanical, thermal, and creep strains for both the concreteand the reinforcing steel and the transient strain in concreteare accounted for in the analysis. There is limited informationin the literature on the complex phenomenon of creep inadhesives. Therefore, creep in the adhesive is not accounted forin the numerical model. The model generates various criticaloutput parameters, such as cross-sectional temperatures, stresses,strains, deflections, fire-induced axial restraint force, and momentcapacity for each time increment. These parameters are checkedagainst pre-designated failure limits, which include thermal andstructural considerations. Although both thermal and structurallimit states are built into the model, for FRP-strengthened RCbeams strength failure often governs the failure, and it shouldbe used in the analysis. Full details of the numerical procedure,

A. Ahmed, V. Kodur / Engineering Structures 33 (2011) 2201–2211 2209

w w

insulation

insulation

FRP laminate

254 mm

25 mm

25 mm

406 mm

Finermesh

Fire exposure

Ls (length of beam segment)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

L

Fig. 11. Layout of a typical FRP-strengthened RC beam, its idealization, and discretization for analysis.

Fig. 12. High-temperature thermal properties of the insulation.

including the derivation of appropriate equations, are presentedin [16].

The model has been previously validated against test datafor FRP-strengthened RC beams under standard fire exposure[16]. However, the model could not be validated against FRP-strengthened RC beams under design (non-standard) fire andrestraint conditions and under different insulation systems due tothe lack of fire test data. This validation is carried out here usingthe above-generated fire test data.

6.1. Material properties

For modeling the response of FRP-strengthened beams, high-temperature properties of the concrete, steel, FRP, adhesive, andinsulation are required. These properties include thermal,mechan-ical, and deformation properties that vary as a function of tem-perature. In the literature, reliable data on the high-temperatureproperties of concrete and steel have been reported. However,knowledge is limited on the high-temperature properties of theFRP, adhesive, and insulation. For the concrete and steel, the prop-erties suggested by the ASCE Manual [20] and for the FRP thesemi-empirical relationships suggested by Bisby [21] have beenincorporated into the model. For the insulation, the high-temperature material properties obtained from experiments con-ducted at MSU have been incorporated in the model (refer toFig. 12).

The thermal properties of the constitutive materials dependon the maximum temperature attained during the tests. For thisanalysis, it is assumed that the thermal properties do not change

for both the heating and cooling phases. For the mechanicalproperties, linear interpolation between the high-temperaturestrength and residual strength is used to estimate the strengthof both the concrete and reinforcing steel in the cooling phase.The residual strengths for any concrete element or steel rebarare determined based on the maximum temperature attainedduring the test. Since the maximum temperature recorded in therebars remained below 400 °C for the entire duration of the test,the residual strength is assumed to be similar to that at roomtemperature.

6.2. Model validation

The above-described model has been further validated usingthe above test data on four FRP-strengthened RC beams. Eachbeamwas analyzed under corresponding fire scenario and supportconditions. A summary of the results is presented in Table 1.

The validity of the model is established by comparing the mea-sured temperatures, deflections, and axial restraint force in thetest with predictions from the model. Fig. 13 provides a compar-ison of the temperatures at the FRP/concrete and FRP/insulationinterfaces, and at three different locations (TC5, TC6, and TC9) inthe beam cross section. TC5 represents temperature in compres-sion reinforcement, TC6 represents the corner rebar temperature(flexural reinforcement), and TC9 represents the mid-depth of thebeam cross section (203mm), as shown in Fig. 1. It can be seen (re-fer Fig. 13(a)–(d)) that the measured and predicted temperaturesare in good agreement throughout the fire duration time in all fourbeams.

