Bond between Textile Reinforced Mortar (TRM) and Concrete Substrate

................................................................................................................ i List of Tables ............................................................................................................. ix List of Figures ............................................................................................................ xi List of abbreviations ............................................................................................... xix List of Notations ...................................................................................................... xix INTRODUCTION .................................................................................. 1 1.1 Background ............................................................................................................ 1 1.2 Research Aim and Objectives ................................................................................ 3 1.3 Thesis Outline ........................................................................................................ 3 LITERATURE REVIEW ...................................................................... 5 2.

TRM comprises high-strength fibres in form of textiles embedded into inorganic matrices such as cement-based mortars. TRM offers well-established advantages such as: fire resistance, low cost, air permeability, and ability to apply on wet surfaces and at ambient of low temperatures.
It is well known that the effectiveness of any external strengthening system in increasing the flexural capacity of concrete members depends primarily on the bond between the strengthening material and member's substrate. This PhD Thesis provides a comprehensive experimental study on the bond behaviour between TRM and concrete substrate and also provides a fundamental understanding of the flexural behaviour of RC beams strengthened with TRM.
Firstly, the tensile properties of the textile reinforcement were determined through carrying out tensile tests on bare textiles, and TRM coupons. Secondly, the bond behaviour between TRM and concrete substrates both at ambient and, for the first time, at high temperature was extensively investigated. A total of 148 specimens (80 specimens tested at ambient temperature and 68 specimens tested at high temperatures) were, fabricated, and tested under double-lap shear. Parameters investigated at ambient temperature comprised: (a) the bond length; (b) the number of layers; (c) the concrete surface preparation; (d) the concrete compressive strength; (e) the textile surface condition; and (f) the anchorage through wrapping with TRM jackets. Whereas, the parameters examined at high temperatures included: (a) the strengthening systems (TRM versus FRP); (b) the level of temperature at which the specimens were exposed; (c) the number of FRP/TRM layers; and (d) the loading conditions. The results of ambient temperature tests indicated that the bond at the TRM-concrete interface is sensitive to parameters such as: the number of layers, the ii textile surface condition, and the anchorage through wrapping with TRM. On the other hand, the results of high temperature tests showed that TRM exhibited excellent bond performance with concrete (up to 400 0 C) contrary to FRP which practically lost its bond with concrete at temperatures above the glass trainset temperature (Tg).
The flexural strengthening of RC beams with TRM at ambient and for the first time at high temperature was also examined carrying out 32 half-scale beams. The examined parameters were: (a) the strengthening system (TRM versus FRP); (b) the number of layers; (c) the textile surface condition; (d) the textile fibre material; (e) the end-anchorage system of the external reinforcement; and (f) the textile geometry. The results of ambient temperature tests showed that TRM was effective in increasing the flexural capacity of RC beams but its effectiveness was sensitive to the number of layers. Furthermore, a simple formula used for predicting the mean FRP debonding stress was modified for predicting the TRM debonding stress based on the experiment data available. The results of high temperature tests showed that TRM maintained an average effectiveness of 55%, of its effectiveness at ambient temperature, contrary to FRP which has totally lost its effectiveness when subjected to high temperature.
Finally, a stress reduction factor of TRM flexural effectiveness (compared to its ambient effectiveness) when subjected to high temperature was also proposed.
iii Acknowledgment I feel the words are few to express my upmost gratitude and thanks to my supervisor, Dr. Dionysios Bournas. I have been so lucky working under his supervision. His engineering experience, advices, critical thinking, and unlimited support, was of great importance in guiding me to the right direction and completing my PhD. I would like also to thank Dr. Savvas Triantafyllou for his support and assistance in the last year of my PhD study.
I would like to thank the Higher Committee for Education Development in Iraq (HCED) for giving me this chance to complete my PhD study.
I also deeply thank Dr. Lampros Koutas for his valuable assistance during my PhD study. His scientific knowledge and experience, lab skills, positive thinking, and personal characters helped me in completion my PhD.

List of Tables
Over the last three decades, the use of fibre-reinforced polymer (FRP) for retrofitting concrete and masonry structures, has gain popularity over other conventional strengthening systems (such as steel/RC jacketing). This is due to the favourable properties offered by FRP such as: resistance to corrosion, high strength to weight ratio, ease and speed of application and minimal change in the geometry.
However, some drawbacks have been observed with the use of FRP, which are mainly associated to the use of epoxy resins. These drawbacks including: high cost, unsafe for manual workers, low permeability to water vapour, and poor behaviour at high temperatures .
Almost a decade ago, an innovative cement-based composite material, the socalled textile-reinforced mortar (TRM), was introduced in the field of structural retrofitting (Bournas et al., 2007) as an alternative solution to FRP, addressing some of FRP's drawbacks. Since that time, TRM progressively attracts the interest of the structural engineering community. TRM is a composite comprising fibre rovings made of carbon, basalt or glass in form of textiles embedded into inorganic materials such as cement-based mortars. TRM is relatively low-cost materials (mortars are generally lower cost compared to epoxy resins), compatible with concrete or masonry substrates, friendly materials for manual workers, can be applied at low temperatures or on wet surfaces, and resistant at high temperatures (Tetta and Bournas, 2016). In the last few years, a significant number of studies have been directed towards investigating the effectiveness of TRM as a mean of external strengthening. The results have indicated that TRM is a promising alternative to FRP in retrofitting structures.
It is well known that the effectiveness of any externally bonded strengthening system in increasing the load-carrying capacity of concrete members depends primarily on the bond between that strengthening material and the member's substrate.
Therefore, the study of the bond behaviour between TRM materials and concrete is of crucial importance, because it helps understanding the complex mechanisms of transferring stresses from the textile reinforcement to the surrounding matrix and eventually to the concrete substrate. It is also a fundamental step towards the development of design models to be used in real strengthening applications.

Research Aim and Objectives
The aim of this PhD study is to provide comprehensive understanding on the bond behaviour between TRM and concrete substrates both at ambient and high temperatures. The current study also aims to evaluate the effectiveness of TRM in flexural strengthening of RC beams at ambient and high temperature. To achieve these aims, the following objectives were set:

Thesis Outline
This Thesis comprises eight chapters including the current introductory chapter.

Introduction
Over the last decades, the issue of upgrading and rehabilitation of existing reinforced concrete (RC) infrastructure has become of great importance due to their continues deterioration. Fibre-reinforced polymers (FRP) has gain popularity as a means of external strengthening, however some drawbacks (see Section 1.1) which are mainly associated to the epoxy resins have limited the use of FRP.
To overcome these drawbacks, researchers have suggested to replace the organic materials (epoxy resins) with inorganic materials (such as modified cement mortar). However, due to granularity of cement mortars, the impregnation (wetting) of continues fibre sheets was difficult to achieve, hence, the bond has become an issue.
To overcome that issue, it was suggested replacing the continuous fibre sheets by an open mesh configuration in the form of textiles, thus the bond condition between fibres and mortar could be improved. This new cement-based composite material is known as textile-reinforced mortar (TRM) (Bournas et al., 2007). TRM is a composite comprising high-strength fibres made of carbon, basalt or glass in form of textiles embedded into inorganic materials such as cement-based mortars. TRM is also identified in the literature with the following acronyms: Textile Reinforced Concrete (TRC) (Brameshuber, 2006b); and Fibre Reinforced Cementitious Matrix (FRCM) (Carloni et al., 2016).
The textiles used as a reinforcement typically consist of long woven, unwoven, or knitted rovings fabricated at least in two directions (typically orthogonal). Figure   2.1a-d shows photos of some types of textiles which were fabricated from different fibres' materials and geometries. As shown in this Figure, the quantity, materials, and spacing between rovings in both orthogonal directions can be controlled independently which resulted in textiles with different geometries and materials in the two orthogonal directions. In the last decade, significant research effort has been put to take advantage of the use of textile cement-based composite materials for construction of new structural elements (Brameshuber, 2006a) or as a means of external strengthening of existing structures (Triantafillou and Papanicolaou, 2005). In the next sections, the experimental studies on the use of TRM as a means of external strengthening of concrete and masonry members are reported.

Using TRM for Strengthening and Seismic Retrofitting of RC and Masonry Members
This section presents the relevant experimental studies on the use of TRM as a means of external strengthening of concrete and masonry members subjected to static or cyclic loading. Indicative studies for each case of strengthening application are described in detail.

Confinement of RC columns with TRM
To begin with, ; Bournas et al. (2007); Peled (2007);Ortlepp et al. (2010) studied the effectiveness of TRM as a means of confining reinforcement of concrete cylinders and short columns. In specific,   TRM has also been investigated as a means of confining reinforcement of RC columns subjected to seismic loading (Bournas et al., 2007;Bournas et al., 2009;Bournas and Triantafillou, 2010;Bournas and Triantafillou, 2013;Ombres and Verre, 2015). In the study of Bournas et al. (2007) higher, equal and slightly less effective than the equivalent FRP jacket when the internal reinforcement was continuous, long lap-splice length, and short lap-splice length, respectively.

Strengthening of RC slabs with TRM
TRM was also assessed as a measure of enhancing the flexural capacity of one-way (Jesse et al., 2008;Schladitz et al., 2012;Loreto et al., 2013) and two-way RC slabs (Papanicolaou et al., 2009;Koutas and Bournas, 2016). In all of these studies TRM was found to be a promising strengthening system in increasing the load-carrying capacity of retrofitted slabs. In specific, Papanicolaou et al. (2009); Koutas and Bournas (2016) examined the effectiveness of TRM in increasing the strength and deformation capacity of two-way RC slabs. Different parameters were investigated, namely: the textile materials (carbon and glass fibres textile), and the number of TRM layers (Papanicolaou et al., 2009;Koutas and Bournas, 2016), the strengthening configurations (fully or partially covering of the tension face of the slab with TRM), and the presence of cracks in the slab . It was mainly concluded that application of TRM considerably enhanced the load-carrying capacity of the slabs. This enhancement was also found to be sensitive to the number of layers, but it was comparable to the textiles having approximately the same axial stiffness (Papanicolaou et al., 2009;Koutas and Bournas, 2016). It was also observed that the presence of cracks (pre-cracked slab) reduced slightly the effectiveness of TRM in increasing the load-carrying capacity compared to the un-cracked slab, and finally, the flexural capacity of fully covered face slab was higher than that of partially covered face slab .  investigated the effectiveness of TRM as a means of shear strengthening of RC beams. Parameters examined were: the strengthening configuration (typical wrapping versus spirally applied), the number of strengthening layers (1 or 2), and the type of strengthening system (TRM versus FRP).

Shear strengthening of RC beams with TRM
For one layer strengthened beams, it was found that TRM jacket is 45% less effective than the equivalent FRP jacket but still effective in increasing the shear capacity of strengthened beam (providing 40 kN) compared to the control. Moreover, both 2 layers TRM and FRP jackets provided substantial gain to the shear capacity (more than 60 kN compared to the control) of strengthened beams, however, the effectiveness of TRM versus FRP and also the strengthening configuration (typical wrapping versus spirally applied) were not evaluated due to the failure of that beams in flexure. Koutas et al. (2014) studied the potential of using TRM as a means of seismic retrofitting of nearly full scale three-story masonry infilled RC frame subjected to cyclic loading. The results of experimental test indicated that application of TRM enhanced the global response of the frame in terms of lateral strength (56% increase) and deformation capacity (52% increase) at the top of the frame compared to the unretrofitted frame.

Strengthening of masonry members with TRM
TRM was also used as a means of external strengthening of masonry walls for enhancing their in-plane Papanicolaou et al., 2011) and outof-plane (Papanicolaou et al., 2008;Harajli et al., 2010;Papanicolaou et al., 2011;Babaeidarabad et al., 2013) loading capacity. In particular, Papanicolaou et al. (2007) performed experimental work to examine the effectiveness of TRM in enhancing the Finally, TRM was also used as a confining reinforcement of masonry columns.
In particular, Ombres (2015b) measured the effectiveness of one and two layers of TRM as a means of confining reinforcement of masonry columns subjected to eccentrical loading (the eccentricity_ ⁄ _varied between 0 to 0.2). It was found that the enhancement of the load carrying capacity of strengthened specimens was depending on the value of eccentricity. In specific, the ultimate load of strengthened column subjected to axial compression load was increased by 78% compared to the unretrofitted specimen, whereas, the corresponding ultimate load increase varied between 20 to 42.9% for those columns strengthened with one and two layers and subjected to eccentric load, respectively. Contrary to the failure mode of unretrofitted specimen which was sudden and brittle, the TRM retrofitted specimens failed gradually due to rupture of the textile fibres.

Bond between TRM and Concrete
As mentioned previously, the effectiveness of any external strengthening systems in transferring stresses substantially depends on the bond between the strengthening material and concrete substrate. The stresses transfer between TRM and concrete is a complex phenomenon depending on several factors including: (a) the bond between a single fibre and the matrix Banholzer, 2004;Banholzer et al., 2006;Hartig et al., 2008;Zastrau et al., 2008;D'Ambrisi et al., 2013), (b) the degree of penetration of cement matrix into single roving Banholzer, 2004;Hegger et al., 2004;Xu et al., 2004;Banholzer et al., 2006;Hartig et al., 2008;Zastrau et al., 2008;D'Ambrisi et al., 2013); (c) the nature of the bond between the external fibres and the internal fibres in a single roving (Hartig et al., 2008); (d) the bond between the new matrix and the old concrete substrate (Ortlepp et al., 2004;D'Ambrisi and Focacci, 2011). All the aforementioned factors are depending on: (a) the fibre surface condition (dry or coated) (Hegger et al., 2006;Peled et al., 2008;Aljewifi et al., 2010), (b) the geometry of: single fibre/ roving Peled et al., 1999), or textile Peled et al., 1997;Peled and Bentur, 2000;Soranakom and Mobasher, 2009), (c) the composition of the matrix and the degree of grain fineness, and (d) the quality of concrete surface preparation (D'Ambrisi et al., 2013). Therefore, it is of crucial importance to study the bond behaviour between TRM and concrete in order to understand the factors affecting stresses transferring from the textile reinforcement to the matrix and eventually to the concrete substrate. In the next two sections, the test setups adopted, the analytical and experimental studies conducted to investigate the bond of TRM-to-concrete are described in detail.

Test setups for investigation the bond of TRM-to-concrete
Past studies on the bond of FRP-to-concrete have mainly adopted two distinct test setups namely, single-lap shear and double lap shear test setups (Yao et al., 2005). In the single-lap shear test (see Figure 2.2a), the FRP strip was externally bonded to a concrete block, and then a classical pull out test was performed on the FRP strip while the concrete block was fixed. In the double-lap shear test on the other hand (see Figure   2.2b), FRP strips are externally bonded on two opposite sides of two concrete prisms which are connected only by the FRP strips, and then a tensile force is applied up to failure.  (Yao et al., 2005). Undoubtedly, the stresses transferred from concrete to the composite materials in real strengthened RC beams is best simulated by bending tests of full-scale beams.
However, such tests are expensive and time consuming. Therefore, the single and double-lap bond tests which is simple, economic, fast, and offer less bond strength compared to the bending test were adopted in the previous research to perform parametric study. Among these two test setups, the double-lap shear test is preferable due to its simplicity (Yao et al., 2005); symmetry and better control of normal stresses (Serbescu et al., 2013); and less scattered of the measured ultimate loads .

(a) (b)
Thus, the double-lap shear test setup which is a modification of the set-up proposed by Serbescu et al. (2013) was adopted in the current study to investigate the bond of TRM-to-concrete. The selection of this test setup was also deemed necessary for testing more than one TRM layers, as with such a set up the stresses are transferred from the concrete to the composite material indirectly, simulating realistically realword applications. In contrast, in single-lap tests the load is applied directly to the composite material, which means that shear stresses between layers cannot be developed in case of more than one TRM layer.

