Review on Bond Properties b etween Wood and Fiber Reinforced Polymer

: Retro ﬁ tting of existing ancient and modern timber structures has been an important project recently. And it triggers a need of excellent strengthening methods, so does the strengthening of newly built architecture. Traditional strengthening methods have shortcomings such as high costing and destroying the aesthetic of the structure, many of which can be overcome by means of using ﬁ ber reinforced polymer (FRP) composites. However, the behavior of FRP-to-wood systems has yet to be thoroughly researched compared with their FRP-to-concrete or FRP-to-steel counterparts. As FRP retro ﬁ tting and strengthening timber structures has a promising future, better understanding of their failure modes will enable more precise designs balancing safety and cost. Three of the most common FRP-to-wood systems in the literature are discussed in this paper, namely, the externally bonded reinforcement (EBR), the near-surface mounted (NSM) and the glued-in rods (GiR) techniques. Debonding of the FRP from the substrate is one of the most common failure modes, which exhibits the signif-icance of the interface bond between FRP laminates and wood. Hence, bond properties and behavior of FRP-to-wood composite systems are described, parameters in ﬂ uencing the composite action are summarized in this paper, previous works on the bond interface of FRP and timber element are reviewed and future topics are also suggested. This work can provide a reference for future research and engineering applications.


Introduction
Due to its outstanding properties such as high elastic modulus, high fatigue performance, high stiffness and strength-to-weight ratio and superior resistance, fiber reinforcement polymer (FRP) combining high strength fibers and a resin matrix, has been widely used in practice [1]. Its versatility has been demonstrated by a wide variety of industrial applications, particularly for strengthening of concrete structures [2][3][4] and steel structures [5][6][7][8]. More recently, timber structures have also been considered as an area of application. In contrast with FRP, wood as well as bamboo has been broadly utilized in construction for thousands of years and has numerous applications in structural engineering [9][10][11][12].
Although being renewable, recyclable, relatively inexpensive and architecturally attractive, wood possesses inherent defects, such as biodegrade over time and be dimensionally unstable under alternating environmental conditions.
To solve this problem, research programs focused on the reinforcement of wood beams have examined the use of FRP applied on the tension side. FRP materials have excellent mechanical properties and exhibit very good characteristics especially in relation to long-term behavior such as corrosion resistance [13,14]. FRPs have been employed either to improve flexural and shear characteristics of existing structures or to reduce the dimension of new timber structures. Three of the most common FRP-to-wood systems in the literature are: the externally bonded reinforcement (EBR), the near-surface mounted (NSM) and the glued-in rods (GiR) techniques. EBR consists of FRP laminates bonded on the surface of the timber element, which is usually used to retrofit existing timber structures. While NSM is an efficient method to strengthen newly built structures due to its greater efficiency in flexural and shear strengthening. Groove is cut near the wood surface before inserting FRP bar into it with adhesive in the NSM system. As illustrated in Fig. 1. GiR have been used in the construction engineering for decades based on steel rods, however, FRP rods have multiple advantages, such as improved mechanical properties, high resistance and increased compatible with resin and timber which point towards FRP GiR as a suitable strengthening method for wood.
Bond behavior has always been a crucial issue in strengthening techniques since it has a significant impact not only on the ultimate load-carrying capacity of the composite strengthening system but also on serviceability aspects such as deformation and crack width. Although FRP strengthening wood appears more and more frequently as a research subject in literature, the study on the bond behavior is still in its infancy as illustrated in Fig. 2.  This paper presents a review on the bond properties and behavior of FRP as a strengthening alternative for timber structures including a summary of previous tests conducted in literature and the main parameters influencing the bond strength. Future research topics are also suggested.
2 Bond of FRP-to-Wood System 2.1 Bond of FRP Laminates to Wood The feasibility of the FRP-wood bonding technique [15] such as mechanical enhancement [16][17][18][19], failure mode [20], selection of adhesive [21][22][23], selection of FRP [24] and effects from environmental conditions [25,26] has been covered by a number of studies in the literature. Biscaia et al. [27] revealed that the FRP-wood interfaces had the highest strength among the interfaces between FRP and three substrates: concrete, steel and wood. From all these studies, a general agreement that the FRP-to-wood is indeed an efficient approach to strengthen the wood material can be reached.
In spite of the many merits of FRP reinforcing or retrofitting structures, there are still some fundamental disadvantages in the external bonding application to structural members. Much of the success of this technique is heavily dependent on the interfacial performance between the FRP composite and the substrate. As a result, the effectiveness of the stress transfer between materials for binding reinforcement method is vital for the bonding strengthening method. In the case of concrete structures, many studies suggest that premature debonding of the FRP composite from substrate occurs [28][29][30]. Debonding can be defined as the single most important failure mechanism of retrofitted beams [31] that occurs at much low FRP strains before the beam reaching its ultimate strain, which directly impacts the total integrity of the structure with the subsequent outcome that the ultimate capacity and desirable ductility of the structure may not be achieved.

