..28.06.2017.. 11

European Network for Durable Reinforcement and Rehabilitation Solutions

Contract Number: MC-ITN-2013-607851

Behaviour of FRP Reinforcement at High Temperatures

Alessandro Proia

The University of Ghent



In the last years, fiber reinforced polymer (FRP) materials have been applied to make new structural concrete elements and strengthen existing RC members.

FRP are formed by fibre elements (carbon, glass, aramid, basalt etc.) kept together through a matrix polymer (epoxy resin).

There are two main ways to apply FRP material to strengthen concrete members: Externally Bonded Reinforcement (EBR) and Near Surface Mounted (NSM) systems (Fig.1a).

Sheets or laminates may be glued on the external side of the beam (EBR) or embedded inside the beam by means of precut grooves (NSM) and filler (epoxy is mainly used). Thus, the bearing capacity of strengthened members can be influenced by:

  • Material properties of the beam
  • FRP types
  • Adhesive (EBR) and filler (NSM) behaviour

Accidental situations such as fire (Fig.1b) can modify the behaviour of strengthened members because epoxy as adhesive/filler and FRP matrix is sensitive to elevated temperatures.

Understanding FRP behaviour at elevated temperature is needed to increase and make safe the application of these polymer based materials.

The topic “Behaviour of FRP Reinforcement at High Temperatures” will study concrete beams strengthened with EBR/NSM system under high temperatures. Analytical and numerical modelling will allow to investigate strengthened beam behaviour and evaluate post-fire residual capacity.

Fig.1 - A beam reinforced with NSM (a) and a beam subjected to fire action (b)


Aims and Objectives

This research will study strengthened beams during accidental situation with special focus on fire. Analytical and numerical models will be made to take into account material properties, bond stress-slip relationship and thermal effects. Material properties as a function of temperature will be implemented to follow the RC member behaviour under high temperatures. Based on the obtained results, this study will contribute to develop design guidelines for FRP reinforcement subjected to fire.

This work is composed by several objectives:

  • Study of the behaviour of strengthened beam elements under elevated temperature (concrete, steel, FRP, adhesive/filler)
  • Analysis of the strengthening systems and their failure modes at high and room temperature
  • Study of bond stress-slip relationship at room and elevated temperature
  • Implementing of thermal analysis and numerical models to capture beam behaviour
  • Comparison between analytical/numerical solutions with fire test database.


Methodology and Results

In order to acquire knowledge, this work started with a literature review on FRP materials, strengthening systems and bond stress-slip relationships. A part of this study is reported.

The influence of concrete strength on bond behaviour at room and high temperature was studied and as shown in Figs.2 (with reference to Table 1). It could be noted that the maximum shear stress at room temperature increases with concrete tensile strength, yet only in those cases where the failure mode involves a proper activation of the concrete substrate.

Fig.2 - Bond strength as a function of temperature for different concrete type (left) and normalized bond strength as a function of temperature for different concrete type (right)


Table 1 - Summary tests reported in Fig.2


Fig.3 - Ultimate stress CFRP spirally wound ∅7.5mm with different K ratio and groove type


The decrease in shear strength for initial increase of temperature (20°C-60°C range) tends to be higher for higher concrete classes. On the other hand for CFRP, an increased bond strength is observed at temperatures lower than Tg. This is due to thermal shear pre-stress in the adhesive and axial compression stresses in the substrate when there is a significant difference between CTE of concrete and FRP material. At elevated temperature beyond 70°C the bond strength becomes indifferent to the concrete strength. This behaviour is due to the loss of mechanical properties when the adhesive is subjected to a temperature beyond to the glass transition temperature (Tg) and related shift to adhesion controlled failure modes.

It was observed that NSM strengthening may be influenced by filler type, groove roughness and groove dimensions (Figs. 5 and 6). In NSM systems with epoxy filler of the groove, the groove dimensions influence the bond strength. Hereby, the failure load has a tendency to increase with K-ratio. For cement mortar filler however, this is generally not the case. Similar order of magnitude have been observed for smooth and rough grooves. Yet, groove roughness (if done properly) positively influences the bond strength.