Bond between TRM and concrete; analytical and experimental studies
The bond between TRM and concrete has been analytically analysed in order to suggest a model for predicting important parameters which can be used for design. In In the next paragraphs, the experimental studies that focused on the bond between TRM and concrete substrate are described in detail. A summary of the main findings of these studies is also presented at the end of this section.
The first experimental study on the bond between TRM and concrete was that of D' Ambrisi et al. (2013). In that study, double-lap shear tests were carried out on twelve specimens strengthened with PBO-TRM. The test specimen comprised two concrete prisms (with cross section of 100 x 100 mm and length of 250 mm for each prism) which were connected only by PBO-TRM. The parameters investigated were: the bond length (50, 100, 150, 200, and 250 mm), and the number of layers (1, and 2 layers).
The following conclusions were drawn: (a) by increasing either the bond length or number of layers, the ultimate load was also increased, mode. The textile reinforcement used were carbon, glass and steel fibres textiles. The results showed that the ultimate load was sensitive to the textile fibres materials. In specific, the carbon and steel textile strengthened specimens had the highest load followed by the glass fibres textile strengthened specimens. The failure mode was also sensitive to the textile materials. In particular, it was slippage of the fibres through the mortar for the carbon fibres textile strengthened specimens, whereas, it was debonding of the composite at the concrete-matrix interface for the steel textile retrofitted one and rupture of the textile fibres for the glass textile reinforced specimens. under single-lap shear. It was observed that the failure mode was partially affected by the concrete surface condition; specifically, four out of eighteen specimens had untreated concrete surface failed due to debonding of the whole or part of the composite from concrete substrate due to the poor bond at that interface, whereas, the failure of the remaining fourteen specimens was identical (slippage of the fibres through the mortar) to the corresponding specimens that had the same bond length but with a treated concrete surface. It was also found that the concrete compressive strength had no effect on the measured ultimate load. To sum up, the main investigated parameters of the aforementioned studies were: the number of layers (1 and 2), the bond length (50-450 mm), the textile materials mainly PBO, and rarely carbon and glass fibres textiles, the concrete compressive strength, and the concrete surface preparation (treated and untreated with sandblasting). The main findings were: (a) the effective bond length varied between 200-300 mm, (b) the common failure mode was slippage of the fibres within the mortar except form (Ombres, 2015a) who concluded that increasing the number of layers from 1 to 2 alters the failure mode from slippage to debonding at the mortar-concrete interface without including parts of concrete cover, (c) the concrete compressive strength (Tran et al., 2014; and the concrete surface preparation  had very limited effect on the bond capacity. However, the conclusion in (c) was based on very limited number of specimens in which the failure was within the composite due to slippage, hence the concrete compressive strength and the concrete surface condition were not involved in that failure. Finally, it is noted that data is missing on: the influence of the number of layers beyond two, the textile surface condition (i.e. dry versus coated), and the concrete compressive strength if the failure was within the concrete cover on both the load response and failure mode.

Flexural Strengthening of RC Beams with TRM
The flexural capacity of RC beams is one of the most critical requirement when assessing the serviceability of beams in real structures. Due to corrosion of internal reinforcement, deterioration of concrete strength induced aging or environmental conditions, or the need to increase in the applied load, the external strengthening is becoming progressively needed. In the next paragraphs, the available experimental studies on the flexural strengthening of RC beams with TRM are described in detail.
A summary of the main finding is also presented at the end of this section.
The first study on the flexural strengthening of RC beams with TRM was carried out by Triantafillou and Papanicolaou (2005). In this study, the effectiveness of TRM versus FRP in increasing the flexural capacity of RC beams was compared. A high strength carbon fibres textile was used as a reinforcement for the strengthened beams.
It was found that TRM system is 30% less effective than FRP system in increasing the flexural capacity of RC beams. Moreover, the failure mode of the FRP strengthened beam was due to rupture of the fibres, whereas, the corresponding failure mode of the counterpart TRM-strengthened beam was due to interlaminar debonding. In Ombres (2011), the effectiveness of PBO-TRM in increasing the flexural capacity of RC beams was evaluated whereas, in Ombres (2012) the debonding behaviour of PBO-TRM strengthened beams was analysed. Parameters examined were: the number layers (1, 2, and 3) Ombres (2011;2012), the ratio of internal reinforcement (Ombres, 2011), and the bond length of TRM (applied to the entire length of beams, and only at the constant moment zone) (Ombres, 2012). It was observed that: (a) application of TRM resulted in flexural capacity increases varying between 10-44 % depending on the ratio of internal reinforcement (Ombres, 2011), and the number of layers (Ombres, 2011;2012), (b) the failure mode of strengthened beams was sensitive to the number of layers and the provided bond length. In particular, it was slippage of the fibres through the mortar for the 1 layer strengthened beam, whereas it was debonding at the concrete-matrix interface for the two and three layers retrofitted beams (Ombres, 2011). Finally, providing TRM along the entire length of the beams resulted in gradual failure, whereas, a sudden and catastrophic failure was observed when providing inadequate bond length (Ombres, 2012). Elsanadedy et al. (2013) examined the effectiveness of basalt-TRM in increasing the flexural capacity of RC beams. The type of mortar (polymer modified cement versus cementitious mortar), the number of layers (5 and 10 layers), and the strengthening systems (TRM versus FRP) were parameters under investigation. The results showed that the polymer modified cement mortar exhibited higher performance in enhancing the flexural capacity than the cementitious one. Particularly, the specimen that received the former failed due to textile rupture, whereas the counterpart specimen that received the latter failed due to end debonding. When applying ten layers of basalt textile, the flexural capacity was increased by 90%. Finally, TRM was less effective than FRP in enhancing the flexural capacity, but more effective in increasing the deformation capacity. Furthermore, two failure modes were observed depending on the number of layers: the beam strengthened with 1 and 2 layers failed due slippage of the fibres through the mortar, whereas the beams strengthened with 3 layers of carbon-TRM failed due to delamination of TRM from concrete substrate without including concrete cover.
To sum up, the investigated parameters in these studies were: the textile-fibre materials, the number of layers, the strengthening configuration, the concrete compressive strength, the type of textile-fibre materials, and the strengthening system (i.e. TRM versus FRP). The main conclusions were: (a) application of TRM to RC beams considerably enhanced their flexural capacity; (b) increasing the number of layers enhanced the flexural capacity and altered the failure mode (Ombres, 2011;Ebead et al., 2016). Regarding the effectiveness of TRM versus FRP, Triantafillou and Papanicolaou (2005) reported on the basis of two specimens, that TRM was 30% less effective than FRP. Whereas, Elsanadedy et al. (2013) found that TRM was slightly less effective than FRP in increasing the flexural capacity but more effective in enhancing the deformation capacity. This conclusion was made based on two tested specimens, one with five layers of TRM in form of U-shaped jacket made of basaltfibre textile and another retrofitted with one layer of basalt FRP. Based on the above studies, it is clear that more research is needed to cover the subject of the effectiveness of TRM versus FRP in flexural strengthening of RC beams.

Performance of TRM at High Temperatures
As mentioned previously, some drawbacks have been observed with the use of FRP system mainly the poor performance at high temperature, as under loading, epoxy resins normally lose their tensile capacity. Therefore, unless protective (thermal insulation) systems are provided (Kodur et al., 2006), the bond strength between the FRP and concrete substrate significantly deteriorates at temperatures above the glass transition temperature (Tg). A review on the behaviour RC members strengthened with FRPs and subjected to fire or high temperature was recently conducted by Firmo et al. (2015b).
TRM could outperform FRP systems at high temperatures or fire due to the breathability, non-combustibility, and non-flammability offered by mineral-based cement mortars used as binding materials. In general, research on the performance of TRM systems at high temperature or under fire scenario and comparison with FRP systems is extremely limited. This is attributed to the inherent experimental difficulties applying simultaneously loading and high temperature, even for medium or smallscale specimens. For this reason, the past studies have mainly focused on determination of the residual strength of TRM coupons after being exposed to high temperatures and cooled down to the ambient temperature. In the next paragraphs, the relevant literature on the behaviour of TRM at high temperatures is summarised. However, above these temperatures, the residual tensile strength was gradually decreased with the increase of temperature due to the deterioration of tensile strength of the textile fibres themselves. Ombres (2015a), examined the bond between TRM and concrete at elevated temperatures. Parameters investigated were the number of PBO-TRM layers (1, and 2 layers), and the exposed temperature (50, and 100 0 C). The specimens were firstly exposed to predefined temperatures (50 and 100 0 C) for 8 hrs, cooled down to the ambient temperature, and then subjected to single-lap shear test. It was found that the bond between TRM and concrete was significantly affected from the elevated temperatures. Particularly, the ultimate load was dropped (compared to the ambient load) by 0%, and 36% for one layer; and 28%, and 38% for two layers strengthened specimens when subjected to 50, 100 0 C, respectively.
Regarding the performance of TRM versus FRP at high temperatures, the only studies reported in the literature on the effectiveness of TRM versus FRP at high temperature were that of Tetta and Bournas (2016) and Bisby et al. (2013). In Tetta and Bournas (2016), the effectiveness of TRM versus FRP in shear strengthening of half-scale rectangular beams and full-scale T-beams at high temperatures was compared. The investigated parameters were: the temperature at which the specimens were exposed (20 0 C, 100 0 C, 150 0 C and 250 0 C), the number of strengthening layers (2 and 3), and the strengthening configuration. The results indicated that TRM jackets had far better performance in increasing the shear capacity of strengthened beams at high temperature than FRP jackets which totally lost effectiveness when subjected to temperature above Tg. Finally, in Bisby et al. (2013), FRP and TRM flexurally retrofitted beams were subjected to a sustained load and then exposed to increasing high temperature up to failure. In those specimens, the end anchorage zones were kept cold assuming that the debonding at this zone due to high temperature is prevented by a mechanical means. It was concluded that both strengthening systems (TRM and FRP) can have the same performance at high temperature if the anchorage zones of the beams were kept cold. However, in that study, the effect of high temperature on the debonding mechanism was not addressed because the bond condition was not realistically simulated due the cold anchorage zones. Hence, the effectiveness of the strengthening materials in increasing the flexural capacity of beams subjected to high temperature was not adequately evaluated.

Real Applications of TRM: Selected Case Studies
In the last decade, TRM has been successfully used worldwide for strengthening and seismic retrofitting of concrete and masonry structures. Selected case studies of real applications of TRM in the construction field can be found in Bournas (2016)  Other examples of real applications of TRM for retrofitting of concrete structures can be summarised as follows: • Flexural strengthening of RC slabs, and shear reinforcement of unreinforced masonry walls of school building in Karystos, Greece (2007) using carbon-TRM.

(a) (b)
• Retrofitting of the cooling towers of The Niederaussem Power Station in Germany (2012) using PBO-TRM.
As mentioned previously, TRM has also been used for retrofitting of masonry structures, examples of these structures as follows: • Strengthening and seismic retrofitting of the historical San Roque Church located in Spain (2008) which experienced an earthquake that induced an out of plane separation of the exterior walls.
• Retrofitting of masonry chimney with a total height of 38m and diameter ranging from 3.6 m at the bottom to 1.7 m at the top. This chimney located in Gerardmer, France, represents symbol of industrial heritage. Carbon TRM was selected for strengthening where it was applied in the vertical and transversal directions of the chimney.
• Strengthening of the main dome of the Molla Celebi Mosque in Turkey (2013) using four layers of basalt TRM.

Contribution of this Study
TRM is a new strengthening material and more experimental studies are required to better understand its behaviour. Based on the literature survey, the following points can be highlighted: 1. It is obvious from the literature survey described in Section 2.3 that the bond between TRM and concrete has not covered adequately yet. In particular, the main focus of the previous studies was on PBO textile fibres with the maximum number of layer being equal to two. The common conclusion was that the failure occurred within the composite materials due to slippage of the fibres through the mortar.
Contrary to the failure mode observed in FRP which is often debonding from concrete substrate. Thus, Chapter 4 of this PhD Thesis examines systematically a great variety of parameters the majority of them were not investigated previously including: (a) the number of TRM layers from 1 to 4, which is beyond the current limit of two, (b) the bond length (50-450 mm), (c) the concrete surface preparation, (d) the concrete compressive strength, (e) the coating of the textile, which has not been investigated before in comparison with uncoated textiles, and (f) the anchorage of tested bond length through wrapping with TRM jackets (see Chapter 4), which is again a parameter investigated for the first time.

Research on the bond between TRM and concrete at high temperatures is very
scarce. The only available study (see Section 2.5) is that of Ombres (2015a)

Cement Mortar
The matrix used for specimens retrofitted with TRM was an inorganic modified cement mortar comprising cement and polymers. The ratio of cement to polymers is 8:1 by weight, whereas, the water cement ratio was 0.23:1, resulting in a mix with a very good workability and plastic consistency. The flexural and compressive strength of the mortar were experimentally obtained on the day of testing. The test was conducted according to BS EN 1015-11 (1999) on three standard mortar prisms with dimensions of 40x40 mm cross section and 160 mm length.
It is generally recommended that the cement matrix should meet the following requirements: no shrinkage; high level of workability so as to allow for using a trowel during application; high viscosity in order to facilitating the application of mortar on overhead surfaces; and slowly rate of losing workability which allows for application the mortar layer while the previous one is still in a fresh state (Triantafillou, 2011).

Epoxy Resin
For those specimens strengthened with FRP system, a commercial epoxy resin (Sikadur ® 330) was used as a binding material. This epoxy consisted of two epoxy parts, the mixing ratio of these two parts was 4:1 by weight. According to the product datasheet, the tensile strength, modulus of elasticity, and the glass transition temperature (Tg) of this adhesive was 30 MPa, 3.8 GPa, and 68 0 C, respectively.

Textile Fibre Materials
Seven different types of textile fibre materials were used in this study as means of external reinforcement. Three of them were fabricated with fibre rovings (made of the same fibres materials) distributed equally in two orthogonal directions, namely: dry carbon_ fibre textiles (C), coated basalt_ fibre textile (BCo), and dry glass_ fibre textile (G) (Figure 3.1). Details of the textiles, such as mesh size, weight, density, equivalent thickness, tensile strength and modulus of elasticity (according to the manufacturer datasheets) of each textile material, are also given in Figure 3.1. It is noted that the textile fibres thickness (tf) in each direction was calculated based on the equivalent smeared distribution (the ratio of areal weight to density) of fibres. The dry carbon fibres textile was coated using low viscosity two-part epoxy resin in order to investigate the effect of textile surface condition (dry versus coated carbon fibre textile) on the performance of the textile. The acronym used for the coated carbon fibres textile is CCo. The procedure for application of coating included, impregnation the textile with low viscosity epoxy resin using a plastic roll and leaving the textile for two days (prior to use) at the ambient temperature for curing. According to the manufacturer data sheets, the tensile strength and the modulus of elasticity of the adhesive used for coating were equal to 72.4 MPa and 3.18 GPa, respectively. The remaining four types of textiles were hybrid and fabricated (by a UK company) using two different types of fibre materials in the two-orthogonal directions.
The longitudinal direction (direction of loading) comprising carbon fibre rovings, whereas, the transversal direction consisting of glass fibre rovings (Figure 3.2a-d). All four types of the hybrid textiles had the same quantity of carbon fibres in the direction of loading compared to the dry carbon fibre textile (i.e. C). The only difference was the spacings between rovings; in specific, two hybrid textiles had the same area/distance (i.e. 10 mm) of the carbon rovings in the loading direction compared to the dry carbon textile (C), whereas, the remaining two had double spacing (i.e. 20 mm)/area of the carbon rovings in the loading direction compared to the dry carbon fibres textile. The transversal rovings on the other hand, comprising glass fibres with two different spacing between rovings, namely, 20 and 40 mm which resulted in textile geometries with the following acronyms and details: • F10x20: Hybrid textile had 10 mm spacing between the longitudinal carbon rovings and 20 mm spacing between the transversal glass rovings (Figure 3.2a).
• F10x40: Hybrid textile had 10 mm spacing between the longitudinal carbon rovings and 40 mm spacing between the transversal glass rovings (Figure 3.2b).
• F20x20: Hybrid textile had 20 mm spacing between the longitudinal carbon rovings and 20 mm spacing between the transversal glass rovings (Figure 3.2c).
• F40x40: Hybrid textile had 20 mm spacing between the longitudinal carbon rovings and 40 mm spacing between the transversal glass rovings (Figure 3.2d).
The weight, density, nominal thickness, and modulus of elasticity of the carbon fibres in the loading direction for all four types of the hybrid textiles were the same and equal to 174 g/m 2 , 1.83 g/m 3 , 0.095 mm, and 225 GPa, respectively (according to the manufacturer datasheets).

Tensile Tests on Bare Textiles
Uniaxial   bf is the textile width (50 mm) and tf is the nominal thickness of the textiles in the direction of loading (see Figure 3.1). The modulus of elasticity (Ef,tex.) was calculated by dividing the ultimate tensile stress (ffu) to the corresponding ultimate tensile strain ( fu) because the behaviour of the textile is linear up to failure (see Figure 3.4a and b).