Composite Action
How well stress transmits between FRP composite and the substrate, which degree of strain can be transferred to an FRP, and how much slip occurs in the adhesive, are the keys to maintaining composite action at all stages up to failure. This is one of the most important aspects of externally strengthened wood beams that will determine the forces in each material and the overall resistance of the section. Adhesive is an important carrier for FRP bonded wood system to form effective bond and transfer shear stress and normal stress in contact interface and lap joint. The performance of the adhesive directly determines the performance of the composite. Barbero et al. [32] concluded that using structural adhesive to bond FRP sheets can ensure the effective transfer of interfacial stress. While Vahedian et al. [33] regarded FRP-to-wood width ratio (the ratio of the width of FRP to the width of wood) as crucial for effective stress transfer. The low FRP-to-wood width ratio leads to a non-uniform stress distribution across the width of wood and interfacial failure at lower load level, resulting in a higher stress in the bond at failure. Furthermore, plated length has also been identified as a desirable characteristic because shear stress transfers within the bond more uniformly as the plated length increases [34].
Raftery et al. [35] concluded the dissimilarities between the composite action are: a) moduli of elasticity, b) surface properties, c) reaction to creep loading, and d) response to moisture and to alternating environmental conditions.

Bond Strength Models
Several models (tabulated in Tab. 1) based on empirical relations, fracture mechanics theories or modified equation for FRP-concrete surface with many parameters calibrated with experimental data [33,[36][37][38][39][40][41] have been proposed for the bond strength between FRP laminates and wood. However, there is still a lack of bond strength models of FRP-to-wood compared with FRP-to-concrete.

Model name Model
Vahedian Model [33,41] Note: τ is the bond stress; τ max is the maximum shear stress along the bond length; P u is the maximum load. b f , E f and t f are width, elastic modulus and thickness of FRP plate or sheet, respectively. b w , E w and t w are width, elastic modulus and thickness of wood block, respectively; f ts is tensile strength of wood block. G a and t a are shear modulus and thickness of adhesive layer, respectively. L b and L e are the bond length and effective length, respectively. A, B and a are constants. a is a parameter relating the prestress at the bottom fiber of the beam, σ p , to the tensile stress in the fiber composite, σ f (σ p = -aσ f ). The parameters c 1 and c 2 are obtained by experimental calibration with a value of 0.7 and 10, respectively. The factor k b represents the anchor zone geometry that has a range of 1-1.29. The surface preparation effect is represented by the factor k c with a range in between 0.67 and 1. The factor K μ represents the strengthening degree which can be considered as 1. γ w and γ a are referred to timber sides and adhesive types, respectively, in which γ w is equal to 0.1 and 0.08 for LVL (Laminated Veneer Lumber) and hardwood, respectively. s is corresponding slip at specific location and C N is referred to elastic stiffness; C w and C f are axial stiffness coefficients of the timber prism and FRP plate, respectively; D w is bending stiffness of the timber substrate; and P is externally applied tensile force. τ 1 , τ 2 and s 1 , s 2 , s 3 are shown in Fig. 3, which includes four stages: elastic stage which corresponds to the first branch of the bond-slip model where slips are less than s 1 ; softening stage where the bond stress decays linearly once this stage begins until the slip s 2 is reached; constant stage where a uniform distribution of the bond stress can be observed and the slips are between s 2 and s 3 which designated as the ultimate slip; and debonded where no interfacial bond stress is transferred between materials once the final slip is reached. According to this model, the bond stress increases with the slips within the interface and reaches a maximum value (τ 1 ) when the slip is s 1 . Afterwards, the interface shows a softening behavior where the bond stress decreases linearly with the interfacial slips until the interfacial slip s 2 . Finally, between the interfacial slips s 2 and s 3 the bond stress remains constant. Beyond the interfacial slip s 3 , the CFRP composite completely debonds from the timber substrate and therefore, the bond-slip model has no bond stresses at all. x is the distance from the free end, x = 0 corresponds to the free end and x = L represents the loaded end.