The effect of temperature on bond stress-slip relationships was analysed. Looking to the CFRP NSM specimen C_SC in Fig.7, a shear peak gain of 30% is observed between 20˚C and 50˚C. Between 20˚C and 65˚C the peak stress is similar. For the GFRP specimen G_SW (Fig.8) the peak stress seems to decrease more quickly. This different behaviour is due to the different CTE between carbon and glass fibres. Thermal effects create axial compression stresses into the substrate, increasing the shear strength. Looking into the post-peak branch of the bond stress-slip curves, it can be observed that there is a difference between EBR (Fig.4c) and NSM (Fig.4a, b and d) for T=20˚C - 100˚C.


Fig.4 - Bond stress-slip curves for varying test temperatures of specimens C_SC (a), G_SW (b), C_STR (d) and CFRP laminates (c)


In NSM the bars and strips embedded in the concrete groove tend to slip upon bond failure creating shear stresses related to the friction coefficient between concrete-adhesive, while EBR does not show residual shear stress. Temperature changes the NSM failure mode and this can be observed from the magnitude of the last branch of bond stress-slip curve. At 20˚C the failure load involves sliding between a concrete-adhesive interface and at 100˚C the bar slips in the adhesive with a different friction coefficient.

As the adhesive stiffness decreases with increasing temperature, the bond transfer occurs over a longer length.

To study the beam behaviour a fibre beam model was made. This modelling involves three different levels: section, element, structure. Non-linear analysis is implemented to reach solutions.


March 2016

In Table 2 the research project is summarized into 5 steps. For each of them a brief description is reported.

Table 2 - Methodology


June 2016

Experimental results reveal that the epoxy glass transition temperature (Tg) affects significantly the bearing capacity of FRP strengthened concrete beams exposed to elevated temperature. The glass transition temperature of epoxy resin may be influenced by the following factors:

  • composition (different amount of component A and B)
  • curing time
  • curing temperature
  • ageing
  • thermal history
  • moisture
  • thickness

A study will be performed to investigate the influence of moisture absorption, ageing at different environmental conditions and thermal cycles on Tg.

Material and samples

Three commercially available epoxy resins used for civil engineering application have been chosen in this experimental campaign. Components A and B are mixed as reported by the manufacturers (Table 3). Afterwards, the mixture is laid on a silicone mold to make samples 100mmx100mm and thickness ~1mm. Samples will cure three days before testing (Table 4). In particular, Sikadur 30 will be cured at 20°C and 70°C.

Table 3 - Epoxy resins

Table 4 - Curing condition


The test program (Table 5) will deal with three factors which are able to influence Tg: moisture absorption, ageing and thermal cycles. Two extra tests (Test 4 and 5) will be performed to investigate the influence of extended curing time at high temperature and different amount of component B (+50%) on moisture uptake. To draw the absorption curves, all of samples will be weigh every 5 days. This interval has been chosen to reduce the influence of weighing on measurements. Sample dimensions (w/h>=100 – w: nominal length of one side, h: nominal thickness), sample weight (>=5g) and balance accuracy (0.1mg) have been defined according to ASTM D5229/D5229M “Standard Test Method for Moisture Absorption Properties and Equilibrium Conditioning of Polymer Matrix Composite Materials”. Uptake can happen in two different ways: by Vapor Exposure (Test 2 and 3) and by liquid immersion (Test 1, 4 and 5). Experimental results show the property of epoxy to uptake higher level of moisture at temperature higher than Tg as the free volume tends to be filled by water (vapor or liquid). Increasing the moisture content, Tg degradation might be forced.


Fig.5 - Volume-Temperature relationship


The glass transition temperature will be investigated by means of Differential Scanning Calorimetry (DSC) according to the time schedule in Table 5. DSC will be performed with a heating rate 10°C/mm between 0°C and 120°C.

