A Systematic Review of Anchored and Unanchored EB-FRP Systems for Tension Strengthening of Concrete Structures Authors: Junrui Zhang, Enrique del Rey Castillo, Mohammad Sadegh Salimian Rizi, Tingting Yu Published: 2026 ABSTRACT Externally bonded fiber-reinforced polymer (EB-FRP) systems have been extensively investigated for tension strengthening concrete structures. Interpretation of the available evidence remains challenging because experimental methods, specimen scales, material systems, anchorage configurations, and reporting practices vary substantially across the literature. This systematic review synthesized 174 peer-reviewed studies published between 1994 and 2026, comprising 3908 experimental test results and 42 analytical formulations addressing unanchored and anchored EB-FRP systems. Review findings showed that bond performance in unanchored systems is governed primarily by FRP stiffness, bond geometry, concrete properties, adhesive behavior, surface preparation, and environmental exposure. These parameters influence bond capacity, debonding strain, effective bond length, and failure mode. Anchored configurations consistently enhanced force transfer, delayed premature debonding, and improved load-carrying capacity relative to unanchored systems. Unanchored systems dominated the available evidence base with 3162 test results, whereas only 96 multi-anchor system tests were identified, highlighting limited understanding of anchor interaction and load redistribution mechanisms. CFRP represented the dominant material system, while substantially fewer studies investigated GFRP, BFRP, and AFRP systems. Existing strength models generally captured specific failure mechanisms within their calibration ranges but demonstrated limited transferability across different geometries, loading conditions, anchorage configurations, and environmental conditions. Limited evidence remains available for scale transfer, durability degradation, anchor strip interaction, and multi-anchor load sharing under field-representative conditions. Future research should focus on standardized benchmarking procedures, large-scale validation programs, durability-informed design approaches, experimentally validated numerical modeling, and unified design provisions for EB-FRP strengthening systems. 1. Introduction Externally bonded fiber-reinforced polymer (EB-FRP) systems have become popular in the rehabilitation of deteriorated concrete infrastructure due to their high tensile strength, corrosion resistance, and ease of installation without adding significant weight [1,2]. These systems offer practical advantages for flexural strengthening of beams [3,4,5], axial confinement of columns [6,7,8], and tensile enhancement of slabs [9,10,11] and shear walls [12,13,14]. EB-FRP systems are also widely applied in seismic and blast retrofitting, where rapid installation and minimal section enlargement are critical [15,16]. A typical EB-FRP configuration uses unidirectional fibers bonded to concrete substrates with epoxy resin, as shown in Figure 1. Load is transferred through shear stress along the adhesive–concrete interface, where stiffness mismatch between the materials concentrates stress near the loaded end of the sheet, often triggering premature debonding failures [17,18]. Debonding failures typically occur in three principal forms: cohesive failure within the concrete substrate, cohesive failure within the adhesive layer, and adhesive failure at the concrete, adhesive, or adhesive–FRP interface [19,20]. Among these, failure within a thin surface layer of concrete, commonly referred to as concrete cover separation, is the most frequently observed and often governs overall system performance. Adhesive and interfacial failures are less prevalent but can become critical when surface preparation, adhesive quality, or bond continuity is inadequate. All three modes are brittle and result in limited energy dissipation, which reduces the effective utilization of FRP tensile strength. A clear understanding of the conditions governing each mode is essential for effective FRP system design. Research has established that EB-FRP system performance is further influenced by parameters such as strip geometry and the mechanical properties of the FRP and concrete materials [11,21]. Insufficient control of these variables can diminish bond capacity and cause premature failure. Environmental conditions, including elevated temperature and moisture exposure, also accelerate degradation, particularly in adhesive-dependent systems [22,23]. These effects are especially problematic in high-performance applications involving stiff or thick FRP laminates, which intensify interfacial stress gradients and reduce resistance to debonding. Several test methods have been developed to assess EB-FRP bond performance under tension. These include direct tension tests [24], tension-face flexural tests (e.g., three- or four-point bending) [25,26], and single or double lap-shear tests [27,28]. Each method imposes distinct interfacial stress states that influence bond characteristics and failure mechanisms. A systematic understanding of how test setup, material processing, and surface preparation affect test results is essential to ensure reliable performance evaluation. Fiber anchors have been introduced to enhance bond capacity by providing mechanical confinement and delaying debonding [29,30,31]. Among various configurations such as bow-tie [32], circular [33], or through-anchors [34], the fan-shaped anchor remains the most widely adopted due to its proven effectiveness and ease of fabrication [35]. These anchors are commonly constructed from the same FRP material and are embedded into the concrete to form a dowel/fan system. Figure 2 shows the typical layout of the dowel/fan anchor system embedded in concrete. The fan portion distributes load to surrounding concrete, while the dowel provides axial restraint. Anchor efficiency is governed by embedment depth, insertion angle, fan width, and spatial arrangement. Many existing studies have evaluated shallow and small-scale anchors, constrained by the physical limitations of retrofit applications [29,32,36,37,38,39,40,41,42]. For instance, slab toppings often range from 50 to 75 mm in thickness, restricting embedment depth [43]. Anchor performance is typically evaluated through direct tension (pullout) [38] or lap-shear tests [43]. Improper test configuration or design detailing can lead to undesired failure modes such as anchor pullout or concrete cone detachment, rather than the more predictable fiber rupture at the fan–concrete interface [44]. Multi-anchor systems introduce additional complexity by altering load paths and reducing the effective bond length per anchor. Figure 3 shows a multi-anchored EB-FRP system used in concrete diaphragm retrofitting. The interaction between anchor fans creates stress redistribution effects that remain poorly characterized. Parameters such as FRP stiffness, anchor density, sheet overlap, and concrete modulus collectively influence strain gradients and peak interfacial stress [45,46]. Current design codes remain inadequate for guiding EB-FRP systems that incorporate anchors. ACI 440.2R-17 [47] adopts simplified assumptions that omit anchorage mechanics and fail to reflect the nonlinear interfacial behavior observed in experiments [14]. Similarly, the CNR-DT 200 [48] guideline proposes empirical bond strength equations but lacks provisions for anchored systems [20]. The fib Bulletin 90 [49] extends the fib Bulletin 14 model database but retains a primary focus on unanchored, small-scale configurations. Similarly, existing fracture mechanics-based theoretical models, primarily developed from tests on thin and short FRP strips and small, shallow anchors under bending or lap-shear configurations [50], lack general applicability to realistic field-scale conditions. Most available predictive models for isolated anchored systems are empirical and calibrated to narrow parameter ranges, limiting their usefulness for broader structural applications. These limitations highlight the need for a systematic review to consolidate findings, strengthen modeling generalizability, and support more practical and transferable design approaches. Data acquisition systems (DAQs) play a critical role in understanding EB-FRP anchorage behavior. Conventional instrumentation such as strain gauges and LVDTs captures only localized responses and fails