Behavior and design of GFRP bar adhesive anchors under direct tension for deteriorated concrete bridge barrier replacement

Abstract

The application of post-installed GFRP bars is notably illustrated in replacing deteriorated bridge barriers. Fourteen steel-reinforced concrete slabs were fabricated, into which 120 GFRP bars were anchored using epoxy adhesive. These anchored GFRP bars were exposed to outdoor conditions throughout one winter to assess the impact of freeze-thaw cycles before undergoing testing to evaluate their ultimate pullout load. The experimental program included two types of adhesives, various diameters of bars, two categories of GFRP bars (sand-coated and ribbed-surface), and three nominal embedment depths for each bar diameter. For each bar combination, five identical samples were constructed. The predominant observed failure modes were pullout failures, which contributed to the development of an analytical model. A regression analysis was conducted to investigate the influence of embedment depth and bar diameter on pullout capacity, resulting in a best-fit formula that may be used in design practices.

Keywords

pullout capacity adhesive post-installed anchors glass fiber fiber-reinforced polymer bars regression analysis

Introduction

Corrosion of steel reinforcement due to environmental effects is a major cause of deterioration problems in concrete structures, as depicted in Figure 1. Currently, fiber-reinforced polymer (FRP) bars have been considered an adequate construction material to solve this problem owing to their outstanding characteristics such as lower self-weight, higher tensile strength, and better chemical resistance (Lu et al., 2023; Sun et al., 2014; Zhu et al., 2023). Consequently, replacing the reinforcing steel bars in concrete with FRP bars has become an attractive means to solve the problem of reinforcement corrosion (Hao et al., 2024). Thus, researchers have studied the environmental durability, damage evolution mechanism, and strength degradation of GFRP reinforcement (Azimi et al., 2014a, 2014b; Zhang et al., 2022). The findings indicated that following a degradation cycle of 183 days, the elastic modulus of GFRP bars remained largely unchanged, exhibiting a variation within ±8%. Conversely, microscopic analyses revealed that the alkaline conditions of seawater induced a hydrolysis reaction in the resin matrix, which compromised the bond between the resin and the fibers, ultimately resulting in a reduction of the tensile strength of the GFRP bars (Sun et al., 2014).

Figure 1.

Observed corrosion of steel bars and spalling of concrete in bridge barriers.

Recent improvements in the field of Fiber Reinforced Polymer (FRP) bar-reinforced structural members have significantly impacted the construction industry by enhancing the performance and durability of concrete structures. Thus, researchers studied the combination of the FRP bars with different concrete types and matrices such as High-Strength Concrete (HSC), Ultra High-performance Fiber-Reinforced Concrete (UHPFRC), Ultra High-Strength Engineered Cementitious Composites (UHS-ECC) and Geopolymer concrete (Gyawali et al., 2024; Khalafalla and Sennah, 2021; Lu et al., 2022; Peng et al., 2023; Sayed-Ahmed et al., 2024a, 2024b; Zhu et al., 2024a, 2024b). This highlights key concepts and original insights into the latest developments in FRP reinforcement technology, focusing on several noteworthy advancements that have the potential to revolutionize current construction practices.

Many applications require suitable anchoring of FRP bars to the substrate concrete, particularly in rehabilitation projects. One specific application is to replace a damaged concrete barrier with a new one. The TL-5 barrier should be properly anchored to the existing deck overhang where TL-5 stands for Test Level 5 for heavy traffic highways. An economical approach involves the utilization of post-installed adhesive anchors, which entails drilling holes into the existing concrete and subsequently inserting Glass Fiber fiber-reinforced polymer (GFRP) bars into these holes. An adhesive is a bonding agent, facilitating the connection between the bars and the concrete, as depicted in Figure 2. Recent research conducted on GFRP-reinforced barriers (Dervishhasani and Sennah, 2018; Khederzadeh and Sennah, 2014; Rostami et al., 2017, 2019a, 2019b; Sennah et al., 2018; Sennah and Hedjazi, 2018; Sennah and Khederzadeh, 2014; Sennah and Mostafa, 2018) led to a cost-effective design of TL-5 bridge barrier using either sand-coated or ribbed-surface GFRP bars with headed end or end with 180° hook for new construction. Results showed that a barrier wall connected to non-deformable concrete slab using post-installed straight-ended GFRP bars showed comparable behavior with monolithic barrier (Rostami et al., 2017, 2019a, 2019b).

Figure 2.

Schematic diagram of GFRP-reinforced barrier with post-installed anchors.

A comprehensive study on metallic adhesive anchors has been carried out by various researchers (Cook et al., 1998; Eligehausen et al., 2006a, 2006b), and detailed design procedures are available in the relevant design standards (CSA, 2014; ACI 318-19, 2019). Recently, Charney et al. (2013) provided background into adhesive anchor design and development length provisions in ACI 318, as well as the provisions of post-installed reinforcing steel bars in international standards, followed by recommendations for the development of a new procedure that applies to the design of post-installed reinforcing bars.

Also, Piccinin et al. (2013) studied the breakout capacity of headed steel anchors embedded in prestressed concrete. The findings led to recommendations aimed at improving existing guidelines to account for the breakout capacity of headed anchors installed in confined and unconfined concrete.

The research on FRP adhesive anchors can be divided into two categories: FRP sheet anchors and FRP bar anchors. FRP sheet anchors provide extra anchorage capacity for FRP sheets attached to the concrete surface. Several anchor dowels are formed at the end of the sheets and then inserted in holes filled with adhesive (Kim and Smith, 2010; Ozbakkaloglu and Saatcioglu, 2009). The FRP bar adhesive anchors are installed similarly to reinforcing steel adhesive anchors. Figure 3 shows five possible failure modes generally accepted for metallic adhesive anchors. Similar failure modes are expected for GFRP bars (Ahmed et al., 2008).

Figure 3.

Possible failure modes of adhesive anchors after Cook et al. (1998): (a) concrete cone breakout; (b) adhesive/concrete interface pullout; (c) anchor/adhesive interface pullout; (d) combined adhesive/concrete and anchor/adhesive interface anchor/adhesive interface; and (e) anchor rupture.

Research on fiber-reinforced polymer (FRP) anchors is generally less extensive than on metallic anchors. A few researchers have conducted experimental investigations and analytical modeling specifically on FRP sheet anchors (Kim and Smith, 2009, 2010; Ozbakkaloglu and Saatcioglu, 2009; Özdemir, 2005). In comparison, there exists a limited number of experimental studies concerning FRP bar anchors (Ahmed et al., 2008). To the best of the authors’ knowledge, no design model has been suggested for these anchors.

In regions with cold climates, concrete structures are subjected to repetitive freeze-thaw cycles which can lead to deterioration of the concrete. When replacing deteriorated concrete bridge barriers with GFRP bar adhesive anchors, understanding the impact of freeze-thaw cycles on the performance of these anchors is essential. During freeze-thaw cycles, water penetrates the concrete and freezes, causing an increase in volume and generating internal pressure. This cyclic loading can induce cracking and spalling in the concrete, compromising the bond between the GFRP bar adhesive anchors and the concrete substrate. Therefore, it is crucial to consider the effects of freeze-thaw cycles on the durability and load-carrying capacity of the adhesive anchors. In designing GFRP bar adhesive anchors for deteriorated concrete bridge barriers, special attention should be paid to selecting appropriate adhesive products that can withstand the effects of freeze-thaw cycles and provide sufficient bond strength between the GFRP bars and the concrete substrate. This will help in understanding the behavior of the anchors, identifying potential failure modes, and determining the necessary design modifications to enhance their performance in deteriorated concrete structures subjected to freeze-thaw cycles. Thus, this study aims to investigate the performance of post-installed adhesive GFRP bars exposed to outdoor conditions of one winter in Canada and to develop an analytical model for design purposes.