Table 5 - Test program

Experimental investigation

Moisture Absorption

Moisture absorption will be experimentally determined with the following equation:

  • Wi: current specimen mass (g)
  • Wb: baseline specimen mass (g)

Differential Scanning Calorimetry (DSC)

DSC measures the heat absorbed by the polymer against temperature. Figure 6 shows a typical DSC curve for semi-crystalline polymer. In this case the second order transition, crystallization and first order transition (melting) are well identifiable. In 100% Amorphous polymer crystallization does not happen.

For epoxy resin cured at room temperature, low glass transition temperature are usually recorded (~60-70°C). According to ISO 11357-2 “Determination of Glass transition temperature” heating should be approximately 30°C higher than the Tefg (extrapolated end temperature) so 120°C are reasonably assumed as the maximum temperature for the DSC tests.


Fig.6 - Typical DSC curve


The glass transition temperature will be evaluated by DSC results as the approximate midpoint of the temperature (Tmg) range over which the glass transition temperature takes place (Figure 7). 


Fig.7 - Example of characteristic glass transition temperature determination


September 2016

Experimental work: preliminary results

Three commercially available epoxy resins used for civil engineering application have been investigated in order to observe the influence of thermal history, ageing and moisture on the glass transition temperature (Tg). In this preliminary study the behaviour of the epoxy samples in terms of water absorption is analysed. Moisture is experimentally determined with the following equation as a function of the time t: 

  • Wi: current specimen mass (g)
  • Wb: baseline specimen mass (g)

The moisture content M of the material during absorption can be expressed:

  • G: time dependent parameter
  • Mm: maximum moisture content (%)
  • Mi: initial moisture content (%)

In this study the percent moisture content versus time is monitored for a material specimen that is maintained in a steady-state environment at a known temperature and moisture exposure level until the material reaches effective moisture equilibrium. Specimens are not dried before conditioning (for convenience Mi will be equal to 0). For a material exposed on two sides to the same environment, as in this study, s is equal to the thickness (s=h). D is the diffusivity of the material in the direction normal to the surface. The thickness is calculated by means of the density values provided by the manufacturers, while the diffusivity D is obtained from the initial slop (t<tL) of the MM versus √t curve.

Tab.6 - Density of epoxy resins provided by manufacturers


Fig.8 - Example of the change of moisture content as a function of √t


The results obtained so far for samples conditioned to test 1, 3 and 4 are reported below.

Tab.7 - Test conditioning


Test 1

In Figure 9, 10, 11 the behaviour of the three different epoxy is shown. For Sikadur30 (0112003M270) and Epicol U (0132003T270) has been possible to define the analytical curve obtained by the Fick’s model. Because of the slight difference in their thickness (absorption is significantly influenced by the thickness) we are not able to compare straightaway M(t) versus √t for these samples but, looking at the diffusivity values, absorption looks to be a bit faster in Epicol U than Sikadur30. Furthermore, Epicol U shows an higher value of Mm than Sikadur30 because of free volume (as reported in the previous report). PC 5800 (Figure 10), in contrast with Sikadur30 and Epicol U, shows two different phases. At the beginning (up to 6 days), its weight tend to increase because of the uptake process but then the loss in mass starts. Figure 3 reveals a not stable behaviour of PC5800 immersed in hot water at 70°C for more than 6 days. Xiao and Shanahan (1997) observed degradation during absorption subjected to hot water at 90°C. They ascribed degradation of epoxy to hydrolysis process which leads to chain scission. After 100 days in hot water, PC5800, although deteriorated, did not show any visible damage to the material.