Fig. 13(e)–(h) provides a comparison between the predicted andmeasured temperatures at the FRP/concrete and FRP/insulation(VG) interfaces. These temperatures are critical indicators of theperformance of the FRP under elevated temperatures. The modelpredicts the temperature fairly well up to 40 min of fire exposuretime. Beyond this, the model underpredicts the temperature atthe FRP/insulation interface and overpredicts the FRP/concreteinterface temperatures for all FRP-strengthened RC beams. Thiscould be attributed to the fact that measured temperature at theFRP/insulation interface increases rapidly after 40 min due tolocalized burning of the epoxy as a result of crack propagationin the insulation. Due to this localized burning, the measuredtemperatures are higher as compared to those predicted by themodel. In contrast, the increase in temperatures recorded at theFRP/concrete interface is slightly lower than that predicted by

2210 A. Ahmed, V. Kodur / Engineering Structures 33 (2011) 2201–2211

(a) Beam B1. (b) Beam B2.

(c) Beam B3. (d) Beam B4.

(e) Beam B1. (f) Beam B2.

(g) Beam B3. (h) Beam B4.

Fig. 13. Comparison of measured and predicted temperatures for beams B1–B4.

the model. The possible reason for this temperature lag couldbe due to the formation of a char layer as a result of thermaldecomposition of the epoxy (pyrolysis process) that acts as athermal barrier and restricts the heat flow to the inner interface.The model predictions for beam B3 do not match the measuredtemperatures since a portion of insulation fell off when the FRPdelaminated at around 38 min, and the model could not account

for the falling off of insulation. Overall, the model predicts thetemperature progression reasonably well.

The predicted and measured mid-span deflections of testedFRP-strengthened RC beams are compared in Fig. 14. There is agood agreement between the measured and predicted deflectionsfor all four FRP-strengthened RC beams. Compared to the observedtime of FRP debonding between 20 and 25min, the model predicts

A. Ahmed, V. Kodur / Engineering Structures 33 (2011) 2201–2211 2211

Fig. 14. Measured and predicted mid-span deflections of tested FRP-RC beams.

it to be around 30 min. This variation can be attributed to thediscrepancy between the measured and predicted temperaturesat the FRP interface, as discussed above. In the numerical analysis,the failure of the beamswas checked against pre-assigned thermal(the critical temperature in the corner rebar reaches 593 °C)and structural (the strength and deflection limit state) failurecriterion [16]. The strength limit state is defined as the timewhen the structural member cannot resist the applied service load(moment), while failure under the deflection limit state is takenfrom BS 476 [22]. As illustrated in Table 1, the model predictsno failure in any of the four FRP-strengthened RC beams. For thecontrol RCbeam (B01), themodel predicts failure at 182min (basedon the strength limit state) as compared to 180 min measuredduring the test. Overall, the model provides reasonable estimatesof the cross-sectional temperatures and deflections.

7. Conclusions

Based on experimental studies, the following conclusions canbe drawn,

• FRP-strengthened RC beams, supplemented with 25 mm thickspray-applied Tyfo r⃝ WR Advanced Fire Protection system, cansurvive failure under non-standard fire (design fire) conditionscomprising 3 h of ASTM E119 growth phase followed by a decay(cooling) phase.

• In FRP-strengthened RC beams, failure occurs significantly afterreaching the glass transition temperature. The extent to whichthe failure is prolonged is mainly dependent on the rebartemperature level, which in turn depends on the insulationproperties.

• The presence of cooler anchorages outside the fire zone helps incontributing to the load-carrying capacity of FRP-strengthenedRC beams through a cable mechanism, which is provided byunbonded continuous fibers at the beam soffit.

• The fire-induced axial restraint force significantly increasesthe fire resistance of FRP-strengthened RC beams. Therefore,appropriate boundary conditions are to be considered forevaluating the fire resistance of FRP-strengthened RC beams.

• Insulation contributes effectively in reducing the deflections inFRP-strengthened RC beams by controlling the rise of tempera-ture in the concrete and in the steel reinforcements.

• Debonding of the FRP may expose part or all of the beam sof-fit to direct heat flux. Thus, a proper configuration and careful

application of insulation is essential to achieve good fire resis-tance in FRP-strengthened RC beams.

Acknowledgements

The research presented in this paper is supported by theNational Science Foundation (Grant No. CMMI 0855820) andMichigan State University through Strategic Partnership Grant(Award No. SPG 71-4434). The authors would also like tothank Fyfe Company LLC, USA, for supplying the CFRP sheetsand fire insulation. Any opinions, findings, and conclusions orrecommendations expressed in this paper are those of the authorsand do not necessarily reflect the views of the sponsors.