Research significance

Post-installed adhesive anchors are extensively used in the rehabilitation of structures. The application of GFRP bars has increased mainly due to their corrosion-resistant nature. There is a lack of comprehensive information regarding the design of adhesive anchors for GFRP bars compared to those for reinforcing steel. To the authors’ knowledge, no design model currently exists. The results of an experimental investigation of GFRP adhesive anchors are presented to investigate the behavior of such anchors and develop a simplified design model.

A review of analytical models for metallic anchors

Few analytical models are proposed in the literature for failure load of bonded steel anchors with large spacing and edge distance considering different failure modes. Cook et al. (1998) and Eligehausen et al. (2006a) did a comprehensive literature review. A summary of these prediction models is given in this section, considering the main modes of failure.

Concrete cone models

Concrete cone failure occurs for anchors with small embedment depths. The general form of the widely used equation for concrete cone models to calculate the failure load N_c is as follows:

N_{c} = k h_{e f}^{n} \sqrt{f_{c}^{'}}

(1)

where k is a constant,

h_{e f}

is effective embedment depth and n is taken as 1.5 for the majority of cases, including cast-in-place headed anchors, post-installed undercut or expansion anchors (CSA, 2014; Eligehausen, 2006b) and post-installed adhesive anchors (ACI 355.4M-19, 2019), while it was taken as two for post-installed adhesive anchors (Eligehausen et al., 1984),

f_{c}^{'}

is concrete compressive strength.

Eligehausen et al. (2006a) presented the following equation:

N_{c} = 0.92 h_{e f}^{2} \sqrt{f_{c}^{'}}

(2)

Further investigations by Fuchs et al. (1995) led to a design model developed for cast-in-place headed steel anchors as follows:

N_{c} = 16.5 h_{e f}^{1.5} \sqrt{f_{c}^{'}}

(3)

The basic concrete breakout strength of a single anchor in tension in cracked concrete, N_cb, is given by ACI 318-19 (2019) as follows:

N_{b c} = k_{c} λ_{a} \sqrt{f_{c}^{'}} {h_{e f}}^{1.5}

(4)

where

k_{c}

is taken as 10 and seven for cast-in and post-installed anchors, respectively. Higher values of

k_{c}

may be used for post-installed anchors but shall in no case exceed 10 and

λ_{a}

is modification factor to reflect the reduced mechanical properties of lightweight concrete in certain concrete anchorage applications.

Bond models

Bond models can be generally divided into two groups: (i) elastic bond stress and (ii) uniform bond stress. The elastic bond stress model includes the elastic stress distribution along the depth of the anchor (Cook, 1993) as follows:

N_{b} = π τ_{\max} d_{0} [\frac{\tanh (λ h_{e f})}{λ}]

(5)

where

λ

is determined from the experiment representing anchor flexibility instead of anchor stiffness,

τ_{\max}

is the maximum shear stress in the adhesive layer, h_ef is the effective embedment depth of the anchor and

d_{0}

is the hole diameter.

And the uniform bond model is obtained using a constant stress distribution (McVay et al., 1996):

N_{b} = π τ d_{0} h_{e f}

(6)

Where

τ

is the uniform bond strength. It was shown by McVay et al. (1996), using non-linear finite element analysis, that the elastic bond model is suitable for low loads where the anchor has not yet experienced significant concrete cracking or other sources of nonlinearity. On the other hand, they showed that the uniform stress bond model is more appropriate for the failure of the anchor and can be better used in the strength design method. They also concluded that the uniform bond stress model is appropriate for shallow and deep embedment depths.

Experimental results on the chemically bonded steel anchors revealed that $τ$ in equation (6) can have a wide range of 6.3 to 14.6 MPa for various adhesive types obtained from experimental results of 280 tests (Cook, 1993). A simplified bond model was later proposed by Cook et al. (1998) to be used for steel adhesive anchors as follows:

N_{b} = π τ^{'} d_{b} h_{e f} Ψ_{c}

(7)

where

τ^{'}

is the characteristic uniform bond strength of anchors embedded in concrete with the compressive cylinder strength of 20 MPa, d_b is the bar diameter, and

Ψ_{c}

is a modification factor to account for the effect of compressive strength other than 20 MPa.

τ^{'}

and

Ψ_{c}

are given by:

τ^{'} = τ_{{f^{'}}_{c} = 20 M P a} (1 - k C O V); Ψ_{c} = \sqrt[n]{\frac{f_{c}^{'}}{20}}

(8)

where

τ_{{f^{'}}_{c} = 20 M P a}

is the average experimental bond strength of anchors embedded in concrete with

f_{c}^{'}

= 20 MPa, k is a constant used for the calculation of characteristic value, COV is the coefficient of variation of the test results, and n is a product-dependent constant obtained from an individual adhesive product tested in various concrete strengths. The model given in equation (7) is only valid for the following range of parameters:

4.5 \leq h_{e f} / d_{b} \leq 25.0

d_{b}

\leq

50 mm, bond area

π d_{b} h_{e f} \leq

55,000 mm², and

f_{c}^{'}

= 13.0 to 68.0 MPa (Cook et al., 1998).

The uniform bond stress model given in equation (7) was used by Cook and Kunz (2001) for an extensive experimental program. They tested 20 different adhesive products from 12 manufacturers considering various parameters affecting anchorage performance before and after installation. They performed 756 tests and concluded that product-specific testing should be used for a reliable prediction of the ultimate capacity of adhesive anchors. They reported 15.4 MPa as bond strength ( $τ$ ) obtained during experiments and 12.0 MPa as characteristic bond strength ( $τ^{'}$ ) assumed for design.

ACI 318-19 (2019) recommends the basic strength of a single metallic adhesive anchor as:

N_{b} = π λ_{a} d_{a} h_{e f} τ

(9)

where

λ_{a}

is a modification factor for lightweight concrete equal to 1.0 since normal-weight concrete was used, d_a is outside diameter of anchor and

τ^{'}

is the cracked (

τ_{c r}^{'}

) or uncracked (

τ_{u n c r}^{'}

) characteristic bond strength depending on the application. ACI 318-19 (2019) recommends

τ_{c r}^{'}

and

τ_{u n c r}^{'}

be taken as 1.4 and 4.5 MPa for outdoor applications, and 2.1 and 7.0 MPa for indoor applications, respectively.

Combined cone/bond models

These models assume a shallow conical failure at the surface of the concrete and a failure at the interface of the concrete and the anchor. The depth of conical failure, $h_{c}$ , is important in these models and could be obtained by Cook (1993):

h_{c} = \frac{π τ d_{0}}{1.84 \sqrt{f_{c}^{'}}}

(10)

If $h_{e f}$ is less than $h_{c}$ , equation (10) can be used to calculate failure load due to conical failure only, otherwise, the summation of conical failure and bond failure can be used as follows (Cook, 1993):

N_{c b} = 0.92 h_{c}^{2} \sqrt{f_{c}^{'}} + τ π d_{0} (h_{e f} - h_{c})

(11)

The bond model used in the above formula is based on the uniform bond stress. Uniform bond strength for various adhesive types was investigated and reported by Cook (1993), ranging from 6.3 to 14.6 MPa. Using the elastic bond stress distribution combined with uniform bond stress, the failure load given in equation (11) will be modified. However, the resulting model is relatively complicated for design practices and was not recommended by Cook et al. (1998).