Tab.8 - Thickness, diffusivity and maximum moisture content (Mm)


Fig.9 - Sample 0112003M270: test 1 - resin type 1 (Sikadur30) – cured at 20°C for 3 days – deadline 270 days after curing


Fig.10 - Sample 0122003M270: test 1 - resin type 2 (PC 5800) – cured at 20°C for 3 days – deadline 270 days after curing


Fig.11 - Sample 0132003T270: test 1 - resin type 3 (Epicol U) – cured at 20°C for 3 days – deadline 270 days after curing


Test 3

Comparing Tables 8 and 9 is immediately clear that absorption in Epoxy resins is strongly influenced by the temperature. In terms of diffusivity Sikadur30 shows a more stable behaviour than Epicol U, indeed, Epicol U has a double diffusivity in water at 20°C (in other words the absorption in water at 20°C looks faster than 70°C as shown in Figure 13). The Change of temperature leads to a different behaviour in maximum moisture content and, at 20°C, Mm of Sikadur30 looks double than Epicol U. Looking at the ratio of Mm for test 1 (70°C) and test 3 (20°C) we are able to notice the higher influence of the temperature on Epicol U. In addition, at 20°C PC5800 did not lose weight during the conditioning, it may mean that hydrolysis process is significantly affected by temperature.

Tab.9 - Thickness, diffusivity and maximum moisture content (Mm)


Fig.12 - Comparison between samples 0132003T270 and 0332003T240 in terms of normalised absorption.


Fig.13 - Sample 0312003T270: test 3 - resin type 1 (Sikadur30) – cured at 20°C for 3 days – deadline 270 days after curing


Fig.14 - Sample 0322003T270: test 3 - resin type 2 (PC 5800) – cured at 20°C for 3 days – deadline 270 days after curing


Fig.15 - Sample 0332003T240: test 3 - resin type 3 (Epicol U) – cured at 20°C for 3day – deadline 240days after curing


Test 4

In test 4, moisture was applied by air. Tables 9 and 10 show that epoxy resins subjected to different moisture application (water and air) respond in various ways. In particular, the diffusivity of Sikadur30 appears much more higher than that one measured in tests 1 and 3. As expected, every sample tends to uptake less moisture than in water condition but, as regards Mm, Sikadur30 and Epicol U show an opposite behavior compared with test 3.

Tab.10 - Thickness, diffusivity and maximum moisture content (Mm)


Fig.16 - Sample 0412003T120: test 4 - resin type 1 (Sikadur30) – cured at 20°C for 3 days – deadline 120 days after curing


Fig.17 - Sample 0422003M270: test 4 - resin type 2 (PC5800) – cured at 20°C for 3 days – deadline 270 days after curing


Fig.18 - Sample 0432003M270: test 4 - resin type 3 (Epicol U) – cured at 20°C for 3 days – deadline 270 days after curing


Dissemination Activities

Conferences and meetings 

  • Gent endure Meeting
  • Kaiserslautern endure/Cost Meeting
  • FRPRCS 12 – Nanjing 14-16 December 2015


  • Kaiserslautern endure/Cost Meeting


  • Proia A, Matthys S. Literature review on reinforced concrete members strengthened with FRP at room and elevated temperature. In: Proc of the 7th Biennial conference on advanced composites in construction (ACIC). Cambridge: St John’s College; 2015.
  • Proia A, Matthys S, Palmieri A. Influence of bond stress-slip relationship on bond strength prediction. In: Proc of the 3rd Conference on Smart Monitoring, Assessment and Rehabilitation of Structures (SMAR). Antalya; 2015
  • Proia A, Matthys S, Cassaert C. Double bond shear tests at elevated temperature on NSM FRP systems with epoxy and grout adhesive. In: Proc of the 12th International Symposium on Fiber Reinforced Polymer for Reinforced Concrete Structures (FRPRCS-12). Nanjing; 2015.
  • Proia A, Palmieri. A., Matthys S and Taerwe L. Fire testing of insulated RC beams strengthened with near surface mounted FRP reinforcement. In: Proc of the 9th International Conference Structures in Fire (SIF). Princeton University; 2016.