to resolve distributed slip or crack development [51,52,53]. Advanced systems such as digital image correlation (DIC) and embedded fiber optic sensors enable full-field strain visualization and high-resolution tracking [54,55,56]. Despite their advantages, these tools remain underutilized due to cost, setup complexity, and limited access in large-scale testing. Current studies often employ hybrid data acquisition approaches to combine the strengths of multiple systems, but a lack of standardization and limited literature coverage leave open questions regarding optimal configurations [57,58,59,60]. Further investigation is required to assess how integrated DAQ strategies contribute to performance evaluation. This systematic review addresses these limitations by synthesizing and categorizing experimental studies using a PRISMA-guided approach [61,62,63,64]. Key research questions include the influence of FRP anchors on interfacial behavior, the generalizability of existing bond–slip models, and the limitations of current design standards. A Complexity of Investigation (COI) classification is applied to benchmark experimental rigor, instrument precision, and model sophistication across the published literature. The goal is to establish a foundational database that enables improved analytical models, informed design guidelines, and optimized anchorage configurations for EB-FRP systems under tension. The research gap addressed by this review is therefore not only the absence of additional test data, but the lack of a traceable, metric-based synthesis that connects test configuration, FRP material, bond geometry, environmental exposure, anchorage configuration, and model validation to comparable performance outcomes. In particular, the literature still lacks normalized evidence for scale transfer, durability-sensitive bond degradation, anchor–strip interaction, multi-anchor load sharing, and the accuracy limits of empirical strength models. 2. Methodology 2.1. PRISMA-Guided Review Framework This systematic literature review synthesizes and evaluates experimental research on EB-FRP systems used to strengthen concrete structures in tension. The review focuses on parameters influencing interfacial performance and experimental test configurations. A systematic review approach was adopted to reduce the subjectivity often found in narrative reviews and to improve reproducibility in study identification, coding, and synthesis. Through this approach, the review maps research trends and links database outputs to ongoing developments in EB-FRP system behavior. The methodology follows the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 checklist [60,61,62,63], which supports structured evidence identification and synthesis. Five main stages guided the review process: (1) framing research questions, (2) identifying relevant studies, (3) screening for quality and relevance, (4) conducting full-text eligibility checks, and (5) summarizing and interpreting findings. These stages are outlined in Figure 4 and elaborated in the following subsections. 2.2. Framing Research Questions To guide the review, three primary research questions (RQs) were formulated to evaluate the performance of EB-FRP systems for concrete strengthening, with and without the use of fiber anchors: RQ1. How do non-anchored EB-FRP systems perform in strengthening concrete under axial tension, measured by ultimate bond capacity, bond strength or capacity per bonded width, debonding strain or strain profile, effective bond length, fracture energy or load–slip response where reported, and observed failure mode? RQ2. How do fiber anchors influence bond behavior and load-transfer efficiency in EB-FRP systems? RQ3. What are the current limitations and future research needs for EB-FRP systems, both with and without fiber anchors? The research questions define the scope of the review by focusing on the performance of bonded FRP strips, the contribution of FRP anchors to interfacial performance, and areas where further research is needed. Figure 5 shows a mind map that links related questions to major themes in the literature, including fracture mechanics, geometric effects, anchor layout, and bond failure mechanisms. For traceability, RQ1 was evaluated through extracted performance indicators rather than a single generic performance label. Direct comparisons used reported ultimate load, normalized capacity per FRP width where sufficient geometry was available, FRP strain or debonding strain where measured, effective bond length, bond–slip or load–slip response, fracture energy, and failure mode. RQ2 used the same base indicators and added anchor geometry, embedment depth, fan angle, insertion angle, dowel diameter, anchor spacing, and single- versus multi-anchor configuration. RQ3 was derived from evidence gaps revealed by these coded variables, especially missing large-scale, durability, and model-validation data. The literature search used two academic databases, Scopus and Web of Science. Identical search strategies were applied to both databases using structured keyword combinations tailored to the research topic. This dual-database approach ensured comprehensive coverage and helped identify studies across different publishers. In addition to database searches, reference lists of selected papers were reviewed to identify further relevant studies through citation tracking. 2.3. Literature Search Strategy The literature search targeted experimental studies on EB-FRP systems for concrete strengthening, with and without FRP anchors. Scopus and Web of Science were selected as the primary and secondary databases, respectively. Identical search strategies were applied across both platforms to ensure consistency. The search included all relevant journal articles published up to 17 February 2026, without restrictions on earlier publication dates. Only journals indexed in the SCImago Journal Rank (SJR) database were considered. The SJR system provides broader coverage than other ranking indicators, including Journal Impact Factor (JIF), Eigenfactor Score (ES), and H5 [65]. A total of 302 civil and structural engineering journals were included, consisting of 80 Q1, 80 Q2, 75 Q3, and 67 Q4 journals. Strict eligibility criteria were applied to maintain the quality and relevance of the included studies. Each article was reviewed based on its title, abstract, and methodology. Studies were retained only if they met all of the following conditions:Published in any year, up to 17 February 2026.Focused on EB-FRP strengthening of concrete, with and without FRP anchors, as indicated in the title, abstract, or keywords.Reported experimental work based on direct tension or lap-shear tests; studies using numerical models, simulations, or finite element analysis (FEM) were excluded.Written in English and published in final form as peer-reviewed journal articles. Published in any year, up to 17 February 2026. Published in any year, up to 17 February 2026. Focused on EB-FRP strengthening of concrete, with and without FRP anchors, as indicated in the title, abstract, or keywords. Focused on EB-FRP strengthening of concrete, with and without FRP anchors, as indicated in the title, abstract, or keywords. Reported experimental work based on direct tension or lap-shear tests; studies using numerical models, simulations, or finite element analysis (FEM) were excluded. Reported experimental work based on direct tension or lap-shear tests; studies using numerical models, simulations, or finite element analysis (FEM) were excluded. Written in English and published in final form as peer-reviewed journal articles. Written in English and published in final form as peer-reviewed journal articles. The search process was designed around the three primary research questions. Keywords were selected in advance to reflect core themes of EB-FRP behavior, interfacial bonding, and anchorage performance. In the first stage, broad search terms such as “FRP and concrete” and “Fiber/Fiber Reinforced Polymer/Plastics and Concrete” were used to ensure topic coverage. This initial stage yielded 24,069 results from Scopus and 17,529 from Web of Science, reflecting the growing interest in FRP systems over recent decades. Terms such as “CFRP” or “carbon fiber” were intentionally excluded from the early stages to avoid excluding studies involving GFRP or BFRP that may not use those specific terms. The remaining stages applied Boolean operators (AND, OR, and NOT) to progressively refine the results, targeting experimental work on EB-FRP under tension or lap-shear loading and eliminating irrelevant topics. Scopus was prioritized due to its broader coverage, which includes approximately 99.11% of the publications indexed by Web of Science [66]. The full breakdown of search terms and filtering stages is shown in Table 1, which documents the sensitivity of the search result to each keyword-refinement stage. The table should be read from top to bottom: broad FRP–concrete terms establish coverage, EB-FRP and bond-behavior terms increase specificity, material and anchorage terms retain relevant CFRP/GFRP/BFRP/AFRP and anchor studies, and exclusion terms remove numerical, machine learning, non-EBR, and non-bond-focused records before deduplication. This staged reporting makes the effect of each keyword family transparent without changing the final eligibility criteria. Quantitative examples were extracted retrospectively from the final included studies after completion of the screening process to improve transparency regarding keyword sensitivity. Table 2 preserves the original coverage explanations and residual limitations for each alternative term family. The additional information does not alter the PRISMA workflow, the eligibility criteria, or the screening process. The final study selection process and the number of included studies are presented in the following section. Percentages reported in Table 2 were calculated using the final set of 174 included studies. In the second search stage, the keywords “externally bonded” and “EB-FRP” were applied to isolate studies focused specifically on the externally bonded reinforcement (EBR) method. This step aimed to eliminate studies on alternative techniques such as near-surface mounted (NSM), grooving methods (GM), or hybrid bonding (HB) systems. Subsequent stages incorporated keywords such as “tension test” and “lap-shear test” to capture experimental configurations relevant to axial and shear performance. During the fourth and fifth stages, additional filters were introduced to capture studies investigating FRP anchor effects and interfacial bond behavior. These terms aligned directly with the review’s first two research questions. Notably, in Stage 5, the Boolean operator used for Web of Science was changed from “OR” to “AND” when combining keywords. This adjustment was necessary due to differences in search engine algorithms, which produced an excessive number of unrelated results when “OR” was used in Web of Science. Scopus retained the original “OR” logic, while Web of Science served a supplementary role in this round. Stage 6 excluded irrelevant studies involving numerical analysis, finite element modeling, machine learning, and algorithm-based approaches. These were filtered using the “NOT” operator applied to keywords such as “finite element,” “simulation,” “machine learning,” and related terms. No restrictions were imposed on publication year or journal tier. The final filter in Stage 7 limited the dataset to English-language peer-reviewed journal articles or review papers. Other types of publications, including conference proceedings, book chapters, and editorials, were excluded. The complete Boolean search strings used in the final round are shown below: Scopus search string: TITLE-ABS-KEY ((“FRP” OR “FIBER REINFORCED POLYMER”) AND (CONCRETE OR “REINFORCED CONCRETE” OR RC) AND (“EXTERNALLY BONDED” OR EB-FRP OR “BONDED REINFORCEMENT”) AND (“PULLOUT TEST” OR “LAP SHEAR” OR “DIRECT TENSION” OR “BOND-SLIP” OR “LOAD-SLIP” OR DEBOND*) AND (EXPERIMENT* OR TEST* OR “BOND BEHAVIOR” OR “INTERFACIAL BEHAVIOR” OR “FAILURE MODE” OR “BOND STRENGTH”) AND (“CFRP” OR “CARBON FIBER” OR “STRAIN PROFILE” OR DIC OR “FAN ANGLE” OR “EMBEDMENT DEPTH” OR “FRP ANCHOR*” OR “SPIKE ANCHOR*” OR “FRP TIE*” OR “MECHANICAL ANCHORAGE”)) AND NOT TITLE-ABS-KEY (NSM OR “NEAR SURFACE MOUNTED” OR “MACHINE LEARN*” OR “DEEP LEARN*” OR ALGORITHM* OR SIMULAT* OR “FINITE ELE*”) AND (LIMIT-TO (DOCTYPE,”ar”) OR LIMIT-TO (DOCTYPE,”re”)) AND (LIMIT-TO (LANGUAGE, “English”)). Web of Science search string: TS = ((“FRP” OR “FIBER REINFORCED POLYMER”) AND (CONCRETE OR “REINFORCED CONCRETE” OR RC) AND (“EXTERNALLY BONDED” OR EB-FRP OR “BONDED REINFORCEMENT”) AND (“PULLOUT TEST” OR “LAP SHEAR” OR “DIRECT TENSION” OR “BOND-SLIP” OR “LOAD-SLIP” OR DEBOND*) AND (EXPERIMENT* OR TEST* OR “BOND BEHAVIOR” OR “INTERFACIAL BEHAVIOR” OR “FAILURE MODE” OR “BOND STRENGTH”) AND (“CFRP” OR “CARBON FIBER” OR “STRAIN PROFILE” OR DIC OR “FAN ANGLE” OR “EMBEDMENT DEPTH” OR “FRP ANCHOR*” OR “SPIKE ANCHOR*” OR “FRP TIE*”)) NOT TS = (NSM OR “NEAR SURFACE MOUNTED” OR “MACHINE LEARN*” OR “DEEP LEARN*” OR ALGORITHM* OR SIMULAT* OR “FINITE ELE*”) AND DT = (“ARTICLE” OR “REVIEW”) AND LA = (“ENGLISH”). The search was limited to titles, abstracts, and keywords. After deduplication across both databases, a total of 334 unique journal articles remained for screening and further analysis. 2.4. Eligibility Criteria and Screening Process After finalizing the initial search, the screening process began by reviewing each article’s title and abstract to determine its relevance based on predefined inclusion and exclusion criteria. This process followed a two-round approach to ensure consistency and minimize omission, and the inclusion and exclusion criteria are summarized in Table 3. Each study was independently reviewed at least twice by the authors. The first round focused on screening titles and keywords, while the second round evaluated abstracts. A shared decision protocol ensured consistency. Any disagreements were resolved through discussion and mutual agreement. Table 4 shows the number of papers excluded at each round along with the reasons. Numerical and FEM studies were excluded from the primary evidence base because this review compares experimental performance metrics and failure observations that can be traced to physical specimens. Simulation studies are valuable for mechanism exploration and parameter studies, but they depend on assumptions about interface laws, fracture energy, mesh strategy, and calibration data; including them in the same quantitative pool as tests would blur measured outcomes with model outputs. Numerical work is therefore discussed only where it supports interpretation of experimental models, and its systematic synthesis is identified as a separate future research need. After the first-round screening, 247 articles remained. In the second round, 54 additional articles were removed, leaving 193. Following full-text checks and final consensus between reviewers, 19 more articles were excluded. A total of 174 studies remained for full analysis. 2.5. Data Extraction and Thematic Coding After applying the eligibility criteria, a total of 174 articles were retained for full-text analysis. The process leading to this selection was outlined in Figure 4. Although the initial search identified 41,598 articles across Scopus and Web of Science, most were excluded through keyword refinement and eligibility screening. Articles were eliminated if they used unrelated materials (e.g., bamboo fiber, textile-reinforced mortar, and engineered cementitious composites) or if full-text access was not available [66]. To extract structured information from the included studies, this review adopted a thematic template analysis approach [67]. This method offers flexibility in analyzing qualitative text while being grounded in predefined research questions. Unlike fixed-form coding schemes, the template approach allows iterative adjustment of themes as new patterns emerge during review. Compared to alternative methods like analytical induction [68], interpretive phenomenological analysis [69], discursive psychology [70], and grounded theory [71], template analysis offers improved structure and adaptability for large-scale technical reviews. The PRISMA 2020 checklist was used as a guiding protocol to ensure transparency and reproducibility in the data extraction process [62]. A modified version of the checklist was created for this review (Table 5); items related to formal meta-analysis were not applicable, while the engineering quality controls used for screening, extraction, comparability, and limitations are reported in the methods and synthesis sections. After refining the checklist, 17 applicable items were retained to structure the review process. Following the modified PRISMA checklist, a thematic