Bond of FRP Bars Bonded into Wood
Since surface preparation except grooving is no longer required, the amount of installation work could be reduced in NSM or GiR compared with EBR. Moreover, debond failures which are quite common in EBR can be avoided by FRP bars reinforcement as bars are anchored into the substrate. And FRP bars are less exposed to external service environment resulting in the unchanged aesthetic of the strengthened structure and a better long-term performance [42].
Due to high mechanical properties, ease of application, a high stiffness-to-weight ratio (10-to 15-times higher than the steel) and good long-term behavior of FRP bars, the use of FRP has been encouraged as a substitute for steel. However, current research involving FRP reinforcement has concentrated on the FRP bar strengthening concrete, and the restricted data are accessible on the FRP bar reinforcing wood particularly the response of the interface bond between FRP bar and wood block. Meanwhile, design guidelines such as that provided in Eurocodes for steel bars cannot be completely mirrored for this purpose due to essential differences in surfaces deformations and mechanical properties.
The enhancement of FRP bar strengthening wood members has been proven by many researchers. Johnsson et al. [43] found that GiR reinforcement method increased the short-term flexural load-carrying capacity of glulam beams by 49%-63% on average. The reinforced beams demonstrated a moderate enhancement in ultimate moment capacity and stiffness [44,45].
According to Jahreis et al. [46], stress in the interface of FRP bar and wood is not distributed uniformly. Madhoushi et al. [47] reported that shear stress of bar/adhesive interface is much larger than the one of adhesive/wood. Raftery et al. [48] concluded that bond behavior of bonded-in FRP rod reinforcing beams can be improved by reducing the effects of stress concentrations.
Note: P u,v,k is the pull-out capacity; the strength parameters f ws and f wl are given as 520 N/mm 1.5 and 37 N/mm 2 respectively for epoxy; d = min [d r , d h ], d r and d h are the diameter of rod and drilled hole, respectively; ρ k is the characteristic density of the timber members; l b is the bonded length of rod; f v,k representing the characteristic shear strength of the timber around the hole for softwoods for all angles between the rod and the fiber direction; the equivalent diameter of the rod d equ = min [d h , 1.25 d r ]; P u,mean,k is the mean pull-out capacity; f v,mean = 5.7 N/mm 2 ; k 1 and k 2 are factors for joint stiffness; E c A c is the stiffness of the connector; E t A t is the stiffness of the timber; E a A a is the stiffness of the adhesive; The parameter for stiffness of adhesive k = 1.1-1.2.

Bond Test Methods
Different bond testing methods were adopted by various researchers for different purposes of study, as illustrated in Fig. 4. They can be categorized into three types: The Type 1 testing method involves contoured double-cantilever beam (CDCB, see Fig. 4a) bilayer specimens, which are designed by the Rayleigh-Ritz method to conduct Mode I fracture tests of bonded FRP-wood interfaces.
To conduct Mode II fracture test, three different setups of Type 2 testing method are recommended as follows. A shear-block test specimen composed of FRP and a timber block as described in ASTM D-905 is involved in the test (see Fig. 4b(i)). However, this method is only applicable to FRP plate (e.g., Barbero et al. [32]) because compression loading is applied on FRP directly. A modified double-notched test specimen shown in Fig. 4b(ii) is analogous to the one above. It consists of two timber blocks and a piece of FRP plate or sheet, where the block on the left side is fixed while load is applied to the block on the right side. Raftery et al. [35], Crews et al. [53] have adopted this kind of set up to conduct the test. As FRP is sandwiched between two timber blocks, FRP plate surface strains are difficult to monitor, not to mention the shear stress distribution and bond-slip responses. Another set up allows detailed monitoring and inspection of the failure process, due to only one possible path for debonding, as shown in Fig. 4b(iii) adopted by Wan et al. [39,54], Biscaia et al. [40] and Vahedian et al. [33]. This method is also consistent with that used in studying the bond between FRP and concrete or steel. However, it is a challenge to be sure that the alignment is maintained to minimize load eccentricity. This method may not be applicable for FRP sheets with the difficulty in gripping the sheets.
The Type 3 test method involves a piece of FRP sheet/plate attached to the tensile flange of a timber beam. The loading is applied on the beam to create a pure bending zone. This type of testing closely replicates the adhesive shear and peel stresses that are induced by flexural loads. This method consists of two cases, one in which the FRP is stuck only at the middle section of the beam (see Fig. 4c(i), Vahedian et al. [34]) and the other extends to the both ends of beam (see Fig. 4c(ii)). It is worth noting that the second case seems to be the only method when prestressed FRP laminates as external reinforcement of wood beams is encountered [36,47].
Based on the above discussions, it is recommended that the test set up illustrated in Fig. 4b(ii) be used for FRP sheets, and that in Fig. 4b(iii) be used for FRP plates to conduct Mode II fracture test in establishing the bond-slip relationship between FRP and wood in tension as surface strain can be easily monitored. The failure modes can be separated into two categories based on the duration of composite action between the materials. Failure will occur in wood or FRP, i.e., (a) or (e), (f), when composite action is maintained until the ultimate load is reached. However, when composite action is not maintained until the ultimate load is reached, premature failure results from debonding of the FRP laminates, termed  interfacial debonding ((b), (c) and (d)). Interfacial debonding is the most common mode of failure for FRP bonded wood system. It should be noted that for multi-layer FRP sheets, it is not practical to single out failure modes (c) and (d). A schematic view of failure modes is given in Fig. 5, while some examples are given in Figs. 6-8.