A review of analytical models for post-installed FRP adhesive anchors

Despite the extensive research on the cast-in-place FRP reinforcing bars used to develop bond-slip properties and development lengths, there is limited research on the post-installed adhesive FRP bar anchors and FRP sheet anchors. Comparable to steel anchors, the pullout capacity of cast-in-place FRP bars is believed to be different from that of adhesive anchors due to the presence of the adhesive material. The failure modes of post-installed adhesive FRP bars can be considered similar to those for steel anchors (Ahmed et al., 2008) which is shown in Figure 3. Limited experimental studies exist regarding the tensile capacity of post-installed adhesive FRP bars (Ahmed et al., 2008), and to the authors’ knowledge, no analytical model has been introduced yet.

The tensile capacity of adhesive FRP sheet anchors was studied experimentally considering the following parameters: anchor diameter, embedment depth, fiber content, and concrete compressive strength (Kim and Smith, 2009; Ozbakkaloglu and Saatcioglu, 2009; Özdemir, 2005). The observed failure modes by Özdemir (2005) included: FRP rupture, concrete cone failure, and combined concrete cone/bond failure. In addition to the three previously mentioned modes of failure, Ozbakkaloglu and Saatcioglu (2009) identified concrete splitting failure. Likewise, Kim and Smith (2009) documented the same four failure modes previously discussed. Hence, the failure modes observed in adhesive FRP sheet anchors are similar to those of metallic anchors despite the fundamental differences between FRP and metallic anchors (Kim and Smith, 2010).

Analytical models have been proposed for single post-installed adhesive FRP sheet anchors. Özdemir (2005) proposed the following for either concrete cone failure ( $N_{c}$ ) or a combination of concrete cone and bond ( $N_{c b}$ ):

N_{c} = 0.33 π \sqrt{f_{c}^{'}} (d_{0} + h_{e f}) h_{e f} i f h_{e f} < (h_{c} = 50 m m)

(12a)

N_{c b} = 0.33 π \sqrt{f_{c}^{'}} (d_{0} + 50) 50 + π τ d_{0} (h_{e f} - 50) i f h_{e f} \geq (h_{c} = 50 m m)

(12b)

Parameters used for experimental investigation were embedment depth with nominal depth = 50, 70, and 100 mm and concrete cylinder compressive strength of 10 and 16 MPa. Similarly, Ozbakkaloglu and Saatcioglu (2009) proposed the following:

N_{c b} = π f_{c t} h_{c} (d_{0} + h_{c}) + π τ d_{0} (h_{e f} - h_{c})

(13)

where

f_{c t}

is the tested concrete splitting tensile strength. Moreover, Kim and Smith (2010) used a data bank of 43 test results that either experienced concrete cone failure or combined cone-bond failure. Variables considered were nominal anchor hole diameter 12.7, 15.9, 19.1 mm, nominal embedment depth 25, 50, 75, 100 mm, and nominal concrete strength 30, and 50 MPa. The following equations were derived:

N_{c} = 9.68 h_{e f}^{1.5} \sqrt{f_{c}^{'}}

(14a)

N_{c b} = 4.62 π d_{0} h_{e f} i f f_{c}^{'} < 20 M P a

(14b)

N_{c b} = 9.07 π d_{0} h_{e f} i f f_{c}^{'} \geq 20 M P a

(14c)

N_{r} = 0.59 w_{F R P} t_{F R P} f_{F R P}

(14d)

where

N_{r}

= anchor rupture capacity,

w_{F R P}

and

t_{F R P}

= width and thickness of FRP sheets (mm), and

f_{F R P}

= FRP tensile rupture strength (MPa).

Experimental program

GFRP bars

Two types of GFRP bars were used, including V-ROD GFRP bar with sand-coated surface produced by Pultrall Inc. (2012) and ComBAR® GFRP bar with ribbed surface produced by Schoeck Canada Inc. (2013), as shown in Figure 4. Bar sizes used were 16 and 19 mm (#5 and #6) for sand-coated bars and 12 and 16 mm (#4 and #5) for ribbed surface bars. Table 1 shows the mechanical properties of GFRP bars as provided by the manufacturer and the size of drill bits used for drilling the concrete. Mechanical properties listed in Table 1 were obtained using the available test methods in CSA-S806 and CSA-S807 Standards (CSA, 2019, 2021).

Figure 4.

GFRP bars used in the experimental program: (a) sand-coated bars, (b) ribbed-surface bars.

Table 1.

Properties of the GFRP bras and drill bit sizes used for drilling holes.

Bar type	US size	Nominal bar diameter (mm)	Core diameter (mm)	Exterior diameter (mm)	Drill bits used (mm)	Minimum guaranteed tensile strength (MPa)	Elastic modulus (GPa)	Ultimate elongation (%)	Bond strength (MPa)
Ribbed surface	#4	12	12.00	13.5	17.50	1350^a	64.0	1.67	12.2
Ribbed surface	#5	16	16.00	18.0	22.33	1188^b	64.0	1.67	12.2
Sand-coated surface	#5	16	15.87	19.2	24.00	1184^c	62.6 ± 2.5	1.89	14.0
Sand-coated surface	#6	19	19.05	22.4	26.99	1105^c	63.7 ± 2.5	1.73	14.0

^aMean value, Shoeck (2013).

^bMean value, Azimi et al. (2014a).

^cMinimum guaranteed tensile strength, Pultrall (2012).

Adhesives

Two different commercial adhesives were used. The first one is called Hilti HIT-RE 500 (adhesive Type I), an epoxy-based adhesive with a high strength specifically designed to fasten in solid-based materials. It could be used underwater and for holes up to a maximum of two times larger than the bar diameter but limited to 76 mm. The manufacturer reports a minimum bond strength of 12.4 MPa (Hilti Canada Corp., 2011). Sika AnchorFix®-2001 (adhesive Type II) is the second adhesive used for the experiments. It is a VOC-compliant epoxy acrylate hybrid anchoring adhesive (Sika Canada, 2013). The manufacturer describes it as suitable for medium and heavy load applications with a short setting time in dry or damp holes, achieving high early strength in various base materials. The properties of both adhesives are presented in Table 2. Properties listed in Table 2 were obtained using the available ASTM test methods (ASTM, 2015, 2020, 2022a & b).

Table 2.

Adhesive specifications.

AdhesiveName	Compressive strength	Tensile strength	Elastic modulus	Bond strength	Absorption	Resistance	Min cure temperature
AdhesiveName	(MPa)	(MPa)	(MPa)	(MPa)	(%)	(Ω/m)	°C
Type I	82.7	43.5	1493	12.4	0.06	6.6 × 10¹³	−5
Type II	69.0	13.3	NA	NA	NA	5.0 × 10¹³	−5

NA = not applicable.

Concrete

The ready-mix concrete of 35 MPa target compressive strength was utilized to cast reinforced concrete slabs measuring 30 m in length, 1.35 m in width, and 300 mm in thickness. The concrete was cast in an off-campus facility in outdoor conditions in Toronto, Canada. The concrete was of normal weight with a nominal 6.5% air content. After pouring the concrete, it was cured for 2 weeks and was exposed to outdoor conditions for almost 1 year. The compressive strength of concrete was measured to be 40.2 MPa by performing compression tests on several core samples taken from the slabs immediately after pullout tests.