coding template was developed and applied across all 174 studies (Table 6). This template included both general metadata and research-specific content. Each paper was categorized according to six main analytical domains. All coding was done manually through full-text reading, using a spreadsheet-based matrix to maintain consistency and traceability. Studies were grouped according to theme, research question relevance, and contribution type. This structured classification provided the basis for the analytical synthesis that follows in subsequent sections. Quantitative synthesis and normalization were performed at the test-result level where the source papers reported sufficient information. Extracted values were first retained in their reported units, then grouped by COI level, test method, FRP type, bond geometry, loading configuration, and failure mode. Direct pooling was avoided when studies used incompatible boundary conditions or missing geometry. When geometry was available, comparisons emphasized normalized or scale-aware indicators, including capacity per bonded FRP width, FRP stiffness ratio, bond length relative to effective bond length, debonding strain, strain profile shape, load–slip or bond–slip response, fracture energy, normalized anchor area, embedment-depth-to-dowel-diameter ratio, and anchor spacing or count. Where studies reported only ultimate load or qualitative failure observations, those data were used for stratified trend interpretation rather than direct numerical ranking. The normalization framework, comparison criteria, and links between the research questions and extracted evidence are summarized in Table 7. 2.6. Complexity of Investigation (COI) Assessment All 174 included articles were evaluated using a structured thematic template and categorized based on their contribution to the core research questions. To systematically assess the depth, workload, and execution difficulty of each experimental study, a Complexity of Investigation (COI) matrix was developed. This matrix classified studies into three levels, CoI-A, CoI-B, and CoI-C, based on four evaluation aspects: (1) Objectives of the experimental program; (2) data acquisition methods; (3) theoretical contribution or model development; and (4) failure mode characterization and interpretation. The classification criteria and descriptions for each COI level are shown in Table 8. Each study was assigned a COI level based on its alignment with the four evaluation dimensions. COI-A denotes baseline investigations focused on unanchored EB-FRP systems tested under monotonic direct tension or lap-shear conditions. These studies employed minimal instrumentation and lacked model development or validation. COI-B represents intermediate studies involving FRP-anchored systems applied to concrete, typically with single-anchor configurations. These investigations included limited parameter variation and some use of empirical models or comparative analyses. COI-C reflects advanced studies that combined FRP sheets with multiple anchors, examined system-level behavior under variable conditions, and incorporated detailed instrumentation (e.g., strain field mapping and digital image correlation). These studies often validated or modified analytical models and provided comprehensive failure mode analysis. The COI classification enabled systematic comparison across studies and helped evaluate the experimental and analytical depth contributing to the review’s synthesis. Since some studies addressed more than one objective, thematic categories were treated as analytical labels rather than mutually exclusive bins. Each paper received a primary label based on its dominant experimental purpose and secondary labels for additional variables or model contributions. For example, a study on anchored lap-shear behavior that also proposed a strength equation was counted under the appropriate anchor-configuration objective and also considered in the model-validation discussion. This approach explains why sub-objective classifications can exceed the number of unique studies while the PRISMA corpus remains fixed at 174 papers. 2.7. Interpretation and Synthesis Assessing the overall complexity of an experimental study using only a single COI score can be misleading. Many papers contain multiple test programs that vary in detail, instrumentation, or modeling depth. For instance, a study classified as COI-A might include a basic FRP sheet test and also present extensive parameter variation or high specimen replication in a parallel series, reflecting higher complexity in specific dimensions. The synthesis applies a multidimensional interpretation at the test level to capture differences across experimental programs that a single COI score may overlook. Each test component is evaluated independently. A practical example illustrates the benefit of this approach. A traditional classification based only on study counts might conclude that unanchored EB-FRP systems dominate the available evidence because COI-A contains substantially more test results than COI-B or COI-C. However, the multidimensional interpretation shows that several questions of direct design relevance, including anchor spacing effects, load sharing among anchors, and system-level force redistribution, can only be addressed using COI-C studies. Consequently, the framework distinguishes between the quantity of available evidence and the scope of conclusions that can be drawn from that evidence. This distinction directly influenced the synthesis by preventing high-volume COI-A datasets from being interpreted as evidence for phenomena that can only be evaluated through higher-complexity anchored-system investigations. The four dimensions from the COI framework are repurposed as analytical categories to identify trends, methodological emphasis, and research gaps across all test programs:(1)Experimental objectives and parameter selection.(2)Instrumentation and loading protocol.(3)Data acquisition and post-processing.(4)Failure mode analysis and theoretical development. (1)Experimental objectives and parameter selection. Experimental objectives and parameter selection. (2)Instrumentation and loading protocol. Instrumentation and loading protocol. (3)Data acquisition and post-processing. Data acquisition and post-processing. (4)Failure mode analysis and theoretical development. Failure mode analysis and theoretical development. A test-level, aspect-based framework supports systematic comparison and helps structure findings for RQs 1 and 2. It also allows for cross-study comparison and supports the organization of bond models, anchorage schemes, and parameter ranges that may inform future benchmarking or database development. Figure 6 shows the synthesis framework, outlining how test-level findings were structured according to these four analytical categories. Several prior review studies informed the development of this framework. Ref. [50] evaluated over 1000 tests from 143 studies, focusing on axial tensile performance in EB-FRP systems with and without anchors, and benchmarked results against ACI 440.2R, CNR-DT 200, and fib Bulletin 90. Ref. [67] analyzed the influence of FRP anchor configurations, confirming their effect in delaying premature debonding and enhancing force transmission. Ref. [68] compared EB-FRP systems with and without grooving techniques, highlighting how interface geometry, composite stiffness, width ratio, and area ratio influence failure mechanisms. Insights from these reviews guided the development of the analytical dimensions used to interpret the current dataset. The 174 experimental test programs were synthesized using four analytical dimensions, enabling direct comparison of bond behavior in unanchored and anchored EB-FRP systems under tension or lap-shear loading. The interpretation focuses on performance patterns, anchor effectiveness, and experimental limitations, directly addressing RQs 1 and 2, while also contributing to RQ3 by identifying unresolved challenges and outlining future research directions. 