Tensile Tests on TRM Coupons
Uniaxial tensile tests were conducted on TRM coupons comprising one and two textile layers, in order to evaluate the tensile properties of the composite materials. Three identical specimens (TRM coupons) made of one and two layers for each type of textile material (described in Section 3.43. 3) were fabricated and tested. The geometry of the coupons is shown in Figure 3.5a, whereas the test setup is depicted in Figure 3.5b. The TRM coupons had a dumbbell shaped which was a modification of the setups adopted by Brameshuber (2006a); Orlowsky and Raupach (2008). The coupon was gripped to the tensile machine using special steel fixtures (see Figure 3.5b) that were used to fit the curved parts of the coupons and also used to apply the tension load. Each coupon was instrumented with two LVDTs (one on each side) which were mounted at the centre of the coupon to measure the tensile strain of the composite material in a gauge length of 240 mm. The load was applied monotonically under displacement control at a rate of 2 mm/min using a 200-kN capacity universal testing machine up to failure. All coupons failed due to rupture of the fibres at the central region of the gauge length (see Figure 3.6). During the test and after initiating of the first crack, as the load increase more cracks were appeared and developed until the failure occurred. It is noted that the crack pattern of the two layers' TRM coupons was denser than that of one layer, hence better stress distribution was achieved when the number of layers increased. The results of the tensile tests were presented in form of stress-strain curves.

Tensile Tests on FRP Coupons
The FRP composite used as external strengthening of RC beams comprised three different types of textile materials namely: the dry carbon, the coated basalt and the dry glass fibre textiles in combination with epoxy resin described in Section 3.2. As in the case of TRM system, uniaxial tensile tests were carried out on FRP coupons consisted of one layer of textile reinforcement in order to determine their tensile properties in the direction of loading. The FRP coupons had a rectangular shape and were designed according to the requirements of ACI 440. 3R-04 (2008). Three identical coupons were tested for each type of textile material. The geometry of the coupons is shown in Figure 3.11a, whereas, the test setup is depicted in Figure 3.11b.
The coupon was gripped to the tensile machine using two aluminium plates (see Figure   3.11a) which were glued to the ends of each coupon. Two LVDTs were fixed at the centre of the coupon (one on each side) to measure the tensile strain of the composite material in a gauge length of 300 mm. The load was applied monotonically under loading control at a rate of 5 kN/min (440)   and modulus of elasticity (Ef,FRP). The tensile strength of FRP coupons was calculated in the same way described for the TRM coupon, whereas the elastic modulus of FRP coupons was calculated directly from the stress-strain curves by dividing the ultimate stress (ffu) to the corresponding ultimate strain ( fu) because the behaviour of the stressstrain curves almost linear up to failure.

Summary
The main conclusions drawn from this chapter are summarized below: • In general, the tensile properties obtained from testing one layer of bare textiles coupons were approximately identical to that measured from the equivalent TRM coupons. Hence, for design purposes, it is suggested that both tests can be used to determine the tensile properties of TRM as a composite material.
• Increase the number of TRM layers from one to two, reduced slightly the ultimate tensile strength (average of 4%), and increased marginally the ultimate tensile strain (average of 4%).
• All hybrid textiles showed approximately identical tensile behaviour in both tests, the bare textile and the TRM composite. Moreover, their tensile properties (i.e. ultimate tensile stress, ultimate tensile strength and modulus of elasticity) were approximately the same to that of the dry carbon fibre textile (C). Such behaviour indicated that the effect of the fibre's materials (carbon or glass) and the distance between the rovings in the transversal direction was not significant. •

Test Specimens and experimental parameters
The main objective of this chapter is to investigate the bond between TRM and concrete substrate considering different parameters. A total of 80 specimens were fabricated, strengthened and subjected to double-lap shear test. The details of the specimens are provided in Figure 4.1a-f. Each specimen comprised two RC prisms with cross sectional area of 100x100mm, and length of 265, or 515 depending on the tested bond length.
The procedure for specimen's preparation was as follows: an acrylic plate with dimensions of 100x100 mm cross sectional was fixed at the middle of a steel mould (Figure 4.1a) in order to isolate the two prisms during casting stage. The acrylic plate was provided with two acrylic rods with 10 mm diameter fixed at the position shown in Figure 4.1b so as to create holes into concrete mass of each prism. Each prism was reinforced with a steel cage with the details shown in Figure 4.1c to prevent the failure of prisms due to concrete splitting during the test. A 16-mm bar was fitted at the centre of each prism to allow for the application of the load during the test (Figure 4.1d).
After 24-hour of casting, the specimen (two prisms) was removed from the mould, the acrylic plate was removed from the central zone, and the two prisms were reconnected to each other's using a 10 mm diameter acrylic rods that were inserted into the premade holes (see Figure 4.1d). The purpose of these two acrylic rods was to ensure fully alignment between the two prisms and reduce the error in the measurements during the test resulted from possible bending of specimen due to misalignment between the two prisms. Finally, full details of the design of the test specimen and a 3D overview is shown in Figure 4.1e and f, respectively.  Table 4.1.

Materials and strengthening procedure
The RC prisms were cast in different groups and dates. For all tested specimens, the targeted concrete compressive strength was 30 MPa except for group LN_X_Ls (twelve specimens) where the targeted compressive strength was lower and equal to 15 MPa. The compressive strength of all specimens was measured on the day of the testing (average value of three 150x150x150 mm cubes) and is given in Table 4.1.
The textile reinforcement used for strengthening was the dry carbon fibres textile (C) described in Section 3.3. The binding material comprising the inorganic cement mortar described in Section 3.1. The compressive and flexural strength of the mortar (average value from 3 prisms) were experimentally obtained on the day of testing using prisms with dimensions of 40x40x160 mm according to BS EN 1015-11 (1999 and are reported in Table 4.1. Prior to strengthening, the concrete surface was prepared by removing a thin layer of concrete (using of a grinder) and creating a grid of groves (with a depth of approximately 3 mm_Figure 4.2a). This procedure was performed for all specimens, except for those of group LX_N_S, where the concrete surface was sandblasted ( Figure 4.2b). After cleaning and dampening the concrete surface, the first layer of mortar with approximately 2 mm thickness was placed on the concrete surface using a metallic trowel (Figure 4.3a). Then the first textile layer was applied and pressed slightly into the mortar, which protruded through the perforations between the fibre rovings as shown in Figure 4.3b. This procedure was repeated until the required number of TRM layers was applied. Finally, an external layer of mortar with approximately 3 mm thickness was applied and levelled by trowel (Figure 4.3c). Of crucial importance in this method was the application of each mortar layer while the previous one was still in a fresh state.
The specimens in group LX_N_CCo were retrofitted using the coated carbon textile (CCo) described in Section 3.3. For the specimens that received wrapping, namely the main TRM reinforcement was anchored through TRM jackets wrapped around the concrete prism (group LX_N_W), additional surface preparation was made prior to strengthening including rounding of the prism corners to a radius of 10 mm.
After applying the required number of main TRM layers, the prism side under investigation was wrapped with two TRM layers following the strengthening procedure described previously. The width of the textile used for wrapping was 100 mm which was equal to the bond length of the main TRM reinforcement (Figure 4.3d).
It is worth mentioning that the bond width of TRM reinforcement for all tested specimens was the same and equal to 80 mm.

Experimental setup and procedure
All specimens were tested after a curing period of six weeks (same curing conditions were applied to all specimens). As mentioned previously (see Section 2.3.1) the double-lap shear test was adopted in the current study. The experimental setup included two steel clamps which were fixed at one side (restrained side) of the specimen to ensure that failure would occur in the monitored side (Figure 4.4). The TRM composite was left un-bonded at a 100 mm-long central zone (50 mm at each prism) of the specimen (Figure 4.1f) to prevent concrete-edge failure which could have adverse effects. This was achieved by wrapping the central zone (prior to strengthening) with a plastic tape in order to isolate the strengthening materials from the concrete prisms at this zone and prevent any possible attachment with the concrete surface. All tests were carried out using a universal testing machine of 250-kN capacity. The specimens were gripped to the tensile machine using the 16 mm steel bars fitted at the centre of each prism during casting (these bars were terminated at the interface between the two prisms). The load was applied monotonically under displacement control with rate of 0.2 mm/min. Two LVDTs were mounted to the unstrengthened sides of the specimens to measure the relative displacement between the two prisms ( Figure 4.4).

Experimental Results
Figure 4.5 shows the free body diagram of the tested side of the specimen. By assuming perfect symmetry (up to peak load) between the two TRM strip in the tested side, each side will carry half of the measured ultimate load (Pu.), whereas, the relative displacement between the two concrete prisms measured at ultimate load is the average of the two LVDTs' readings (i.e. δmax = (δ1+δ2)/2). Key results of all tested specimens are presented in Table 4.2 which includes: 1. the maximum load (Pu) carried out by the TRM strips for both twin specimens S1 and S2.
2. the displacement (average of two LVDTs readings) which corresponds to the maximum load ( max).
3. the average ultimate load (Pav) of the two twin specimens.
4. the average displacement (δav) of the two twin specimens.
5. the average bond strength developed at the concrete-matrix interface ( ).
6. the average tensile stress in the textile reinforcement ( ).
Initial position of the prism before application of the load

Pmax./2
The average bond strength ( ) and the average tensile stress in the textile reinforcement ( ) were calculated from Eq. 4.1 and Eq. 4.2, respectively: where . is the average ultimate load, is the bond length, bf is the bond width (bf=80 mm), n is the number of TRM layers, tf is the nominal thickness of the textile in the loading direction (tf =0.095mm).
Eq. 4.2 was used to calculate the effective stress of the fibres excluding the contribution of the mortar. This is typical in the case of TRM systems, and is valid for the ultimate capacity, since the matrix has already been cracked. At this load level, all the tension force is carried by the textile reinforcement.
Starting from the specimens LX_N that were strengthened with one up to four TRM layers at bond lengths of 50, 100, 150, 200 and 250 mm, the maximum load recorded (average from twin specimens) was (see also   to the number of TRM layers applied. It is noted that the trend of the curves of twin specimens was similar in all the cases (see "S1" and "S2" columns in Table 4.2). A common characteristic of all curves is their behaviour up to the maximum load. In specific, a first ascending linear branch with high axial stiffness is followed by a second ascending non-linear branch with progressively decreasing stiffness due to mortar cracking.  For the specimens strengthened with one or two TRM layers, the failure mechanism was controlled by slippage and partial rupture of the longitudinal fibres through the mortar at the loaded end, where a single crack was developed (at an early loading stage) and further opened at the end of the test (Figure 4.7). After failure, a residual strength was recorded which was attributed both to the contribution of friction between the inner filaments themselves and the outer filaments with the surrounding matrix. When TRM debonding from the concrete substrate occurred, it was accompanied by removal of a thin concrete cover layer (Figure 4.8). Failure was initiated by the formation of a longitudinal crack at the loaded end; this crack was continuously propagating towards the free end as the load was increasing. At peak load, propagation of the crack up to free end caused full detachment (debonding) of the TRM composite from the concrete surface and the load dropped to zero. A noticeable difference between the specimens failed due to fibres slippage and those specimens failed due to TRM debonding is that in the latter case several transversal cracks developed on the TRM face as shown in Figure 4.9. Hence, a better distribution of stresses along the bond length was achieved in these cases due to better activation of the textile reinforcement when the number of layers increased.
After debonding occurred, a rotation of the specimen with respect to the longitudinal axes was observed (Figure 4.9). This is because the failure was control by one of the two monitored sides of the concrete prism. However, this rotation had no effect on the bond behaviour because it happened after reaching the ultimate load. As shown in Table 4.2, specimens with low concrete strength (LX_N_Ls) reached an ultimate load of 29.9, 30.7 and 34.9 kN for three layers, and 32.2, 35.1 and 37.7 kN for four layers, for bond lengths of 100, 150 and 200 mm, respectively. As also illustrated in Figure 4.10b, the global behaviour of this group of specimens in terms of force-displacement curves was very similar to their counterparts with higher concrete strength (i.e. group LX_N). The failure mode was also identical to their counterpart equivalent specimens including debonding of TRM from the concrete substrate accompanied with removal of concrete particles which remained attached to the debonded TRM strip (Figure 4.11b). It is observed that the quantity of concrete cover being peeled off was thicker than that of the corresponding specimens, and this is due to the weaker concrete surface resulted from lower concrete strength. The load-displacement curves of the specimens retrofitted with coated textiles (LX_N_CCo) are presented in Figure 4.10c. The ultimate load measured for one TRM layer was 21.9 kN and 23.9 kN for 150 and 200 mm bond length, respectively, which is substantially higher with respect to their counterpart's specimens strengthened using dry carbon textile. The corresponding ultimate load of the two TRM layers was 29.5 and 31.9 kN for 150 and 200 mm bond length, respectively. As shown in Figure 4.10c the post-peak behaviour of LX_N_CCo specimens was different from their counterparts from group LX_N, owing to the different failure mode observed. In particular, all specimens with coated textiles failed due to debonding of TRM due to fracture the surface at the textile-mortar interface (Figure 4.11c). This failure mode was different from their counterpart's specimens which experience slippage of the textile fibres through the mortar (Figure 4.7). Coating the textile with epoxy significantly enhance the bond between the inner and the outer filaments in a single roving. As a result, failure due to slippage of the fibre through the mortar was prevented, and damage was shifted to the textile-mortar interface, which seems the weakest among all interfaces. This type of failure mode can also be described as interlaminar shearing. A denser crack pattern was observed in all specimens with the coated textiles, indicating a better activation of the textile fibres in tension.
Finally, the load-displacement curves for specimens LX_N_W, which were wrapped with two TRM layers in order to provide better anchorage, are shown in and 50.8 kN for three and four layers, respectively (for 100 mm bond length). In terms of ultimate load response, they performed better than their counterparts (see Table 4.2) due to delay the premature debonding, whereas a change on the failure mode was also observed. Wrapping of the prism did not allow for debonding of the TRM strips and damage was localized in the loaded-end, where a single transversal crack appeared

Discussion
In terms of the various parameters investigated in this experimental programme, an examination of the results in terms of ultimate loads and failure modes revealed the following information.

Influence of the bond length and the number of layers
The effect of the bond length and the number of layers on the load-carrying capacity is depicted in Figure 4.13. The curves in Figure 4.13 clearly demonstrate that by increasing either the bond length or the number of layers, the bond capacity increases in a non-proportional way. Similar to the bond behaviour of FRP strips (Yao et al., 2005), after a certain bond length the anchorage force tends to reach a constant value which is considered as the maximum anchorage force. This length is called "effective bond length" (Leff) and according to the curves provided in Figure 4.13 is in the range of 200 and 300 mm for the number of layers (one to four) investigated. For the same bond length, increasing the number of layers resulted in an increase in the load-carrying capacity. This effect was more pronounced for the transition from one to two layers, whereas for more layers it was gradually becoming less significant.
Almost the same trend was followed for all examined bond lengths between 50 and related to the change in the failure mode. In particular, as explained in Section 4.2, specimens of LX_N group strengthened with one or two layers failed due to slippage of the textile fibres through the mortar (see Figure 4.7), whereas specimens with three or four layers failed due to TRM debonding from the concrete substrate with peeling off of a part of the concrete cover (Figure 4.8).
The above finding adds new information to the existing knowledge, because in all previous studies on bond between TRM and concrete (where the maximum number of layers examined was two), failure occurred either at the interface between fibres and mortar or at the interface between concrete and mortar without involving the concrete cover. It is noted that failure of TRM involving peeling off of the concrete cover has also been reported in the study of Tetta et al. (2015), where RC beams were retrofitted in shear with TRM U-jackets. This type of failure is very common in case of FRP bonded to concrete (Yao et al., 2005), indicating that TRM can behave similar to FRP by increasing the number of strengthening layers.
The bond length had also an effect on the residual bond strength of the specimens failed due to slippage of the fibres, which is related to the friction developed between the inner and the outer filaments of each individual fibre roving. Table 4.3 shows the percentage of residual load compared to the maximum load recorded for specimens one and two TRM layers. It is generally concluded that the larger the bond length, the higher the slipping surfaces become, so the residual strength do. The bond length had also effect on the bond strength ( _calculated from Eq. 4.2) at the concrete-matrix interface. It is noted that the bond strength was calculated only for those specimens failed due to debonding of TRM from concrete substrate (see Table 4.2). As shown in Figure 4.14, as the bond length increase, the bond strength at the concrete-mortar interface decrease (approximately in proportional way). This is typical behaviour because increasing the bond length led to increase the area of interface that resist the applied load. Furthermore, it is noted that the effect of the number of layers on the bond strength at the concrete-matrix interface was very limited attributed to the identical observed failure mode (debonding). Finally, Figure 4.15 shows the variation of the tensile stress in the textile fibres reinforcement (calculated from Eq. 4.2) with the bond length for different number of TRM layers. It is generally observed that by increasing the number of layers, the normal stress decreases, which is consistent with the behaviour of FRP bonded plates to concrete (Yao et al., 2005).  where no differences were observed between specimens with untreated and sandblasted concrete surfaces, strengthened with one PBO-fibres TRM layer.