Failure Modes
Observations of fracture surface of the Type 1 test specimen in Fig. 6a shows that interfacial adhesive failure was the most common failure mode. While some failure occurred within the continuous strand mat layer of the FRP accompanied by interfacial adhesive failure. In addition, substantial fiber bridging was eminent during the fracture process in several specimens. An example is shown in Fig. 6b and a schematic view of fiber bridging is shown in Fig. 6c [55,59].
Compared to failure modes in Mode I fracture tests, those in Mode II show greater diversity [33,39,53,54]. Among all five modes shown in Fig. 7, timber-adhesive interface in timber failure is  Failure initiating at the interface of the FRP and timber substrate of the end of FRP laminates is found to occur frequently in wood beams strengthened using FRP. Such failure is characterized by the formation of an oblique crack from the soffit of the beam to the level of natural texture of wood. Cracking proceeds along the level of the tensile flange until the FRP laminate is completely separated from the wood beam. This failure mode is referred to as ripping or end peel failure (see Fig. 8) which is found to occur frequently in beams where the FRP laminate is terminated far from the supports.

Bond Test Methods
There are several test configurations seen in the literature that can be used to evaluate pull-out capacity of a bar glued into timber. Pull-pull, pull-push, pull-pile foundation, pull-beam and pull-bending have been conducted mostly. While they can also be classified as: single-shear test, double-shear test and beam test. The schematics of most common setups are shown in Fig. 9. It needs to be pointed out that this figure shows the GiR specimens, which can be replaced with NSM specimens. Beam pull-out tests (BPT, shown   [57]; (c) FRP composite sheet rupture [60] in Fig. 9b) are more likely to represent the actual conditions than the direct pull-out tests (DPT, shown in Fig. 9a), but significant limitations are also present for this setup as it cannot be applied to other sections directly [61]. It is worth noting that numerical and experimental investigations have demonstrated that different test setups can produce different results, and small variations in setup may have significant effects [62].