3. Bibliometric Analysis and Research Landscape 3.1. Keywords and Regional Research Network Mapping Keyword co-occurrence analysis was conducted using VOSviewer (version 1.6.20) to identify dominant research topics and thematic structures across the 174 included studies. This tool applies statistical and network-based methods to generate scientific landscape visualizations that reveal conceptual relationships between key terms. The analysis focused on experimental studies related to EB-FRP systems under direct tension and lap-shear testing. A custom thesaurus spreadsheet was developed to clean and consolidate the keyword dataset. Redundant terms were merged to improve clarity and avoid duplicate entries. For example, variations such as “single lap-shear test” and “single lap-shear tests” were unified, as were “delaminating” and “delamination”. Abbreviations and synonymous terms were consolidated where appropriate. For example, “Glass fiber/fiber,” “GFRP,” and “Basalt fiber reinforced polymer” were grouped under the general term “FRP.” However, CFRP was intentionally retained as a distinct keyword in the network. Due to its dominant role in EB-FRP research, particularly in externally bonded systems, CFRP frequently appears in published studies and was not merged into a broader category. Maintaining CFRP as a separate node allowed its influence and connections in the literature to be visualized more accurately. Terms related to anchorage, such as “spike anchors” and “anchor(s)”, were merged as “FRP anchors” or “fiber anchors”. Generic or irrelevant terms, such as “repair,” “issue,” and “sun,” were excluded. Figure 7a shows the keyword network map after term consolidation. A minimum threshold of three occurrences per keyword was applied, resulting in 214 keywords selected from the original 1453. Prominent terms include “FRP/CFRP”, “strengthening”, “externally bonded”, “(single-) lap shear tests”, “bond behavior”, “anchorage systems”, “concrete/reinforced concrete (RC)”, “debonding”, and “digital image correlation (DIC)”, all of which align with the core themes of this review. In parallel, geographic publication trends were examined through country-level co-authorship mapping. Figure 7b shows the international distribution of EB-FRP research contributions. Node size reflects the number of publications, while link thickness represents co-authorship intensity. China, Iran, and the United States emerged as the most active contributors. Strong collaboration clusters were observed between institutions in Asia, North America, and Europe, with notable co-authorship linkages among China, Iran, Italy, and the United States. The map also reveals regions with limited representation, indicating geographic variation in research activity and uneven levels of engagement with EB-FRP experimental studies. The country-level map was used to describe collaboration and publication concentration, not to infer national differences in test philosophy. Although regional clusters may reflect different equipment, code environments, or preferred test methods, the extracted variables did not support a reliable country-by-country methodological ranking. The review therefore limits country interpretation to research activity and collaboration patterns and treats methodological differences through test type, COI level, and coded experimental variables. A preliminary comparison of the leading country clusters suggests differences in predominant experimental emphases, although regional trends should not be interpreted as fixed national scientific schools. China and the broader East Asian research cluster accounted for approximately 26.0% of the included studies, followed by the European cluster led by Italy (18.1%), the United States and North American institutions (17.7%), Iran (13.0%), and Australia/New Zealand (5.8%). These five research clusters contributed more than 80% of the included studies, indicating that EB-FRP bond and anchorage research has been concentrated within a limited number of geographical regions. Studies from the China and broader East Asian cluster frequently investigated FRP–concrete bond behavior, bond–slip modeling, and plate or sheet interface testing. Iranian studies commonly examined CFRP sheet-to-concrete bond, EBROG and EBR comparisons, aggregate and concrete effects, and environmental or surface-condition variables. North American studies emphasized design-oriented strengthening, anchored FRP systems, mechanically or fiber-anchored configurations, and structural-member behavior. European contributions, particularly Italian studies, focused on externally bonded reinforcement terminology, interface and fracture mechanics, durability, and guideline-oriented interpretation. Country attribution was based on the first-author or corresponding author affiliation of the included studies. The extracted database does not support a rigorous country-by-country methodological ranking because collaboration networks, co-authorship patterns, and experimental variables remain highly heterogeneous. Regional interpretation is therefore presented only as a descriptive trend. Table 9 summarizes the approximate representation, research emphases, and predominant experimental tendencies associated with the major regional clusters. 3.2. Publication Patterns The publication sources of the 174 included articles are summarized in Table 10, categorized by journal title, publisher, SJR quartile ranking, H-index, impact factor, and publication year. All journals were indexed in the SCImago Journal Rank (SJR) system, with a total of 39 journals represented. Among these, 21 were Q1 journals, five were Q2, and three were Q3; no articles appeared in Q4 journals. The majority of papers, 156 in total (89.7%), were published in Q1 journals, reflecting the field’s emphasis on high-impact, technically rigorous outlets. Q2 journals accounted for 14 papers (8.0%) and Q3 journals for four papers (2.3%). Publication activity increased substantially over time. Only six papers (3.5%) were published before 2003, while more than 83% appeared after 2013. The period from 2018 to early 2026 saw the highest concentration, with 99 articles (56.9%) published, indicating growing research interest in the interfacial behavior of EB-FRP systems under direct tension and lap-shear testing. Journals most frequently publishing on this topic include Composites for Construction (ASCE, 34 papers), Composite Structures (Elsevier, 27), Construction and Building Materials (Elsevier, 25), and Composites Part B: Engineering (Elsevier, 21). Together, these four journals contributed over 60% of the reviewed literature. Other notable contributors include Engineering Structures, Engineering Fracture Mechanics, Journal of Reinforced Plastics and Composites, and ACI Structural Journal. Elsevier was the leading publisher, accounting for 56.3% of the included papers. The American Society of Civil Engineers (ASCE) followed with 21.8%. Other publishers included SAGE (5%), ACI (4%), Springer (3%), MDPI (2%), Hindawi (2%), and ICE Publishing (2%). Single-paper contributions came from ASTM, Taylor & Francis, Techno Press, Wiley-Blackwell, and the National Research Council of Canada. Figure 8 shows the temporal distribution of publications, grouped in five-year intervals from 1994 to 2026. Power and exponential regression models were applied to assess publication growth. The power model produced an R2 value of 0.9577, suggesting a strong fit. Extrapolating from this model, publication volume is expected to continue rising, though a plateau may begin to emerge beyond 2026. As of mid-February 2026, 50 papers had already been published during the current five-year window. Overall, the data confirm both the concentration of EB-FRP bond research in high-ranking journals and its accelerating trajectory in recent years. These trends reflect the field’s increasing technical maturity and broadening relevance within structural engineering research. 