Influence of concrete compressive strength
The concrete compressive strength was selected to be investigated only for three and four TRM layers. This is because of the failure mechanism observed in LX_N specimens. In particular, TRM debonding from the concrete substrate involving part of the concrete cover (a failure mechanism which is associated to the concrete strength) layers and 8%, 7.4% and 9.2% four TRM layers, and for bond lengths equal to 100, 150, and 200 mm, respectively. As expected, the lower (by 50%) compressive strength resulted in a decrease in the ultimate load which on average was equal to approximately 7.5%. This reduction, though, cannot be considered as significant as it may be in the range of the statistical error. It is noted that the insignificant effect of the concrete strength on the load capacity has also been reported by D'Antino et al. (2015).
However, in their study the concrete was not directly involved in the failure mode which was at the interface between the matrix and the fibres.

Influence of coating
Coating the textile fabric with epoxy resin was investigated only for specimens with one and two TRM layers, in order to prevent the premature failure due to slippage of the fibres through the mortar that observed in these specimens with uncoated textiles.
According to the results, the effect of coating was twofold:  Coating the textile with epoxy resin made the textile more stable and easy-toapply, while at the same time it increases its rigidity. When a good level of impregnation of the fibres with resin is achieved, the inner filaments of the rovings are better bound to the outer filaments. As a result, the mechanism of transferring stresses from the fibres to the matrix is improved providing better mechanical interlock conditions. Ultimately, the textile fibres are better utilized in carrying tensile forces and the load capacity increases. A more uniform distribution of stresses is also achieved (something that is indicated by the formation of several transversal cracks) and the failure mode changes from local slippage of the fibres to global debonding of the TRM strips with the failure surface though being within the TRM thickness (textile-mortar interface).

Influence of anchorage through wrapping
The influence of anchorage through confinement (full wrapping) was investigated for a short bond length (100 mm) and for 3 and 4 TRM layers. The idea behind that was to improve the bond conditions when a short bond length (less than the effective bond length) is provided, by preventing early TRM debonding. As shown in Figure 4.19, the load capacity was increased by 28% and 45% when three and four TRM layers, respectively were anchored through wrapping with TRM jackets. Note that the bond length was equal to 100 mm whereas two TRM layers were used for wrapping. As expected, the failure mode changed from TRM debonding to partial rupture and slippage of the fibres across a single crack developed at the loaded end (Figure 4.12b).

Specimen name
A conclusion that must be highlighted is that the anchored TRM strips with a short bond length (100 mm) not only reached, but exceeded the load capacity of nonanchored strips with much higher bond length. Particularly, by comparing specimen L100_3_W with specimens L200_3 and L250_3, an increase of the maximum load of 11.1% and 5.2%, respectively, is observed. Similarly, by comparing specimen L100_4_W with specimens L200_4 and L250_4, the increase in the maximum load reaches 22.3% and 21.4%, respectively. Therefore, wrapping with TRM jackets is recommended to improve the bond conditions when the available length for anchorage of TRM reinforcement is limited.

Summary
This • By increasing the bond length, the bond capacity increases in a non-proportional way for all the number of TRM layers examined (1 to 4). After a certain bond length, the so-called effective bond length, the increase in the bond capacity was not significant. This length is ranging between 200 to 300 mm for the examined number of layers and for the materials used in this study.
• By increasing the number of TRM layers for the same bond length, the bond capacity increases in a non-proportional way. The increase was more pronounced for the transition from one to two layers due to the change in the failure mode, whereas for more layers it was gradually becoming less significant.
• The number of layers has a significant effect on the failure mode. For one and two TRM layers the failure was due to slippage of the textile fibres through the mortar at a single crack close to the loaded end. For three and four TRM layers the failure was attributed to debonding at the mortar/concrete interface including detachment of a thin concrete layer, similarly to EB FRP systems.
• Different concrete surface preparation methods (grinding and formation of a grid of grooves versus sandblasting) did not influence the bond characteristic between TRM and concrete, suggesting that both methods are suitable.
• The use lower concrete compressive strength marginally affected the bond strength of the TRM to concrete. A 50% reduction in concrete's compressive strength resulted in an average decrease of the ultimate bond capacity of 7.5%, without affecting the failure mode.
• Coating the textile with an epoxy adhesive has a twofold effect: (a) change in the failure mode from slippage through the mortar to TRM debonding at textile-mortar interface, and (b) increase the ultimate load by 75% and 15% (comapred to their counterpart speicmens strengthened wiht dry textile) for specimens retrofitted with one and two layers, respectively. •

Strengthening Procedure
The main aim of this chapter is to investigate the bond behaviour of TRM-to-concrete at high temperatures, and also to compare the bond of TRM versus FRP with concrete at different high temperatures and loading conditions. In total 68 specimens (34 twin specimens) were constructed, strengthened and tested under double-lap shear test. The specimen's setup is described in detail in Section 4.1.1 supported with Figure 4.1a-f.
Each specimen comprised two RC prisms with dimensions of 100x100mm cross section and 265 mm length. The two prisms were connected only by FRP/TRM layers which were bonded on two opposite sides of the prisms.
The parameters examined were: (a) the matrix used to impregnate the fibres, namely resin or mortar, resulting in two strengthening systems (TRM or FRP), (b) the temperature to which the specimens were exposed (50, 75, 100, 150 0 C) for FRP and (50,75,100,150,200,300,400 and 500 0 C) for TRM retrofitted specimens (c) the number of layers (3 and 4), and (d) the loading condition, namely steady state test and transient test conditions. In the steady state test, 56 (28 twin) specimens were heated up to a predefined temperature (see Table 5.1), kept at this temperature for 60 min., and then loaded monotonically up to failure. In the transient test, 12 (6 twin) specimens were first loaded (at ambient temperature) up to a load fraction equal to 25%, 50%, and 75% of the bond strength of the corresponding specimens tested at ambient temperature and then the specimens were heated up to failure.
The notation used for the specimens is BN_T, where B represents the type of binding material (R for epoxy resin and M for cement mortar), N refers to the number of FRP/TRM layers, whereas T denotes the exposed temperature for steady state tests, and the loading fraction of specimens tested at ambient for transient test condition. For example, M4_400 refers to a specimen strengthened with 4 TRM layers and tested monotonically (in steady state condition) at 400 0 C; whereas, M4_75% denotes to a 4 layers TRM specimen, subjected to a load fraction of 75% of the bond strength measured at ambient temperature, and then exposed to high temperature up to failure.
Details for each parameter of all specimens are also presented in  Figure 5.1).
The specimens were cast in different groups using the same mix design. The concrete compressive strength was obtained on the day of the testing.   The strengtheneing procedure for TRM speicmens was presented in Section 4.1.2 and including the following steps: (a) preparation the concrete surface ( Figure   5.3a); (b) application of the first layer of mortar followed by the first layer of textile ( Figure 5.3b). For specimens that received FRP, the concrete surface was prepared by removing a thin layer of concrete cover followed by roughening the surface ( Figure   5.3c), then the first layer of the textile fibres was applied on a thin layer of epoxy resin and impregnated using a plastic roll (Figure 5.3d). For both strengthening systems, the above procedure was repeated until the required number of layers (3 or 4 layers) was applied.
Note that, before application of strengthening materials, a 100 mm-long central zone was wrapped with a foil tape ( Figure 5.3a and c) in order to isolate the strengthening materials from the concrete prisms at this zone and prevent any possible attachment with the concrete surface. This was performed in order to prevent concreteedge failure as described in Chapter 4 (Section 4.1.3). Note also that the bond width of FRP/TRM reinforcement was the same for all tested specimens and was equal to 80 mm.

Test Setup, Instrumentations and Procedure
The specimens were positioned inside a furnace with inner chamber dimensions of 600 mm x 400 mm x 400 mm and maximum temperature of 600 0 C. The furnace was installed into a universal testing machine of 250-kN capacity, as shown in Figure 5.4a.

Figure 5.4. (a) Details of the test setup; and (b) details of test specimen.
The instrumentations used for specimens tested in steady state condition included: (i) Two high temperature LVDTs, fixed to the specimens' un-strengthened sides to measure the relative displacement between the two prisms (

Experimental Results
As already mentioned in Section 4.2 supported with Figure 4.5, by assuming a perfect symmetry (up to peak load) between the two TRM strip in the tested side, each side will carry half of the measured ultimate load (Pu.), whereas, the relative displacement between the two concrete prisms measured at ultimate load will be the average of the two LVDTs' readings; (i.e. δmax= (δ1+δ2)/2_see Figure 4.5).
The main experiment results of all specimens tested in both loading conditions are presented in Table 5.2 and Table 5.3. Table 5.2, reports the results of the steady state test including: (1) the ultimate load (Pu) recorded for twin specimens S1 and S2; (2) the relative displacement (  It is worth mentioning that the measurements of the strain gages at high temperatures were not reliable and therefore are not presented here.   Figure 5.13a); D: Debonding of FRP/TRM from the concrete substrate with peeling off part of the concrete cover (see Figure 5.13b and c for FRP specimens and Figure 5.13d-f for TRM specimens).

Temperature profile
Figure 5.6 presents a typical temperature-time curve obtained from the two thermocouples affixed at the concrete-matrix interface, for a specimen tested in steady state condition and heated up to 400 0 C. Since the readings (in all tests) were identical, the average (of the two thermocouples) temperature was used. Figure 5.7 displays the actual temperature-time curves for all FRP and TRM-strengthened specimens tested in steady state condition. It can be observed that: (a) the heating rate is identical between all specimens and (b) all specimens were exposed to predefined temperature for one hour before application of the load, and then tested under displacement control up to failure. Any further exposure time (more than one hour) was related to the time required to test the specimens up to failure. Note that the consistency in the heating procedure for all tested specimens is important to reduce errors, obtain reliable, and comparable results

(d)
This is due the progressive decreasing in the stiffness of adhesive resulted from increasing the temperature at the concrete-adhesive interface.
Two types of failure modes were observed for FRP-strengthened specimens: (a) deboning of FRP from the concrete substrate including parts of the concrete cover

Transient test: time, temperature at failure, and failure mode
As reported in Table 5  For TRM strengthened specimens, premature adhesive failure modes were prevented due to the better resistance of mortar than resin at temperatures above Tg, with all specimens failing due to debonding including part of the concrete cover ( Figure 5.13d-f).

Discussion
In terms of the various parameters investigated in this experimental programme, an examination of the results (Table 5.2 and Table 5.3) revealed the following information.

Matrix materials (TRM versus FRP)
The matrix material (epoxy resin or mortar) significantly affects the bond performance of FRP and TRM composites with concrete at ambient and especially at high temperatures. At 20 0 C, although both FRP and TRM-strengthening specimens failed due to debonding including part of concrete cover, the bond performance of FRPstrengthened specimens was considerably better than TRM ones. In particular, the bond strength of 3 and 4 layers FRP specimens was 1.4, and 1.5 times higher than that of counterpart TRM specimens respectively, (see Table 5.2). This is attributed to the excellent bond between FRP composite and concrete substrate which is confirmed by the amount of concrete being peeled off (see Figure 5.10a and c for FRP specimens and Figure 5.11a and Figure 5.12a for three and four TRM specimens, respectively).
However, at high temperatures, the TRM system exhibited excellent bond performance with concrete, which was superior to that of FRP systems. In particular, in steady-state tests, the TRM specimens retained an average of 85% of their ambient bond strength up to 400 0 C. On the contrary, the FRP systems maintained approximately 17% of their ambient bond strength at 150 0 C due to the premature adhesive bond failure at the concrete-resin interface. In the next sections a comparison between the effectiveness of FRP versus TRM materials at high temperatures is made in terms of the exposed temperature, the number of layers, and the loading condition. and 4% when the temperature attained 300 and 400 0 C, respectively. The highest reduction in the bond strength was 48% for TRM specimens tested at 500 0 C ( Figure   5.14a) seems to be attributed to the reduced tensile and compressive strength of the mortar by 87% and 68% at that temperature ( Figure 5.14b). The observation that the reduction of bond strength is associated with the mortar strength is better explained if someone compares the quantity of concrete being peeled off. All TRM-strengthened specimens tested at ambient and high temperature failed due to deboning, but the concrete cover detached at high temperature was thinner than Finally, an attempt was made to examine the bond performance of TRM at 600 0 C; however, when the interface temperature reached 550 0 C, the specimen failed due to spalling of the concrete cover in an explosive manner. It is worth noting though that the TRM was still bonded to the concrete substrate even after the specimen's failure as illustrated in Figure 5.15. Such a type of failure was also observed by Chowdhury et al. (2007) in FRP strengthened column tests under fire scenario.

Influence of the number of layers
As depicted in Figure 5.14a, when the number of layers increased from 3 to 4, the ultimate load increased by 1.21 and 1.15 for FRP and TRM specimens tested at ambient temperatures, respectively. However, at high temperatures, the influence of the number of layers on the bond strength was more pronounced for the TRM than FRP specimens. As shown in Figure 5.14a, for FRP specimens, the effect of number of layers on the bond strength was almost disappeared above the Tg, as it was controlled by the properties of the epoxy resin.
The influence of the number of TRM layers on the bond strength was not that clear, nevertheless, specimens retrofitted with 4 TRM layers showed an overall higher bond strength for all temperatures investigated. It is worth mentioning that Rambo et al. (2015)observed similar results in TRM coupon tensile test, in which the tensile behaviour at high temperature of TRM coupons made of 3 and 5 fabric layers was better than the tensile performance of a TRM coupon made of one layer. Furthermore, Tetta and Bournas (2016) concluded that by increasing from 2 to 3 TRM layers the bond of TRM to concrete at high temperatures increases considerably. As it can be observed from Figure 5.16 for both FRP and TRM specimens tested in transient condition, when the load fraction level was increased, the time to reach failure was decreased and consequently the temperature did. Also, it is illustrated that the TRM outperformed their FRP counterparts for all load fractions. Particularly, the time required to reach failure of the TRM specimens was 3.3, 3.5 and 1.58 times higher for the low, moderate and high load fractions, respectively. Correspondingly, the attained temperature at failure was 3.3, 4,4 and 1.58 higher in the TRM-strengthened specimens. Another interesting observation from Figure 5.16a is that the bond strength attained at different temperatures was nearly identical for both loading conditions for the FRP-strengthened specimens. This confirmed that the temperature at the concreteresin interface controlled the bond behaviour rather than the loading condition, as also reported by Firmo et al. (2015a). This was not the case for the TRM system which was sensitive to the loading conditions. In fact, the TRM specimens had increased bond strengths at higher temperatures in the steady state in respect with the transient tests.

Loading conditions
As illustrated in Figure 5.16b, the measured bond strength of M4_300 which was subjected to 300 0 C, was almost double and triple the predefined bond strengths of specimens M4_50% and M4_25%, respectively which failed at around 300 0 C. ambient and high temperatures. The main findings of the current study are summarized below:

Summary
• The bond between TRM strengthening system and concrete substrate remains excellent at high temperatures up to 400 0 C.
• In steady state test the reduction in bond strength of FRP-strengthened specimens was significantly higher than for the TRM-retrofitted specimens with the increase of the temperature. The average reduction in the bond strength of FRP-concrete interface was about 83% when the temperature reached 150 0 C. Whereas the corresponding values in TRM-concrete interface was about 15% when the temperature attained 400 0 C.
• Two types of failure modes were observed in the FRP strengthened specimens tested in steady state condition. At ambient and moderate temperature (50 0 C), cohesive failure was observed with parts of the concrete cover remaining attached to the adhesive. Whereas, at elevated temperatures (i.e. 75, 100, and 150 0 C), adhesive failure at the concrete-resin interface was occurred. On the other hand, for TRM specimens subjected to temperatures up to 500 0 C, the failure was due to TRM debonding with parts of concrete cover peeling off.
• The bond strength at the FRP-concrete interface was nearly identical for the same temperature regardless of the loading condition (transient or steady state). On the contrary, the bond behaviour at the TRM-concrete substrate was sensitive to the loading condition, and resulted to considerably higher bond strengths (for nearly the same temperature) in the steady state in respect with the transient tests.