Failure Modes
Tab. 3 is a summary table for failure modes of FRP bars bonded into wood in the literature [47,[64][65][66][68][69][70], which includes both NSM and GiR specimens and both test methods: DPT and BPT. Among all the modes, wood shear and adhesive failure have been observed in both cases. Whilst, both NSM and GiR have their exclusive failure modes: FRP bar tensile failure for NSM specimen and wood splitting for GiR specimen.
Lee et al.'s [65] concuded that NSM specimens showed the phenomenon of tensile fracture of epoxy which was different from debonding failure of EBR specimens in DPT, and the failure pattern was same each other regardless of the bonded length, width and depth. Corradi et al. [66] reported three failure modes of NSM specimens shown in Fig. 10a in DPT, which were: bar pull-out (i.e., CFRP bar/adhesive failure), timber shear failure and CFRP bar tensile failure (appeared in NSM specimens only). More bond failures were observed in Sena-Cruz et al.'s [69] beam pull-out tests because not only shear but also bending force were applied to FRP-to-wood system.
For GiR specimens, adhesive and wood shear failure are prevalent as much as NSM specimens. However, wood splitting is only observed in GiR specimen as one end of the FRP bar is completely embedded in the wood. Fig. 11 shows the failure modes of GiR specimens according to O'Neill et al. [68]. O'Neill et al. [70] also carried out tests to assess the bond strength of FRP-wood in GiR beam. They found that 64% of specimens were failing in wood shear and concluded it as wood is the weakest element in the system. Splitting of wood also appeared and the length of the splitting is equal to the bond length of FRP bar.     [71] research, a higher ultimate load was recorded for FRP-timber joints made from hardwood when compared with the joints made from LVL (usually using softwood species), with an increase of 8%. Higher tensile strength of the hardwood species leads to the improvement of the bond strength. And they also found that specimens made from LVL exhibited a degree of ductile behavior, failing gradually; while joints made from hardwood exhibited brittle behavior, failing suddenly. According to Wan et al. [39], all softwood joints failed predominantly in the timber, whereas the hardwood joints exhibited failure at different interfacial positions.
Wan et al. [54] experimentally evaluated the influence of different bond surfaces of wood on the interfacial strength. Sides A and B bonded FRP specimens illustrated in Fig. 12 were designed for tests, where the annual growth rings of wood were predominantly oriented parallel to the FRP plate of Side A specimen and perpendicular to the FRP plate of Side B specimen. The weakness of FRP-wood interfacial strength for Side B specimen has been demonstrated by the debonded surface of the FRP plate in which a large portion was covered in thick wood of up to 5 mm thickness (see Fig. 7a), while the wood attached to the plate of Side A specimen was considerably less. Furthermore, the effect of pith was only eminent in the Side A specimen. Reasons may be due that the radius of the growth rings in the immediate vicinity of the pith is smallest leading to more perpendicular intersections of the FRP plate and growth rings, and the denser older wood next to the pith contributes to higher interfacial strength which was particularly evident in the large range of the results for the 120 mm and 180 mm bond length specimens.
The effect of grain orientation of the substrate has also been assessed by Subhani et al. [72]. Three groups of specimens were used to test the bond between CFRP and LVL by applying CFRP composite parallel (Group 1 and 2) or perpendicular (Group 3) to the grain on laminate face (Group 1) or the grain face (Groups 2 and 3) of LVL. The schematic view of Groups 1-3 is posted in Fig. 13. The result shows that the maximum shear strength of Group 1 and 2 are quite similar expect that Group 1 was more ductile than Group 2. However, Group 3 showed a poor bond performance compared to Groups 1 and 2 due to the weak material properties of timber perpendicular to the grain. Therefore, it can be concluded that surface characteristics of wood prism been used need to be known for determining the bond strength when FRP is bonded to wood.

Moisture Content
The moisture content of wood affects the physical and mechanical properties of wood as well as the material itself, which also affects the interface bonding performance of FRP-to-wood. Water adsorbed by  [54] research wood exists between the microfibril in the cell wall and acts as a lubricant, allowing certain slippage or relative displacement between the microfibril. When the water is lost, the microfibril draw closer to each other and attract to each other, making a strong frictional resistance to the sliding displacement. Therefore, when the moisture content is lower than the fiber saturation point, the wood strength decreases with the increase of moisture content. And the strength reaches the minimum value as soon as the moisture content arrives at the fiber saturation point. As moisture content is higher than the fiber saturation point, the free water content increases while the strength remains stable. The research results in the literature show that the interfacial bonding property of the dry case for FRP sheets and wood substrate is better than that of the wet case under the same condition. Barbero et al. [32] concluded that using structural adhesive to bond FRP sheets could ensure the effective transfer of interfacial shear force, but the moisture content had a substantial adverse effect on the interfacial bonding strength. The interface shear stress of the wet FRP-wood specimen was only 43% of that of the dry one, and the shear strength of the wet specimen was about 53% of that of the dry specimen. The results also showed that due to the influence of moisture content, the strain caused by the mismatch between FRP and wood layers could be predicted by the finite element model. The swelling coefficient of wood was also determined, and the relationship between moisture content and strain of wood in the tangential and radial direction was established, which is expressed as follows: where, MC is the moisture content.
Another research has been conducted by Zhou et al. [57] through performing experimental tests and molecular dynamics simulations to investigate the effect of moisture on the fracture behavior and the mechanical properties of FRP-to-wood composite. The interface fracture in moisture conditioned samples implied the weakening of the epoxy-wood interface. By performing molecular dynamics simulations, the adhesion energy of cellulose and epoxy was measured under dry and wet conditions. It is observed that water molecules diffused within the bilayer connections and led to significant decrease in adhesion energy. The adhesion energy in wet case dropped to one third of that in dry case. The findings revealed  [72] research that the bond strength between epoxy and wood decreased with moisture conditioning and the mechanical performance of FRP-reinforced wood would be significantly degraded.