Specimen preparation

Table 3 shows the test matrix of the experimental program. As shown in Table 3, five identical anchor samples were installed for each specific combination of bar type, adhesive type, bar diameter, and embedment length. It is mandated by CSA S807-19 (2019) that at least five identical samples must be prepared. Hence, a total of 120 bars were installed in concrete. 14 concrete panels were constructed and reinforced at the top and bottom with 15M steel bars with 300 mm spacing in each direction to consider the influence of confinement of internal reinforcement on the pullout capacity of GFRP bars.

Table 3.

Test matrix.

Adhesive type	Bar type, diameter	Embedment length, mm			No. of samples
TYPE I	#4 Ribbed GFRP bar	100	150	200	3 × 5 = 15
Type II	#4 Ribbed GFRP bar	100	150	200	3 × 5 = 15
TYPE I	#5 Ribbed GFRP bar	150	200	250	3 × 5 = 15
Type II	#5 Ribbed GFRP bar	150	200	250	3 × 5 = 15
TYPE I	#5 Sand-coated GFRP bar	150	200	250	3 × 5 = 15
Type II	#5 Sand-coated GFRP bar	150	200	250	3 × 5 = 15
TYPE I	#6 Sand-coated GFRP bar	150	200	250	3 × 5 = 15
Type II	#6 Sand-coated GFRP bar	150	200	250	3 × 5 = 15

After pouring the concrete, it was cured for 2 weeks and exposed to an open environment for 1 year. The layout of the GFRP post-installed anchors was made per ASTM E E488/E488M–10 (ASTM, 2010), having a clear distance of 2L_e as a circle around the bar that is not interfering with the 2L_e distance of other bars, where L_e is the embedment depth. The chosen diameters for bits are shown in Table 1. To install GFRP anchors, holes were drilled using rotary hammer pits as shown in Figure 5(a). All the procedures for bar installation are illustrated in Figure 5.

Figure 5.

Step-by-step procedure of bar installation.

For the adhesive Type I, the holes were cleaned by wire brushes and compressed air although this type of epoxy performs well in moisture. On the other hand, the holes for Type II adhesive were cleaned and dried by torch, wire brushes, and compressed air since this series of bars were installed when the temperature of concrete was between 0 and -5°c. In addition, ribbed bars were cleaned using Aceton to ensure proper bonding after placement of the bar inside the holes. Prior to bar placement, each bar was marked to ensure proper embedment length is satisfied. Once the holes and bars were cleaned, the two-component adhesive package was installed in the dispenser and injected into the holes. Then, bars were pushed into the holes in a screwing fashion. The slabs were left outdoor during winter so that anchors were installed and then tested after 7 months as depicted in Figure 6. The GFRP adhesive anchors were tested during the last week of March as shown in Figure 6.

Figure 6.

Air temperature during the GFRP bar installation and testing periods.

Test setup and procedure

A portable test setup was designed considering a minimum 2 $h_{e f}$ distance between the support of the test setup and the bar being tested as specified by ASTM (2010) and Ahmed et al. (2008). The test setup consisted of a manual hydraulic jack, four hollow steel sections (HSS), a potentiometer, a load cell designed for both compression and tension, as well as a specially designed grip to secure the bar. Figure 7 shows a schematic drawing and views of the test setup and grips. The anchors were pulled using the hydraulic jack connected to a manual pump. Each anchor was instrumented with one potentiometer at the bottom of the bar to measure bar displacement at different loading stages. The data acquisition system was accurate to 0.01 kN up to the failure load.

Figure 7.

Details of test setup: (a) schematic view, (b) test setup view, (c) grip assemblage.

Experimental results and discussions

Observed modes of failure

Five modes of failure were possible in the pullout tests. Schematics of these modes are given in Figure 3. Modes of failure ‘a’ and ‘e’ can be easily specified during the experiment. Specifying the mode of failure ‘c’ is also feasible since the surface cone can be easily removed and the bar can be exposed. If no adhesive was attached to the bar, it is a mode of failure ‘c’ and cannot be the mode of failure ‘b’ or ‘d’. If adhesive was left attached to the bar, failure could be either type ‘b’ or ‘d’. If experiments were continued after bar failure until removing the bar from the concrete, it would be possible to distinguish between type ‘b’ and ‘d’ modes of failure. However, experiments were stopped after failure, which is the case for most experimental investigations. Representative modes of failure observed during the experiment are shown in Figure 8.

Figure 8.

Observed failure modes: (a) adhesive/concrete interface, failure mode ‘b’ or ‘d’, (b) ribbed GFRP/adhesive interface, failure mode ‘c’, (c) sand-coated GFRP/adhesive interface, failure mode ‘c’, (d) bar rupture, failure mode ‘e’.

To further investigate the failure zone, several core samples were taken. Bars were cut at the bottom, and then a coring machine was placed on top of a tested bar (having the bar in the center). Then, a 300 mm thick slab was scored. Core samples were then cut in half in the laboratory. Representative sliced core samples are shown in Figure 9. Most bar slippages occurred at the GFRP/adhesive interface rather than the adhesive/concrete interface. A mixed interface mode of failure (type ‘d’ failure) was observed more than type ‘b’ failure. Some samples of failure type ‘d’ are shown in Figure 9(d), (g), and (h).

Figure 9.

Details of the failure zones.

Figure 10 compares failure modes observed for different bar sizes and types (sand-coated and ribbed surfaces). It can be concluded from the figure that regardless of adhesive or bar type and size, no cone failure (type ‘a’ failure) was observed in any of the performed pullout tests. This is mainly due to the relatively large embedment depths used in the experiments and the presence of the top layer of steel reinforcement. Figure 10(a) demonstrates that #4 ribbed bars mainly failed in either type ‘c’ or ‘e’ failure mode. Bar rupture occurred for cases with an embedment depth of more than 150 mm, showing that the embedment depth of 200 mm was enough to develop full bar capacity. It should be mentioned that each bar chart presented in Figure 10 consists of the modes of failure observed in 30 tests. The custom grip assembly experienced slippage before the bar failure in a few tests. The results of these tests were discarded, and the corresponding specimen was not used again. This is why the summation of GFRP bars is less than 30 in Figure 10(b)–(d).

Figure 10.

Occurrence of the various types of failure modes. Note: for failure type a, b, c, d, and e, see Figure 2.

Modes of failure ‘b’, ‘c’, and ‘d’ are generally considered pullout modes of failure and treated similarly in developing the analytical model. This is because it is challenging to distinguish between these modes of failure during experiments (Cook et al., 1998). Figure 10 depicts that for #5 and #6 bars, the majority of modes of failure are pullout mode.

Mode of failure ‘c’ is the failure at the interface of the adhesive and bar. The number of modes of failure ‘b’ or ‘d’ was more than the mode of failure ‘c’ in #5 sand-coated bars compared to #5 ribbed bars, probably due to the stronger bond of the sand-coated surface with adhesives. Similarly, #6 sand-coated bars predominantly failed in modes ‘b’ or ‘d’, confirming a strong bond between the sand-coated surface and adhesive.