Test Specimens and experimental parameters
The   Table 6.1, with the support of Figure 6.2, provide a description of the tested specimens. The notation of the strengthened specimens is BN_F, where B represents the binding materials (R for epoxy resin, and M for cement mortar), N refers to the number of TRM or FRP layers and F denotes the type of textile fibres (C for dry carbon fibres, CCo for coated carbon fibres, BCo for coated basalt fibres and G for glass fibres). For the specimens retrofitted with Ujackets at their ends, an additional suffix (EA, standing for end-anchorage) is added to the notation. The description of the specimens follows: • CON: unstrengthened beam which served as control specimen.
• R3_C_EA and M3_C_EA: 3 dry carbon FRP and TRM layers strengthened beam, anchored at their ends with two dry carbon FRP and TRM layers, respectively.

Textile material
It is noted that seven layers of glass-fibre or basalt-fibre textile have approximately same axial stiffness of one dry carbon textile layer. The axial stiffness is expressed by the product n . tf . Ef, where n is the number of textile layers, tf is the nominal thickness and Ef is the elastic modulus of textile according to manufacturer data sheet (see Figure 3.1). Using this expression to calculate the axial stiffness of seven layers of coated basalt or glass yields approximately same value of axial stiffness of one layer of carbon fibres. Table 6.1 gives the normalized axial stiffness of the textile reinforcement used in all specimens (normalized to one layer of carbon-fibre textile).

Materials properties
The beams were cast in different groups using the same mix design of concrete. The those beams that received FRP, the epoxy resin described in Section 3.2 was used as a binding material.

Strengthening procedure
The strengthening material (TRM or FRP) was externally bonded to the bottom of the beams over a length of 1350 mm (see Figure 6.1a). The strengthening procedure for both strengthening systems had the characteristics of a typical wet lay-up application and comprised the following steps: • Prior to strengthening, the concrete surface was prepared as follows: for FRP strengthened beam, the surface was roughened using a grinding machine and the resulted concrete surface was cleaned from dust with compressed air (Figure 6.3a); For TRM-strengthened specimens, a 50-mm grid of grooves with a depth of approximately 3 mm was made using a grinding machine, as a means of improving the bond. Finally, the concrete surface was cleaned with compressed air ( Figure   6.3b).
• The procedure for application of TRM materials included: (i) dampening the concrete surface with water ( Figure 6.3b); (ii) application of a layer of mortar with approximately 2-3 mm-thickness (Figure 6.3c); (iii) application of the textile into the mortar, and gently pressing with hand to ensure good impregnation with cement mortar (Figure 6.3d).
• The procedure for FRP-retrofitted specimens included: application of the textile over a thin layer of resin and then impregnated with resin using a plastic roll ( Figure   6.3e).
• The above procedure for both strengthening systems was repeated in case of more than one textile layers were applied.
• For TRM-retrofitted beams, the final layer of textile was covered with a final layer of mortar with approximately 3 mm thickness and levelled (Figure 6.3f).
Similar surface preparation was used for the specimens that received U-shaped FRP or TRM end strips as an anchorage system (R3_C_EA and M3_C_EA), as shown in Figure 6.3g and h, respectively. The application of the two layered U-jackets commenced immediately after the application of the longitudinal external reinforcement.

Experimental setup and procedure
All beams were tested as simply supported and were subjected to four-point bending.
As shown in Figure 6.1b, the flexural span was 1500 mm, and the selected configuration resulted in a 340 mm-long constant moment zone and a 580 mm-long shear span. Calculations were made to ensure that a sufficient anchorage length of the FRP and TRM reinforcements was provided at the ends of beams. Details of these calculations are presented in Appendix A. The load was applied using a 100 kNcapacity servo-hydraulic actuator which was vertically fixed on a stiff reaction frame.
A picture of the test setup is shown in Figure 6.4.
All specimens were loaded monotonically up to failure, under displacement control with a rate of 1 mm/min. In addition to the internal LVDT (linear variable differential transformer) of the actuator, two LVDTs were fixed at the mid-span of the beam (one on each side) to measure the mid-span deflection. Two bearing plates with square dimensions of 100 mm and 25 mm thickness were fixed under the points of load application in order to prevent the local failure of the specimen due to concrete crushing. During the test, the load and displacement data were recorded at a sampling rate of 4 Hz, using a fully-automated data acquisition system.

Experimental results
The main results of all tested beams are presented in Table 6.2, including: (1) The cracking load (Pcr). (2)

Load-deflection curves
The idealized load-displacement curve for strengthened beams is presented in Figure   6.5. The curves of strengthened beams were characterised by three distinct stages

Ultimate loads and failure modes
The values of maximum loads and the observed failure modes of all tested beams are presented in Table 6.2, supported by Figure 6.7. The reference beam (CON) failed in flexure after the formation of large flexural cracks at the constant moment region. The failure was due to yielding of the tensile reinforcement followed by concrete crushing at the compression zone (Figure 6.7a). This type of failure mode is typical for under-reinforced beams. The yield and ultimate load was 30.1 kN and 34.6 kN, respectively, at corresponding mid-span deflection of 6.1 mm and 30.0 mm, respectively.
All FRP strengthened beams also failed in flexure at loads substantially higher than the control beam ( Two distinct failure modes were observed in the FRP-retrofitted beams. Specimens retrofitted with one and three layers of carbon-fibre reinforcement (R1_C and R3_C), failed due to debonding of the FRP composite from the concrete surface.
Debonding was initiated from an intermediate shear crack (Figure 6.7b and c) which caused debonding of the FRP composite from the concrete and propagated from the mid-span towards the end of the beam. Eventually, the FRP strip completely debonded from the beam's soffit with parts of concrete cover being attached. Details of the failure mode for those beams are also provided in Figure 6.7b and c). This kind of failure mode is brittle and quite common for FRP reinforced beams (Commitee, 2008). The beams strengthened with seven layers of coated basalt-fibre reinforcement (R7_BCo), seven layers of glass-fibre reinforcement (R7_G), and three layers of carbon-fibre reinforcement anchored at the beam's ends (R3_C_EA), failed due to fibres rupture at the constant moment region of the beam (Figure 6.7d-f, respectively). Five different failure modes were observed in the TRM-retrofitted beams depending on the number of TRM layers and the textile fibres material: -Loss of composite action due to slippage of the fibres within the mortar accompanied by partial rupture of the fibres, at a single crack within the maximum moment region (Figure 6.7g). This type of failure mode was not brittle (see the postpeak curve in Figure 6.6a) and was observed in specimen M1_C which retrofitted with one layer of dry carbon-fibre textile. A progressive load-drop was recorded as a result of the fibres slippage through the cement matrix. This type of failure mode was consistent with that observed in TRM to concrete bond tests presented in Section 4.2, for the same number of TRM layers and the same textile fibre materials (i.e. dry carbon), it is also reported in .
-Debonding of TRM due to fracture the surface at the textile-mortar interface.
This kind of failure mode was observed in specimen M1_CCo (strengthened with one layer of coated carbon fibre-textile). Debonding was initiated at the intermediate shear crack and propagated towards the end of the beam (Figure 6.7h). This kind of failure, which can also be described as interlaminar shearing, is attributed to the effect of coating. Coating the textile with epoxy leads to a strong bond between the inner and the outer filaments of each roving, which increases the rigidity of the textile in both directions and creates strong joints in the junctions between the longitudinal and transversal fibre rovings. As a result, failure due to slippage of the fibre through the mortar was prevented, and damage was shifted to the textile-mortar interface, which was the weakest among all interfaces. The same failure mode was also observed in the TRM to concrete bond tests for the same number of TRM layers and the same textile fibre materials (see Section 4.3.4). A detailed picture of the TRM failure surface is also given in Fig. 8h.
-Debonding of TRM from the concrete surface accompanied with part of the concrete cover. The debonding initiated from an intermediate shear crack (Figure 6.7i) and propagated from the constant moment zone towards one end of the TRM reinforcement. Eventually TRM debonded form the concrete surface with a part of concrete cover being peeled off (Figure 6.7i). This failure mode was observed in specimen M3_C and M5_C, and it was the same as in its counterpart FRP-retrofitted beam (R3_C). Again, the same failure mode was also observed in TRM to concrete bond tests for three layers of the same materials (see Section 4.3.1 and ).
-Fibres rupture in the region of maximum moment (Figure 6.7 j and k). This type of failure mode was noted in specimens M7_BCo and M7_G, strengthened with seven layers of coated basalt and glass-fibre textile, respectively.
-Debonding of TRM from the concrete substrate (part of the concrete cover was also included) at an intermediate shear crack (Figure 6.7l), followed by slippage of the fibres at a different region. This failure mode was observed in specimen M3_C_EA which was retrofitted with three layers of dry carbon-fibre textile and anchored with TRM U-jackets at their ends to provide anchorage. It is noted that providing U-jacket at the ends of the beam prevented debonding of TRM, but slippage of fibres finally occurred at the region where the longitudinal TRM meets the TRM U-jacket ( Figure   6.7l). The same failure mode was also observed in in TRM to concrete bond tests for the same number of TRM layers and the same textile fibre materials (see Section 4.3.5).

Bending stiffness and crack pattern
The bending stiffness of the tested beams at several stages (pre-cracking, cracking and post-yielding) is reported in Table 6.3. It was calculated form the load versus mid-span deflection curves as the tangent stiffness of the pre-cracking, cracking and postyielding stages. As shown in Table 6.3, the application of strengthening (TRM or FRP) enhanced the cracking and post-yielding stiffness compared to the reference beam. It is noted that the increase in the cracking and post-yielding stiffness was sensitive to the investigated parameters such as the strengthening system (TRM or FRP), the number of TRM/FRP layers, the textile fibre material, and the strengthening configuration. Percentage increase (%) in stiffness with respect to CON included in parentheses.

Discussion
All strengthened specimens responded as designed and failed by the loss of strengthening after yielding of the internal steel reinforcement. On the basis of the various parameters investigated in this experimental programme, an examination of the results (Table 6.2) in terms of strength, stiffness and failure modes, revealed the following information.

Number of strengthening layers
The effect of the number of layers on the beams flexural capacity was investigated for the case of dry carbon-fibre textiles, and is depicted in Figure 6.8a. For FRPstrengthened beams, tripling the amount of reinforcement (from one to three layers) resulted in almost proportional increase in the flexural capacity, namely 2.8 times. The corresponding enhancement in the TRM-strengthened beams was equal to 4.7 times (non-proportional increase). To further investigate the effect of increasing the number of TRM layers on the flexural capacity increase, a beam strengthened with five TRM layers was also tested. As shown in Figure 6.8a, applying five layers of TRM resulted in 6.3 times increase compared to one TRM layer. The non-proportional increase observed in the TRM strengthened (especially for the transition from one to more layers) is associated to the different failure modes observed, as described below.
The cracking and post-yielding stiffness were enhanced by increasing the number of layers for both strengthening systems in an identical manner (Figure 6.8b).
In the FRP-strengthened beams, tripling the number of layers resulted in an increase of 1.2 and 2.7 times in the cracking and post-yielding stiffness, respectively. The corresponding enhancement in TRM-retrofitted beams was similar, namely 1.1 and 3 times, respectively. It seems that the increase in the post-yielding stiffness was almost directly proportional to the number of layers even for the case of M5_C (4.5 times compared with M1_C). This is attributed to the fact that the only mechanism contributing to the flexural capacity increase is the activation of the externally applied materials in tension. The failure mode of FRP strengthened specimens was not sensitive to the number of layers; it was always debonding of FRP from the concrete substrate including part of concrete cover (Figure 6.7b and c). However, in the case of TRMretrofitted beams, the failure mode was sensitive to the number of layers. In particular, the failure mode altered when three or five layers of dry carbon-fibre textile were applied instead of one. With 3 or 5 layers, slippage of the fibres through the mortar was prevented and the failure, as in the case of FRP, was attributed to TRM debonding including part of concrete cover (Figure 6.7i). This behaviour is identical with the observations made by Tetta et al. (2015) in shear strengthening of RC beams with TRM when the number of layers increased from 1 to 2, also noted in double-lap shear TRM-to-concrete bond tests (see Section 4.3.1). Improved mechanical interlock between the increased number of textile layers and the surrounding mortar is believed to be the main reason for this behaviour.

Textile-fibres coating
Beam M1_C, strengthened with one TRM layer of dry (uncoated) carbon-fibre textile, failed prematurely due to local slippage of the fibres through the mortar, and since the coated textile exhibited good performance in the bond tests, it was decided to retrofit a beam using the same textile but with coated fibres (M1_CCo). As a result of that, the flexural capacity was further increased by 52% (compared to beam M1_C).
Additionally, the failure mode was changed from slippage of the fibres through the mortar to debonding of TRM due to fracture at the textile-mortar interface ( Figure   6.7h; interlaminar shear failure. Such a failure mode was also observed in bond tests when the same textile with the same coating was used (see Section 4.3.4 and also presented in ). As illustrated in Figure 6.9a-c, although the performance of the beam M1_C was poor compared to its counterpart FRPstrengthened specimen (R1_C), when coated textile was used, the behaviour of TRM became comparable to FRP. Coating the textile leads to improved bond between the inner and the outer filaments of each roving of the textile. Hence, the textile develops higher tensile stresses, and the matrix is called to transfer higher shear stresses, which leads to shear failure of the mortar (interlaminar shearing).

Textile-fibres material
According to the results (Figure 6.9a), in both TRM and FRP strengthening systems, the highest flexural capacity increase was achieved in the beams retrofitted by the coated basalt-fibre reinforcement. In TRM-strengthened beams, specimen M7_BCo recorded a 45% higher capacity increase compared to specimen M1_CCo with equivalent axial stiffness of the strengthening layers, and beam M7_G recorded a 49% higher capacity increase compared to beam M1_C. Note that the above comparisons were made on the basis of similar textile surface conditions (dry or coated textiles).
Similarly, in FRP-retrofitted beams, the flexural capacity increase of beam R7_BCo was 52% and 30 % higher than that of beams R1_C and R7_G, respectively. This disparity in the flexural capacity increase between beams with external reinforcement of approximately the same axial stiffness, can be attributed to the influence of the numbers of layers (one layer of TRM reinforcement was less effective than multiple no. of layers as discussed in Section 6.1.3.1), and to the fact that the basalt-fibre textile was coated, which was beneficial at least in the case of the TRM strengthening system.

End-anchorage with U-jackets
An end-anchorage system comprising U-jackets at both ends of the beams was applied only for specimens strengthened with three layers of carbon TRM or FRP, as a means of preventing premature debonding from the concrete substrate. As illustrated in Figure 6.9a, in the case of beam R3_C_EA, the strengthening efficiency was substantially increased (by 90%) compared to the beam without end-anchorage (R3_C). However, this enhancement was limited in the case of the TRM strengthened beam (only 9%). The difference in the behaviour between specimens R3_C_EA and M3_C_EA is attributed to the difference in the failure mode observed. Beam R3_C_EA failed due to rupture of the textile fibres (Figure 6.7f) achieving full composite action. In contrary, in beam M3_C_EA even if TRM debonding was prevented, a full composite action was not achieved due to slippage of the textile fibres at the junction where the longitudinal TRM meets the U-jacket (Figure 6.7.l). Table 6.4 reports the values of the TRM versus FRP effectiveness factor (k), which is defined as the ratio of the flexural capacity increase achieved by TRM to the increase achieved by the equivalent FRP. This factor varied between 0.46 and 0.80 for the different parameters examined in this study.

TRM versus FRP effectiveness factor
Increasing the number of dry carbon-fibre textile layers from one to three, resulted in enhancement of the effectiveness factor from 0.47 to 0.80, which was associated to the change in the failure mode of TRM retrofitted beams (from slippage of the fibres to debonding from the concrete substrate). Coating the carbon textile with epoxy resin in the case of 1 TRM layer increased the k factor from 0.47 to 0.73, as a result of prevention of fibres slippage.
The effectiveness factor for the specimens retrofitted with either coated basalt, or glass-fibre textiles was the same and equal to 0.63. In this case, although both FRP and TRM-retrofitted specimens failed due to rupture of textile fibres, the reduced effectiveness of TRMs can be attributed to the lower tensile strength of TRM composites compared to FRPs (as shown from the results of the coupons tensile tests_ see Table 3.2 and Table 3.3).
Finally, in terms of strengthening configuration, specimen M3_C_EA recorded an effectiveness factor of 0.46. This low value of k factor was due to the presence of slippage at the junction where the longitudinal TRM reinforcement meets the U-jacket ( Figure 6.7l). This slippage considerably reduced the TRM effectiveness and prevented a full composite action.