FRP Gluing 4.3.1 FRP Types
Nadir et al. [60] carried out a block shear test of GFRP-and CFRP-wood specimens and reported a shear strength of 5.61 MPa and 5.52 MPa, respectively. They also experimentally tested the performance of strengthened laminated beam. For strengthened laminated wood specimens with single layer and two layers of GFRP or single layer of CFRP composite sheets, all specimens showed flexural failure occurring in the outermost wood lamination on the tension side of the beam prior to the rupture of FRP composite sheet. While in one specimen strengthened with two layered CFRP composite sheets, FRP composite sheet from the adjacent wood lamination debonded with traces of wood right after the crack occurring in the outermost wood lamination on the tensile flange of the beam. Sliding failure occurred between timber laminae accompanied by the delamination in CFRP sheet at some places in other specimens. It seems like FRP has a negligible effect on the bond strength or the failure mode of the joint on the basis of Nadir et al.'s research [60], whereas more studies are needed to confirmed it.
According to the SEM (scanning electron microscope) images of fracture surface shown in Fig. 14, the aramid fibers are not as well immersed in epoxy matrix as basalt and carbon fibers are [73]. Future researches are required to investigate whether or how fibers immersion would impact the bond performance of FRP-towood joint.

FRP Stiffness
Biscaia et al. [40] concluded that the performance of FRP-to-wood interface is affected by the stiffness of the FRP composite. With the increase of the CFRP stiffness, the maximum load increased, and the plateaux observed in the load-slip curves decreased which is because the effective bond region was extended to satisfy the demand of the interface strength. However, if there is not enough bond region to meet the demand, the debonding happens with the increase of FRP stiffness.

Plate Thickness
In addition to the mechanical properties of FRP materials, the thickness, width and length of FRP plate will also affect the bond of the joint. Wan et al. [39] found that joints strengthened with pultruded plates (i.e., CFRP plates) experienced higher strengths than joints strengthened with wet layup plates (i.e., carbon sheets). One reason can be highlighted that pultruded plates are much thicker, although plate efficiencies were generally higher for sheets. Vahedian et al. [33] revealed that the brittle failure was more evident in the joints strengthened with two layers of FRP composite compared to single layer, which can be attributed to the ineffective gluing of FRP.

Plate Width
Plate width has been proven to have significantly impacts on the bond strength by results from the tests with different bond widths, namely, 35 mm, 45 mm, and 55 mm [33]. As FRP-to-wood width ratio increase, the interfacial bond strength of the joint increased and maximum shear stress decreased [41]. This finding is consistent with the FRP-to-concrete one in the literature [74][75][76]. Furthermore, the local slip at the same level of applied load decreased with width ratio increasing. One reason may be due that when the width ratio is low, the force transferred from FRP to wood results in a non-uniform of stress distribution across the width of wood, thus causing premature bond failure. And a low width ratio may lead to a high level of stress at failure, direct stress from the bonded zone of the substrate to the unbonded zone. These findings are in agreements with the previous researches conducted by Xu et al. [76] and Hollaway [77].

Plate Length
It was also observed by Vahedian et al. [34] that with the increase of the plated length, the ultimate bending strength increases, and conversely mid-span defection of wood beam decreases, signifying that the reinforcement leads to higher stiffness values. A noticeable decrease in shear stress at failure was obtained, too. This enhancement provides improved behavior at failure leading a more ductile collapse. That is because, shear stress transfers within the bond more uniformly and the strengthened wood beam will not collapse completely since FRP prevents crack opening and restricts local rupture.
Nevertheless, many experimental studies [31,78,79] and fracture mechanics analyses [80,81] have confirmed that extending the bond length beyond a certain length will not contribute to a better mechanical behavior of the joints where there is no increase in the bond strength. The certain length is called effective bond length.

Conclusion
In conclusion, FRP affects the performance of the joint superficially since the strength or stiffness of FRP materials can easily satisfy the demand of the interface performance usually, it is still the bonding that plays the decisive role. The gluing region should be large enough to avoid ineffective gluing, otherwise the reinforcing effect can be quite limited.

Prestressing
In many cases, the load-carrying capacity of the FRP is not reached as failure occurs in the wood component when FRP is in a slack state. Prestressing the FRP plate is a solution to this situation. By the means of applying a tensile force to the FRP plate prior to bonding, the prestressing force can induce compressive stresses in the flange of wood beam to offset against the tensile stresses aroused by the loads.
The effect of prestressing the FRP laminates on beam behavior was investigated by Triantafillou et al. [36], Brunner et al. [58]. Prestressing the FRP significantly improved the strength and stiffness compared to the non-stressed sheet.
Although prestressing is tested to be a good strategy to improve the properties of the composite, the premature failure of the prestressed beam caused by delamination of the laminate should be noticed.