Pullout strength

Table 4 presents the summary of test results obtained for ultimate loads. It includes average values of five identical samples in each case, the corresponding coefficient of variations, and the minimum specified failure load. For the test results to be used in design purposes, CSA S807-19 (2019) specifies the following equation to determine the minimum specified failure load for a group of test samples:

F_{k} = {\bar{F}}_{t e s t} (\frac{1 - 1.645 C_{v}}{1 + \frac{1.645 C_{v}}{\sqrt{n}}})

(15)

where

{\bar{F}}_{t e s t}

is the mean value,

C_{v}

is the coefficient of variation, n is the number of test samples. Characteristic value is calculated based on the ultimate load of five identical samples. This value considers the deviation of ultimate loads from the mean ultimate load of the five tested identical samples.

Table 4.

Summary of pullout test results.

GFRP bar type	Adhesive type	Embedment depth (mm)	Average ultimate load (kN) (4)	Dominant mode of failure	Coefficient of variation (%)	Minimum specified failure load (kN) CSA S807-19 (2019) (6)	(4)/(6)
#4 Ribbed	TYPE I	100	71.3	c	14.4	49.2	1.45
		150	98.3	c, e	4.7	87.7	1.12
		200	98.4	e	1.8	94.2	1.04
	TYPE II	100	80.9	c	5.0	71.6	1.13
		150	91.6	b, c, e	2.8	85.7	1.07
		200	99.0	e	3.7	90.5	1.09
#5 Ribbed	TYPE I	150	150.5	c	11.2	113.4	1.33
		200	169.9	b, d	2.8	157.7	1.08
		250	157.0	b, d, e	3.2	144.8	1.08
	TYPE II	150	143.9	c	10.8	109.5	1.31
		200	152.6	c, b, d	5.1	134.8	1.13
		250	154.3	b, d	4.7	137.1	1.13
#5 Sand-coated	TYPE I	150	159.6	c	6.9	134.7	1.18
		200	169.7	b, d	18.8	103.0	1.65
		250	164.5	b, d	16.8	106.0	1.55
	TYPE II	150	137.6	c	15.5	92.0	1.50
		200	159.2	c, b, d	3.2	147.1	1.08
		250	170.5	b, d	13.6	119.0	1.43
#6 Sand-coated	Type I	150	122.3	c	16.2	80.2	1.52
		200	197.2	b, d	20.0	115.2	1.71
		250	207.1	b, d	10.1	160.9	1.29
	Type II	150	152.8	b, d	8.5	123.8	1.23
		200	189.4	b, d	14.0	130.6	1.45
		250	203.9	b, d	10.0	158.5	1.29

a = concrete cone breakout; b = adhesive/concrete interface pullout; c = anchor/adhesive interface pullout; d = combined adhesive/concrete and anchor/adhesive interface anchor/adhesive interface; and e = anchor rupture.

Effect of embedment depth on ultimate load

Figure 11(a) illustrates the average failure load obtained from experiments considering different adhesive types and embedment depths. It can be noted that the average ultimate load increases with an increase in embedment depth. However, increasing the embedment depth from 200 mm to 250 mm has less influence on ultimate load capacity compared to embedment depth increment from 150 to 200 mm.

Figure 11.

Comparison with existing models and the influence of effective embedment depth on: (a) mean ultimate load, (b) minimum specified ultimate load.

As shown in Figure 11(a), results obtained from prediction models presented by Eligehausen et al. (1984) and Fuchs et al. (1995) given in equations (2) and (3), respectively, are considerably higher than experimental results, particularly for higher embedment depths. The reason is that these models were presented for concrete conic failure, which did not occur in the experiments. Conic failure occurs if the embedment depth is relatively small. If those prediction models are used for larger embedment depths that did not undergo conic failure, they tend to produce unrealistic large values as shown in Figure 11(a). It should also be noted that Eligehausen et al. (1984) proposed the model for conic failure of adhesive steel anchors as opposed to the model for cast-in-place headed steel anchors given by Fuchs et al. (1995). Generally, the expected conic failure of post-installed steel anchors is less than that of cast-in-place headed steel anchors (Eligehausen et al., 2006b), which can also be seen in Figure 11(a).

In addition, the results obtained from the prediction model given in equation (11) proposed by Cook (1993) are depicted in Figure 11(a). The model is for post-installed steel adhesive anchors and considers a combined cone/bond failure. Results are given for two cases of lower and upper values of ultimate bond strength (i.e. 6.3 and 14.6 MPa) given by Cook (1993) for various adhesive types. As shown in Figure 11(a), results obtained from the lower-bound value of Cook (1993) are significantly smaller than experimental results, while the upper-bound value of Cook (1993) yields a better match with experimental results, particularly for embedment depth of 150 mm for #4 ribbed bar and 200 mm embedment depth of other bars.

Figure 11(b) shows the experimental results of the minimum specified tensile strength obtained from equation (15), together with design models recommended by ACI 318-19 (2019) and Kim and Smith (2010). The models given by (ACI 318-19, 2019; Kim and Smith, 2010) are for the concrete conic failure of adhesive steel and FRP anchors, respectively. As mentioned above, no conic failure mode was observed during experiments due to large embedment depths; hence, models based on conic failure are expected to provide greater results if they are used for larger embedment depths. This can be seen in Figure 11(b). It is worth mentioning that results obtained from Kim and Smith (2010) are very close to those obtained from ACI 318-19 (2019) and Kim and Smith (2010) even though they are for FRP and steel adhesive anchors, respectively.

Effect of embedment depth on uniform bond strength

Figure 12 shows each adhesive type’s bond stress versus embedment depth, considering different bar types and sizes. The uniform bond stress was calculated as follows:

τ = \frac{F_{u}}{π d_{b} h_{e f}}

(16)

where

F_{u}

is the failure load obtained from the experiments. A critical parameter in the development of analytical models is the type of anchor diameter used in the formula. Various researchers have chosen to focus on either the hole diameter, the outer bar diameter, or the core bar diameter in their proposed models, as outlined in equation (4) through 14. In practical design, engineers typically utilize the core diameter to determine the ultimate tensile strength of anchors rather than count on the outer diameter. Moreover, hole diameter can vary from one product to another depending on specific manufacturer’s recommendations. Therefore, using anchor core diameter in the analytical model of post-installed adhesive anchors is more practical.

Figure 12.

Comparison with existing models and the influence of effective embedment depth on: (a) mean bond stress, (b) minimum specified bond stress.

As shown in Figure 12(a), the mean bond stress decreases with the increase of embedment depth, which is usually the case for steel adhesive anchors (Eligehausen et al., 2006a) and GFRP bar adhesive anchors (Ahmed et al., 2008). Although assuming constant bond strength is convenient for design practices, it is shown in Figure 12(a) that it is not quite representative of the anchor behavior tested in this study. For instance, both #5 ribbed and sand-coated bars depicted almost similar behavior for both adhesive types, as shown in Figure 12(a). For these bars, the difference of bond strength obtained for 150 and 250 mm embedment depths are 60% and 40%, respectively. The constant bond model has been widely used for steel adhesive anchors (Cook et al., 1998; Cook and Kunz, 2001). Cook and Kunz (2001) reported 15.4 MPa for the average bond strength of tested steel adhesive anchors, which is close to the bond strength of #4 ribbed bars with 150 mm embedment depth and other bars with 200 mm embedment depth.

Minimum specified bond strengths were calculated from bond strengths of test results using equation (15) and are shown in Figure 12(b). As shown in equation (15), the minimum specified value depends on the coefficient of variation of test results. The scatter trend illustrated in Figure 12(b) is anticipated, as indicated by the diverse coefficients of variation obtained from the experiments, which range from 1.8% to 20%, as presented in Table 4. Despite the variation in results, it can be concluded that the minimum specified bond strength generally decreases with the increase in embedment depth.