References

[1] American Concrete Institute (ACI). Guide for the design and constructionof externally bonded FRP systems for strengthening concrete structures ACI440.2R-08. ACI Committee 440. Farmington Hills (MI); 2008. p. 1–45.

[2] Deuring M. Brandversuche an nachtraglich verstarkten tragern aus beton.Research Report EMPA No. 148,795 Dubendorf (Switzerland): Swiss FederalLaboratories for Materials Testing and Research 1994.

[3] Blontrock H, Taerwe L, Vandevelde P. Fire tests on concrete beamsstrengthened with fibre composite laminates. In: Third Ph.D. symposium.2000.

[4] Barnes R, Fidell J. Performance in fire of small-scale CFRP strengthenedconcrete beams. J Compos Constr 2006;10:503–8.

[5] Williams B, Kodur VKR, Green MF, Bisby L. Fire endurance of fiber-reinforcedpolymer strengthened concrete T-beams. ACI Struct J 2008;105(1):60–7.

[6] Williams B, Bisby L, Kodur VKR, Green M, Chowdhury E. Fire insulationschemes for FRP-strengthened concrete slabs. Composites Part A 2006;37(8):1151–60.

[7] ACI 318-08. Building code requirements for reinforced concrete and commen-tary. Farmington Hills (MI): American Concrete Institute 2008.

[8] American society for testing and materials (ASTM). Standard practicefor plastics: dynamic mechanical properties: determination and report ofprocedures. ASTM D4065. West Conshohocken (PA); 2006.

[9] Xian G. Comparison of glass transition temp and critical temp of Tyfo S–T andTyfo S based FRPS. In: Inc FL, editor. Technical Report. 2008. p. 1–4.

[10] Eurocode 1. EN 1991-1-2: actions on structures. Part 1–2: general actions —actions on structures exposed to fire. Brussels (Belgium): European Committeefor Standardization 2002.

[11] Buchanan A. Structural design for fire safety. Chichester: Wiley; 2002.[12] Davies JM, Wang YC, Wong FMH. Polymer composites in fire. In: Proceedings

of advanced composites for structural applications in construction. 2004. p.3–17.

[13] Eurocode 2. EN 1992-1-2: design of concrete structures. Part 1–2: generalrules — structural fire design. Brussels (Belgium): European Committee forStandardization 2004.

[14] Dwaikat MB, Kodur VKR. Response of restrained concrete beams under designfire exposure. J Struct Eng 2009;135:1408.

[15] Griffis CA, Masumura RA, Chang CI. Thermal response of graphite epoxycomposite subjected to rapid heating. J Compos Mater 1981;15(5):427.

[16] Kodur VKR, Ahmed A. A numerical model for tracing the response of FRP-strengthened reinforced concrete beams exposed to fire. ASCE J Compos 2010;14(6):730–42.

[17] Cook RD, Malkus DS, Plesha ME, Witt RJ. Concepts and applications of finiteelement. 4th ed. NY (USA): John Wiley & Sons, Inc.; 2007.

[18] Ahmed A. Behavior of FRP-strengthened reinforced concrete beams under fireconditions. Ph.D. dissertation. East Lansing: Michigan State University; 2010.p. 261.

[19] Ahmed A, Kodur VKR. Effect of bond degradation on fire resistance of FRP-strengthened reinforced concrete beams. Composites Part B: Engineering2011;42(2):226–37.

[20] Lie TT. Structural fire protection Editor. New York (NY): ASCE Committee onFire Protection, Structural Division, American Society of Civil Engineers; 1992.p. 225–9.

[21] Bisby LA. Fire behavior of fibre-reinforced polymer (FRP) reinforced orconfined concrete. Doctoral thesis. Kingston (Canada): Queen’s University;2003. p. 365–7.

[22] British standard institute (BSI). Fire test on building materials and structures— part 20: method for determination of fire resistance of elements ofconstruction (general principles). BS476-20. London (UK); 1987.