Surface Preparation
The wood surface and FRP surface should be pretreated to roughen the surface and achieve a better combination of the composite before the FRP is pasted on the substrate (e.g., Barbero et al. [32], the FRP composite and the wood were hand sanded and wiped or air-cleaned to make the surface free from dirt and other impurities prior to bonding). Surface preparation is one of the most important processes in the whole process of reinforcement. Because the bonding is mainly based on the adhesion of the adhesive to the wood surface and the FRP surface, the surface preparation of the material may be the main factor to determine the bonding strength and durability of the interface.
There are two different factors of surface preparation, one is the use of coupling agent or not, the other is the roughness of the surface.
Two completely different coupling agents, HMR (methyl benzodiazepines) and RF (resorcinol formaldehyde) were used for wood surface preparation by Davalos et al. [82]. The results showed that the application of HMR can effectively reduce the delamination rate of FRP-wood composite, and significantly improve the bond strength in wet-dry cycling environment as well. Superior performance of HMR has also been proven by Vick et al. [83,84] carrying out an experimental study on the bond performance between three types of adhesive and five wood species.
A comparative study on surface roughness has been conducted by Lyons et al. [85]. The wood surface was polished with sandpaper of 100-grit. It was found that the roughness and smoothness of the surface had little influence on the bond strength between FRP and wood.

Adhesive
Adhesive is an important carrier to form effective bond and transmit shear and normal stress between the contact interface of FRP-to-wood system. The performance of the adhesive directly determines the performance of the composite. Factors such as the service environment of the joint, the rigidity or flexibility of the materials, and surface conditions of the members should be taken into consideration when selecting adhesive.
The epoxy adhesive has been confirmed to be more applicable than traditional formaldehyde-based adhesive for improving the bonding performance of FRP-wood interface [32,[86][87][88][89]. Raftery et al. [56] pointed out that the epoxy adhesive studied in the experimental program was considered the most suitable adhesive for the FRP-wood interface due to the excellent quality bonds under ambient conditions. Gardner et al. [90] carried out a test to evaluate the performance of three adhesives, namely RF, epoxy resin and emulsion polymer isocyanate and concluded that all three adhesives behaved well in a dry environment performance, but only the RF still had a good performance in the wet or wet-dry cycling environment.
How to apply adhesive is the crux of the bond. Although it seems to be impossible to obtain a uniform adhesive thickness as recommended by the manufacturer due to the unevenness of the composite surfaces, complete coverage on the surfaces can be ensured by forcing excess adhesive out at the sides of the composite.
Considering linear elastic fracture mechanics (LEFM), Custódio et al. [91] reported that the interfacial brittle fracture energy G f can be determined for a given adhesive layer thickness t a based on Eq. (3): where τ v and G a are the adhesive shear resistance and the adhesive shear modulus, respectively.

Temperature
Zhou et al. [73] concluded that the exposure temperature changed the failure modes of FRP-wood systems as the microstructure of both wood and FRP has been affected by temperature (Figs. 15a and  15b), and the wood deteriorated more rapidly than the interface with the increasing of temperature. The failure mode of the BFRP joint shifted from a mix of adhesive/cohesive failure at interface between FRP and wood at ambient temperature to cohesive failure at wood at elevated temperature. Meanwhile, the interfacial fracture energy of the joint was reduced by elevated temperatures. The equations proposed by Vahedian et al. [41] (in Tab. 1) were modified via relating the effect of temperature to the coefficient α in Zhou et al.'s [73] research, and α:

Preservative Treatment
Some oil preservatives such as coal tar and creosote are almost inert to wood. And it does not affect the wood strength because chemical reaction will not occur in the wood after the injection. On the other hand, water-borne preservatives can enhance the compressive strength and hardness and weaken impact strength slightly with prescriptive concentration. Although the wood preservative itself has no significant effect on the wood strength when the preservative is injected into the wood, the wood strength may be significantly reduced if the temperature, pressure and other variables are not appropriate. In particular, when the pressure infusion method is adopted, the wood strength will be substantially weakened if the high temperature and high-pressure treatment is maintained for a long time.
Studies have shown that preservative treatment has complex effects on the longitudinal elastic modulus, longitudinal tensile properties and interlaminar shear properties of materials. Tascioglu et al. [92] studied the adverse effect of preservative on interface bonding performance by accelerated cyclic exposure test. The results showed that the interfacial bonding property of preservative treatment prior to reinforcement is obviously inferior to that of preservative treatment after reinforcement. Tascioglu et al. [92] also found that brown rot fungus and white rot fungus commonly existing in wood could grow in CFRP sheet as well. The deterioration of the interfacial bonding performance due to wood-rot fungi can be detected by non-destructive technologies such as interlayer shear test and ultrasonic and scanning electron microscope.