The minimum specified bond strength (or characteristic bond strength as stated in (ACI 355.4M-19, 2019) is used for design practices. Hence, the experimental results in Figure 12(b) are compared with corresponding design models in the literature. Since experimental results are calculated from equation (16) using anchor core diameter, results from design models that use diameter other than core diameter should be normalized accordingly for proper comparison. The design model by Cook and Kunz (2001) in equation (7) uses core diameter; hence no modification is required. The model presented by ACI (2019a), as indicated in equation (9), employs the outer diameter, whereas the bond model proposed by Kim and Smith (2010), as shown in equation (14), utilizes the hole diameter. The ratios of hole diameter and outer diameter to core diameter are slightly different for various bar types used in this study when calculated from Table 1. However, an approximate value may be sufficient for simplicity and comparison with experimental results. The ratios of 1.15 and 1.40 were used in Figure 12(b) to normalize ACI 318-19 (2019) and Kim and Smith (2010) models, respectively. To normalize the results, it is necessary to multiply the proposed design models by the appropriate ratio. ACI 355.4M-19 (2019) recommends 4.5 MPa bond strength for uncracked concrete used in outdoor conditions. Kim and Smith (2010) proposed 9.07 MPa for a design model with a hole diameter. Hence, the normalized values in Figure 12(b) are 5.2 and 12.7 MPa, respectively. Analytical models of Kim and Smith (2010) and Cook and Kunz (2001) are in better agreement with experimental results compared to ACI 318-19 (2019), which are considerably smaller. It is important to note that the estimated normalized ratios of 1.15 and 1.40 were regarded as the average values derived from the outer diameter to the core diameter and the hole diameter to the core diameter for the four sizes of GFRP bars utilized, as detailed in Table 1.

Effect of bar diameter

The effect of bar diameter on the average ultimate load and bond strength is illustrated in Figure 13, which is consistent with others (Islam et al., 2015; Lu et al., 2021). Except for the sand-coated bar with 150 mm embedment depth, an increase in GFRP bar diameter leads to a rapid increase in average ultimate load. The rate of increase was almost the same for ribbed bars and sand-coated bars, as shown in Figure 13(a). Similarly, bond strength recorded for ribbed bars slightly increases with the increase in bar diameter, while sand-coated bars depict an inconsistent behavior. Except for the sand-coated bar with 150 mm embedment depth, it can be concluded from Figure 13(b) that the increase of bar diameter from #5 to #6 sand-coated bars has a negligible effect on the bond strength.

Figure 13.

Effect of core bar diameter on: (a) mean ultimate load, (b) mean bond strength.

Effect of adhesive type

Figure 14 shows the influence of adhesive tape on the mean ultimate load. In most cases, adhesive Type 1 showed slightly higher results with an average increase of 2% and a coefficient of variation of 9%. Hence, the effect of adhesive type can be neglected. The main difference in adhesives’ application was the temperature at the installation time. For adhesive Type 1, the installation was done at a temperature above 10°c, whereas for adhesive Type 2, the temperature of concrete was between −5°c and 0°c. Although this was within the practice recommendation of the manufacturer, the low workability of the adhesive at low temperatures and harder conditions for hole cleaning and moisture removal were the two main reasons contributing to the less pullout capacity of Type 2 adhesive.

Figure 14.

Comparison of the ultimate load of the two adhesives used.

Regression analysis

The formula type should be determined to carry out a regression analysis. The following formula type was considered in this study:

F_{u} = α d_{b}^{β} h_{e f}^{γ}

(17)

where α, β, and γ are constants to be determined from regression analysis. If all ultimate loads obtained from test results are considered, various values for α, β, and γ were used in equation (17) to predict the experimental ultimate load. Regression analysis was undertaken to maximize the coefficient of determination (denoted by R²) with varying α for a pair of values assumed for β and γ. If β = γ = 1, then the model corresponds to a uniform bond model similar to those proposed by Cook et al. (1998), ACI 318-19 (2019), and Kim and Smith (2010), for which the regression result is shown in Figure 15(a). The resulting R² is 0.477, which is relatively small, implying that the uniform bond stress model does not correspond well with the experimental results. Many other combinations of β and γ were considered to obtain the best test results prediction. One of the best combinations is β = 1 and γ = 0.5, which is shown in Figure 15(b). In this case, R² of 0.782 is much higher than 0.477 discussed earlier. If the model obtained in Figure 15(a) is considered, the uniform bond strength can be obtained by dividing α factor over π using the general formula in equation (16). Hence, the uniform bond strength for test results will be determined as 15.2 MPa for all bars.

Figure 15.

Regression analysis results for all test results: (a) γ = 1, (b) γ = 0.5.

For design purposes, the minimum specified ultimate load should be used. The regression analysis was then used for the same formula in equation (17). Results are given in Figure 16 for two values equal to 0.5 and one for ribbed surface bars and all bars (i.e. sand-coated and ribbed bars). Results show that similar to regression analysis for all tested ultimate loads given in Figure 15, regression analysis of γ = 1 does not correlate well with those obtained from experiments. One of the best combinations of β and γ that yields the closest correlation is β = 1 and γ = 0.5. The proposed model for ribbed surface bars with γ = 0.5 correlates well with the minimum specified load obtained from experiments with R² = 0.888, as shown in Figure 16(b). Due to the inconsistency of results observed for sand-coated bars, the model proposed for all bars has less R² and less α factor shown in Figure 16(d), compared to the model obtained for ribbed bars shown in Figure 16(b).

Figure 16.

Regression analysis results for minimum specified ultimate loads: (a) Ribbed bars γ = 1, (b) Ribbed bars γ = 0.5, (c) All bars γ = 1, (d) All bars γ = 0.5.

If the models given in Figure 16(a) and (c) are compared with equation (16), the minimum specified uniform bond strength will be determined as 13.2 MPa for ribbed bars and 11.9 MPa for all bars. These values are close to those obtained from the manufacturer’s specifications in Table 1. In addition, these values are comparable with those obtained for steel adhesive anchors reported in the literature. As mentioned earlier, Cook and Kunz (2001) reported 15.4 MPa as bond strength obtained during experiments and 12.0 MPa as characteristic bond strength assumed for design, which is comparable with the results obtained in this study being 15.2 MPa as bond strength obtained during experiments and 11.9 MPa as the minimum specified bond strength for design purposes.

Design recommendations

Based on the extensive experimental investigation conducted, including 120 tests, the following model is proposed for test results:

F_{u} = 690 d_{b} \sqrt{h_{e f}}

(18a)

and the following model is proposed for design purposes:

F_{u} = 542 d_{b} \sqrt{h_{e f}}

(18b)

These formulas are applicable for embedment depths ranging from 100 mm to 250 mm and for core bar diameters of 12 mm, 16 mm, and 19 mm. The main limitation of this formula is the concrete compressive strength. All the test results were obtained for the same concrete with $f_{c}^{'}$ = 40.2 MPa. The effect of concrete may be treated by adding a new factor similar to equation (7) done by Cook et al. (1998) for steel adhesive anchors. More tests on a wider range of bar diameter, embedment depth, adhesive type, and concrete compressive strength will provide a design model with more confidence.