Analytical calculations
To calculate the effective stress, eff, of the TRM or FRP reinforcement, an inverse analysis method was used. The effective stress is defined here as the tensile stress of the composite material in the region of maximum moments at the instant of ultimate load. By using the experimental values of the flexural moment of resistance, , (Table 6.5), a standard cross section analysis was performed for each of the retrofitted beams. The procedure for the calculation of in this method is built on the equilibrium of internal forces and strains compatibility described in Triantafillou (2006) (Figure 5.10a-d). Also, the following assumptions were adopted: • There is perfect bond between the FRP/TRM strengthening layers and the concrete substrate.
• The ultimate compressive allowable strain of concrete ( ) is 0.0035.
• The strengthening material behaves linearly up to failure.
From equilibrium condition, the resultant of compression forces is equal to the resultant of the tension forces ( . . = ): The compression and tension forces can be expressed as follow (see also Figure   6.10c): where; , 2 , 1 , and is the compression force provided by concrete, the compression force provided by steel (in compression zone), the tensile force provided by steel (in tension zone), and the tensile force provided by FRP or TRM, respectively.
The above-mentioned terms (i.e. , 2 , 1 , and ) are the product of the following expressions: should not exceed 0.4; otherwise = 0.4, is the concrete compressive strength (see Table 6.1), is the beam's width= 100 mm, is the depth of the neutral axis, 2 , , and 2 is the area, the modulus of elasticity ( = 200 GPa), and the strain of the steel in compression. 1 , is the area and yielding stress of the steel in tension. , , and is the area, the modulus of elasticity, and the tensile stain of the TRM or FRP.
From the strain compatibility (see Figure 6.10b), the strain in the compression steel ( 2 ) can be expressed in terms of the strain in the fibres ( ) as follows: The theoretical ultimate moment can be calculated from Eq. 6.9 (by taking a moment about the centroid of the concrete block_ see Figure 6.10d): 6.9 Now to determine the effective strain ( ) in the fibres, the value of and is assumed to meet the two following conditions: If this condition is achieved, then the effective stress in the textile reinforcement ( . ) is calculated using Eq. 6.10 as follows: . = 6.10 All the above procedure is summarized in the following flow chart: It is worth mentioning that the mechanical properties of the external reinforcement (Ef and ffu) were taken from Table 3.2 (for TRM composite) and Table   3.3 (for FRP composite). Assume a value for x and f and then calculate C c , C s2 , T s1 , T f

Check if M u,theor ≈ M u , exp
Compute the effective stress in the textile reinforcement ( eff ) from Eq. 6.10 Check if: C c +C s2 ≈ T s1 +T f

NO YES
Calculate M u,theor from Eq. 6.9

YES NO
The experimental values of the effective stress, , resulted from the inverse analysis, are presented in Table 6.5. As shown in the same Table, the ratio of the effective stress ( , ) to the ultimate stress obtained from coupon test (ffu) was always less than one, except for the beam R3_C_EA (probably due to the effect of the end-anchorage system).
The theoretical values of the debonding stress of the composite material, . ℎ were calculated according to Eq. 6.11 [fib model code (2010)] equation for flexural strengthening with FRP) and are presented in Table 6.5 (without safety factors). Note that Eq. 6.11 can only be used for debonding failures occurring at the concrete substrate.

6.11
In the above equation, is the mean debonding stress of the composite material; is the intermediate crack factor and equal to 2; is the matrix factor and equal to 0.25 for the case of epoxy bonded CFRP system (the same value was used here for the case of the carbon-TRM system); is the shape factor (calculated from Eq. 6.12 below); ℓ is the length factor which can be taken equal to 1; is the elastic modulus the composite material (obtained from coupon test); is the equivalent thickness of the textile and is the concrete compressive strength.

6.12
where; is the width of the composite, and b is the width of the beam.
Eq.6.11 was used here to calculate both FRP and TRM debonding stresses in the cases of debonding failures. A comparison of the stresses calculated according to Eq. 6.11 with that stress developed in the FRP and TRM composite calculated based on cross section analysis using the ultimate moment obtained experimentally is presented Table 6.5. It was found that the debonding stress calculated by Eq.6.11 is in a good agreement with the experimental results of that beams reinforced with high FRP and TRM reinforcement ratio (M3_C, M5_C and R3_C) and failed due to debonding of FRP or TRM from the concrete substrate including part of the concrete cover. Figure 6.11 shows the relationship between the effective stress obtained experimentally , and the product f f; together with the curve corresponding to Eq. 6.11. Where f is the textile fibres reinforcement ratio (ρf =Af /bh), and f is the modulus of elasticity of the composite material obtained from coupon tests. It is clear from this figure that the effective stress developed in the textile fibres reinforcement is inversely proportional to the product f f when the failure is associated to debonding of the externally bonded reinforcement, regardless the binding material (epoxy resin or mortar). This trend of the effective stress is consistent with the trend of the theoretical stress calculated by Eq. 6.11 and shown in Figure 6.11a and b.
Based on the above findings, in design of flexural retrofitting with TRM system, the effective stress can be the minimum value obtained from coupon tests (ffu) and Eq. 6.11, applying the same safety factors as in FRP systems until more data become available and a semi-probabilistic approach can be applied to obtain TRMspecific safety factors. Nevertheless, this design approach is suggested to be used only when the failure mode is either TRM debonding at an intermediate crack or fibres rupture. According to the results of the present experimental study, this applies when more than 2 TRM layers are used for retrofitting.   • The effectiveness of TRM system in increasing the loading carrying capacity of retrofitted beams was less than that of FRP. Nevertheless, TRM effectiveness was sensitive to the number of layers. It was found that the effectiveness factor increased from 0.47 to 0.80 when the number of TRM layers increased from 1 to 3.
• Coating the carbon fibres textile with epoxy adhesive significantly enhanced the performance of TRM materials. When one layer of coated carbon textile was used instead of one layer of dry carbon textile, the flexural capacity gain increased from 12.7 to 19.7% (about 55% enhancement).
• Different textile fibres materials (carbon, coated basalt, and glass) having approximately the same axial stiffness resulted in different flexural capacity increases. In both strengthening systems, seven coated basalt-fibre textile layers recorded the highest flexural capacity increase, followed by seven dry glass-fibre textile layers, and finally by one carbon-fibre textile layer. This variance in the performance was related to the effect of number of layers (in both FRP and TRM strengthening systems), but also to the textile surface condition (dry or coated textiles) in TRM strengthening system.
• Providing end-anchorage with U-jackets to FRP-retrofitted beams resulted in 90% enhancement in the flexural capacity compared to non-anchorage beam. However, the corresponding enhancement in TRM-retrofitted beam was limited (9%) and was attributed to the presence of slippage of the textile at the U-jacketlongitudinal TRM sheets.
• Two types of failure mode were observed in the FRP-retrofitted beams, these failure modes were: debonding from concrete substrate (for specimens M1_C and M3_C), and fibres rupture at the constant moment zone (for specimens M7_BCo, M7_G and M3_C_EA). Whereas, in the TRM-retrofitted beams five different failure modes were observed, namely slippage of the rovings through the surrounding cement mortar (specimen M1_C), fracture the surface at the textile-matrix interface (interlaminar debonding-specimen M1_CCo), debonding of TRM from the concrete with peeling off parts concrete cover (specimen M3_C and M5_C), rupture of the textile fibres at the constant moment zone (M7_BCo and M7_G), and debonding of TRM from the concrete substrate followed by slippage of the fibres at a different region (specimen M3_C_EA). These failure modes were found to be sensitive to the number of TRM layers, the textile fibres materials (carbon, coated basalt or glass fibres), and the textile surface condition (dry or coated fibres).
• The failure modes observed in the TRM strengthened beams were identical to the failure modes noted in the bond tests (described in Chapter 4) for the same number of TRM layers and the same textile fibre materials.
• For both strengthening systems (TRM and FRP), the cracking and post-yielding stiffness of strengthened beams was substantially enhanced compared to the unstrengthened beam (up to 72% and 1298%, respectively).
• A formula proposed by fib model code (2010) was used to predict the debonding stress in FRP reinforced for those specimens failed due to debonding of FRP from concrete substrate. This formula was also used to predict the debonding stress of TRM reinforcement for those specimens that have same failure mode (i.e. debonding). It was found that this formula is in a good agreement with the effective stress calculated based on the experimental results providing that TRM properties are obtained from coupon tests.

Influence of Textile Geometry on the Performance of TRM in
Flexural Strengthening of RC Beams

Experimental programme
This section evaluates the effect of textile geometry on the flexural behaviour of RC beams strengthened with TRM. For this purpose, eleven beams were fabricated, strengthened and tested under four point-bending. The geometry, internal reinforcement of the test beams is identical to that beams presented in Section 6.1.1.1 (see Figure 6.1a-b). The investigated parameters were: (a) the number of TRM layers (1 and 3), (b) the geometry of textiles namely: the area of a single carbon roving in the direction of loading, and the material (carbon and glass) and spacing between rovings in the transversal direction. The textiles reinforcement used were the hybrid group F10x20, F10x40, F20x20, and F20x40 and the dry carbon fibres textile (C) described in Section 3.3. Figure 6.12 shows the textile characteristics namely: the mesh size, the area of carbon rovings in the loading direction, and the material and spacing between the transversal rovings. It is worth mentioning that the area of the carbon fibres in the loading direction is equal for all five types of textiles.
Two out of the eleven specimens have already been presented in Section 6.1.
These specimens are M1_C, and M3_C which were strengthened with 1 and 3 layer of the dry carbon textile (C). The details of the remaining nine specimens were as follows: one specimen was left without strengthening and served as a control beam (CON), whereas, the remaining eight specimens were strengthened with 1 and 3 layers of the four hybrid textiles. The notation of the hybrid strengthened specimens is FX_N, where X denotes to the type of hybrid textile (F10x20, F10x40, F20x20, and F20x40-see Figure 6.12), N refers to the number of TRM layers. Table 6.6, with the support of Figure 6.13, provides a description of the tested specimens.

=0.95
A rov.   Table 6.6. The tensile properties of the steel bars used for flexural and shear reinforcement were the same of that reported in Section 6.1.1.2.

=0.95
The strengthening procedure has also already been described in Section 6.1.1.3 and also documented here in Figure 6.14a-d. Finally, the beams were tested as simply supported and were subjected to fourpoint bending (see Figure 6.4).

Experimental results
The response of all tested beams is presented in Figure 6.15 in form of load-deflection curves, whereas key results are reported in Table 6.7. As in the case of the loaddeflection curves of the beams described in Section 6.1.2.1. The curves of tested beams were also characterised by three distinct stages up to the maximum load. These stages were (see also Figure 6.5): un-cracked stage, cracked stage, and post-yielding stage.
After reaching the ultimate load, the load was dropped to the level of the un-retrofitted (CON) beam indicating that the effect of strengthening had totally been lost.  The reference beam (CON) failed as designed in flexure at an ultimate load of 34.8 kN. The failure was due to yielding of the tensile reinforcement concrete followed by concrete crushing at the compression zone. As mentioned in Section 6. in the flexural capacity due to application of strengthening compared to the control beam. The failure of specimens F10x20_1, and F10x40_1 was due to slippage of the fibres through the mortar accompanied by partial rupture of the fibres (Figure 6.16a and c). This failure mode was identical to the failure observed in the counterpart specimen M1_C which was strengthened with one layer of dry carbon fibres textile (see Figure 6.7g). The failure of specimens F10x20_3 and F10x40_3 on the other hand was due to debonding of the TRM from concrete substrate accompanied by peeling off part of the concrete cover ( Figure 6.16b and d). This type of failure mode was also identical to the failure of specimen M3_C which was strengthened with three layers of dry carbon fibres textile (see Figure 6.7i).

Discussion
All strengthened specimens responded as designed and failed by the loss of strengthening after yielding of the internal steel reinforcement. Based on the various parameters investigated in this experimental programme, an examination of the results (Table 6.7) in terms of flexural capacity increase and failure modes, revealed the following information.

Influence of number of layers
The influence of number of layers on the flexural capacity increase of beams strengthened with the dry carbon fibres textile (C) and the four-hybrid textiles (F10x20, F10x40, F20x20, and F20x40) is depicted in Figure 6.17a. Tripling the number of layers resulted in dramatical improvement (non-proportional increase to the number of layers) in the effectiveness of TRM in enhancing the flexural capacity. The cracking and post-yielding stiffness were also influenced by the number of layers. As shown in Figure 6.17b, increasing the number of layers from 1 to 3 resulted in slight enhancement in the cracking stiffness but considerable improvement in the post-yielding stiffness (see also Table 6.8). The failure mode was also sensitive to the number of layers; for one TRM layer strengthened specimens, the failure was due a local damage as a result of slippage the fibres through the mortar. With three layers, the local damage was completely as in the case of specimens M3_C, F10x20_3, and F10x40_3 (see Figure 6.16b and d) or partially as in the case of specimens F20x20_3, and F20x40_3 ( Figure 6.16f and h) prevented, and the failure was shifted to the concrete substrate. Improving the mechanical interlocking due to the overlapping the textile layers is believed to be the reason of such behaviour.

Influence of textile geometry
This paragraph discusses the influence of textile geometry on the flexural capacity enhancement, cracking and post-yielding stiffness, and also, the failure mode. The term "textile geometry" includes: (a) the area of a single roving in the direction of loading, and (b) the materials and spacing between rovings in the transversal direction.
As described in Section 3.3, the dry carbon textile and the four types of the hybrid textiles had the same quantity of carbon fibres in the direction of loading. The only difference is the area of rovings (Arov_ Figure 6.12); in specific, the area of a single carbon roving of the hybrid textiles F20x20 and F20x40 is equal to Arov= 1.90 mm 2 which is double than the area of a single roving of the dry carbon textile (C), F10x20, and F10x40 (Arov= 0.95 mm 2 ). The materials and spacing between rovings in the transversal direction were also different; in particular, the dry carbon fibres textile (C) had carbon fibres rovings with 10 mm spacing, whereas, the hybrid textiles F10x20, and F10x40 had glass fibres rovings of 20, and 40 mm spacing in the transversal, respectively. Similarly, the hybrid textiles F20x20, and F20x40 had also glass fibres rovings of spacing equal to 20 and 40 mm in the transversal direction, respectively.
The influence of the area of a single roving (Arov) in the direction of loading on the flexural capacity increase is presented in Figure 6.18. Doubling the area of the rovings in the loading direction, resulted in slight reduction in the TRM effectiveness in enhancing the flexural capacity of both 1 and 3 TRM strengthened specimens. This is attributed to the degree of impregnation of a single roving into mortar; with smaller area of rovings, the degree of impregnation is better achieved resulted in improving the bond characteristics between the mortar and the textile reinforcement, consequently the flexural capacity of strengthened beams was also enhanced.  According to these findings, it seems that main function of the transversal rovings is to maintain the stability of the overall geometry of the textile. This conclusion leads to a significant reduction of the overall cost of textile reinforcement due to the considerable savings in the material of fibres in the transversal direction. • Increasing the number of layers from 1 to 3, dramatically enhanced the flexural capacity and also altered the failure mode.
• Doubling the area of a single roving in the direction of loading resulted in limited but adverse effect on the flexural capacity enhancement.
• The materials and spacing between rovings in the transversal direction had no effect on both the flexural capacity increase and the failure mode providing that the same amount of carbon fibres in the loading direction is used. This would significantly reduce the overall cost of the textile reinforcement due to the considerable saving of fibres' materials in the transversal direction of the textile reinforcement.
According to the above findings, and to produce an effective textile geometry combining acceptable flexural performance and low cost, the following suggestions could be beneficial: (a) the area and spacing between rovings in the loading direction would be as smaller as possible. This creates a denser mesh pattern which would improve the bond characteristics between the textile and mortar. In any case, the size of perforations between the rovings should be sufficient to ensure that cement mortars protruded through them; and (b) low-cost fibre materials can be used in the transversal direction with a spacing between rovings large as possible but ensuring the stability of the overall textile geometry. The main objective of this study was to compare the effectiveness of TRM versus FRP in enhancing the flexural capacity of RC beams at high temperature. A total of 23 halfscale rectangular section RC beams were constructed, strengthened and tested under 4-point bending load as follows: eleven beams were tested at high temperature (150 0 C), whereas the remaining twelve beams were tested at ambient temperature and presented in Section 6.1. The geometry, internal reinforcement of the test beams is identical to that beams presented in Section 6.1.1.1 (see Figure Figure 7.1a, provide description of the tested specimens and strengthening configurations. The strengthened specimens were named following the notation BN_F_T, where B denotes the type of bonding agent (M for cement mortar and R for epoxy resin); N the number of TRM or FRP layers; F the type of textile fibres material (C for dry carbon fibres, CCo for coated carbon fibres, BCo for basalt fibres and G for glass fibres); and T denotes the temperature at which the specimens were exposed (20 0 C or 150 0 C). For the specimens that received U-jackets at their ends (Figure 7.1b), an additional suffix (EA-End anchorage) is added to the notation. For example, 'M3_C_20' refers to a beam strengthened with 3 layers of dry carbon TRM and tested at 20 0 C, whereas 'R3_C_EA_150' refers to a beam strengthened with 3 layers of carbon FRP, anchored at its ends using two layers of Ushaped jacket, and tested at temperature of 150 0 C. The beams were cast in different groups using the same concrete mix-design.
The compressive and splitting tensile strength of the concrete were determined on the day of testing and the results are presented in Table 7.1. The binding material used for TRM strengthened beams was the cement mortar described in Section 3.1. The compressive and flexural strength (at 20 and 150 0 C) of the mortar are given in Table   7.1.  The tensile properties of the steel bars used for flexural and shear reinforcement were the same of that reported in Section 6.1.1.2. For those beams retrofitted with FRP, the epoxy resin described in Section 3.2 was used as a binding material. The strengthening material (TRM or FRP) was bonded to the beams' soffit over a length of 1350 mm. The strengthening procedure for both strengthening systems had the characteristics of a typical wet lay-up application as described Section 6.1.1.3 (see Figure 6.3a-h).