Embedded Length
For NSM specimens, the failure load increased as embedded length increased, so well as loaded end slip. While bond stress showed a decreasing trend with increasing embedded length [66,69]. The same results were obtained from GiR specimens by O'Neill et al. [70]. However, Yeboah et al. [61] revealed that interfacial stress is in relation to the direction of the wood fibers with respect to the longitudinal axis of the joint. As the bonded length increased, GiR specimens loaded perpendicular to the grain increased in interfacial stress, while the case loaded parallel to the grain went the other way.

Groove Depth
Groove depth (the difference between the embedded length and groove depth is illustrated in Fig. 16) is another factor that influences the bond strength of NSM specimens. It has a benefit for bond strength with groove deeper, particularly when the bond length is large [69].

FRP Bar Surface
Both NSM and GiR specimens can achieve a better bond performance with a rougher surface of FRP bar because adhesive can penetrate the surface more easily [64,69]. Two different bars used in Sena-Cruz et al.'s [69] research can be seen in Fig. 17.

Adhesive and Glue-Line Thickness
The bonded strength of the GiR specimen used with poly-urethane adhesive was 2.9-4.0 times greater than the one used with resorcinol adhesive in the tests conducted by Lee et al. [93]. They also pointed out that the bond performance in the case of the glue-line thickness of 2 mm improved by 17%-29% in comparison to the case when the glue-line thickness was 1 mm. Harvey et al. [64] argued that failure load of GiR specimen showed an upward trend with increasing glue-line thickness, which seemed to have no effect on interfacial shear stress though.

Wood Types
Lee et al. [65] and Madhoushi et al. [47] confirmed that different wood types (e.g., laminated veneer lumber, glulam and pine) can result in different bond performance of NSM or GiR specimens.

Direction of the Wood Grain with Respect to the Longitudinal Axis of the Joint
As discussed above, the direction of the wood grain with respect to the longitudinal axis of the joint has impact on the interfacial stress. Moreover, the stress-slip behavior of GiR specimen loaded perpendicular to the grain exhibited more ductile than the corresponding one parallel to the grain [60]. De Lorenzis et al. [94] pointed out that the bond strength is higher for the joints with rods perpendicular to the grain than for rods parallel to the grain, and splitting bond failure is more critical for rods parallel to the grain. The splitting bond stress of specimen loaded parallel to the grain is: (5) and the splitting bond stress of specimen loaded perpendicular to the grain is: where r = c/d b , c is the minimum radius of the calculation model when specimen is loaded parallel to the grain, d b is the nominal diameter of the bar, f u,t is the tensile strength of wood in the transverse direction.
6 Other Parameters Influencing Bond Strength of FRP to Wood Though many parameters influencing bond strength of FRP to wood have been discussed, there is still a lack of researches on other parameters. Therefore, further studies are required for this area. Fig. 18 presents a list of parameters affecting the mechanical behavior of FRP-to-wood joint which is modified from Serrano et al.'s [95] paper.

Summary
The behavior of FRP-to-wood systems has yet to be thoroughly researched compared with their FRP-toconcrete or FRP-to-steel counterparts. As FRP rehabilitation and strengthening of timber structures has a promising future, better understanding of their failure modes will enable more precise designs balancing safety and cost. One of the most common failure modes of wood strengthened by FRP composite is debonding of the FRP from the substrate.
Composite action between the bonded FRP and wood is very much related to the bond-slip behavior between the two materials. The currently available models for estimating bond strength of the bonded FRP to wood are based on empirical relations or fracture mechanics theories with many parameters A series of test setups for evaluating bond strength of FRP to wood have been involved in the literature. It is recommended that the test setup illustrated in Fig. 4b(ii) be used for FRP sheets, the setup in Fig. 4b(iii) for FRP plates to conduct Mode II fracture test and BPT in Fig. 9b for FRP bars to conduct NSM and GiR test in establishing the bond-slip relationship between FRP and wood.
The failure modes of wood strengthened with FRP are shown above. Although bond has a substantial impact on the performance of FRP-to-wood joints, many studies have confirmed that wood is still a weak component part in the joint. Increasing the strength of the wood material is crucial to improve the integral performance of the joint.
Parameters influencing the bond strength of FRP to wood has been discussed, while other parameters which have not been researched in literature are also suggested for future projects.

Conflicts of Interests:
The authors declare that they have no conflicts of interest to report regarding the present study.