Several factors, including the uniformity of adhesive coverage, the surface preparation of the bar, and the embedment depth, influence adhesive anchors’ bond strength. An uneven adhesive layer can cause localized stress concentrations during loading, which might reduce the overall bond strength. However, the experimental setup and procedure in Figures 5 and 7 aimed to minimize such effects through standardized installation procedures. Bars were inserted in a screwing motion, and care was taken to ensure that the adhesive filled the drilled holes adequately. Additionally, the test matrix included multiple replicates for each condition to account for variability, including potential deviations in adhesive distribution. For the current study, any potential reduction in bond strength due to tilting is embedded within the overall variability, as evidenced by the coefficient of variation (Table 4). This variability has been accounted for in the regression models and design recommendations, ensuring conservative and reliable predictions for practical applications.

Conclusions

Based on the results of the experimental tests and analytical formulation taking into consideration the studied parameters and the mechanical properties of the used materials, the following conclusions can be drawn:

1. Test results showed that regardless of adhesive or bar type and size, no cone failure (type ‘a’ failure) was observed in any of the performed tests. This is mainly due to the relatively large embedment depths used in the experiments and the top layer of steel reinforcement.

2. Bar rupture occurred for most cases of #4 GFRP ribbed bars with 200 mm embedment depth showing that such embedment depth is enough to develop full bar capacity. For #5 and #6 GFRP bars, modes of failure were predominately either ‘b,’ ‘c,’ or ‘d’, which are all considered as pullout modes of failure treated similarly in the development of the analytical model.

3. The pullout load increased with embedment depth from 150 to 200 mm. However, increasing the embedment depth from 200 to 250 mm less influences pullout load capacity.

4. Mean and minimum specified bond stress decreased with the increase of embedment depth. Thus, assuming a constant bond strength, usually done for steel adhesive anchors, is not quite representative of the GFRP bar adhesive anchors. For instance, #5 ribbed bars showed a 60% smaller mean bond strength for 250 mm embedment depth than 150 mm embedment depth.

5. If a constant bond strength model is used due to its simplicity, the results obtained in this study are comparable with those for steel adhesive anchors. In this study, the mean bond strength and minimum specified bond strength were obtained as 15.2 and 11.9 MPa, respectively, which are close to 15.4 and 12.0 MPa for steel adhesive anchors reported in the literature.

6. An increase in GFRP bar diameter resulted in a rapid increase in the mean ultimate load. The rate of increase was almost the same for ribbed bars and sand-coated bars. On the other hand, the effect of GFRP bar diameter on the uniform bond strength should be addressed.

7. This study used two adhesive types that showed similar behavior. Adhesive Type 1 demonstrated a slightly higher mean ultimate load in the majority of cases, with an average increase of 2% and a coefficient of variation of 9%.

8. Using regression analysis, a new analytical model was developed for GFRP bar adhesive anchors given in equation (18), which can be used for design purposes. However, the range of parameters used for experimental investigation should be considered in any application.

9. The freeze-thaw cycle is a crucial condition that must be taken into account in the behavior and design of GFRP bar adhesive anchors for deteriorated concrete bridge barrier replacement. By considering the impact of freeze-thaw cycles and implementing appropriate design measures, engineers can ensure the long-term durability and reliability of adhesive anchor systems in challenging environmental conditions. The current paper considers one winter of Canada; however, more cycles of freeze-thaw should be considered in future research.

Footnotes

Acknowledgments

The financial support of the Ministry of Transportation of Ontario (MTO) in the form of research grants and site use is greatly appreciated. Also, the authors acknowledge the financial support from the Ontario Centre of Excellence (OCE). Moreover, the authors thank Pultrall Inc., Schock Canada Inc., Hilti Inc., and Sika Canada for sponsoring this project and in-kind contributions. The first author sincerely acknowledges the post-doctoral financial support Fonds Québécois de la Recherche sur la Nature et les Technologies (FQRNT) provided.

Author contributions

Michael Rostami: Methodology, Validation, Formal analysis, Investigation, Data curation, Writing – original draft, Visualization, Project administration. Khaled Sennah: Conceptualization, Investigation, Data curation, Writing – review & editing. Hossein Azimi: Formal analysis, Investigation, Writing – review & editing. Hamdy M. Afefy: Formal analysis, Investigation, Writing – review & editing.

Declaration of conflicting interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Ontario Centre of Excellence (OCE) financial support, Ministry of Transportation of Ontario (MTO) (research grants and site use), Pultrall Inc., Schock Canada Inc., Hilti Inc., and Sika Canada (sponsoring this project and in-kind contributions).

ORCID iDs

Khaled Sennah

Hamdy M Afefy

Data availability statement

Data will be made available on request.*

References

Ahmed

El-Salakawy

Benmokrane

(2008) Tensile capacity of GFRP post installed adhesive anchors in concrete. Journal of Composites for Construction 12(6): 596–607.

American Concrete Institute (ACI 318-19) (2019a) Building Code Requirements for Structural Concrete. Farmington Hills, MI: American Concrete Institute.

American Concrete Institute (ACI 355-4M-19) (2019b) Qualification of post-installed adhesive. Anchors in Concrete. Farmington Hills, MI: American Concrete Institute.

ASTM (2010) Standard Test Methods for Strength of Anchors in Concrete Elements - E488/E488M. Conshohocken, PA, US: American Society for Testing and Materials.

ASTM (2015) ASTM D695-15 - Standard Test Method for Compressive Properties of Rigid Plastics. West Conshohocken, PA, USA: ASTM International.

ASTM (2020) ASTM C882/C882M-20 - Standard Test Method for Bond Strength of Epoxy-Resin Systems Used with Concrete by Slant Shear. West Conshohocken, PA, USA: ASTM International.

ASTM (2022a) ASTM D638-22 - Standard Test Method for Tensile Properties of Plastics. West Conshohocken, PA, USA: ASTM International.

ASTM (2022b) ASTM D570-22 - Standard Test Method for Water Absorption of Plastics. West Conshohocken, PA, USA: ASTM International.

Azimi

Sennah

Sayed-Ahmed

, et al. (2014a) Bridge deck-guardrail anchorage detailing for sustainable construction. ASCE Journal of Bridge Engineering 19(10): 04014039.

10.

Azimi

Sennah

Tropynina

, et al. (2014b) Anchorage capacity of concrete bridge barriers reinforced with GFRP bars with headed ends. Journal of Bridge Engineering 19(9): 04014030.

11.

Canadian Standards Association (CSA) (2014) Design of concrete structures. Mississauga, ON, Canada: CSA standard, Vol. A23.3-14.

12.

Canadian Standards Association (CSA) (2019) “Specification for fiber-reinforced polymers” CSA S807-19. Toronto, ON, Canada: Canadian Standards Association.

13.

Canadian Standards Association (CSA) (2021) Design and Construction of Building Components with Fiber Reinforced Polymers. CSA S806-12. Toronto, ON, Canada: Canadian Standards Association, Re-approved in 2017.

14.

Charney

Pal

Silva

(2013) Recommended procedures for development and splicing of post-installed bonded reinforcing bars in concrete structures. ACI Structural Journal 110(3): 437–447.

15.

Cook

(1993) Behavior of chemically bonded anchors. Journal of Structural Engineering 119(9): 2744–2762.

16.

Cook

Konz

(2001) Factors influencing bond strength of adhesive anchors. ACI Structural Journal 98(1): 76–86.

17.

Cook

Kunz

Fuchs

, et al. (1998) Behavior and design of single adhesive anchors under tensile load in uncracked concrete. ACI Structural Journal 95(1): 9–26.

18.