Testing protocol and instrumentations
All beams were simply supported and subjected to four-point bending. The flexural span was 1500 mm, and the selected configuration resulted in a 340 mm-long constant moment zone and a 580 mm-long shear span (see Figure 6.1). The load was applied using a 100 kN-capacity servo-hydraulic actuator which was fixed on a stiff reaction frame. A picture of the test setup is shown in Figure 7.2c. The beams were loaded monotonically up to failure at a displacement rate of 1 mm/min. Two LVDTs were fixed at the mid-span of the beam (one on each side) to measure independently the mid-span deflection.
For the beams tested at high temperature, five type K thermocouples were mounted to the concrete surface prior the application of the strengthening materials in order to monitor the temperature at concrete-adhesive interface. As shown in Figure  to ensure that the targeted temperature (i.e. 150 0 C) is uniformly reached along that span. The test procedure at high temperature included the following steps: the heating system was placed underneath the specimen; the height of the legs was adjusted in order to achieve a distance of 100 mm (to allow for beam's deflection) between the heaters and the beam's soffit. The specimen was heated up to the predefined temperature (i.e. 150 0 C), and then loaded monotonically up to failure, while the temperature at the concrete-adhesive interface was approximately kept constant at 150 0 C. The data of the tests was recorded using a fully-computerized data acquisition system.   and 150 0 C. The results of ambient temperature tests (also presented in Table 6.  D: TRM debonding from concrete substrate; AF: adhesive failure at the concrete-resin interface; DS: Debonding of TRM from concrete substrate, followed by slippage of the fibres at the region where the longitudinal TRM meets the TRM U-jacket; and AS: adhesive failure at the concrete-resin interface in the nonanchorage zone followed by partial rupture and slippage of the fibres at the region where the longitudinal FRP meets the FRP U-jacket.

Load-displacement curves
The response of all beams tested at ambient and high temperature is presented in Figure   7.4a-c in the form of load-displacement curves. As shown in Figure 7.4a-c, the load-displacement curves are characterized by three distinct stages: (1) Stage I: un-cracked beam; (2) Stage II: initiation of cracking up to steel yielding; and (3) Stage III: post-yielding response up to failure. The observed gain in flexural strength is due to the contribution of TRM/FRP reinforcement, and is completely lost after the peak-load (when this reinforcement is lost), with the load capacity dropped to the un-retrofitted (CON) beam level (the postpeak behaviour of the load-displacement curves was removed for the sake of clarity).

Ultimate load and failure mode
The control specimen (CON) sustained a peak load of 34.6 kN (Table 7.2) and failed in flexure. After yielding of the longitudinal reinforcement, the concrete in the compression zone crushed (Figure 7.5a).

FRP strengthened beams
All FRP-strengthened beams tested at ambient temperature failed in flexure at an ultimate load substantially higher than that of the control beam. The peak load recorded for specimens R1_C_20, R3_C_20, R7_BCo_20, R7_G_20, and R3_C_EA_20 was 43.9, 60.4, 54.2, 48.2, and 83.7 kN, respectively, yielding 26.9, 74.6, 56.6, 39.3, and 141.9 % gain in load-carrying capacity, respectively (Table 7.2). Two different failure modes were observed, namely: debonding of FRP from the beam's soffit including part of the concrete cover (Figure 7.5b -specimens R1_C_20 and R3_C_20), and rupture of the fibres at the constant moment region of the beam (Figure 7.5d, f and h specimens R7_BCo_20, R7_G_20, and R3_C_EA_20, respectively). All FRP-retrofitted beams tested at 150 0 C failed also in flexure but at ultimate loads significantly lower (except from specimen R3_C_EA_150) than their counterpart specimens tested at 20 0 C. The peak load attained by specimens R1_C_150,R3_C_150,R7_BCo_150,and R7_G_150 was 35.9,36.7,36.5,and 35.8 kN (Table 7.2), respectively, resulting in negligible increases in the flexural capacity equal to 3.8, 5.8, 5.5, and 3.5%, respectively. Thus, the effectiveness of FRP reinforcement in increasing the flexural capacity of the beams was decreased (in average) by 90% at 150 0 C in comparison with ambient temperature. In all of these specimens, adhesive failure at the concrete-resin interface was observed (Figure 7.5c, e and g), namely the FRP composite detached from concrete substrate without including any parts of concrete cover. This is attributed to the poor bond behaviour of epoxy resin at temperatures above Tg. Finally, specimen R3_C_EA_150 having an anchorage system provided by U-shaped FRP strip at the ends of the beam attained an ultimate load of 57.5 kN, which yields 53.4% reduction in the effectiveness of the FRP reinforcement compared to its corresponding ambient temperature. Failure of this specimen initiated by adhesive debonding at the concrete-resin interface in the mid-span which propagated to the anchorage zones, and then followed by slippage and partial rupture of the rovings through the resin, which lost its strength at high temperature (Figure 7.5i).

TRM strengthened beams
Similar to the FRP strengthened beams, the TRM ones tested at ambient temperature, sustained considerably higher loads than the control beam. The ultimate load-carrying capacity of specimens M1_C_20, M1_CCo_20, M3_C_20, M7_BCo_20, M7_G_20 and M3_C_EA_20 was 39, 41.3, 55.3, 46.9, 43.2, and 57.1 kN, respectively, resulting an increase in the flexural capacity of 12.7, 19.4, 59.8, 35.5, 24.9, and 65.0% in comparison with the control beam. Five different failure modes were observed depending on the investigated parameters. In particular, failure of specimen M1_C_20 was attributed to partial rupture and slippage of the fibres within the mortar ( Figure   7.6a), whereas specimen M1_CCo_20 failed due to debonding of TRM at the textilemortar interface (interlaminar shearing) (Figure 7.6c). Failure of specimen M3_C_20 was identical to R3_C_20, namely due to TRM debonding including part of the concrete cover (Figure 7.6e). Failure due to rupture of textile glass and basalt fibres was respectively observed in both M7_BCo_20 and M7_G_20 specimens (Figure 7.6g and i). Finally, specimen M3_C_EA_20 failed due to TRM debonding from concrete substrate, followed by slippage of the fibres at the region where the longitudinal TRM meets the TRM U-jacket (Figure 7.6k).
The performance of the TRM-strengthened beams tested at 150 0 C was far better compared to their FRP counterparts. In particular, specimens M1_C_150, M1_CCo_150, M3_C_150, M7_BCo_150, M7_G_150 and M3_C_EA_150, reached an ultimate load of 37.7, 38.3, 44.7, 41.1, 38.8, and 46.2 kN, respectively, resulting in 9, 10.7, 29.2, 18.8, 12.1, and 33.5% increase in the flexural capacity. Consequently, the effectiveness of the TRM at 150 0 C was decreased in average by about 45% in comparison with its performance at 20 0 C. Specimen M1_C_150, failed identically to its counterpart tested at 20 0 C due to partial rupture and slippage of the fibre rovings through the mortar (Figure 7.6b). Specimen M1_CCo_150 failed due to debonding of TRM at the textile-mortar interface (Figure 7.6d) similar to its counterpart specimen tested at ambient temperature. Specimen M3_C_150 failed also identically to its counterpart M3_C_20, namely TRM debonding from the concrete substrate involving parts of concrete cover ( Figure 7.6f), indicating the good bond between the concrete substrate and the TRM reinforcement even at high temperature. Specimens M7_BCo_150 and M7_G_150 had different failure modes compared to their counterpart specimens tested at 20 0 C, as they failed due to slippage of textile fibres (although some debonding was observed in specimen M7_G_150) through the mortar (Figure 7.6h and j). The alteration of failure mode is attributed to the reduction of the mortar strength at high temperature (see Table 7.1). Finally, the failure mode of specimen M3_C_EA_150 was also identical to its counterpart M3_C_EA_20 that is debonding of TRM from concrete substrate, followed by slippage of the fibres at the region where the longitudinal TRM meets the TRM U-jacket (Figure 7.6l).  Table 7.3 reported the bending stiffness of the tested beams at high temperature in precracking, cracking and post-yielding stages. As shown in this Table, the application of the FRP or TRM resulted in enhancing of both the cracking and post-yielding stiffness compared to the control beam. The average percentage increase in the cracking stiffness of both strengthening systems was approximately the same (17%). However, the percentage increase of the post-yielding stiffness of TRM strengthened beams was dramatically higher than that of the corresponding FRP-reinforced beams (see Table   7.3).    In the next sections a comparison between the effectiveness of FRP versus TRM materials at high temperatures in terms of the number of layers, the textile fibres materials, and the end-anchorage system is made. The effect of textile coating on the performance of TRM strengthened specimens in increasing the flexural capacity is also discussed.

Number of strengthening layers
The effect of the number of TRM layers on the beams flexural capacity enhancement at high temperature was investigated only for the case of dry carbon-fibre textiles, and is depicted in Figure 7.9. Increasing the number of layers from 1 to 3 layers, resulted in an almost proportional enhancement in the flexural capacity of 3.25 times. For FRP specimens the corresponding increase was nearly zero as can be seen in Figure 7.7a and Figure 7.9. When the number of TRM layers was increased from one to three, the failure mode altered from local fibre slippage to TRM debonding with concrete cover due to the better mechanical interlock, for both ambient and high temperatures, indicating that the failure mode was not affected from the increase of the temperature.
For FRP strengthened specimens however, the increase in the number of layers did not affect the failure mode, which was adhesive failure at the concrete-resin interface ( Figure 7.5c), attributed to the deterioration of the epoxy tensile strength above the Tg, as also reported in bond of FRP-to-concrete tests presented in Chapter 5.

Textile fibre coating
Coating was applied to the dry carbon-fibres textile to prevent the premature failure due to slippage of the fibre that was observed with dry carbon fibres textile. As a result of coating, the flexural capacity of specimen M1_CCo_150 was further increased by 19% compared to specimen M1_C_150. In fact, the effectiveness of the coated carbon textile was dropped compared to its effectiveness at ambient temperature (52%), most possibly due to the adverse effect of high temperature on the properties of the epoxy resin that used for coating.
The failure mode was altered from slippage to TRM debonding at the textilemortar interface (Figure 7.6d) because coating the textile improved the bond between the inner and outer filaments, and hence, prevented slippage. Identical failure mode was also observed at ambient temperature (see Section 6.1.3.1), indicating that the failure mode was not affected by the increase of the temperature.

End-anchorage with U-jackets
Providing end-anchorage enhanced significantly the effectiveness of FRP at high temperature (11.4 times compared to the non-anchorage beam). The corresponding enhancement for the TRM-strengthened beams was ten time less (only 1.14) due to the observed failure modes observed (sections 7.2.2.1 and 7.2.2.2, respectively). Table 7.4 reports the values of TRM versus FRP effectiveness factor (hT) at high temperature, which is defined as the ratio of the TRM to FRP in terms of flexural capacity enhancement. This factor was varying between 0.5 and 5.1 depending on the investigated parameter. Increasing the number of layers from one to three (for dry carbon fibres textile), resulted in an increase of the hT factor from 2.4 to 5.1 (2.12 times) due to the change in failure mode, as discussed in Section 4.2. On the other hand, coating the dry carbon textile in the case of one TRM layer increased the hT factor from 2.4 to 2.8 due to prevention of slippage of the fibres. The effectiveness factor hT for both specimens retrofitted with 7 basalt or glass TRM (M7_BCo_150 and M7_G_150) was approximately the same (about 3.4) due to their identical failure mode (slippage of fibres through the mortar). Finally, the low value of 0.5 for specimen M3_C_EA_150 that received end-anchorage, is related to the observed failure mode (see section 3.2.2).

Effective Stress Reduction Factor for FRP and TRM
The effective stress is defined here as the tensile stress of the composite material in the region of maximum moment at the instant of ultimate load. For all tested beams, the effective stress of the FRP or TRM reinforcement for both ambient ( eff) and high ( eff, high) temperature was calculated following the same procedure described in section 6.1.4 and using the experimental values of the flexural moment of resistance, Mu,exp (Table 7.4) and the mechanical properties of the TRM and FRP reinforcement (Ef and ffu) reported in Table 3.2 and Table 3.3, respectively.
The effective stress of FRP or TRM jackets at high temperature, σeff, high, is a reduced value of their effective stress, σeff, at ambient temperature. It is expressed by the following equation: ,ℎ ℎ =

7.1
The values of the effective stress of TRM and FRP jackets at both ambient and high temperature, σeff and σeff, high, respectively are given in Table 7.4. The calculated stress reduction factor, k varies with the strengthening material (TRM, FRP) and investigated parameter (see Table 7.4). For the FRP strengthened beams, the average values of k was quite low and equal to 0.29, whereas, the corresponding values of k for TRM strengthened beams was far higher and equal to 0.7.  Hence, for design purposes, FRP is not recommended for flexural strengthening of RC beams when fire or high temperature is a critical issue, unless proper protective (thermal insulation) systems are provided. For TRM strengthened beams on the other hand, and based on the limited experimental results presented in this study in the flexural design of beams strengthened with TRM and exposed to high temperature (up to e 150 0 C), the effective stress for those specimens failed due to debonding can be the minimum value obtained from coupon tests (ffu) and Eq. 6.11, after applying a reduction factor k equal to 0.5. It is worth mentioning that a reduction factor of 0.4 was proposed by Tetta and Bournas (2016) for shear design of beams strengthened with TRM jacketing and exposed to high temperature up to 250 0 C.  • TRM showed far better effectiveness than FRP in increasing the flexural capacity of RC beams subjected to high temperature. TRM sustained an average of 55% of its ambient temperature effectiveness, contrary to FRP which totally lost its effectiveness.
• Increasing the number of TRM layers (from 1 to 3) enhanced the flexural capacity and altered the failure mode. Whereas, the corresponding effect of the number of FRP layers was negligible due to the premature adhesive failure.
• Coating the dry carbon fibres with epoxy adhesive improved the TRM effectiveness in increasing the flexural capacity (approximately 20% compared to the dry one).
• The effect of textile materials (having approximate same axial stiffness) in the FRPstrengthened beams disappeared due to their identical adhesive failure at the concrete-resin interface.
• Providing end-anchorage to the FRP-retrofitted beam significantly enhanced the flexural capacity increase (compared to the non-anchorage beam). This enhancement was limited in the corresponding TRM-reinforced beam due to the witnessed failure mode.
• Different types of failure modes were observed in the TRM-retrofitted beams including: slippage of the fibres, interlaminar shear and debonding of TRM including parts of concrete cover. On the other hand, the only observed failure mode in the FRP strengthened specimens (except from specimen R3_C_EA_150) was adhesive failure.
• The fib model code (2010) formula, which predicted the experimental TRM debonding effective stress with good accuracy, can be also used in the flexural design of beams strengthened with TRM and exposed to high temperature (up to 150 0 C), after halving the ambient temperature effective stress. It is well known that the effectiveness of any externally bonded strengthening system in increasing the load-carrying capacity of concrete members substantially depends on the bond between that strengthening material and the member's substrate.
Due to granularity of cement mortars, the impregnation of a single roving is difficult to achieve, hence, the bond between TRM and concrete has become an issue.
This PhD Thesis provides extensive experimental study on the bond between textile reinforced and concrete substrate and, also evaluates the effectiveness of TRM in flexural strengthening of RC beams. Firstly, the tensile properties of the bare textiles reinforcement and TRM coupons were experimentally determined. Secondly, the bond behaviour between TRM and concrete substrate at ambient and for the first time at high temperature was extensively studied by conducting 148 double-lap shear tests (80 at ambient and 68 at high temperature). The key investigated parameters at ambient temperature were: (a) the bond length (50-450 mm); (b) the number of layers (1 to 4 which is beyond the current limit of two); (c) the concrete surface preparation Finally, a simple formula used for predicting the mean FRP debonding stress was modified for predicting the TRM reinforcement effective stress based on the experiment data available.