3.3. COI-Based Categorization Figure 9 shows the classification of 174 publications by Complexity of Investigation (COI). Dark columns represent COI levels, and light columns reflect sub-objectives within each level. Most publications (150 out of 174) fell under COI-A, representing studies on unanchored EB-FRP systems subjected to axial tension. Most investigations focused on interfacial bond behavior, including debonding mechanisms, FRP strip size effects, and environmental degradation. Frequently addressed sub-objectives included debonding (25 publications), geometric parameters (23), environmental exposure (22), and concrete strength (18). Fewer studies examined EBROG and EBR systems (15), dynamic effects (14), aggregate properties (eight), adhesives (six), fatigue (six), mixed-mode fractures (six), surface treatments (five), and general reviews (two). Although substantial in quantity, most of these studies investigated isolated parameters under simplified testing conditions. Limited integration across variables restricted the extrapolation of findings to real applications. The studies involving FRP anchors under axial tension, grouped under COI-B, remained comparatively underrepresented, with 33 publications (19.3%). This category included both isolated anchor tests and single-anchor EB-FRP systems. COI-B1 contained 19 studies that examined isolated anchors without integration into EB-FRP assemblies. Most employed pullout tests or single-lap shear configurations. The primary objectives addressed anchor configuration (eight publications), design methodology (six), and anchorage typologies (three). Two studies provided review-based summaries. These investigations typically relied on idealized test conditions. Although effective in characterizing basic anchor performance, the absence of system-level interaction limited the applicability of results to practical strengthening scenarios. COI-B2 included 14 studies that incorporated single anchors within EB-FRP systems. These investigations examined anchorage typologies (six publications), configuration (three), anchor positioning (three), and design parameters (two). Although more representative of field conditions than COI-B1, inconsistencies in geometry, installation, and substrate conditions prevented meaningful synthesis and comparison across studies. This variation impeded the development of broadly applicable design recommendations. COI-C represented the highest level of complexity, encompassing 16 studies (9.4%) involving EB-FRP systems with multiple FRP anchors. Research objectives within this group addressed anchor layout strategies (six publications), configuration effects (four), number and positioning of anchors (three), anchorage typologies (three), and system-level reviews (three). These studies introduced added complexity by accounting for scale effects, anchor-to-anchor interactions, and more realistic boundary conditions. Despite these advancements, limited sample sizes and diverse methodologies hindered a comprehensive evaluation of overall system performance. While the total number of reviewed publications remained fixed at 174, many studies investigated more than one objective. For instance, articles categorized under the concrete strength objective in COI-A often also examined environmental exposure or FRP geometry. This overlap led to a total number of sub-objective classifications exceeding the number of unique studies, highlighting the multidimensional nature of the investigations and the need for integrated frameworks capable of capturing parameter interactions across COI levels. 3.4. Types of DAQs, FRPs, and Experimental Programs Accurate measurement of deformation or strain is critical in FRP-to-concrete bond testing. Various data acquisition (DAQ) strategies were employed in the reviewed literature, as shown in Figure 10a. The most common method was the use of strain gauges (SGs), adopted in approximately 62.1% of studies (108 out of 174). A significant portion of these studies (52.8%, or 57 out of 108) used SGs only as auxiliary tools to validate measurements at discrete points. Digital image correlation (DIC) was the second most widely used technique, applied in 56 studies (32.7%), either in 2D or 3D configurations, depending on the number of cameras used. DIC enabled full-field strain mapping and often served as the primary DAQ system in recent works [11,21,43,69]. Linear Variable Differential Transformers (LVDTs) were used in 67 studies (39.2%), though only a small subset (19 studies) relied solely on LVDTs for all deformation measurements. Two interferometry-based approaches were identified, with Electronic Speckle Pattern Interferometry (ESPI) and Moire Fringe Interferometry (MFI). Both were used in conjunction with SGs for hybrid strain field capture. Only one study employed S-type load cells as the sole deformation recording tool. The number and combinations of DAQ devices are summarized in Figure 10b. Just over half of the reviewed studies (58.5%, or 100 out of 151) used a single type of DAQ device. A total of 64 studies (37.4%) employed two DAQ systems simultaneously, while only two studies utilized three. Most combinations (37 publications) involved traditional devices, such as SGs and LVDTs, used to validate data collected from full-field optical systems. Notably, 2D-DIC, 3D-DIC, and Particle Image Velocimetry (PIV-DIC) have shown increasing adoption in configurations involving more than two DAQ systems (27 publications) due to their capacity to resolve continuous strain fields across the bond zone. Varied FRP materials were used in the reviewed experimental programs, as shown in Figure 11. Carbon fiber-reinforced polymer (CFRP) dominated the selection, appearing in 143 studies (83.6%). Other types included Glass FRP (GFRP) in 23 studies (13.5%), Basalt FRP (BFRP) in 18 studies (10.5%), and Aramid FRP (AFRP) in seven studies (4.1%). Some publications employed more than one FRP type for comparative analysis. As shown in Figure 11b, 147 studies (86%) used a single FRP type, while 17 studies (9.9%) used two types, and two studies (1.2%) used three. The unequal distribution of FRP types affects the strength of material-specific conclusions. CFRP provides the highest-confidence evidence base because it appears in 143 studies and dominates both unanchored and anchored datasets. GFRP, BFRP, and AFRP are represented by 23, 18, and seven studies, respectively, so trends for these materials should be interpreted as indicative rather than definitive. Available studies suggest that differences in stiffness and material behavior may influence load-transfer demand, interfacial stress distribution, and debonding response [70], but the limited number of direct comparative studies prevents robust quantitative comparison among all FRP families. Conclusions that use the general term FRP, therefore, mainly reflect CFRP-dominated evidence, unless a specific comparison among GFRP, BFRP, and AFRP is stated. A summary of the evidence distribution, interpretive confidence, and principal implications for each FRP material system is provided in Table 11. The experimental programs adopted across the reviewed studies are shown in Figure 12. Single-lap shear configurations dominated, as reported in 123 publications (71.9%). Double-lap shear tests appeared in 21 studies (12.3%), primarily investigating interfacial bond behavior between FRP sheets and concrete substrates, between fan-shaped FRP anchors and bonded strips, or isolated bent fiber anchors under shear, corresponding to COI-A, COI-B2, and COI-C levels. Direct tension tests, reported in eight studies (4.7%), applied tensile force through embedded reinforcement in concrete (COI-A) to examine the bond between FRP configurations and reinforced concrete. These tests aimed to replicate realistic loading patterns where concrete acted as the primary structural element. Pullout tests, reported in 12 publications (8.2%), applied uniform axial tension directly on the anchor dowel to isolate anchorage performance (COI-B1). These tests typically assessed anchorage effectiveness, dowel geometry, and fan configurations under controlled slip conditions, or they investigated specific failure patterns. 4. EB-FRP Without Anchors (COI-A) 4.1. Overview A systematic analysis of unanchored EB-FRP strengthening systems for concrete is presented in this section under COI-A classification. The discussion addresses RQ1 and integrates bibliometric context to interpret trends in research focus, experimental methods, and theoretical developments. Studies are organized by experimental parameters, mechanical behavior, failure modes, and model contributions, as summarized in Table 12. 4.2. Test Methods Indirect and direct shear tests are widely used to investigate interfacial behavior and failure mechanisms between FRP and concrete in COI-A studies. Both methods evaluate bond performance and identify conditions that lead to debonding or fracture. The notched beam test shown in Figure 13a is an indirect method where the FRP strip response can be affected by the concrete prism deformation because of the presence of the bending moment [20]. Single-lap and double-lap shear tests are direct shear tests widely adopted in previous research. The single-lap shear test shown in Figure 13b is commonly used for evaluating interfacial behavior due to its relative simplicity [137]. In this configuration, the loaded end refers to the beginning of the bonded area where the load is applied, while the opposite end is known as the free end. The concrete prism is subjected to compression forces at both the loaded and free ends by a restraining frame, such that the concrete becomes a rigid body that cannot crack. An unbonded area is often introduced near the edge of the specimen to prevent undesired concrete cone failure. The double-lap shear tests (Figure 13c) were introduced as an alternative method to eliminate the eccentricity caused by the tensile load applied to the FRP and the reaction force on the concrete, which can lead to out-of-plane deformations that may affect the results. Despite the symmetry of the double-lap test configuration, debonding typically initiates at the weakest interface due to material variability, causing an asymmetry in the interfacial bond. This asymmetry ultimately leads to rotation of the specimen during testing, affecting the accuracy of the results [138]. A comparative summary of the principal test methods, their advantages, limitations, and relevance to practical applications is presented in Table 13. The test methods are therefore not interchangeable. Single-lap shear tests provide simple force measurement and are well suited to comparative bond-capacity screening, but eccentricity and fixture restraint can affect the local stress field. Double-lap shear tests reduce global eccentricity and are useful for symmetric load transfer, although debonding commonly initiates at the weaker side and may still create rotation. Direct tension tests better represent axial load transfer in members, but are less common and harder to standardize. Pullout tests isolate anchor dowel behavior and are useful for straight-anchor capacity, but they do not by themselves represent anchor–strip interaction in an EB-FRP system. Flexural or notched-beam arrangements can reproduce structural bending effects but introduce additional concrete-prism deformation and are less direct for pure interface comparison. No single test method can be regarded as universally superior because each isolates a different aspect of EB-FRP behavior. Direct tension tests generally provide the closest representation of the axial force-transfer mechanisms encountered in tension-strengthened members because they directly capture force transfer between the FRP and concrete substrate under tensile loading. In contrast, lap-shear and pullout tests remain indispensable for mechanistic investigation because they isolate specific bond and anchorage variables that cannot be easily distinguished in larger structural tests. Flexural and notched-beam configurations provide additional insight into the influence of bending and cracking on interfacial behavior, although the resulting stress state is more complex and less suitable for isolating pure bond mechanisms than direct-tension or shear-based tests. Results from the tests provide the basis for interfacial behavior models referenced in design standards, including CNR DT 200 [48], CNGB [139], ACI 440.2R-17 [47] and fib bulletin 90 [49]. The ASTM D8337/D8337M-21 [140] designates the single-lap shear test as the preferred method for determining bond capacity due to straightforward setup and direct measurement of applied forces (Figure 14). 4.3. COI-A Dataset 4.3.1. Collected Test Results A total of 3162 test results under COI-A were compiled in Ref. [28]. The dataset was built from experimental data extracted from the following included studies [9,22,23,24,27,30,51,52,53,55,56,57,58,59,60,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,141,142,143,144,145,146,147,148,149,150,151,152,153]. The collected parameters include specimen geometry, FRP strip dimensions, material properties (e.g., elastic modulus and tensile strength), and recorded ultimate capacity. Box-and-whisker plots were used to illustrate the spread of the data and highlight the distribution of key experimental parameters, as shown in Figure 15. These visualizations reveal trends and limitations in the existing research. Most studies were based on small-scale test configurations, typically employing short FRP ties (median length: 200 mm), thin laminates (median thickness: 0.17 mm), and low stiffness ratios, with the average value of kf/Ec approximately 1.4. Here, kf denotes the FRP stiffness (NEftf), and Ec is the modulus of elasticity of concrete. These test conditions do not adequately represent real-scale structural behavior. Prior studies [11,69] have shown that large-sized (thicker and longer) single-lap shear tests exhibit markedly different performance characteristics, including increased load capacity and extended post-debonding deformation. The lack of validation and limited availability of large-scale EB-FRP test data continue to constrain the development of models for practical field applications [50]. Parameter-level synthesis indicates that unanchored EB-FRP performance is governed by interacting material, geometric, interface, and exposure variables rather than by any single property. Higher FRP stiffness, greater strip thickness, and wider bonded areas can increase capacity but also intensify interfacial stress gradients, making debonding strain, effective bond length, and failure mode sensitive to adhesive stiffness and concrete surface quality. Increasing bond length improves force transfer only until the effective bond length is reached; beyond that point, additional length may add limited capacity but can change post-peak load–slip behavior. Concrete strength and aggregate structure influence substrate fracture and concrete-cohesive debonding, while surface treatment and adhesive properties control whether failure localizes in the adhesive, at an interface, or in the near-surface concrete. Environmental exposure, including elevated temperature, wet or hygrothermal aging, freeze–thaw cycling, and long-term outdoor exposure, mainly affects adhesive-dependent load transfer and residual bond capacity; because these studies use different exposure protocols, their results are best interpreted as durability risk trends rather than a single pooled degradation factor. The principal exposure mechanisms, recommended reporting metrics, and current comparability limitations are summarized in Table 14. Field-transfer limits were also made explicit because most included tests remain on a laboratory scale. Laboratory coupon tests isolate mechanisms, but field strengthening must contend with substrate variability, surface preparation quality, adhesive curing, access constraints, construction tolerances, in-service moisture and temperature, and multi-strip or multi-anchor load paths. These factors do not invalidate small-scale tests, but they limit direct extrapolation unless the model or design rule is checked against larger members or field-representative system tests. 4.3.2. Fracture Mode II Theory and Existing Models Fracture Mode II is a commonly used theory to analyze lap shear tests and evaluate interfacial fracture based on fracture mechanics principles. Figure 16 shows the mechanisms of Mode II fracture within an FRP–concrete interface bonded with adhesive. The load applied, denoted as P(s), is transmitted from the FRP to the concrete through to the bonded region, generating both normal and shear stresses along the interface. The normal stresses are represented as σf(x) in the FRP and σc(x) in the concrete, while τzx(x) represents the shear stress at the interface. As the slip s(x) between the FRP and the concrete increases, these stresses change, making them critical in predicting debonding behavior [154]. The normal strain ε(y) distribution across the width of the interface has parabolic peaks at the center and diminishes toward the edges [155,156], as shown at the top left in Figure 16. The central region carries the majority of the load, while the edges experience minimal deformation. The shear strain γxy is maximal at the edges and decreases toward the center, highlighting that shear deformation is most prominent along the edges of the bond. This emphasizes that load transfer occurs predominantly in the central region of the bond, with shear stresses being concentrated near the edges. The stress distribution within the adhesive joint along the length is shown in the lower left corner of Figure 16 in a green box, which shows two cases. Case 1, the blue solid line (σa1 and τa1), represents a scenario with a short bond length, while case 2, the dashed black line (σa2 and τa2), represents a scenario with a long bond length. As bond length increases, normal stresses σa1(x) become more significant than shear stresses τa1(x) in the middle portion, where shear stresses reduce to nearly zero, while normal stresses remain constant. Shear stress peaks appear at the bond ends Lb, where debonding typically begins. The shear stress τa2(x) curve becomes asymmetrical when the bonded materials have differing rigidity, leading to high stress peaks at the ends of more rigid concrete members, while the less rigid components undergo greater strain, causing strong shear deformation at the edges [157]. The red boxed section in Figure 16 illustrates the equilibrium state at the first bonded line, where normal and shear stresses within the bonded region are balanced. This equilibrium state is crucial for understanding load transfer within the bonded area, and any disruption in this balance leads to debonding at the interface. The interfacial slip s(x), strain ε(x), and stress τ(x) profiles are shown on the lower half of the figure. Here, df(x) represents the displacement on the surface of the FRP ties, while dc(x) refers to the rigid movement and displacement occurring in the concrete blocks. The strain ε(x) is defined as the first derivative of s(x), and the shear stress τ(x) is determined by the second derivative of s(x) and incorporating the material properties of the FRP used. The shear stress τ(x) increases as interfacial slip s(x) increases along the bond length x and reaches its maximum value before the slope of s(x) reaches its peak, which indicates the onset of debonding. The fracture energy Gf is defined as the shaded area under the bond–slip relation, representing the energy dissipated along the entire bond length during the loading process. Bond–slip relation τ(s) is derived from τ(x) and s(x) serves as a key indicator used to calculate the energy required to initiate debonding. The mathematical expressions, highlighted in purple and blue, form the basis for predicting debonding strain and failure conditions in FRP–concrete interfaces along the bonded interface. Lastly, the theoretical debonding force Pdeptheor(s) shown in the black box is derived by integrating the bond–slip relation τ(s), providing an analytical solution for predicting load–slip response under debonding with concrete fracture in EB-FRP-strengthened concrete systems [137]. Substantial experimental, analytical, and computational research has focused on understanding the interfacial behavior between FRP and concrete over the past three decades. The development and refinement of bond–slip models have been central to advancing this understanding. The most commonly used bond–slip models are represented in Figure 17, each with different shapes aimed at capturing the critical aspects of interfacial behavior [57,76,99,158,159,160]. Bilinear models are effective for modeling brittle bond failure as they assume a steep increase in bond stress up to a peak point, followed by a linear decline with increasing slip [158,161]. Tri-linear models introduce a plateau after the peak point to capture a phase of steady stress transfer after the peak bond stress is reached [160], making them suitable for cases where slip behavior is more gradual and debonding occurs progressively. Exponential and nonlinear models represent more complex bond responses [57,159,162], where bond stress rises quickly but decays exponentially or nonlinearly, better reflecting gradual bond degradation under varying conditions. Finally, parabolic/curved models offer a smooth, continuous transition from loading to debonding [97,99], allowing for more accurate predictions of intermediate bond deterioration stages. Each of these models balances simplicity with accuracy, depending on the specific conditions of the FRP–concrete bond. Choosing the appropriate bond–slip model is crucial for predicting debonding behavior and the overall performance of EB-FRP systems, especially in real-world large-sized applications, where accurate modeling of interfacial behavior is essential. In addition to the widely cited models discussed above, Table 15 provides an overview of 18 other models spanning from 1999 to 2025, including analytical [99,159,164,165,166], empirical [145,160,167,168], and semi-empirical [158,161,162,169] approaches. Analytical models often describe bond–slip relationships by deriving and integrating slip and bond stress profiles under Mode II fracture mechanics, which pertains to shear-dominated interfacial failure modes in FRP–concrete systems (more details below), but they tend to be complex and difficult to apply in practical design contexts. In contrast, empirical models rely on experimental data to establish direct correlations through regression and thus may lack generalizability and fail to capture the underlying mechanics of interfacial behavior. The limitations of both analytical and empirical models highlight the need for a more refined approach. Semi-empirical models aim to bridge this gap by incorporating key parameters from both experimental findings and analytical formulations. Most existing models are based on small-scale experimental setups, using short FRP ties (median length: 200 mm), narrow thickness (median value: 0.17 mm), and a low stiffness ratio (kf/Ec) of approximately 1.4 (see Figure 15). Here, kf represents the stiffness of the FRP (NEftf), while Ec is the modulus of elasticity of concrete. These conditions fail to adequately replicate real-world construction scenarios [50]. Research has shown that large-sized (thicker and longer) single-lap shear tests exhibit different characteristics compared to small-scale tests, such as a higher load-carrying capacity and longer post-debonding deformation [11,69]. As a result, this lack of validation and sufficiently accurate bond–slip models for large-sized EB-FRP systems hinders the precise prediction of load–slip responses in practical applications. The accuracy of the empirical bond–slip models is further complicated by several factors, such as experimental conditions, material properties, and surface preparation [96]. A significant challenge in these models is that their coefficients are often calibrated empirically, without a complete understanding of their physical significance. This limitation complicates accurate predictions of debonding behavior, especially when applying the models to real-world conditions. Although experimental studies have clarified key bond indicators between FRP and concrete, achieving precise and efficient debonding load predictions for practical design purposes remains a critical focus. A major factor contributing to the challenge above is the role of material properties in influencing bond behavior. The stiffness of FRP, which is generally much higher than that of concrete, often leads to cracking or fracture at the interface due to the low tensile strength of the concrete [57]. Proper surface preparation is also crucial to ensure effective bonding and minimize the risk of premature failure, especially under tensile loads [51,53,84,170]. However, recent research has shown that when the FRP thickness increases, resulting in stiffer FRP, concrete strength becomes less critical [11,21,69]. In such cases, the debonding load is more influenced by the stiffness of the FRP and the properties of the adhesive layer. Studies have demonstrated that bond strength can be improved with thicker, stiffer FRP and less stiff adhesives [77,151,171]. As a result, models that previously emphasized concrete strength may also need to be reconsidered when addressing large-sized EB-FRP systems. Despite significant advancements in understanding the interfacial behavior between FRP and concrete, current bond–slip models face several limitations. Analytical models often lack practicality due to their complexity, while empirical models fail to generalize across varied configurations, as they are primarily based on small-scale tests. These models do not adequately account for the distinct characteristics of large-sized FRP systems, which are increasingly used in modern construction. Key challenges include accurately predicting debonding loads and capturing load–slip responses under realistic conditions. polymers-18-01598-t015_Table 15Table 15Existing bond–slip models from 1999 to 2025. Bond–Slip Model Ascending Branch 0