Dervishhasani

Sennah

(2018) Investigation on the capacity of TL-5 GFRP-reinforced concrete bridge barrier-deck anchorage subjected to transverse vehicle impact loading. In: Proceedings of 10th Intern. Conf. On Short and Medium Span Bridges. Quebec, Canada: CSCE, 1–10.

19.

Eligehausen

Mallée

Rehm

(1984) Befestigungen mit Verbundankern (Fastenings with bonded anchors). Betonwerk und Fertigteil-Technik 84(10): 682–692.

20.

Eligehausen

Cook

Appl

(2006a) Behavior and design of adhesive bonded anchors. ACI Structural Journal 103(6): 822–832.

21.

Eligehausen

Mallée

Silva

(2006b) Anchorage in Concrete Construction. Berlin, Germany: Ernst & Sohn.

22.

Fuchs

Eligehausen

Breen

(1995) Concrete capacity design (CCD) approach for fastening to concrete. ACI Structural Journal 92(1): 73–94.

23.

Gyawali

Sennah

Ahmed

, et al. (2024) Experimental study of static and fatigue push-out test on headed stud shear connectors in UHPC composite steel beams. Structures 70: 107923. DOI: 10.1016/j.istruc.2024.107923.

24.

Hao

Z-H

Zeng

J-J

Chen

G-M

, et al. (2024) Durability of FRP-to-concrete bonded joints subjected to 110 months accelerated laboratory and field exposure. Engineering Structures 305: 117681.

25.

Hilti Canada Corp (2011) Technical downloads. https://www.hilti.ca

26.

Islam

Afefy

Sennah

, et al. (2015) Bond characteristics of straight- and headed-end ribbed-surface GFRP bars embedded in high-strength concrete. Construction and Building Materials 83: 283–298.

27.

Khalafalla

Sennah

(2021) Ultimate and fatigue responses of GFRP-reinforced, UHPC-filled, bridge deck joints. ACI Special Publication SP-356: Development & Applications of FRP Reinforcements DA-FRPR’21: 347–374. DOI: 10.14359/51737280.

28.

Khederzadeh

Sennah

(2014) Development of cost-effective PL-3 concrete bridge barrier reinforced with sand-coated glass fibre reinforced polymer (GFRP) bars: static load tests. Canadian Journal of Civil Engineering 41(4): 368–379.

29.

Kim

Smith

(2009) Behaviour of handmade FRP anchors under tensile load in uncracked concrete. Advances in Structural Engineering 12(6): 845–865.

30.

Kim

Smith

(2010) Pullout strength models for FRP anchors in uncracked concrete. Journal of Composites for Construction 14(4): 406–414.

31.

Afefy

Azimi

, et al. (2021) Bond performance of sand-coated and ribbed-surface glass fiber reinforced polymer bars in high-performance concrete. Structures 34: 10–19.

32.

Afefy

Azimi

, et al. (2022) Bond characteristics of glass fiber reinforced polymer bars in high-strength concrete. Structures and Buildings 175(10): 48–764.

33.

Zheng

, et al. (2023) Long-term tensile performance of GFRP bars in loaded concrete and aggressive solutions. Journal of Building Engineering 64: 105587.

34.

McVay

Cook

Krishnamurthy

(1996) Pullout simulation of postinstalled chemically bonded anchors. Journal of Structural Engineering 122(9): 1016–1024.

35.

Ozbakkaloglu

Saatcioglu

(2009) Tensile behavior of FRP anchors in concrete. Journal of Composites for Construction 13(2): 82–92.

36.

Özdemir

(2005) Mechanical Properties of CFRP Anchorage. Ankara, Turkey: Middle East Technical University, 74. Master’s thesis.

37.

Peng

Huang

, et al. (2023) Shear strengthening of reinforced concrete beams using geopolymer-bonded small-diameter FRP bars. Composite Structures 305: 116513.

38.

Piccinin

Cattaneo

Biolzi

(2013) Breakout capacity of headed anchors in confined concrete: experimental evidence. ACI Structural Journal 110(3): 469–479.

39.

Pultrall (2012) Technical specifications for V-rod HM straight bars. https://www.pultrall.com/

40.

Rostami

Sennah

Afefy

(2017) Ultimate capacity of barrier-deck anchorage in MTQ TL-5 barrier reinforced with headed-end, high-modulus, sand-coated GFRP bars. Canadian Journal of Civil Engineering 45: 263–278. DOI: 10.1139/cjce-2017-0002.

41.

Rostami

Sennah

Hedjazi

(2019a) GFRP bars anchorage resistance in a GFRP-reinforced concrete bridge barrier. Materials Journal 12(15): 1–24.

42.

Rostami

Sennah

Mostafa

(2019b) Experimental study on the transverse load carrying capacity of TL-5 bridge barrier reinforced with special profile GFRP bars. ASCE Journal of Composites for Construction 23(4): 1–16.

43.

Sayed-Ahmed

Sennah

Afefy

(2024a) Structural behavior of full-depth deck panels having developed closure strips reinforced with GFRP bars and filled with UHPFRC. Journal of Composites Science 8(11): 468.

44.

Sayed-Ahmed

Sennah

Afefy

(2024b) Fatigue and ultimate strength evaluation of GFRP-reinforced, laterally restrained, full-depth precast deck panels with developed UHPFRC filled transverse closure strips. Applied Sciences 14(19): 8806.

45.

Schoeck (2013) Technical information for Schoeck ComBAR. https://www.schoeck-combar.com/

46.

Sennah

Hedjazi

(2018) Structural qualification of a developed GFRP-reinforced PL-3 concrete bridge barrier using vehicle crash testing. International Journal of Crashworthiness 24(3): 1–18.

47.

Sennah

Khederzadeh

(2014) Development of cost-effective PL-3 concrete bridge barrier reinforced with sand-coated glass fibre reinforced polymer (GFRP) bars: vehicle crash test. Canadian Journal of Civil Engineering 41(4): 357–367.

48.

Sennah

Mostafa

(2018) Performance of a developed TL-5 concrete bridge barrier reinforced with GFRP hooked bars: vehicle crash testing. ASCE Journal of Bridge Engineering 23(2): 1–20.

49.

Sennah

Tropynina

Ibrahim

, et al. (2018) Structural qualification of a developed GFRP-reinforced PL-3 concrete bridge barrier using ultimate load testing. International Journal of Concrete Structures and Materials 12(63): 1–22.

50.

Sika Canada (2013) Product data sheet. https://www.can.sika.com/

51.

Sun

J-H

N-J

Z-Q

, et al. (2014) Durability of concrete-encapsulated GFRP bars subjected to seawater immersion. Case Studies in Construction Materials 20: e03249.

52.

Zhang

Xiao

(2022) Durability of FRP bars and FRP bar reinforced seawater sea sand concrete structures in marine environments. Construction and Building Materials 350: 128898.

53.

Zhu

, et al. (2023) Prediction of the long-term tensile strength of GFRP bars in concrete. Buildings 13(4): 1035.

54.

Zhu

Weng

Huang

, et al. (2024a) Ultra-High-Strength Engineered Cementitious Composites (UHS-ECC) panel reinforced with FRP bar/grid: development and flexural performance. Engineering Structures 302: 117193.

55.

Zhu

Weng

Liu

, et al. (2024b) Thin-layer Ultra-High-Strength Engineered Cementitious Composites (UHS-ECC) reinforced with small-diameter FRP bars for structural strengthening. Thin-Walled Structures 205: 112592.