Conclusions
In general, according to the results presented in this Thesis, TRM reinforcement can be considered as a promising strengthening system for retrofitting concrete structures.
In the next sections, the conclusions obtained from the experimental work (grouped according to the chapters of this Thesis) are drawn:

Tensile characterisation of textile reinforcement
This part of the Thesis describes the experimental work carried out to determine the tensile properties (the ultimate tensile stress, the ultimate tensile strain and the modulus of elasticity) of: (a) bare textile reinforcement, (b) TRM coupons made of one and two layers and, (c) FRP coupons made of one layer. The main findings of this part are summarised below: • The tensile properties of the textile reinforcement obtained from testing a bare textile were approximately identical to that measured from TRM composite.
Hence, for design purposes, it is suggested that both tests can be used to determine the tensile properties of TRM as a composite material.
• Increase the number of TRM layers from one to two decreased the ultimate tensile strength by 4%, and increased the ultimate tensile strain by 4%. Thus, the tensile properties of TRM composite obtained from testing one TRM layer can be used for design purposes.
• The influence of textile materials (carbon or glass) and distance between rovings (10, 20 or 40 mm) in the transversal direction (of textiles having the same amount of carbon fibre in the loading direction) on the tensile properties was very limited. This could potentially lead to cost reduction of the textile reinforcement, especially if low-cost fibres are used in the transversal direction.
• The type of binding materials (cement mortar or epoxy resin) significantly affects the tensile properties of the resulted composite materials. FRP composite showed considerably higher tensile properties than that of the corresponding equivalent TRM composite made of the same textile fibres materials.

Bond between TRM and concrete: double-lap shear test at ambient temperature
The main findings of the experimental study conducted to investigate the bond between TRM and concrete substrates are summarized below: • By increasing either the bond length or the number of layers, the failure load increases in a non-proportional way. However, beyond a certain bond length, the increase in the ultimate load was not significant. This bond length so-called effective bond length which was ranging from 200 mm to 300 mm for number of layers up to four and for the textile fibres materials used in this study.
• For the same bond length, increasing the number of TRM layers resulted in nonproportional increase in the ultimate load. The increase was more pronounced when shifting from one to two layers, whereas for three and four layers it was gradually becoming less significant.
• The number of layers has a significant effect on the failure mode; for one and two TRM layers the failure was slippage of the fibres through the mortar, whereas, for three and four TRM layers the failure was debonding of TRM at the concrete-mortar interface including a thin layer of concrete cover. The type of failure mode is consistent with the typical failure mode observed in FRP-to-concrete bonded strip.
• The influence of concrete surface preparation methods (grinding and formation of a grid of grooves versus sandblasting) was very limited on the bond characteristics in terms of ultimate load and failure mode, suggesting that both methods are suitable.
• The effect of concrete compressive strength on the ultimate load was not significant; a 50% reduction in compressive strength of the concrete resulted in 7.5% average decrease in the ultimate bond, and with the same failure mode.
• Coating the textile with an epoxy adhesive has a twofold effect: (a) change in the failure mode from slippage through the mortar to TRM debonding at textile-mortar interface, and (b) the ultimate load by 75% and 15% (comapred to their counterpart speicmens strengthened wiht dry textile) for specimens retrofitted with one and two layers, respectively.
• The anchorage of TRM strips through wrapping with TRM jackets results in a substantial increase of the bond strength (up to 28% and 45% for 3 and 4 TRM layers, respectively), by preventing the premature debonding from the concrete substrate.
According to the above findings, and to improve the bond condition between TRM and concrete in real applications of TRM, the following recommendations can be considered: (a) preventing the premature slippage of the fibres through the mortar; this can be achieved by providing more than two TRM layers, (b) using coated textiles when it is possible, and (c) anchoring TRM strips through wrapping with TRM jackets wherever it is applicable.

Bond between TRM and concrete: double-lap shear test at high temperatures
The main conclusions of the experimental programme performed to examine the bond performance between TRM and concrete interfaces at high temperatures are summarized below: • The bond performance of TRM strengthening system with concrete remains excellent at high temperatures, contrary to FRP.
• In steady state tests, the bond strength at the FRP-concrete interface was significantly deteriorated with the increase of the temperature, however, this was not the case for TRM-retrofitted specimens. The average reduction in the bond strength of FRP-specimens was about 83% when the temperature reached 150 0 C, whereas the corresponding values of bond strength in TRM-concrete interface was about 15% when the temperature attained 400 0 C.
• Two types of failure modes were observed in the FRP specimens tested in steady state condition. At ambient and moderate temperature (50 0 C) cohesive failure was observed; in which parts of concrete cover being removed from concrete substrate, whereas, at elevated temperatures (i.e. 75, 100, and 150 0 C), adhesive failure at the concrete-resin interface was occurred. On the other hand, the only observed failure mode of the TRM specimens subjected to high temperatures up to 500 0 C, was debonding of TRM with parts of concrete cover being peeled off.
• The effect of loading condition on the bond strength of FRP specimens was not significant. The bond strength of the FRP specimens as a function of temperature was nearly identical regardless the loading condition (steady state or transient condition). On the other hand, the bond strength of the corresponding TRM specimens was sensitive to the loading condition. In specific, TRM specimens tested at steady state condition showed higher bond strength compared to the corresponding specimen tested in the transient condition.
Overall, TRM shows a

TRM versus FRP in flexural strengthening of RC beams
The main findings of this part of PhD Thesis are summarized as below: • Generally, TRM system was less effective than FRP in increasing the loading carrying capacity of retrofitted beams. Nevertheless, TRM effectiveness was sensitive to the number of strengthening layers; by increasing the number of layers from 1 to 3, the effectiveness factor (k) increased from 0.47 to 0.80.
• Coating the dry carbon fibres textile with epoxy adhesive significantly enhanced the performance of TRM materials. Providing one layer of coated carbon textile instead of dry one resulted in substantial gain in the flexural capacity increased (about 55% enhancement).
• Different textile fibres materials (carbon, coated basalt, and glass) having approximately the same axial stiffness resulted in different flexural capacity increases. In both strengthening systems, seven layers of coated basalt-fibre textile layers measured the highest flexural capacity enhancement, followed by seven layers of dry glass-fibre textile, and finally by one carbon-fibre textile layer. This variance in the performance was related to the effect of number of layers but also to the textile surface condition (dry or coated textiles) in TRM strengthening system.
• Providing end-anchorage system using U-shaped jackets to FRP-retrofitted beams resulted in 90% enhancement in the flexural capacity compared to non-anchorage beam. However, the corresponding enhancement in TRM-retrofitted beam was not significant recording only 9% due to the observed failure mode.
• Two types of failure mode were observed in the FRP-retrofitted beams, namely; debonding including parts of concrete cover for specimens M1_C and M3_C, and fibres rupture for specimens M7_BCo, M7_G and M3_C_EA. In the TRMretrofitted beams, five different failure modes were noted including: slippage of the fibres (specimen M1_C), debonding at the textile-matrix interface (specimen M1_CCo), debonding with peeling off parts concrete cover (specimen M3_C and M5_C), and rupture of the textile fibres (M7_BCo and M7_G) and debonding of TRM from the concrete substrate followed by slippage of the fibres at a different region (specimen M3_C_EA). These failure modes were sensitive to the number of TRM layers, the fibres materials (carbon, coated basalt or glass fibres), and the textile surface condition (dry or coated fibres).
• The failure mode observed in the TRM strengthened beams tested in flexural was identical to that failure of TRM strengthened specimens tested under double-lap shear, for the same number of TRM layers and the same textile fibre materials.
• For both strengthening systems (TRM and FRP), the cracking and post-yielding stiffness of strengthened beams was enhanced compared to the unstrengthened beam (up to 72% and 1298%, respectively).
• A formula proposed by fib model code (2010) was used to predict the debonding stress in FRP reinforced for those specimens failed due to debonding of FRP from concrete substrate. This formula was also used to predict the debonding stress of TRM reinforcement for those specimens that have same failure mode (i.e. debonding). It was found that this formula is in a good agreement with the effective stress calculated based on the experimental results providing that TRM properties are obtained from coupon tests.
Based on the above conclusions, it is obvious that the number of TRM layers plays important roles in enhancing the TRM effectiveness in increasing the flexural capacity of RC beams. Hence, in real applications of TRM, it is suggested to use more than two TRM layers in order to prevent the premature failure due to slippage of the fibres and consequently enhancing the TRM effectiveness.

Influence of textile geometry on the performance of TRM in flexural strengthening of RC beams
This section studied the effect of the textile geometry on the flexural performance of RC beams strengthened with TRM. The main conclusions of this part are summarized below: • Increasing the number of TRM layers from 1 to 3 resulted in improving the flexural capacity, and altering the failure mode from slippage of the fibres (for 1 layer strengthened beam) to a total or partial debonding of TRM from concrete substrate (for 3 layers retrofitted beams).
• Doubling the area of a single carbon fibres roving in the loading direction reduced slightly the textile effectiveness in enhancing the flexural capacity.
• The materials and spacing between the rovings in the transversal direction had no effect on the textile effectiveness in enhancing the flexural capacity. This conclusion leads to a considerable reduction of the overall cost of the textile reinforcement achieved by saving the materials in the transversal direction of the textile reinforcement.
According to the above findings, and to produce an effective textile reinforcement in terms of cost and flexural performance, the following suggestions are made: (a) smaller area and spacing between rovings in the direction of loading could enhance the flexural performance of the textile reinforcement. This is due to creating a denser mesh pattern that improves the bond condition between the rovings and the mortar; and (b) low-cost fibre materials can be used in the transversal direction with a spacing between rovings as large as possible but ensure the stability of the overall textile geometry.

TRM versus FRP in flexural strengthening of RC beams: behaviour at high temperature
The main findings of the experimental work carried out on RC beams strengthened with TRM and FRP composites subjected to high temperatures are summarized below: • TRM showed far better effectiveness than FRP in increasing the flexural capacity of RC beams subjected to high temperature. TRM maintained an average of 55% of its ambient temperature effectiveness, contrary to FRP which totally lost its effectiveness.
• Increasing the number of TRM layers (from 1 to 3) enhanced the flexural capacity and altered the failure mode. Whereas, the corresponding effect of the number of FRP layers was negligible due to the premature adhesive failure.
• Coating the dry carbon fibres with epoxy adhesive improved the TRM effectiveness in increasing the flexural capacity (approximately 20% compared to the dry one).
• The effect of textile materials (having approximate same axial stiffness) in the FRPstrengthened beams disappeared due to their identical adhesive failure at the resinreinforcement interface.
• Providing end-anchorage to the FRP-retrofitted beam significantly enhanced the flexural capacity increase (compared to the non-anchorage beam). This enhancement was limited in the corresponding TRM-reinforced beam due to the witnessed failure mode.
• Different types of failure modes were observed in the TRM-retrofitted beams including: slippage of the fibres, interlaminar shear and debonding of TRM including parts of concrete cover. On the other hand, the only observed failure mode in the FRP strengthened specimens (except from specimen R3_C_EA_150) was adhesive failure.
• The fib model code (2010) formula, which predicted the experimental TRM debonding effective stress with good accuracy, can be also used in the flexural design of beams strengthened with TRM and exposed to high temperature (up to e 150 0 C), after halving the ambient temperature effective stress.

Recommendations for Future Research
TRM is a promising alternative strengthening materials to FRP, even with experimental evidence presented in this PhD Thesis, further experimental studies are required in order to consolidate the obtained results. In this context, further research could be directed towards the following fields: 1. Investigating the bond of TRM made of different types of textiles and concrete, and deriving analytical expressions for calculating the effective bond length and the effective stress in the TRM composites.
2. Durability issues are important, and research on the durability of TRM is scarce; therefore, it is important to assess the bond performance between TRM and concrete substrate at to extreme environmental conditions such as cycles of freezing and thawing.
3. To generalize the results obtained from the flexural tests, further experimental studies are required on full-scale beams in order to confirm the reliability of existing design models for FRP and/or to develop new reliable ones.
4. Further research could be directed towards examining the fatigue behaviour of RC beams strengthened with TRM and subjected to cyclic loading.
5. More studies are required on the flexural behaviour RC beams strengthened with TRM and subjected to high temperature (above 150 0 C), to transient load histories and to real fires. Testing full-scale beams is also needed for increasing the level of confidence of the obtained results.
6. Finally, developing a fire-resistant cement mortars (such as using a refractory cement) is important as the cement mortar plays a crucial role in increasing the effectiveness of TRM as an external strengthening.
• The concrete compressive strength, = 18 in order to simulate a real situation of deteriorated concrete compressive strength as a result of aging, or degradation induced environmental conditions.
• The average yield stress in the steel rebar is equal to 560 MPa (obtained experimentally_ see Table A.1 and Figure A.2).   It is worth mentioned that the beam was provided with compression reinforcement comprising 2 12 ( 2 = 226 2 ) in order to prevent the failure of the retrofitted beams due to concrete crushing in the compression zone and ensure that the failure would occur within the strengthening material (i.e. full utilization of the strengthening material).
The beam was designed for shear using a factorized load seven times greater than the predicted ultimate load of the control beam (36.1 ). The procedure for the shear design was conducted according to Eurocode2 (2004).
The H8 was used as shear links (shear reinforcement), hence the spacing between the shear links can be calculated from Eq. A.5.

A.2 Calculation of Anchorage Length
The strengthened material was bonded to the beam's soffit over a length of 1350 mm.
Simple calculations were made to ensure that this bond length provides a sufficient anchorage length. The anchorage length (La_ see Figure A.1) is defined as the length just outside the effectively strengthened area. Providing a sufficient anchorage length ensures that the strengthening materials sustained a maximum stress when the failure due to debonding induced intermediate crack (Teng et al., 2002). In the next sections, the calculations of the anchorage length for both the FRP and TRM strengthened beams is provided.

A.2.1 Anchorage length for FRP strengthened beams
The procedure for calculations of the anchorage length was made according to the procedure described in Triantafillou (2006). It is noted that the calculations of the anchorage length were made for the 3 layers FRP strengthened beam (which was the maximum number of FRP layers used for strengthening).
The required anchorage length is calculated from Eq. A.6: Hence, the ultimate moment capacity of the 3 layers strengthened beams is calculated from standard cross section analysis and is equal to 14.3 . .
The procedure for calculation the anchorage length was as follows: 1. Calculating the tension force ( ) provided by the ultimate moment and the tension force provided by tension steel ( ).
2. Dawning the moment diagram (in terms of forces).
3. Intersecting the line corresponding to the tensile force carried by tension steel with the ultimate tensile force envelope provided by the ultimate moment. The section beyond that intersection point is the actually provided anchorage length (residual anchorage length_ Lresd).
4. Checking whether the provided anchorage length is greater than or equal the required anchorage length calculated from Eq. A.6.
To apply the above procedure, the tension force resulted from the ultimate moment

A.2.2 Anchorage length for TRM strengthened beams
According the results of the bond tests described in Chapter 4, the effective anchorage length of TRM is 200 mm. Therefore, it was decided to consider this length as an anchorage length of the TRM strengthened beams. The same procedure described for the FRP strengthened beams is also adopted here so as to ensure that an anchorage length of 200 mm is secured. The only difference in the calculations was the determination of the effective strain ( ) in the TRM reinforcement which was obtained directly from the bond test of the specimen strengthened with 3 TRM layers and had bond length of 200 mm as follows: The ultimate load of the 3 layers TRM strengthened specimen was 36 kN (see Table 4.2). The corresponding effective stress is 790 MPa, hence the corresponding effective strain is: = = 790 225 * 10 3 = 0.0035 Form the cross-section analysis, the ultimate moment of the 3 layer TRM strengthened beam is 12.6 . .
now the tension force resulted from the ultimate moment and from the tension steel reinforcement is: